pre-release

Support "wait before testing" using '+' in pvmove and lvconvert
interval.  Doing so overrides the new default of sleeping after checking
the LV's progress.

Sleeping before checking progress can lead to extraneous polldaemons
being left running.  These polldaemons would have otherwise exited had
they checked before sleeping.  Checking progress before sleeping helps
workaround the subtly unreliable nature of "finished" state checking
in _percent_run.

Update test/t-mirror-names.sh to use '+' when providing its lvconvert
interval.

COPYING

@@ -0,0 +1,340 @@
+		    GNU GENERAL PUBLIC LICENSE
+		       Version 2, June 1991
+
+ Copyright (C) 1989, 1991 Free Software Foundation, Inc.
+     59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+			    Preamble
+
+  The licenses for most software are designed to take away your
+freedom to share and change it.  By contrast, the GNU General Public
+License is intended to guarantee your freedom to share and change free
+software--to make sure the software is free for all its users.  This
+General Public License applies to most of the Free Software
+Foundation's software and to any other program whose authors commit to
+using it.  (Some other Free Software Foundation software is covered by
+the GNU Library General Public License instead.)  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+this service if you wish), that you receive source code or can get it
+if you want it, that you can change the software or use pieces of it
+in new free programs; and that you know you can do these things.
+
+  To protect your rights, we need to make restrictions that forbid
+anyone to deny you these rights or to ask you to surrender the rights.
+These restrictions translate to certain responsibilities for you if you
+distribute copies of the software, or if you modify it.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must give the recipients all the rights that
+you have.  You must make sure that they, too, receive or can get the
+source code.  And you must show them these terms so they know their
+rights.
+
+  We protect your rights with two steps: (1) copyright the software, and
+(2) offer you this license which gives you legal permission to copy,
+distribute and/or modify the software.
+
+  Also, for each author's protection and ours, we want to make certain
+that everyone understands that there is no warranty for this free
+software.  If the software is modified by someone else and passed on, we
+want its recipients to know that what they have is not the original, so
+that any problems introduced by others will not reflect on the original
+authors' reputations.
+
+  Finally, any free program is threatened constantly by software
+patents.  We wish to avoid the danger that redistributors of a free
+program will individually obtain patent licenses, in effect making the
+program proprietary.  To prevent this, we have made it clear that any
+patent must be licensed for everyone's free use or not licensed at all.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+		    GNU GENERAL PUBLIC LICENSE
+   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
+
+  0. This License applies to any program or other work which contains
+a notice placed by the copyright holder saying it may be distributed
+under the terms of this General Public License.  The "Program", below,
+refers to any such program or work, and a "work based on the Program"
+means either the Program or any derivative work under copyright law:
+that is to say, a work containing the Program or a portion of it,
+either verbatim or with modifications and/or translated into another
+language.  (Hereinafter, translation is included without limitation in
+the term "modification".)  Each licensee is addressed as "you".
+
+Activities other than copying, distribution and modification are not
+covered by this License; they are outside its scope.  The act of
+running the Program is not restricted, and the output from the Program
+is covered only if its contents constitute a work based on the
+Program (independent of having been made by running the Program).
+Whether that is true depends on what the Program does.
+
+  1. You may copy and distribute verbatim copies of the Program's
+source code as you receive it, in any medium, provided that you
+conspicuously and appropriately publish on each copy an appropriate
+copyright notice and disclaimer of warranty; keep intact all the
+notices that refer to this License and to the absence of any warranty;
+and give any other recipients of the Program a copy of this License
+along with the Program.
+
+You may charge a fee for the physical act of transferring a copy, and
+you may at your option offer warranty protection in exchange for a fee.
+
+  2. You may modify your copy or copies of the Program or any portion
+of it, thus forming a work based on the Program, and copy and
+distribute such modifications or work under the terms of Section 1
+above, provided that you also meet all of these conditions:
+
+    a) You must cause the modified files to carry prominent notices
+    stating that you changed the files and the date of any change.
+
+    b) You must cause any work that you distribute or publish, that in
+    whole or in part contains or is derived from the Program or any
+    part thereof, to be licensed as a whole at no charge to all third
+    parties under the terms of this License.
+
+    c) If the modified program normally reads commands interactively
+    when run, you must cause it, when started running for such
+    interactive use in the most ordinary way, to print or display an
+    announcement including an appropriate copyright notice and a
+    notice that there is no warranty (or else, saying that you provide
+    a warranty) and that users may redistribute the program under
+    these conditions, and telling the user how to view a copy of this
+    License.  (Exception: if the Program itself is interactive but
+    does not normally print such an announcement, your work based on
+    the Program is not required to print an announcement.)
+
+These requirements apply to the modified work as a whole.  If
+identifiable sections of that work are not derived from the Program,
+and can be reasonably considered independent and separate works in
+themselves, then this License, and its terms, do not apply to those
+sections when you distribute them as separate works.  But when you
+distribute the same sections as part of a whole which is a work based
+on the Program, the distribution of the whole must be on the terms of
+this License, whose permissions for other licensees extend to the
+entire whole, and thus to each and every part regardless of who wrote it.
+
+Thus, it is not the intent of this section to claim rights or contest
+your rights to work written entirely by you; rather, the intent is to
+exercise the right to control the distribution of derivative or
+collective works based on the Program.
+
+In addition, mere aggregation of another work not based on the Program
+with the Program (or with a work based on the Program) on a volume of
+a storage or distribution medium does not bring the other work under
+the scope of this License.
+
+  3. You may copy and distribute the Program (or a work based on it,
+under Section 2) in object code or executable form under the terms of
+Sections 1 and 2 above provided that you also do one of the following:
+
+    a) Accompany it with the complete corresponding machine-readable
+    source code, which must be distributed under the terms of Sections
+    1 and 2 above on a medium customarily used for software interchange; or,
+
+    b) Accompany it with a written offer, valid for at least three
+    years, to give any third party, for a charge no more than your
+    cost of physically performing source distribution, a complete
+    machine-readable copy of the corresponding source code, to be
+    distributed under the terms of Sections 1 and 2 above on a medium
+    customarily used for software interchange; or,
+
+    c) Accompany it with the information you received as to the offer
+    to distribute corresponding source code.  (This alternative is
+    allowed only for noncommercial distribution and only if you
+    received the program in object code or executable form with such
+    an offer, in accord with Subsection b above.)
+
+The source code for a work means the preferred form of the work for
+making modifications to it.  For an executable work, complete source
+code means all the source code for all modules it contains, plus any
+associated interface definition files, plus the scripts used to
+control compilation and installation of the executable.  However, as a
+special exception, the source code distributed need not include
+anything that is normally distributed (in either source or binary
+form) with the major components (compiler, kernel, and so on) of the
+operating system on which the executable runs, unless that component
+itself accompanies the executable.
+
+If distribution of executable or object code is made by offering
+access to copy from a designated place, then offering equivalent
+access to copy the source code from the same place counts as
+distribution of the source code, even though third parties are not
+compelled to copy the source along with the object code.
+
+  4. You may not copy, modify, sublicense, or distribute the Program
+except as expressly provided under this License.  Any attempt
+otherwise to copy, modify, sublicense or distribute the Program is
+void, and will automatically terminate your rights under this License.
+However, parties who have received copies, or rights, from you under
+this License will not have their licenses terminated so long as such
+parties remain in full compliance.
+
+  5. You are not required to accept this License, since you have not
+signed it.  However, nothing else grants you permission to modify or
+distribute the Program or its derivative works.  These actions are
+prohibited by law if you do not accept this License.  Therefore, by
+modifying or distributing the Program (or any work based on the
+Program), you indicate your acceptance of this License to do so, and
+all its terms and conditions for copying, distributing or modifying
+the Program or works based on it.
+
+  6. Each time you redistribute the Program (or any work based on the
+Program), the recipient automatically receives a license from the
+original licensor to copy, distribute or modify the Program subject to
+these terms and conditions.  You may not impose any further
+restrictions on the recipients' exercise of the rights granted herein.
+You are not responsible for enforcing compliance by third parties to
+this License.
+
+  7. If, as a consequence of a court judgment or allegation of patent
+infringement or for any other reason (not limited to patent issues),
+conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot
+distribute so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you
+may not distribute the Program at all.  For example, if a patent
+license would not permit royalty-free redistribution of the Program by
+all those who receive copies directly or indirectly through you, then
+the only way you could satisfy both it and this License would be to
+refrain entirely from distribution of the Program.
+
+If any portion of this section is held invalid or unenforceable under
+any particular circumstance, the balance of the section is intended to
+apply and the section as a whole is intended to apply in other
+circumstances.
+
+It is not the purpose of this section to induce you to infringe any
+patents or other property right claims or to contest validity of any
+such claims; this section has the sole purpose of protecting the
+integrity of the free software distribution system, which is
+implemented by public license practices.  Many people have made
+generous contributions to the wide range of software distributed
+through that system in reliance on consistent application of that
+system; it is up to the author/donor to decide if he or she is willing
+to distribute software through any other system and a licensee cannot
+impose that choice.
+
+This section is intended to make thoroughly clear what is believed to
+be a consequence of the rest of this License.
+
+  8. If the distribution and/or use of the Program is restricted in
+certain countries either by patents or by copyrighted interfaces, the
+original copyright holder who places the Program under this License
+may add an explicit geographical distribution limitation excluding
+those countries, so that distribution is permitted only in or among
+countries not thus excluded.  In such case, this License incorporates
+the limitation as if written in the body of this License.
+
+  9. The Free Software Foundation may publish revised and/or new versions
+of the General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+Each version is given a distinguishing version number.  If the Program
+specifies a version number of this License which applies to it and "any
+later version", you have the option of following the terms and conditions
+either of that version or of any later version published by the Free
+Software Foundation.  If the Program does not specify a version number of
+this License, you may choose any version ever published by the Free Software
+Foundation.
+
+  10. If you wish to incorporate parts of the Program into other free
+programs whose distribution conditions are different, write to the author
+to ask for permission.  For software which is copyrighted by the Free
+Software Foundation, write to the Free Software Foundation; we sometimes
+make exceptions for this.  Our decision will be guided by the two goals
+of preserving the free status of all derivatives of our free software and
+of promoting the sharing and reuse of software generally.
+
+			    NO WARRANTY
+
+  11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
+FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
+OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
+PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
+OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
+TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
+PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
+REPAIR OR CORRECTION.
+
+  12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
+REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
+INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
+OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
+TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
+YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
+PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGES.
+
+		     END OF TERMS AND CONDITIONS
+
+	    How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+convey the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+
+Also add information on how to contact you by electronic and paper mail.
+
+If the program is interactive, make it output a short notice like this
+when it starts in an interactive mode:
+
+    Gnomovision version 69, Copyright (C) year  name of author
+    Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, the commands you use may
+be called something other than `show w' and `show c'; they could even be
+mouse-clicks or menu items--whatever suits your program.
+
+You should also get your employer (if you work as a programmer) or your
+school, if any, to sign a "copyright disclaimer" for the program, if
+necessary.  Here is a sample; alter the names:
+
+  Yoyodyne, Inc., hereby disclaims all copyright interest in the program
+  `Gnomovision' (which makes passes at compilers) written by James Hacker.
+
+  <signature of Ty Coon>, 1 April 1989
+  Ty Coon, President of Vice
+
+This General Public License does not permit incorporating your program into
+proprietary programs.  If your program is a subroutine library, you may
+consider it more useful to permit linking proprietary applications with the
+library.  If this is what you want to do, use the GNU Library General
+Public License instead of this License.

COPYING.LIB

@@ -0,0 +1,504 @@
+		  GNU LESSER GENERAL PUBLIC LICENSE
+		       Version 2.1, February 1999
+
+ Copyright (C) 1991, 1999 Free Software Foundation, Inc.
+ 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+[This is the first released version of the Lesser GPL.  It also counts
+ as the successor of the GNU Library Public License, version 2, hence
+ the version number 2.1.]
+
+			    Preamble
+
+  The licenses for most software are designed to take away your
+freedom to share and change it.  By contrast, the GNU General Public
+Licenses are intended to guarantee your freedom to share and change
+free software--to make sure the software is free for all its users.
+
+  This license, the Lesser General Public License, applies to some
+specially designated software packages--typically libraries--of the
+Free Software Foundation and other authors who decide to use it.  You
+can use it too, but we suggest you first think carefully about whether
+this license or the ordinary General Public License is the better
+strategy to use in any particular case, based on the explanations below.
+
+  When we speak of free software, we are referring to freedom of use,
+not price.  Our General Public Licenses are designed to make sure that
+you have the freedom to distribute copies of free software (and charge
+for this service if you wish); that you receive source code or can get
+it if you want it; that you can change the software and use pieces of
+it in new free programs; and that you are informed that you can do
+these things.
+
+  To protect your rights, we need to make restrictions that forbid
+distributors to deny you these rights or to ask you to surrender these
+rights.  These restrictions translate to certain responsibilities for
+you if you distribute copies of the library or if you modify it.
+
+  For example, if you distribute copies of the library, whether gratis
+or for a fee, you must give the recipients all the rights that we gave
+you.  You must make sure that they, too, receive or can get the source
+code.  If you link other code with the library, you must provide
+complete object files to the recipients, so that they can relink them
+with the library after making changes to the library and recompiling
+it.  And you must show them these terms so they know their rights.
+
+  We protect your rights with a two-step method: (1) we copyright the
+library, and (2) we offer you this license, which gives you legal
+permission to copy, distribute and/or modify the library.
+
+  To protect each distributor, we want to make it very clear that
+there is no warranty for the free library.  Also, if the library is
+modified by someone else and passed on, the recipients should know
+that what they have is not the original version, so that the original
+author's reputation will not be affected by problems that might be
+introduced by others.
+
+  Finally, software patents pose a constant threat to the existence of
+any free program.  We wish to make sure that a company cannot
+effectively restrict the users of a free program by obtaining a
+restrictive license from a patent holder.  Therefore, we insist that
+any patent license obtained for a version of the library must be
+consistent with the full freedom of use specified in this license.
+
+  Most GNU software, including some libraries, is covered by the
+ordinary GNU General Public License.  This license, the GNU Lesser
+General Public License, applies to certain designated libraries, and
+is quite different from the ordinary General Public License.  We use
+this license for certain libraries in order to permit linking those
+libraries into non-free programs.
+
+  When a program is linked with a library, whether statically or using
+a shared library, the combination of the two is legally speaking a
+combined work, a derivative of the original library.  The ordinary
+General Public License therefore permits such linking only if the
+entire combination fits its criteria of freedom.  The Lesser General
+Public License permits more lax criteria for linking other code with
+the library.
+
+  We call this license the "Lesser" General Public License because it
+does Less to protect the user's freedom than the ordinary General
+Public License.  It also provides other free software developers Less
+of an advantage over competing non-free programs.  These disadvantages
+are the reason we use the ordinary General Public License for many
+libraries.  However, the Lesser license provides advantages in certain
+special circumstances.
+
+  For example, on rare occasions, there may be a special need to
+encourage the widest possible use of a certain library, so that it becomes
+a de-facto standard.  To achieve this, non-free programs must be
+allowed to use the library.  A more frequent case is that a free
+library does the same job as widely used non-free libraries.  In this
+case, there is little to gain by limiting the free library to free
+software only, so we use the Lesser General Public License.
+
+  In other cases, permission to use a particular library in non-free
+programs enables a greater number of people to use a large body of
+free software.  For example, permission to use the GNU C Library in
+non-free programs enables many more people to use the whole GNU
+operating system, as well as its variant, the GNU/Linux operating
+system.
+
+  Although the Lesser General Public License is Less protective of the
+users' freedom, it does ensure that the user of a program that is
+linked with the Library has the freedom and the wherewithal to run
+that program using a modified version of the Library.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.  Pay close attention to the difference between a
+"work based on the library" and a "work that uses the library".  The
+former contains code derived from the library, whereas the latter must
+be combined with the library in order to run.
+
+		  GNU LESSER GENERAL PUBLIC LICENSE
+   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
+
+  0. This License Agreement applies to any software library or other
+program which contains a notice placed by the copyright holder or
+other authorized party saying it may be distributed under the terms of
+this Lesser General Public License (also called "this License").
+Each licensee is addressed as "you".
+
+  A "library" means a collection of software functions and/or data
+prepared so as to be conveniently linked with application programs
+(which use some of those functions and data) to form executables.
+
+  The "Library", below, refers to any such software library or work
+which has been distributed under these terms.  A "work based on the
+Library" means either the Library or any derivative work under
+copyright law: that is to say, a work containing the Library or a
+portion of it, either verbatim or with modifications and/or translated
+straightforwardly into another language.  (Hereinafter, translation is
+included without limitation in the term "modification".)
+
+  "Source code" for a work means the preferred form of the work for
+making modifications to it.  For a library, complete source code means
+all the source code for all modules it contains, plus any associated
+interface definition files, plus the scripts used to control compilation
+and installation of the library.
+
+  Activities other than copying, distribution and modification are not
+covered by this License; they are outside its scope.  The act of
+running a program using the Library is not restricted, and output from
+such a program is covered only if its contents constitute a work based
+on the Library (independent of the use of the Library in a tool for
+writing it).  Whether that is true depends on what the Library does
+and what the program that uses the Library does.
+  
+  1. You may copy and distribute verbatim copies of the Library's
+complete source code as you receive it, in any medium, provided that
+you conspicuously and appropriately publish on each copy an
+appropriate copyright notice and disclaimer of warranty; keep intact
+all the notices that refer to this License and to the absence of any
+warranty; and distribute a copy of this License along with the
+Library.
+
+  You may charge a fee for the physical act of transferring a copy,
+and you may at your option offer warranty protection in exchange for a
+fee.
+
+  2. You may modify your copy or copies of the Library or any portion
+of it, thus forming a work based on the Library, and copy and
+distribute such modifications or work under the terms of Section 1
+above, provided that you also meet all of these conditions:
+
+    a) The modified work must itself be a software library.
+
+    b) You must cause the files modified to carry prominent notices
+    stating that you changed the files and the date of any change.
+
+    c) You must cause the whole of the work to be licensed at no
+    charge to all third parties under the terms of this License.
+
+    d) If a facility in the modified Library refers to a function or a
+    table of data to be supplied by an application program that uses
+    the facility, other than as an argument passed when the facility
+    is invoked, then you must make a good faith effort to ensure that,
+    in the event an application does not supply such function or
+    table, the facility still operates, and performs whatever part of
+    its purpose remains meaningful.
+
+    (For example, a function in a library to compute square roots has
+    a purpose that is entirely well-defined independent of the
+    application.  Therefore, Subsection 2d requires that any
+    application-supplied function or table used by this function must
+    be optional: if the application does not supply it, the square
+    root function must still compute square roots.)
+
+These requirements apply to the modified work as a whole.  If
+identifiable sections of that work are not derived from the Library,
+and can be reasonably considered independent and separate works in
+themselves, then this License, and its terms, do not apply to those
+sections when you distribute them as separate works.  But when you
+distribute the same sections as part of a whole which is a work based
+on the Library, the distribution of the whole must be on the terms of
+this License, whose permissions for other licensees extend to the
+entire whole, and thus to each and every part regardless of who wrote
+it.
+
+Thus, it is not the intent of this section to claim rights or contest
+your rights to work written entirely by you; rather, the intent is to
+exercise the right to control the distribution of derivative or
+collective works based on the Library.
+
+In addition, mere aggregation of another work not based on the Library
+with the Library (or with a work based on the Library) on a volume of
+a storage or distribution medium does not bring the other work under
+the scope of this License.
+
+  3. You may opt to apply the terms of the ordinary GNU General Public
+License instead of this License to a given copy of the Library.  To do
+this, you must alter all the notices that refer to this License, so
+that they refer to the ordinary GNU General Public License, version 2,
+instead of to this License.  (If a newer version than version 2 of the
+ordinary GNU General Public License has appeared, then you can specify
+that version instead if you wish.)  Do not make any other change in
+these notices.
+
+  Once this change is made in a given copy, it is irreversible for
+that copy, so the ordinary GNU General Public License applies to all
+subsequent copies and derivative works made from that copy.
+
+  This option is useful when you wish to copy part of the code of
+the Library into a program that is not a library.
+
+  4. You may copy and distribute the Library (or a portion or
+derivative of it, under Section 2) in object code or executable form
+under the terms of Sections 1 and 2 above provided that you accompany
+it with the complete corresponding machine-readable source code, which
+must be distributed under the terms of Sections 1 and 2 above on a
+medium customarily used for software interchange.
+
+  If distribution of object code is made by offering access to copy
+from a designated place, then offering equivalent access to copy the
+source code from the same place satisfies the requirement to
+distribute the source code, even though third parties are not
+compelled to copy the source along with the object code.
+
+  5. A program that contains no derivative of any portion of the
+Library, but is designed to work with the Library by being compiled or
+linked with it, is called a "work that uses the Library".  Such a
+work, in isolation, is not a derivative work of the Library, and
+therefore falls outside the scope of this License.
+
+  However, linking a "work that uses the Library" with the Library
+creates an executable that is a derivative of the Library (because it
+contains portions of the Library), rather than a "work that uses the
+library".  The executable is therefore covered by this License.
+Section 6 states terms for distribution of such executables.
+
+  When a "work that uses the Library" uses material from a header file
+that is part of the Library, the object code for the work may be a
+derivative work of the Library even though the source code is not.
+Whether this is true is especially significant if the work can be
+linked without the Library, or if the work is itself a library.  The
+threshold for this to be true is not precisely defined by law.
+
+  If such an object file uses only numerical parameters, data
+structure layouts and accessors, and small macros and small inline
+functions (ten lines or less in length), then the use of the object
+file is unrestricted, regardless of whether it is legally a derivative
+work.  (Executables containing this object code plus portions of the
+Library will still fall under Section 6.)
+
+  Otherwise, if the work is a derivative of the Library, you may
+distribute the object code for the work under the terms of Section 6.
+Any executables containing that work also fall under Section 6,
+whether or not they are linked directly with the Library itself.
+
+  6. As an exception to the Sections above, you may also combine or
+link a "work that uses the Library" with the Library to produce a
+work containing portions of the Library, and distribute that work
+under terms of your choice, provided that the terms permit
+modification of the work for the customer's own use and reverse
+engineering for debugging such modifications.
+
+  You must give prominent notice with each copy of the work that the
+Library is used in it and that the Library and its use are covered by
+this License.  You must supply a copy of this License.  If the work
+during execution displays copyright notices, you must include the
+copyright notice for the Library among them, as well as a reference
+directing the user to the copy of this License.  Also, you must do one
+of these things:
+
+    a) Accompany the work with the complete corresponding
+    machine-readable source code for the Library including whatever
+    changes were used in the work (which must be distributed under
+    Sections 1 and 2 above); and, if the work is an executable linked
+    with the Library, with the complete machine-readable "work that
+    uses the Library", as object code and/or source code, so that the
+    user can modify the Library and then relink to produce a modified
+    executable containing the modified Library.  (It is understood
+    that the user who changes the contents of definitions files in the
+    Library will not necessarily be able to recompile the application
+    to use the modified definitions.)
+
+    b) Use a suitable shared library mechanism for linking with the
+    Library.  A suitable mechanism is one that (1) uses at run time a
+    copy of the library already present on the user's computer system,
+    rather than copying library functions into the executable, and (2)
+    will operate properly with a modified version of the library, if
+    the user installs one, as long as the modified version is
+    interface-compatible with the version that the work was made with.
+
+    c) Accompany the work with a written offer, valid for at
+    least three years, to give the same user the materials
+    specified in Subsection 6a, above, for a charge no more
+    than the cost of performing this distribution.
+
+    d) If distribution of the work is made by offering access to copy
+    from a designated place, offer equivalent access to copy the above
+    specified materials from the same place.
+
+    e) Verify that the user has already received a copy of these
+    materials or that you have already sent this user a copy.
+
+  For an executable, the required form of the "work that uses the
+Library" must include any data and utility programs needed for
+reproducing the executable from it.  However, as a special exception,
+the materials to be distributed need not include anything that is
+normally distributed (in either source or binary form) with the major
+components (compiler, kernel, and so on) of the operating system on
+which the executable runs, unless that component itself accompanies
+the executable.
+
+  It may happen that this requirement contradicts the license
+restrictions of other proprietary libraries that do not normally
+accompany the operating system.  Such a contradiction means you cannot
+use both them and the Library together in an executable that you
+distribute.
+
+  7. You may place library facilities that are a work based on the
+Library side-by-side in a single library together with other library
+facilities not covered by this License, and distribute such a combined
+library, provided that the separate distribution of the work based on
+the Library and of the other library facilities is otherwise
+permitted, and provided that you do these two things:
+
+    a) Accompany the combined library with a copy of the same work
+    based on the Library, uncombined with any other library
+    facilities.  This must be distributed under the terms of the
+    Sections above.
+
+    b) Give prominent notice with the combined library of the fact
+    that part of it is a work based on the Library, and explaining
+    where to find the accompanying uncombined form of the same work.
+
+  8. You may not copy, modify, sublicense, link with, or distribute
+the Library except as expressly provided under this License.  Any
+attempt otherwise to copy, modify, sublicense, link with, or
+distribute the Library is void, and will automatically terminate your
+rights under this License.  However, parties who have received copies,
+or rights, from you under this License will not have their licenses
+terminated so long as such parties remain in full compliance.
+
+  9. You are not required to accept this License, since you have not
+signed it.  However, nothing else grants you permission to modify or
+distribute the Library or its derivative works.  These actions are
+prohibited by law if you do not accept this License.  Therefore, by
+modifying or distributing the Library (or any work based on the
+Library), you indicate your acceptance of this License to do so, and
+all its terms and conditions for copying, distributing or modifying
+the Library or works based on it.
+
+  10. Each time you redistribute the Library (or any work based on the
+Library), the recipient automatically receives a license from the
+original licensor to copy, distribute, link with or modify the Library
+subject to these terms and conditions.  You may not impose any further
+restrictions on the recipients' exercise of the rights granted herein.
+You are not responsible for enforcing compliance by third parties with
+this License.
+
+  11. If, as a consequence of a court judgment or allegation of patent
+infringement or for any other reason (not limited to patent issues),
+conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot
+distribute so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you
+may not distribute the Library at all.  For example, if a patent
+license would not permit royalty-free redistribution of the Library by
+all those who receive copies directly or indirectly through you, then
+the only way you could satisfy both it and this License would be to
+refrain entirely from distribution of the Library.
+
+If any portion of this section is held invalid or unenforceable under any
+particular circumstance, the balance of the section is intended to apply,
+and the section as a whole is intended to apply in other circumstances.
+
+It is not the purpose of this section to induce you to infringe any
+patents or other property right claims or to contest validity of any
+such claims; this section has the sole purpose of protecting the
+integrity of the free software distribution system which is
+implemented by public license practices.  Many people have made
+generous contributions to the wide range of software distributed
+through that system in reliance on consistent application of that
+system; it is up to the author/donor to decide if he or she is willing
+to distribute software through any other system and a licensee cannot
+impose that choice.
+
+This section is intended to make thoroughly clear what is believed to
+be a consequence of the rest of this License.
+
+  12. If the distribution and/or use of the Library is restricted in
+certain countries either by patents or by copyrighted interfaces, the
+original copyright holder who places the Library under this License may add
+an explicit geographical distribution limitation excluding those countries,
+so that distribution is permitted only in or among countries not thus
+excluded.  In such case, this License incorporates the limitation as if
+written in the body of this License.
+
+  13. The Free Software Foundation may publish revised and/or new
+versions of the Lesser General Public License from time to time.
+Such new versions will be similar in spirit to the present version,
+but may differ in detail to address new problems or concerns.
+
+Each version is given a distinguishing version number.  If the Library
+specifies a version number of this License which applies to it and
+"any later version", you have the option of following the terms and
+conditions either of that version or of any later version published by
+the Free Software Foundation.  If the Library does not specify a
+license version number, you may choose any version ever published by
+the Free Software Foundation.
+
+  14. If you wish to incorporate parts of the Library into other free
+programs whose distribution conditions are incompatible with these,
+write to the author to ask for permission.  For software which is
+copyrighted by the Free Software Foundation, write to the Free
+Software Foundation; we sometimes make exceptions for this.  Our
+decision will be guided by the two goals of preserving the free status
+of all derivatives of our free software and of promoting the sharing
+and reuse of software generally.
+
+			    NO WARRANTY
+
+  15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO
+WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW.
+EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR
+OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY
+KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE
+LIBRARY IS WITH YOU.  SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME
+THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN
+WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY
+AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU
+FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR
+CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE
+LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING
+RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A
+FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF
+SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
+DAMAGES.
+
+		     END OF TERMS AND CONDITIONS
+
+           How to Apply These Terms to Your New Libraries
+
+  If you develop a new library, and you want it to be of the greatest
+possible use to the public, we recommend making it free software that
+everyone can redistribute and change.  You can do so by permitting
+redistribution under these terms (or, alternatively, under the terms of the
+ordinary General Public License).
+
+  To apply these terms, attach the following notices to the library.  It is
+safest to attach them to the start of each source file to most effectively
+convey the exclusion of warranty; and each file should have at least the
+"copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the library's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This library is free software; you can redistribute it and/or
+    modify it under the terms of the GNU Lesser General Public
+    License as published by the Free Software Foundation; either
+    version 2.1 of the License, or (at your option) any later version.
+
+    This library is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+    Lesser General Public License for more details.
+
+    You should have received a copy of the GNU Lesser General Public
+    License along with this library; if not, write to the Free Software
+    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
+
+Also add information on how to contact you by electronic and paper mail.
+
+You should also get your employer (if you work as a programmer) or your
+school, if any, to sign a "copyright disclaimer" for the library, if
+necessary.  Here is a sample; alter the names:
+
+  Yoyodyne, Inc., hereby disclaims all copyright interest in the
+  library `Frob' (a library for tweaking knobs) written by James Random Hacker.
+
+  <signature of Ty Coon>, 1 April 1990
+  Ty Coon, President of Vice
+
+That's all there is to it!
+
+

INSTALL

@@ -0,0 +1,31 @@
+Installation
+============
+
+1) Generate custom makefiles.
+
+   Run the 'configure' script from the top directory.
+
+   If you don't want to include the LVM1 backwards-compatibility code use:
+     ./configure --with-lvm1=none 
+
+   To separate the LVM1 support into a shared library loaded by lvm.conf use:
+     ./configure --with-lvm1=shared
+
+   Use ./configure --help to see other options.
+
+2) Build and install.
+
+   Run 'make' from the top directory to build everything you configured.
+   Run 'make install' to build and install everything you configured.
+
+   If you only want the device-mapper libraries and tools use
+   'make device-mapper' or 'make install_device-mapper'.
+
+3) If using LVM2, create a configuration file.
+
+   The tools will work fine without a configuration file being
+   present, but you ought to review the example file in doc/example.conf.
+
+Please also refer to the WHATS_NEW file and the manual pages for the 
+individual commands.
+

Makefile.in

@@ -0,0 +1,125 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+SUBDIRS = doc include man scripts
+
+ifeq ("@UDEV_RULES@", "yes")
+  SUBDIRS += udev
+endif
+
+ifeq ("@INTL@", "yes")
+  SUBDIRS += po
+endif
+
+SUBDIRS += lib tools daemons libdm
+
+ifeq ("@APPLIB@", "yes")
+  SUBDIRS += liblvm
+endif
+
+ifeq ($(MAKECMDGOALS),distclean)
+  SUBDIRS += daemons/clvmd \
+	     daemons/cmirrord \
+	     daemons/dmeventd/plugins \
+	     daemons/dmeventd \
+	     lib/format1 \
+	     lib/format_pool \
+	     lib/locking \
+	     lib/mirror \
+	     lib/snapshot \
+	     liblvm \
+	     udev \
+	     test/api \
+	     test \
+	     po
+  DISTCLEAN_TARGETS += lib/misc/configure.h lib/misc/lvm-version.h
+  DISTCLEAN_DIRS += lcov_reports*
+endif
+
+include make.tmpl
+
+libdm: include
+lib: libdm
+liblvm: lib
+daemons: lib tools
+tools: lib device-mapper
+po: tools daemons
+
+libdm.device-mapper: include.device-mapper
+daemons.device-mapper: libdm.device-mapper
+tools.device-mapper: libdm.device-mapper
+device-mapper: tools.device-mapper daemons.device-mapper man.device-mapper
+
+ifeq ("@INTL@", "yes")
+lib.pofile: include.pofile
+tools.pofile: lib.pofile
+daemons.pofile: lib.pofile
+po.pofile: tools.pofile daemons.pofile
+pofile: po.pofile
+endif
+
+ifneq ("@CFLOW_CMD@", "")
+tools.cflow: lib.cflow
+cflow: tools.cflow
+endif
+
+ifneq ("@CSCOPE_CMD@", "")
+cscope.out: tools
+	@CSCOPE_CMD@ -b -R
+all: cscope.out
+endif
+
+check: all
+	$(MAKE) -C test all
+
+ifneq ("@LCOV@", "")
+.PHONY: lcov-reset lcov lcov-dated
+
+ifeq ($(MAKECMDGOALS),lcov-dated)
+LCOV_REPORTS_DIR=$(top_srcdir)/lcov_reports-$(shell date +%Y%m%d%k%M%S)
+else
+LCOV_REPORTS_DIR=$(top_srcdir)/lcov_reports
+endif
+
+lcov-reset:
+	$(LCOV) -d $(top_srcdir)/dmeventd --zerocounters
+	$(LCOV) -d $(top_srcdir)/libdm --zerocounters
+	$(LCOV) -d $(top_srcdir)/lib --zerocounters
+	$(LCOV) -d $(top_srcdir)/tools --zerocounters
+
+lcov: all
+	$(RM) -rf $(LCOV_REPORTS_DIR)
+	$(MKDIR_P) $(LCOV_REPORTS_DIR)
+	$(LCOV) -b $(top_srcdir)/libdm -d $(top_srcdir)/libdm -c -o $(LCOV_REPORTS_DIR)/libdm.info
+	$(LCOV) -b $(top_srcdir)/lib -d $(top_srcdir)/lib -c -o $(LCOV_REPORTS_DIR)/lib.info
+	$(LCOV) -b $(top_srcdir)/tools -d $(top_srcdir)/tools -c -o $(LCOV_REPORTS_DIR)/tools.info
+	DMEVENTD_INFO="$(LCOV_REPORTS_DIR)/dmeventd.info" ;\
+	DMEVENTD_INFO_A="-a $$DMEVENTDINFO" ;\
+	$(LCOV) -b $(top_srcdir)/dmeventd -d $(top_srcdir)/dmeventd -c -o $$DMEVENTD_INFO || DMEVENTD_INFO_A="" ;\
+	$(LCOV) $$DMEVENTD_INFO_A -a $(LCOV_REPORTS_DIR)/lib.info \
+		-a $(LCOV_REPORTS_DIR)/libdm.info \
+		-a $(LCOV_REPORTS_DIR)/tools.info \
+		-o $(LCOV_REPORTS_DIR)/lvm.info
+ifneq ("@GENHTML@", "")
+	$(GENHTML) -o $(LCOV_REPORTS_DIR) -p $(top_srcdir) $(LCOV_REPORTS_DIR)/lvm.info
+endif
+
+lcov-dated: lcov
+
+endif

README

@@ -1,2 +1,27 @@
-This is pretty much empty so far...if you can't see subdirectories,
-try 'cvs -f update'
+This tree contains the LVM2 and device-mapper tools and libraries.
+
+For more information about LVM2 read the changelog in the WHATS_NEW file.
+Installation instructions are in INSTALL.
+
+There is no warranty - see COPYING and COPYING.LIB.
+
+Tarballs are available from:
+  ftp://sources.redhat.com/pub/lvm2/
+
+To access the CVS tree use:
+  cvs -d :pserver:cvs@sources.redhat.com:/cvs/lvm2 login
+  CVS password: cvs
+  cvs -d :pserver:cvs@sources.redhat.com:/cvs/lvm2 co LVM2
+
+Mailing list for general discussion related to LVM2:
+  linux-lvm@redhat.com
+  Subscribe from https://www.redhat.com/mailman/listinfo/linux-lvm
+
+Mailing list for LVM2 development, patches and commits:
+  lvm-devel@redhat.com
+  Subscribe from https://www.redhat.com/mailman/listinfo/linux-lvm
+
+Mailing list for device-mapper development, including kernel patches
+and multipath-tools:
+  dm-devel@redhat.com
+  Subscribe from https://www.redhat.com/mailman/listinfo/dm-devel

VERSION

@@ -0,0 +1 @@
+2.02.60(1)-cvs (2010-01-23)

VERSION_DM

@@ -0,0 +1 @@
+1.02.43-cvs (2010-01-23)

WHATS_NEW_DM

@@ -0,0 +1,461 @@
+Version 1.02.43 - 21st January 2010
+===================================
+  Remove bitset, hash and pool headers superceded by libdevmapper.h.
+  Fix off-by-one error causing bad cluster mirror table construction.
+
+Version 1.02.42 - 14th January 2010
+===================================
+  Add support for the "snapshot-merge" kernel target (2.6.33-rc1).
+  Introduce a third activation_priority level in dm_tree_activate_children.
+
+Version 1.02.41 - 12th January 2010
+===================================
+  If DM_UDEV_DISABLE_CHECKING is set in environment, disable udev warnings.
+  Add dm_tree_add_dev_with_udev_flags to provide wider support for udev flags.
+  Add --noudevrules option for dmsetup to disable /dev node management by udev.
+  Fix 'dmsetup info -c -o all' to show all fields.
+  Return errors if dm_tree_*_children functions fail.
+  Fix coredump and memory leak for 'dmsetup help -c'.
+  Disable udev rules for change events with DISK_RO set.
+
+Version 1.02.40 - 19th November 2009
+====================================
+  Fix install_device-mapper Makefile target to not build dmeventd plugins.
+  Support udev flags even when udev_sync is disabled or not compiled in.
+  Remove 'last_rule' from udev rules: honour DM_UDEV_DISABLE_OTHER_RULES_FLAG.
+  Add dmsetup --inactive support.
+  Add dm_task_query_inactive_table to libdevmapper for kernel driver >= 4.16.
+  Fix hash lookup segfault when keys compared are different lengths.
+
+Version 1.02.39 - 26th October 2009
+===================================
+  Remove strict default permissions for DM devices from 95-dm-notify.rules.
+  Add dmsetup udevflags command to decode udev flags in given cookie value.
+  Support udev flags in libdevmapper incl. dm_tree_add_new_dev_with_udev_flags.
+  Make libdm ABI consistent when built with/without selinux support.
+
+Version 1.02.38 - 25th September 2009
+=====================================
+  Export DM_DEV_DIR_UMASK, the default umask for /dev directories created.
+  Handle any path supplied to dm_task_set_name by looking up in /dev/mapper.
+  Add several examples to 12-dm-permissions.rules.
+  Add splitname and --yes to dmsetup man page.
+  Fix _mirror_emit_segment_line return code.
+  Fix dmeventd _temporary_log_fn parameters. (2.02.50)
+
+Version 1.02.37 - 15th September 2009
+=====================================
+  Add dmsetup manpage entries for udevcomplete_all and udevcookies.
+  Check udev is running when processing cookies and retain state internally.
+  Add y|--yes option to dmsetup for default 'yes' answer to prompts.
+  Fix tools Makefile to process dmsetup sources separately.
+  Restore umask when device node creation fails.
+  Check kernel vsn to use 'block_on_error' or 'handle_errors' in mirror table.
+  Add dm-log-userspace.h to tree for cmirrord builds.
+
+Version 1.02.36 - 6th August 2009
+=================================
+  Add udevcookies, udevcomplete, udevcomplete_all and --noudevwait to dmsetup.
+  Add libdevmapper functions to support synchronisation with udev.
+
+Version 1.02.35 - 28th July 2009
+================================
+  Add LOG_LINE_WITH_ERRNO macro.
+  Use log_error macro consistently throughout in place of log_err.
+
+Version 1.02.34 - 15th July 2009
+================================
+  Use _exit() not exit() after forking to avoid flushing libc buffers twice.
+  Rename plog macro to LOG_LINE & add LOG_MESG variant for dm_dump_memory_debug.
+  Change plog to use dm_log_with_errno unless deprecated dm_log_init was used.
+  Add dm_log_with_errno and dm_log_with_errno_init, deprecating the old fns.
+  Fix whitespace in linear target line to fix identical table line detection.
+  Add device number to more log messages during activation.
+
+Version 1.02.33 - 30th June 2009
+================================
+  Don't fallback to default major number: use dm_task_set_major_minor. (1.02.31)
+  Do not fork daemon when dmeventd cannot be found.
+  Add crypt target handling to libdevmapper tree nodes.
+  Add splitname command to dmsetup.
+  Add subsystem, vg_name, lv_name, lv_layer fields to dmsetup reports.
+  Make mempool optional in dm_split_lvm_name().
+
+Version 1.02.32 - 21st May 2009
+===============================
+  Only generate libdevmapper.a when configured to link statically.
+  Export dm_tree_node_size_changed() from libdevmapper.
+  Propagate the table size_changed property up the dm device tree.
+  Detect failure to free memory pools when releasing the library.
+  Fix segfault when getopt processes dmsetup -U, -G and -M options.
+
+Version 1.02.31 - 3rd March 2009
+================================
+  If kernel supports only one dm major number, use in place of any supplied.
+
+Version 1.02.30 - 26th January 2009
+====================================
+  Add "all" field to reports expanding to all fields of report type.
+  Enforce device name length and character limitations in libdm.
+  Replace _dm_snprintf with EMIT_PARAMS macro for creating target lines.
+
+Version 1.02.29 - 10th November 2008
+====================================
+  Merge device-mapper into the LVM2 tree.
+  Split out dm-logging.h from log.h.
+  Use lvm-types.h.
+  Add usrsbindir to configure.
+
+Version 1.02.28 - 18th September 2008
+=====================================
+  Only resume devices in dm_tree_preload_children if size changes.
+  Extend deptree buffers so the largest possible device numbers fit.
+  Generate versioned libdevmapper-event.so.
+  Underline longer report help text headings.
+
+Version 1.02.27 - 25th June 2008
+================================
+  Align struct memblock in dbg_malloc for sparc.
+  Add --unquoted and --rows to dmsetup.
+  Avoid compiler warning about cast in dmsetup.c's OFFSET_OF macro.
+  Fix inverted no_flush debug message.
+  Remove --enable-jobs from configure. (Set at runtime instead.)
+  Bring configure.in and list.h into line with the lvm2 versions.
+
+Version 1.02.26 - 6th June 2008
+===============================
+  Initialise params buffer to empty string in _emit_segment.
+  Skip add_dev_node when ioctls disabled.
+  Make dm_hash_iter safe against deletion.
+  Accept a NULL pointer to dm_free silently.
+  Add tables_loaded, readonly and suspended columns to reports.
+  Add --nameprefixes to dmsetup.
+  Add field name prefix option to reporting functions.
+  Calculate string size within dm_pool_grow_object.
+
+Version 1.02.25 - 10th April 2008
+=================================
+  Remove redundant if-before-free tests.
+  Use log_warn for reporting field help text instead of log_print.
+  Change cluster mirror log type name (s/clustered_/clustered-/)
+
+Version 1.02.24 - 20th December 2007
+====================================
+  Fix deptree to pass new name to _resume_node after a rename.
+  Suppress other node operations if node is deleted.
+  Add node operation stack debug messages.
+  Report error when empty device name passed to readahead functions.
+  Fix minimum readahead debug message.
+
+Version 1.02.23 - 5th December 2007
+===================================
+  Update dm-ioctl.h after removal of compat code.
+  Add readahead support to libdevmapper and dmsetup.
+  Fix double free in a libdevmapper-event error path.
+  Fix configure --with-dmeventd-path substitution.
+  Allow a DM_DEV_DIR environment variable to override /dev in dmsetup.
+  Create a libdevmapper.so.$LIB_VERSION symlink within the build tree.
+  Avoid static link failure with some SELinux libraries that require libpthread.
+  Remove obsolete dmfs code from tree and update INSTALL.
+
+Version 1.02.22 - 21st August 2007
+==================================
+  Fix inconsistent licence notices: executables are GPLv2; libraries LGPLv2.1.
+  Update to use autoconf 2.61, while still supporting 2.57.
+  Avoid repeated dm_task free on some dm_event_get_registered_device errors.
+  Introduce log_sys_* macros from LVM2.
+  Export dm_fclose and dm_create_dir; remove libdm-file.h.
+  Don't log EROFS mkdir failures in _create_dir_recursive (for LVM2).
+  Add fclose wrapper dm_fclose that catches write failures (using ferror).
+
+Version 1.02.21 - 13th July 2007
+================================
+  Introduce _LOG_STDERR to send log_warn() messages to stderr not stdout.
+  Fix dmsetup -o devno string termination. (1.02.20)
+
+Version 1.02.20 - 15th June 2007
+================================
+  Fix default dmsetup report buffering and add --unbuffered.
+  Add tree-based and dependency fields to dmsetup reports.
+
+Version 1.02.19 - 27th April 2007
+=================================
+  Standardise protective include file #defines.
+  Add regex functions to library.
+  Avoid trailing separator in reports when there are hidden sort fields.
+  Fix segfault in 'dmsetup status' without --showkeys against crypt target.
+  Deal with some more compiler warnings.
+  Introduce _add_field() and _is_same_field() to libdm-report.c.
+  Fix some libdevmapper-event and dmeventd memory leaks.
+  Remove unnecessary memset() return value checks.
+  Fix a few leaks in reporting error paths. [1.02.15+]
+
+Version 1.02.18 - 13th February 2007
+====================================
+  Improve dmeventd messaging protocol: drain pipe and tag messages.
+
+Version 1.02.17 - 29th January 2007
+===================================
+  Add recent reporting options to dmsetup man page.
+  Revise some report fields names.
+  Add dmsetup 'help' command and update usage text.
+  Use fixed-size fields in report interface and reorder.
+
+Version 1.02.16 - 25th January 2007
+===================================
+  Add some missing close() and fclose() return value checks.
+  Migrate dmsetup column-based output over to new libdevmapper report framework.
+  Add descriptions to reporting field definitions.
+  Add a dso-private variable to dmeventd dso interface.
+  Add dm_event_handler_[gs]et_timeout functions.
+  Streamline dm_report_field_* interface.
+  Add cmdline debug & version options to dmeventd.
+  Add DM_LIB_VERSION definition to configure.h.
+  Suppress 'Unrecognised field' error if report field is 'help'.
+  Add --separator and --sort to dmsetup (unused).
+  Make alignment flag optional when specifying report fields.
+
+Version 1.02.15 - 17th January 2007
+===================================
+  Add basic reporting functions to libdevmapper.
+  Fix a malloc error path in dmsetup message.
+  More libdevmapper-event interface changes and fixes.
+  Rename dm_saprintf() to dm_asprintf().
+  Report error if NULL pointer is supplied to dm_strdup_aux().
+  Reinstate dm_event_get_registered_device.
+
+Version 1.02.14 - 11th January 2007
+===================================
+  Add dm_saprintf().
+  Use CFLAGS when linking so mixed sparc builds can supply -m64.
+  Add dm_tree_use_no_flush_suspend().
+  Lots of dmevent changes including revised interface.
+  Export dm_basename().
+  Cope with a trailing space when comparing tables prior to possible reload.
+  Fix dmeventd to cope if monitored device disappears.
+
+Version 1.02.13 - 28 Nov 2006
+=============================
+  Update dmsetup man page (setgeometry & message).
+  Fix dmsetup free after getline with debug.
+  Suppress encryption key in 'dmsetup table' output unless --showkeys supplied.
+
+Version 1.02.12 - 13 Oct 2006
+=============================
+  Avoid deptree attempting to suspend a device that's already suspended.
+
+Version 1.02.11 -  12 Oct 2006
+==============================
+  Add suspend noflush support.
+  Add basic dmsetup loop support.
+  Switch dmsetup to use dm_malloc and dm_free.
+
+Version 1.02.10 - 19 Sep 2006
+=============================
+  Add dm_snprintf(), dm_split_words() and dm_split_lvm_name() to libdevmapper.
+  Reorder mm bounds_check code to reduce window for a dmeventd race.
+
+Version 1.02.09 - 15 Aug 2006
+=============================
+  Add --table argument to dmsetup for a one-line table.
+  Abort if errors are found during cmdline option processing.
+  Add lockfs indicator to debug output.
+
+Version 1.02.08 - 17 July 2006
+==============================
+  Append full patch to check in emails.
+  Avoid duplicate dmeventd subdir with 'make distclean'.
+  Update dmsetup man page.
+  Add --force to dmsetup remove* to load error target.
+  dmsetup remove_all also performs mknodes.
+  Don't suppress identical table reloads if permission changes.
+  Fix corelog segment line.
+  Suppress some compiler warnings.
+
+Version 1.02.07 - 11 May 2006
+=============================
+  Add DM_CORELOG flag to dm_tree_node_add_mirror_target().
+  Avoid a dmeventd compiler warning.
+
+Version 1.02.06 - 10 May 2006
+=============================
+  Move DEFS into configure.h.
+  Fix leaks in error paths found by coverity.
+  Remove dmsetup line buffer limitation.
+
+Version 1.02.05 - 19 Apr 2006
+=============================
+  Separate install_include target in makefiles.
+  Separate out DEFS from CFLAGS.
+  Support pkg-config.
+  Check for libsepol.
+
+Version 1.02.04 - 14 Apr 2006
+=============================
+  Bring dmsetup man page up-to-date.
+  Use name-based device refs if kernel doesn't support device number refs.
+  Fix memory leak (struct dm_ioctl) when struct dm_task is reused.
+  If _create_and_load_v4 fails part way through, revert the creation.
+  dmeventd thread/fifo fixes.
+  Add file & line to dm_strdup_aux().
+  Add setgeometry.
+
+Version 1.02.03 - 7 Feb 2006
+============================
+  Add exported functions to set uid, gid and mode.
+  Rename _log to dm_log and export.
+  Add dm_tree_skip_lockfs.
+  Fix dm_strdup debug definition.
+  Fix hash function to avoid using a negative array offset.
+  Don't inline _find in hash.c and tidy signed/unsigned etc.
+  Fix libdevmapper.h #endif.
+  Fix dmsetup version driver version.
+  Add sync, nosync and block_on_error mirror log parameters.
+  Add hweight32.
+  Fix dmeventd build.
+
+Version 1.02.02 - 2 Dec 2005
+============================
+  dmeventd added.
+  Export dm_task_update_nodes.
+  Use names instead of numbers in messages when ioctls fail.
+
+Version 1.02.01 - 23 Nov 2005
+=============================
+  Resume snapshot-origins last.
+  Drop leading zeros from dm_format_dev.
+  Suppress attempt to reload identical table.
+  Additional LVM- prefix matching for transitional period.
+
+Version 1.02.00 - 10 Nov 2005
+=============================
+  Added activation functions to library.
+  Added return macros.
+  Also suppress error if device doesn't exist with DM_DEVICE_STATUS.
+  Export dm_set_selinux_context().
+  Add dm_driver_version().
+  Added dependency tree functions to library.
+  Added hash, bitset, pool, dbg_malloc to library.
+  Added ls --tree to dmsetup.
+  Added dmsetup --nolockfs support for suspend/reload.
+
+Version 1.01.05 - 26 Sep 2005
+=============================
+  Resync list.h with LVM2.
+  Remember increased buffer size and use for subsequent calls.
+  On 'buffer full' condition, double buffer size and repeat ioctl.
+  Fix termination of getopt_long() option array.
+  Report 'buffer full' condition with v4 ioctl as well as with v1.
+
+Version 1.01.04 - 2 Aug 2005
+============================
+  Fix dmsetup ls -j and status --target with empty table.
+
+Version 1.01.03 - 13 Jun 2005
+=============================
+  Use matchpathcon mode parameter.
+  Fix configure script to re-enable selinux.
+
+Version 1.01.02 - 17 May 2005
+=============================
+  Call dm_lib_exit() and dm_lib_release() automatically now.
+  Add --target <target_type> filter to dmsetup table/status/ls.
+  Add --exec <command> to dmsetup ls.
+  Fix dmsetup getopt_long usage.
+
+Version 1.01.01 - 29 Mar 2005
+=============================
+  Update dmsetup man page.
+  Drop-in devmap_name replacement.
+  Add option to compile without ioctl for testing.
+  Fix DM_LIB_VERSION sed.
+
+Version 1.01.00 - 17 Jan 2005
+=============================
+  Add dm_task_no_open_count() to skip getting open_count.
+
+Version 1.00.21 - 7 Jan 2005
+============================
+  Fix /proc/devices parsing.
+
+Version 1.00.20 - 6 Jan 2005
+============================
+  Attempt to fix /dev/mapper/control transparently if it's wrong.
+  Configuration-time option for setting uid/gid/mode for /dev/mapper nodes.
+  Update kernel patches for 2.4.27/2.4.28-pre-4 (includes minor fixes).
+  Add --noheadings columns option for colon-separated dmsetup output.
+  Support device referencing by uuid or major/minor.
+  Warn if kernel data didn't fit in buffer.
+  Fix a printf.
+
+Version 1.00.19 - 3 July 2004
+=============================
+  More autoconf fixes.
+  Fix a dmsetup newline.
+  Fix device number handling for 2.6 kernels.
+
+Version 1.00.18 - 20 Jun 2004
+=============================
+  Fix a uuid free in libdm-iface.
+  Fix a targets string size calc in driver.
+  Add -c to dmsetup for column-based output.
+  Add target message-passing ioctl.
+
+Version 1.00.17 - 17 Apr 2004
+=============================
+  configure --with-owner= --with-group= to avoid -o and -g args to 'install'
+  Fix library selinux linking.
+
+Version 1.00.16 - 16 Apr 2004
+=============================
+  Ignore error setting selinux file context if fs doesn't support it.
+
+Version 1.00.15 - 7 Apr 2004
+============================
+  Fix status overflow check in kernel patches.
+
+Version 1.00.14 - 6 Apr 2004
+============================
+  Fix static selinux build.
+
+Version 1.00.13 - 6 Apr 2004
+============================
+  Add some basic selinux support.
+
+Version 1.00.12 - 6 Apr 2004
+============================
+  Fix dmsetup.static install.
+
+Version 1.00.11 - 5 Apr 2004
+============================
+  configure --enable-static_link does static build in addition to dynamic.
+  Moved Makefile library targets definition into template.
+
+Version 1.00.10 - 2 Apr 2004
+============================
+  Fix DESTDIR handling.
+  Static build installs to dmsetup.static.
+  Basic support for internationalisation.
+  Minor Makefile tidy-ups/fixes.
+
+Version 1.00.09 - 31 Mar 2004
+=============================
+  Update copyright notices to Red Hat.
+  Move full mknodes functionality from dmsetup into libdevmapper.
+  Avoid sscanf %as for uClibc compatibility.
+  Cope if DM_LIST_VERSIONS is not defined.
+  Add DM_LIST_VERSIONS functionality to kernel patches.
+  Generate new kernel patches for 2.4.26-rc1.
+
+Version 1.00.08 - 27 Feb 2004
+=============================
+  Added 'dmsetup targets'.
+  Added event_nr support to 'dmsetup wait'.
+  Updated dmsetup man page.
+  Allow logging function to be reset to use internal one.
+  Bring log macros in line with LVM2 ones.
+  Added 'make install_static_lib' which installs libdevmapper.a.
+  Made configure/makefiles closer to LVM2 versions.
+  Fixed DESTDIR for make install/install_static_lib.
+  Updated README/INSTALL to reflect move to sources.redhat.com.
+  Updated autoconf files to 2003-06-17.

autoconf/install-sh

@@ -1,36 +1,62 @@
 #!/bin/sh
-#
 # install - install a program, script, or datafile
-# This comes from X11R5 (mit/util/scripts/install.sh).
+
+scriptversion=2006-10-14.15
+
+# This originates from X11R5 (mit/util/scripts/install.sh), which was
+# later released in X11R6 (xc/config/util/install.sh) with the
+# following copyright and license.
 #
-# Copyright 1991 by the Massachusetts Institute of Technology
+# Copyright (C) 1994 X Consortium
 #
-# Permission to use, copy, modify, distribute, and sell this software and its
-# documentation for any purpose is hereby granted without fee, provided that
-# the above copyright notice appear in all copies and that both that
-# copyright notice and this permission notice appear in supporting
-# documentation, and that the name of M.I.T. not be used in advertising or
-# publicity pertaining to distribution of the software without specific,
-# written prior permission.  M.I.T. makes no representations about the
-# suitability of this software for any purpose.  It is provided "as is"
-# without express or implied warranty.
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to
+# deal in the Software without restriction, including without limitation the
+# rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+# sell copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+# X CONSORTIUM BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNEC-
+# TION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+#
+# Except as contained in this notice, the name of the X Consortium shall not
+# be used in advertising or otherwise to promote the sale, use or other deal-
+# ings in this Software without prior written authorization from the X Consor-
+# tium.
+#
+#
+# FSF changes to this file are in the public domain.
 #
 # Calling this script install-sh is preferred over install.sh, to prevent
 # `make' implicit rules from creating a file called install from it
 # when there is no Makefile.
 #
 # This script is compatible with the BSD install script, but was written
-# from scratch.  It can only install one file at a time, a restriction
-# shared with many OS's install programs.
+# from scratch.
 
+nl='
+'
+IFS=" ""	$nl"
 
 # set DOITPROG to echo to test this script
 
 # Don't use :- since 4.3BSD and earlier shells don't like it.
 doit="${DOITPROG-}"
+if test -z "$doit"; then
+  doit_exec=exec
+else
+  doit_exec=$doit
+fi
 
-
-# put in absolute paths if you don't have them in your path; or use env. vars.
+# Put in absolute file names if you don't have them in your path;
+# or use environment vars.
 
 mvprog="${MVPROG-mv}"
 cpprog="${CPPROG-cp}"
@@ -41,211 +67,441 @@ stripprog="${STRIPPROG-strip}"
 rmprog="${RMPROG-rm}"
 mkdirprog="${MKDIRPROG-mkdir}"
 
-transformbasename=""
-transform_arg=""
-instcmd="$mvprog"
-chmodcmd="$chmodprog 0755"
-chowncmd=""
-chgrpcmd=""
-stripcmd=""
+posix_glob=
+posix_mkdir=
+
+# Desired mode of installed file.
+mode=0755
+
+chmodcmd=$chmodprog
+chowncmd=
+chgrpcmd=
+stripcmd=
 rmcmd="$rmprog -f"
 mvcmd="$mvprog"
-src=""
-dst=""
-dir_arg=""
+src=
+dst=
+dir_arg=
+dstarg=
+no_target_directory=
 
-while [ x"$1" != x ]; do
-    case $1 in
-	-c) instcmd="$cpprog"
-	    shift
-	    continue;;
+usage="Usage: $0 [OPTION]... [-T] SRCFILE DSTFILE
+   or: $0 [OPTION]... SRCFILES... DIRECTORY
+   or: $0 [OPTION]... -t DIRECTORY SRCFILES...
+   or: $0 [OPTION]... -d DIRECTORIES...
 
-	-d) dir_arg=true
-	    shift
-	    continue;;
+In the 1st form, copy SRCFILE to DSTFILE.
+In the 2nd and 3rd, copy all SRCFILES to DIRECTORY.
+In the 4th, create DIRECTORIES.
 
-	-m) chmodcmd="$chmodprog $2"
-	    shift
-	    shift
-	    continue;;
+Options:
+-c         (ignored)
+-d         create directories instead of installing files.
+-g GROUP   $chgrpprog installed files to GROUP.
+-m MODE    $chmodprog installed files to MODE.
+-o USER    $chownprog installed files to USER.
+-s         $stripprog installed files.
+-t DIRECTORY  install into DIRECTORY.
+-T         report an error if DSTFILE is a directory.
+--help     display this help and exit.
+--version  display version info and exit.
 
-	-o) chowncmd="$chownprog $2"
-	    shift
-	    shift
-	    continue;;
+Environment variables override the default commands:
+  CHGRPPROG CHMODPROG CHOWNPROG CPPROG MKDIRPROG MVPROG RMPROG STRIPPROG
+"
 
-	-g) chgrpcmd="$chgrpprog $2"
-	    shift
-	    shift
-	    continue;;
+while test $# -ne 0; do
+  case $1 in
+    -c) shift
+        continue;;
 
-	-s) stripcmd="$stripprog"
-	    shift
-	    continue;;
+    -d) dir_arg=true
+        shift
+        continue;;
 
-	-t=*) transformarg=`echo $1 | sed 's/-t=//'`
-	    shift
-	    continue;;
+    -g) chgrpcmd="$chgrpprog $2"
+        shift
+        shift
+        continue;;
 
-	-b=*) transformbasename=`echo $1 | sed 's/-b=//'`
-	    shift
-	    continue;;
+    --help) echo "$usage"; exit $?;;
 
-	*)  if [ x"$src" = x ]
-	    then
-		src=$1
-	    else
-		# this colon is to work around a 386BSD /bin/sh bug
-		:
-		dst=$1
-	    fi
-	    shift
-	    continue;;
-    esac
-done
+    -m) mode=$2
+        shift
+        shift
+	case $mode in
+	  *' '* | *'	'* | *'
+'*	  | *'*'* | *'?'* | *'['*)
+	    echo "$0: invalid mode: $mode" >&2
+	    exit 1;;
+	esac
+        continue;;
 
-if [ x"$src" = x ]
-then
-	echo "install:	no input file specified"
-	exit 1
-else
-	true
-fi
+    -o) chowncmd="$chownprog $2"
+        shift
+        shift
+        continue;;
 
-if [ x"$dir_arg" != x ]; then
-	dst=$src
-	src=""
-	
-	if [ -d $dst ]; then
-		instcmd=:
-		chmodcmd=""
-	else
-		instcmd=mkdir
-	fi
-else
+    -s) stripcmd=$stripprog
+        shift
+        continue;;
 
-# Waiting for this to be detected by the "$instcmd $src $dsttmp" command
-# might cause directories to be created, which would be especially bad 
-# if $src (and thus $dsttmp) contains '*'.
-
-	if [ -f $src -o -d $src ]
-	then
-		true
-	else
-		echo "install:  $src does not exist"
-		exit 1
-	fi
-	
-	if [ x"$dst" = x ]
-	then
-		echo "install:	no destination specified"
-		exit 1
-	else
-		true
-	fi
-
-# If destination is a directory, append the input filename; if your system
-# does not like double slashes in filenames, you may need to add some logic
-
-	if [ -d $dst ]
-	then
-		dst="$dst"/`basename $src`
-	else
-		true
-	fi
-fi
-
-## this sed command emulates the dirname command
-dstdir=`echo $dst | sed -e 's,[^/]*$,,;s,/$,,;s,^$,.,'`
-
-# Make sure that the destination directory exists.
-#  this part is taken from Noah Friedman's mkinstalldirs script
-
-# Skip lots of stat calls in the usual case.
-if [ ! -d "$dstdir" ]; then
-defaultIFS='	
-'
-IFS="${IFS-${defaultIFS}}"
-
-oIFS="${IFS}"
-# Some sh's can't handle IFS=/ for some reason.
-IFS='%'
-set - `echo ${dstdir} | sed -e 's@/@%@g' -e 's@^%@/@'`
-IFS="${oIFS}"
-
-pathcomp=''
-
-while [ $# -ne 0 ] ; do
-	pathcomp="${pathcomp}${1}"
+    -t) dstarg=$2
 	shift
+	shift
+	continue;;
 
-	if [ ! -d "${pathcomp}" ] ;
-        then
-		$mkdirprog "${pathcomp}"
-	else
-		true
-	fi
+    -T) no_target_directory=true
+	shift
+	continue;;
 
-	pathcomp="${pathcomp}/"
+    --version) echo "$0 $scriptversion"; exit $?;;
+
+    --)	shift
+	break;;
+
+    -*)	echo "$0: invalid option: $1" >&2
+	exit 1;;
+
+    *)  break;;
+  esac
 done
+
+if test $# -ne 0 && test -z "$dir_arg$dstarg"; then
+  # When -d is used, all remaining arguments are directories to create.
+  # When -t is used, the destination is already specified.
+  # Otherwise, the last argument is the destination.  Remove it from $@.
+  for arg
+  do
+    if test -n "$dstarg"; then
+      # $@ is not empty: it contains at least $arg.
+      set fnord "$@" "$dstarg"
+      shift # fnord
+    fi
+    shift # arg
+    dstarg=$arg
+  done
 fi
 
-if [ x"$dir_arg" != x ]
-then
-	$doit $instcmd $dst &&
+if test $# -eq 0; then
+  if test -z "$dir_arg"; then
+    echo "$0: no input file specified." >&2
+    exit 1
+  fi
+  # It's OK to call `install-sh -d' without argument.
+  # This can happen when creating conditional directories.
+  exit 0
+fi
 
-	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dst; else true ; fi &&
-	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dst; else true ; fi &&
-	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dst; else true ; fi &&
-	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dst; else true ; fi
-else
+if test -z "$dir_arg"; then
+  trap '(exit $?); exit' 1 2 13 15
 
-# If we're going to rename the final executable, determine the name now.
+  # Set umask so as not to create temps with too-generous modes.
+  # However, 'strip' requires both read and write access to temps.
+  case $mode in
+    # Optimize common cases.
+    *644) cp_umask=133;;
+    *755) cp_umask=22;;
 
-	if [ x"$transformarg" = x ] 
-	then
-		dstfile=`basename $dst`
+    *[0-7])
+      if test -z "$stripcmd"; then
+	u_plus_rw=
+      else
+	u_plus_rw='% 200'
+      fi
+      cp_umask=`expr '(' 777 - $mode % 1000 ')' $u_plus_rw`;;
+    *)
+      if test -z "$stripcmd"; then
+	u_plus_rw=
+      else
+	u_plus_rw=,u+rw
+      fi
+      cp_umask=$mode$u_plus_rw;;
+  esac
+fi
+
+for src
+do
+  # Protect names starting with `-'.
+  case $src in
+    -*) src=./$src ;;
+  esac
+
+  if test -n "$dir_arg"; then
+    dst=$src
+    dstdir=$dst
+    test -d "$dstdir"
+    dstdir_status=$?
+  else
+
+    # Waiting for this to be detected by the "$cpprog $src $dsttmp" command
+    # might cause directories to be created, which would be especially bad
+    # if $src (and thus $dsttmp) contains '*'.
+    if test ! -f "$src" && test ! -d "$src"; then
+      echo "$0: $src does not exist." >&2
+      exit 1
+    fi
+
+    if test -z "$dstarg"; then
+      echo "$0: no destination specified." >&2
+      exit 1
+    fi
+
+    dst=$dstarg
+    # Protect names starting with `-'.
+    case $dst in
+      -*) dst=./$dst ;;
+    esac
+
+    # If destination is a directory, append the input filename; won't work
+    # if double slashes aren't ignored.
+    if test -d "$dst"; then
+      if test -n "$no_target_directory"; then
+	echo "$0: $dstarg: Is a directory" >&2
+	exit 1
+      fi
+      dstdir=$dst
+      dst=$dstdir/`basename "$src"`
+      dstdir_status=0
+    else
+      # Prefer dirname, but fall back on a substitute if dirname fails.
+      dstdir=`
+	(dirname "$dst") 2>/dev/null ||
+	expr X"$dst" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	     X"$dst" : 'X\(//\)[^/]' \| \
+	     X"$dst" : 'X\(//\)$' \| \
+	     X"$dst" : 'X\(/\)' \| . 2>/dev/null ||
+	echo X"$dst" |
+	    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+		   s//\1/
+		   q
+		 }
+		 /^X\(\/\/\)[^/].*/{
+		   s//\1/
+		   q
+		 }
+		 /^X\(\/\/\)$/{
+		   s//\1/
+		   q
+		 }
+		 /^X\(\/\).*/{
+		   s//\1/
+		   q
+		 }
+		 s/.*/./; q'
+      `
+
+      test -d "$dstdir"
+      dstdir_status=$?
+    fi
+  fi
+
+  obsolete_mkdir_used=false
+
+  if test $dstdir_status != 0; then
+    case $posix_mkdir in
+      '')
+	# Create intermediate dirs using mode 755 as modified by the umask.
+	# This is like FreeBSD 'install' as of 1997-10-28.
+	umask=`umask`
+	case $stripcmd.$umask in
+	  # Optimize common cases.
+	  *[2367][2367]) mkdir_umask=$umask;;
+	  .*0[02][02] | .[02][02] | .[02]) mkdir_umask=22;;
+
+	  *[0-7])
+	    mkdir_umask=`expr $umask + 22 \
+	      - $umask % 100 % 40 + $umask % 20 \
+	      - $umask % 10 % 4 + $umask % 2
+	    `;;
+	  *) mkdir_umask=$umask,go-w;;
+	esac
+
+	# With -d, create the new directory with the user-specified mode.
+	# Otherwise, rely on $mkdir_umask.
+	if test -n "$dir_arg"; then
+	  mkdir_mode=-m$mode
 	else
-		dstfile=`basename $dst $transformbasename | 
-			sed $transformarg`$transformbasename
+	  mkdir_mode=
 	fi
 
-# don't allow the sed command to completely eliminate the filename
+	posix_mkdir=false
+	case $umask in
+	  *[123567][0-7][0-7])
+	    # POSIX mkdir -p sets u+wx bits regardless of umask, which
+	    # is incompatible with FreeBSD 'install' when (umask & 300) != 0.
+	    ;;
+	  *)
+	    tmpdir=${TMPDIR-/tmp}/ins$RANDOM-$$
+	    trap 'ret=$?; rmdir "$tmpdir/d" "$tmpdir" 2>/dev/null; exit $ret' 0
 
-	if [ x"$dstfile" = x ] 
-	then
-		dstfile=`basename $dst`
+	    if (umask $mkdir_umask &&
+		exec $mkdirprog $mkdir_mode -p -- "$tmpdir/d") >/dev/null 2>&1
+	    then
+	      if test -z "$dir_arg" || {
+		   # Check for POSIX incompatibilities with -m.
+		   # HP-UX 11.23 and IRIX 6.5 mkdir -m -p sets group- or
+		   # other-writeable bit of parent directory when it shouldn't.
+		   # FreeBSD 6.1 mkdir -m -p sets mode of existing directory.
+		   ls_ld_tmpdir=`ls -ld "$tmpdir"`
+		   case $ls_ld_tmpdir in
+		     d????-?r-*) different_mode=700;;
+		     d????-?--*) different_mode=755;;
+		     *) false;;
+		   esac &&
+		   $mkdirprog -m$different_mode -p -- "$tmpdir" && {
+		     ls_ld_tmpdir_1=`ls -ld "$tmpdir"`
+		     test "$ls_ld_tmpdir" = "$ls_ld_tmpdir_1"
+		   }
+		 }
+	      then posix_mkdir=:
+	      fi
+	      rmdir "$tmpdir/d" "$tmpdir"
+	    else
+	      # Remove any dirs left behind by ancient mkdir implementations.
+	      rmdir ./$mkdir_mode ./-p ./-- 2>/dev/null
+	    fi
+	    trap '' 0;;
+	esac;;
+    esac
+
+    if
+      $posix_mkdir && (
+	umask $mkdir_umask &&
+	$doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir"
+      )
+    then :
+    else
+
+      # The umask is ridiculous, or mkdir does not conform to POSIX,
+      # or it failed possibly due to a race condition.  Create the
+      # directory the slow way, step by step, checking for races as we go.
+
+      case $dstdir in
+	/*) prefix=/ ;;
+	-*) prefix=./ ;;
+	*)  prefix= ;;
+      esac
+
+      case $posix_glob in
+        '')
+	  if (set -f) 2>/dev/null; then
+	    posix_glob=true
+	  else
+	    posix_glob=false
+	  fi ;;
+      esac
+
+      oIFS=$IFS
+      IFS=/
+      $posix_glob && set -f
+      set fnord $dstdir
+      shift
+      $posix_glob && set +f
+      IFS=$oIFS
+
+      prefixes=
+
+      for d
+      do
+	test -z "$d" && continue
+
+	prefix=$prefix$d
+	if test -d "$prefix"; then
+	  prefixes=
 	else
-		true
+	  if $posix_mkdir; then
+	    (umask=$mkdir_umask &&
+	     $doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir") && break
+	    # Don't fail if two instances are running concurrently.
+	    test -d "$prefix" || exit 1
+	  else
+	    case $prefix in
+	      *\'*) qprefix=`echo "$prefix" | sed "s/'/'\\\\\\\\''/g"`;;
+	      *) qprefix=$prefix;;
+	    esac
+	    prefixes="$prefixes '$qprefix'"
+	  fi
 	fi
+	prefix=$prefix/
+      done
 
-# Make a temp file name in the proper directory.
+      if test -n "$prefixes"; then
+	# Don't fail if two instances are running concurrently.
+	(umask $mkdir_umask &&
+	 eval "\$doit_exec \$mkdirprog $prefixes") ||
+	  test -d "$dstdir" || exit 1
+	obsolete_mkdir_used=true
+      fi
+    fi
+  fi
 
-	dsttmp=$dstdir/#inst.$$#
+  if test -n "$dir_arg"; then
+    { test -z "$chowncmd" || $doit $chowncmd "$dst"; } &&
+    { test -z "$chgrpcmd" || $doit $chgrpcmd "$dst"; } &&
+    { test "$obsolete_mkdir_used$chowncmd$chgrpcmd" = false ||
+      test -z "$chmodcmd" || $doit $chmodcmd $mode "$dst"; } || exit 1
+  else
 
-# Move or copy the file name to the temp name
+    # Make a couple of temp file names in the proper directory.
+    dsttmp=$dstdir/_inst.$$_
+    rmtmp=$dstdir/_rm.$$_
 
-	$doit $instcmd $src $dsttmp &&
+    # Trap to clean up those temp files at exit.
+    trap 'ret=$?; rm -f "$dsttmp" "$rmtmp" && exit $ret' 0
 
-	trap "rm -f ${dsttmp}" 0 &&
+    # Copy the file name to the temp name.
+    (umask $cp_umask && $doit_exec $cpprog "$src" "$dsttmp") &&
 
-# and set any options; do chmod last to preserve setuid bits
+    # and set any options; do chmod last to preserve setuid bits.
+    #
+    # If any of these fail, we abort the whole thing.  If we want to
+    # ignore errors from any of these, just make sure not to ignore
+    # errors from the above "$doit $cpprog $src $dsttmp" command.
+    #
+    { test -z "$chowncmd" || $doit $chowncmd "$dsttmp"; } \
+      && { test -z "$chgrpcmd" || $doit $chgrpcmd "$dsttmp"; } \
+      && { test -z "$stripcmd" || $doit $stripcmd "$dsttmp"; } \
+      && { test -z "$chmodcmd" || $doit $chmodcmd $mode "$dsttmp"; } &&
 
-# If any of these fail, we abort the whole thing.  If we want to
-# ignore errors from any of these, just make sure not to ignore
-# errors from the above "$doit $instcmd $src $dsttmp" command.
+    # Now rename the file to the real destination.
+    { $doit $mvcmd -f "$dsttmp" "$dst" 2>/dev/null \
+      || {
+	   # The rename failed, perhaps because mv can't rename something else
+	   # to itself, or perhaps because mv is so ancient that it does not
+	   # support -f.
 
-	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dsttmp; else true;fi &&
-	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dsttmp; else true;fi &&
-	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dsttmp; else true;fi &&
-	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dsttmp; else true;fi &&
+	   # Now remove or move aside any old file at destination location.
+	   # We try this two ways since rm can't unlink itself on some
+	   # systems and the destination file might be busy for other
+	   # reasons.  In this case, the final cleanup might fail but the new
+	   # file should still install successfully.
+	   {
+	     if test -f "$dst"; then
+	       $doit $rmcmd -f "$dst" 2>/dev/null \
+	       || { $doit $mvcmd -f "$dst" "$rmtmp" 2>/dev/null \
+		     && { $doit $rmcmd -f "$rmtmp" 2>/dev/null; :; }; }\
+	       || {
+		 echo "$0: cannot unlink or rename $dst" >&2
+		 (exit 1); exit 1
+	       }
+	     else
+	       :
+	     fi
+	   } &&
 
-# Now rename the file to the real destination.
+	   # Now rename the file to the real destination.
+	   $doit $mvcmd "$dsttmp" "$dst"
+	 }
+    } || exit 1
 
-	$doit $rmcmd -f $dstdir/$dstfile &&
-	$doit $mvcmd $dsttmp $dstdir/$dstfile 
+    trap '' 0
+  fi
+done
 
-fi &&
-
-
-exit 0
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "scriptversion="
+# time-stamp-format: "%:y-%02m-%02d.%02H"
+# time-stamp-end: "$"
+# End:

daemons/Makefile.in

@@ -0,0 +1,37 @@
+#
+# Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+.PHONY: dmeventd clvmd cmirrord
+
+ifneq ("@CLVMD@", "none")
+  SUBDIRS = clvmd
+endif
+
+ifeq ("@BUILD_CMIRRORD@", "yes")
+  SUBDIRS += cmirrord
+endif
+
+ifeq ("@BUILD_DMEVENTD@", "yes")
+  SUBDIRS += dmeventd
+endif
+
+include ../make.tmpl
+
+ifeq ("@BUILD_DMEVENTD@", "yes")
+device-mapper: dmeventd.device-mapper
+endif

daemons/clvmd/Makefile.in

@@ -0,0 +1,107 @@
+#
+# Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+CCS_LIBS = @CCS_LIBS@
+CCS_CFLAGS = @CCS_CFLAGS@
+CMAN_LIBS = @CMAN_LIBS@
+CMAN_CFLAGS = @CMAN_CFLAGS@
+CONFDB_LIBS = @CONFDB_LIBS@
+CONFDB_CFLAGS = @CONFDB_CFLAGS@
+CPG_LIBS = @CPG_LIBS@
+CPG_CFLAGS = @CPG_CFLAGS@
+DLM_LIBS = @DLM_LIBS@
+DLM_CFLAGS = @DLM_CFLAGS@
+GULM_LIBS = @GULM_LIBS@
+GULM_CFLAGS = @GULM_CFLAGS@
+QUORUM_LIBS = @QUORUM_LIBS@
+QUORUM_CFLAGS = @QUORUM_CFLAGS@
+SALCK_LIBS = @SALCK_LIBS@
+SALCK_CFLAGS = @SALCK_CFLAGS@
+
+SOURCES = \
+	clvmd-command.c  \
+	clvmd.c          \
+	lvm-functions.c  \
+	refresh_clvmd.c
+
+ifeq ("@DEBUG@", "yes")
+	DEFS += -DDEBUG
+endif
+
+ifneq (,$(findstring gulm,, "@CLVMD@,"))
+	SOURCES += clvmd-gulm.c tcp-comms.c
+	LMLIBS += $(CCS_LIBS) $(GULM_LIBS)
+	CFLAGS += $(CCS_CFLAGS) $(GULM_CFLAGS)
+	DEFS += -DUSE_GULM
+endif
+
+ifneq (,$(findstring cman,, "@CLVMD@,"))
+	SOURCES += clvmd-cman.c
+	LMLIBS += $(CMAN_LIBS) $(CONFDB_LIBS) $(DLM_LIBS)
+	CFLAGS += $(CMAN_CFLAGS) $(CONFDB_CFLAGS) $(DLM_CFLAGS)
+	DEFS += -DUSE_CMAN
+endif
+
+ifneq (,$(findstring openais,, "@CLVMD@,"))
+	SOURCES += clvmd-openais.c
+	LMLIBS += $(CONFDB_LIBS) $(CPG_LIBS) $(SALCK_LIBS)
+	CFLAGS += $(CONFDB_CFLAGS) $(CPG_CFLAGS) $(SALCK_CFLAGS)
+	DEFS += -DUSE_OPENAIS
+endif
+
+ifneq (,$(findstring corosync,, "@CLVMD@,"))
+	SOURCES += clvmd-corosync.c
+	LMLIBS += $(CONFDB_LIBS) $(CPG_LIBS) $(DLM_LIBS) $(QUORUM_LIBS)
+	CFLAGS += $(CONFDB_CFLAGS) $(CPG_CFLAGS) $(DLM_CFLAGS) $(QUORUM_CFLAGS)
+	DEFS += -DUSE_COROSYNC
+endif
+
+
+TARGETS = \
+	clvmd
+
+LVMLIBS = -llvm-internal -lpthread
+
+ifeq ("@DMEVENTD@", "yes")
+	LVMLIBS += -ldevmapper-event
+endif
+ 
+LVMLIBS += -ldevmapper
+
+DEFS += -D_REENTRANT
+CFLAGS += -fno-strict-aliasing
+
+include ../../make.tmpl
+
+INSTALL_TARGETS = \
+	install_clvmd
+
+clvmd: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
+	$(CC) $(CFLAGS) $(LDFLAGS) -o clvmd $(OBJECTS) \
+		$(LVMLIBS) $(LMLIBS) $(LIBS)
+
+.PHONY: install_clvmd
+
+install_clvmd: $(TARGETS)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) clvmd \
+		$(usrsbindir)/clvmd
+
+install: $(INSTALL_TARGETS)
+
+install_cluster: $(INSTALL_TARGETS)
+

daemons/clvmd/clvm.h

@@ -0,0 +1,72 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/* Definitions for CLVMD server and clients */
+
+/*
+ * The protocol spoken over the cluster and across the local socket.
+ */
+
+#ifndef _CLVM_H
+#define _CLVM_H
+
+struct clvm_header {
+	uint8_t  cmd;	        /* See below */
+	uint8_t  flags;	        /* See below */
+	uint16_t xid;	        /* Transaction ID */
+	uint32_t clientid;	/* Only used in Daemon->Daemon comms */
+	int32_t  status;	/* For replies, whether request succeeded */
+	uint32_t arglen;	/* Length of argument below. 
+				   If >1500 then it will be passed 
+				   around the cluster in the system LV */
+	char node[1];		/* Actually a NUL-terminated string, node name.
+				   If this is empty then the command is 
+				   forwarded to all cluster nodes unless 
+				   FLAG_LOCAL is also set. */
+	char args[1];		/* Arguments for the command follow the 
+				   node name, This member is only
+				   valid if the node name is empty */
+} __attribute__ ((packed));
+
+/* Flags */
+#define CLVMD_FLAG_LOCAL        1	/* Only do this on the local node */
+#define CLVMD_FLAG_SYSTEMLV     2	/* Data in system LV under my node name */
+#define CLVMD_FLAG_NODEERRS     4       /* Reply has errors in node-specific portion */
+
+/* Name of the local socket to communicate between libclvm and clvmd */
+//static const char CLVMD_SOCKNAME[]="/var/run/clvmd";
+static const char CLVMD_SOCKNAME[] = "\0clvmd";
+
+/* Internal commands & replies */
+#define CLVMD_CMD_REPLY    1
+#define CLVMD_CMD_VERSION  2	/* Send version around cluster when we start */
+#define CLVMD_CMD_GOAWAY   3	/* Die if received this - we are running 
+				   an incompatible version */
+#define CLVMD_CMD_TEST     4	/* Just for mucking about */
+
+#define CLVMD_CMD_LOCK              30
+#define CLVMD_CMD_UNLOCK            31
+
+/* Lock/Unlock commands */
+#define CLVMD_CMD_LOCK_LV           50
+#define CLVMD_CMD_LOCK_VG           51
+#define CLVMD_CMD_LOCK_QUERY	    52
+
+/* Misc functions */
+#define CLVMD_CMD_REFRESH	    40
+#define CLVMD_CMD_GET_CLUSTERNAME   41
+#define CLVMD_CMD_SET_DEBUG	    42
+#define CLVMD_CMD_VG_BACKUP	    43
+#endif

daemons/clvmd/clvmd-cman.c

@@ -0,0 +1,523 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * CMAN communication layer for clvmd.
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/socket.h>
+#include <sys/uio.h>
+#include <sys/un.h>
+#include <sys/time.h>
+#include <sys/ioctl.h>
+#include <sys/utsname.h>
+#include <syslog.h>
+#include <netinet/in.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stddef.h>
+#include <signal.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <getopt.h>
+#include <errno.h>
+#include <libdevmapper.h>
+#include <libdlm.h>
+
+#include "clvmd-comms.h"
+#include "clvm.h"
+#include "lvm-logging.h"
+#include "clvmd.h"
+#include "lvm-functions.h"
+
+#define LOCKSPACE_NAME "clvmd"
+
+struct clvmd_node
+{
+	struct cman_node *node;
+	int clvmd_up;
+};
+
+static int num_nodes;
+static struct cman_node *nodes = NULL;
+static struct cman_node this_node;
+static int count_nodes; /* size of allocated nodes array */
+static struct dm_hash_table *node_updown_hash;
+static dlm_lshandle_t *lockspace;
+static cman_handle_t c_handle;
+
+static void count_clvmds_running(void);
+static void get_members(void);
+static int nodeid_from_csid(const char *csid);
+static int name_from_nodeid(int nodeid, char *name);
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg);
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid);
+
+struct lock_wait {
+	pthread_cond_t cond;
+	pthread_mutex_t mutex;
+	struct dlm_lksb lksb;
+};
+
+static int _init_cluster(void)
+{
+	node_updown_hash = dm_hash_create(100);
+
+	/* Open the cluster communication socket */
+	c_handle = cman_init(NULL);
+	if (!c_handle) {
+		syslog(LOG_ERR, "Can't open cluster manager socket: %m");
+		return -1;
+	}
+	DEBUGLOG("Connected to CMAN\n");
+
+	if (cman_start_recv_data(c_handle, data_callback, CLUSTER_PORT_CLVMD)) {
+		syslog(LOG_ERR, "Can't bind cluster socket: %m");
+		return -1;
+	}
+
+	if (cman_start_notification(c_handle, event_callback)) {
+		syslog(LOG_ERR, "Can't start cluster event listening");
+		return -1;
+	}
+
+	/* Get the cluster members list */
+	get_members();
+	count_clvmds_running();
+
+	DEBUGLOG("CMAN initialisation complete\n");
+
+	/* Create a lockspace for LV & VG locks to live in */
+	lockspace = dlm_create_lockspace(LOCKSPACE_NAME, 0600);
+	if (!lockspace) {
+		if (errno == EEXIST) {
+			lockspace = dlm_open_lockspace(LOCKSPACE_NAME);
+		}
+		if (!lockspace) {
+			syslog(LOG_ERR, "Unable to create lockspace for CLVM: %m");
+			return -1;
+		}
+	}
+	dlm_ls_pthread_init(lockspace);
+	DEBUGLOG("DLM initialisation complete\n");
+	return 0;
+}
+
+static void _cluster_init_completed(void)
+{
+	clvmd_cluster_init_completed();
+}
+
+static int _get_main_cluster_fd()
+{
+	return cman_get_fd(c_handle);
+}
+
+static int _get_num_nodes()
+{
+	int i;
+	int nnodes = 0;
+
+	/* return number of ACTIVE nodes */
+	for (i=0; i<num_nodes; i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid)
+			nnodes++;
+	}
+	return nnodes;
+}
+
+/* send_message with the fd check removed */
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	int nodeid = 0;
+
+	if (csid)
+		memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
+
+	if (cman_send_data(c_handle, buf, msglen, 0, CLUSTER_PORT_CLVMD, nodeid) <= 0)
+	{
+		log_error("%s", errtext);
+	}
+	return msglen;
+}
+
+static void _get_our_csid(char *csid)
+{
+	if (this_node.cn_nodeid == 0) {
+		cman_get_node(c_handle, 0, &this_node);
+	}
+	memcpy(csid, &this_node.cn_nodeid, CMAN_MAX_CSID_LEN);
+}
+
+/* Call a callback routine for each node is that known (down means not running a clvmd) */
+static int _cluster_do_node_callback(struct local_client *client,
+				     void (*callback) (struct local_client *,
+						       const char *,
+						       int))
+{
+	int i;
+	int somedown = 0;
+
+	for (i = 0; i < _get_num_nodes(); i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid) {
+			int up = (int)(long)dm_hash_lookup_binary(node_updown_hash, (char *)&nodes[i].cn_nodeid, sizeof(int));
+
+			callback(client, (char *)&nodes[i].cn_nodeid, up);
+			if (!up)
+				somedown = -1;
+		}
+	}
+	return somedown;
+}
+
+/* Process OOB messages from the cluster socket */
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg)
+{
+	char namebuf[MAX_CLUSTER_MEMBER_NAME_LEN];
+
+	switch (reason) {
+        case CMAN_REASON_PORTCLOSED:
+		name_from_nodeid(arg, namebuf);
+		log_notice("clvmd on node %s has died\n", namebuf);
+		DEBUGLOG("Got port closed message, removing node %s\n", namebuf);
+
+		dm_hash_insert_binary(node_updown_hash, (char *)&arg, sizeof(int), (void *)0);
+		break;
+
+	case CMAN_REASON_STATECHANGE:
+		DEBUGLOG("Got state change message, re-reading members list\n");
+		get_members();
+		break;
+
+#if defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2
+	case CMAN_REASON_PORTOPENED:
+		/* Ignore this, wait for startup message from clvmd itself */
+		break;
+
+	case CMAN_REASON_TRY_SHUTDOWN:
+		DEBUGLOG("Got try shutdown, sending OK\n");
+		cman_replyto_shutdown(c_handle, 1);
+		break;
+#endif
+	default:
+		/* ERROR */
+		DEBUGLOG("Got unknown event callback message: %d\n", reason);
+		break;
+	}
+}
+
+static struct local_client *cman_client;
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+
+	/* Save this for data_callback */
+	cman_client = fd;
+
+	/* We never return a new client */
+	*new_client = NULL;
+
+	return cman_dispatch(c_handle, 0);
+}
+
+
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid)
+{
+	/* Ignore looped back messages */
+	if (nodeid == this_node.cn_nodeid)
+		return;
+	process_message(cman_client, buf, len, (char *)&nodeid);
+}
+
+static void _add_up_node(const char *csid)
+{
+	/* It's up ! */
+	int nodeid = nodeid_from_csid(csid);
+
+	dm_hash_insert_binary(node_updown_hash, (char *)&nodeid, sizeof(int), (void *)1);
+	DEBUGLOG("Added new node %d to updown list\n", nodeid);
+}
+
+static void _cluster_closedown()
+{
+	destroy_lvhash();
+	dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+	cman_finish(c_handle);
+}
+
+static int is_listening(int nodeid)
+{
+	int status;
+
+	do {
+		status = cman_is_listening(c_handle, nodeid, CLUSTER_PORT_CLVMD);
+		if (status < 0 && errno == EBUSY) {	/* Don't busywait */
+			sleep(1);
+			errno = EBUSY;	/* In case sleep trashes it */
+		}
+	}
+	while (status < 0 && errno == EBUSY);
+
+	return status;
+}
+
+/* Populate the list of CLVMDs running.
+   called only at startup time */
+static void count_clvmds_running(void)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		int nodeid = nodes[i].cn_nodeid;
+
+		if (is_listening(nodeid) == 1)
+			dm_hash_insert_binary(node_updown_hash, (void *)&nodeid, sizeof(int), (void*)1);
+		else
+			dm_hash_insert_binary(node_updown_hash, (void *)&nodeid, sizeof(int), (void*)0);
+	}
+}
+
+/* Get a list of active cluster members */
+static void get_members()
+{
+	int retnodes;
+	int status;
+	int i;
+	int high_nodeid = 0;
+
+	num_nodes = cman_get_node_count(c_handle);
+	if (num_nodes == -1) {
+		log_error("Unable to get node count");
+		return;
+	}
+
+	/* Not enough room for new nodes list ? */
+	if (num_nodes > count_nodes && nodes) {
+		free(nodes);
+		nodes = NULL;
+	}
+
+	if (nodes == NULL) {
+		count_nodes = num_nodes + 10; /* Overallocate a little */
+		nodes = malloc(count_nodes * sizeof(struct cman_node));
+		if (!nodes) {
+			log_error("Unable to allocate nodes array\n");
+			exit(5);
+		}
+	}
+
+	status = cman_get_nodes(c_handle, count_nodes, &retnodes, nodes);
+	if (status < 0) {
+		log_error("Unable to get node details");
+		exit(6);
+	}
+
+	/* Get the highest nodeid */
+	for (i=0; i<retnodes; i++) {
+		if (nodes[i].cn_nodeid > high_nodeid)
+			high_nodeid = nodes[i].cn_nodeid;
+	}
+}
+
+
+/* Convert a node name to a CSID */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (strcmp(name, nodes[i].cn_name) == 0) {
+			memcpy(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN);
+			return 0;
+		}
+	}
+	return -1;
+}
+
+/* Convert a CSID to a node name */
+static int _name_from_csid(const char *csid, char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (memcmp(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN) == 0) {
+			strcpy(name, nodes[i].cn_name);
+			return 0;
+		}
+	}
+	/* Who?? */
+	strcpy(name, "Unknown");
+	return -1;
+}
+
+/* Convert a node ID to a node name */
+static int name_from_nodeid(int nodeid, char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (nodeid == nodes[i].cn_nodeid) {
+			strcpy(name, nodes[i].cn_name);
+			return 0;
+		}
+	}
+	/* Who?? */
+	strcpy(name, "Unknown");
+	return -1;
+}
+
+/* Convert a CSID to a node ID */
+static int nodeid_from_csid(const char *csid)
+{
+        int nodeid;
+
+	memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
+
+	return nodeid;
+}
+
+static int _is_quorate()
+{
+	return cman_is_quorate(c_handle);
+}
+
+static void sync_ast_routine(void *arg)
+{
+	struct lock_wait *lwait = arg;
+
+	pthread_mutex_lock(&lwait->mutex);
+	pthread_cond_signal(&lwait->cond);
+	pthread_mutex_unlock(&lwait->mutex);
+}
+
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
+{
+	int status;
+	struct lock_wait lwait;
+
+	if (!lockid) {
+		errno = EINVAL;
+		return -1;
+	}
+
+	DEBUGLOG("sync_lock: '%s' mode:%d flags=%d\n", resource,mode,flags);
+	/* Conversions need the lockid in the LKSB */
+	if (flags & LKF_CONVERT)
+		lwait.lksb.sb_lkid = *lockid;
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	status = dlm_ls_lock(lockspace,
+			     mode,
+			     &lwait.lksb,
+			     flags,
+			     resource,
+			     strlen(resource),
+			     0, sync_ast_routine, &lwait, NULL, NULL);
+	if (status)
+		return status;
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	*lockid = lwait.lksb.sb_lkid;
+
+	errno = lwait.lksb.sb_status;
+	DEBUGLOG("sync_lock: returning lkid %x\n", *lockid);
+	if (lwait.lksb.sb_status)
+		return -1;
+	else
+		return 0;
+}
+
+static int _sync_unlock(const char *resource /* UNUSED */, int lockid)
+{
+	int status;
+	struct lock_wait lwait;
+
+	DEBUGLOG("sync_unlock: '%s' lkid:%x\n", resource, lockid);
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	status = dlm_ls_unlock(lockspace, lockid, 0, &lwait.lksb, &lwait);
+
+	if (status)
+		return status;
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	errno = lwait.lksb.sb_status;
+	if (lwait.lksb.sb_status != EUNLOCK)
+		return -1;
+	else
+		return 0;
+
+}
+
+static int _get_cluster_name(char *buf, int buflen)
+{
+	cman_cluster_t cluster_info;
+	int status;
+
+	status = cman_get_cluster(c_handle, &cluster_info);
+	if (!status) {
+		strncpy(buf, cluster_info.ci_name, buflen);
+	}
+	return status;
+}
+
+static struct cluster_ops _cluster_cman_ops = {
+	.cluster_init_completed   = _cluster_init_completed,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_cman_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_cman_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-command.c

@@ -0,0 +1,353 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+
+  CLVMD Cluster LVM daemon command processor.
+
+  To add commands to the daemon simply add a processor in do_command and return
+  and messages back in buf and the length in *retlen. The initial value of
+  buflen is the maximum size of the buffer. if buf is not large enough then it
+  may be reallocated by the functions in here to a suitable size bearing in
+  mind that anything larger than the passed-in size will have to be returned
+  using the system LV and so performance will suffer.
+
+  The status return will be negated and passed back to the originating node.
+
+  pre- and post- command routines are called only on the local node. The
+  purpose is primarily to get and release locks, though the pre- routine should
+  also do any other local setups required by the command (if any) and can
+  return a failure code that prevents the command from being distributed around
+  the cluster
+
+  The pre- and post- routines are run in their own thread so can block as long
+  they like, do_command is run in the main clvmd thread so should not block for
+  too long. If the pre-command returns an error code (!=0) then the command
+  will not be propogated around the cluster but the post-command WILL be called
+
+  Also note that the pre and post routine are *always* called on the local
+  node, even if the command to be executed was only requested to run on a
+  remote node. It may peek inside the client structure to check the status of
+  the command.
+
+  The clients of the daemon must, naturally, understand the return messages and
+  codes.
+
+  Routines in here may only READ the values in the client structure passed in
+  apart from client->private which they are free to do what they like with.
+
+*/
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <unistd.h>
+#include <errno.h>
+#include <libdevmapper.h>
+#include <libdlm.h>
+
+#include "locking.h"
+#include "lvm-logging.h"
+#include "lvm-functions.h"
+#include "clvmd-comms.h"
+#include "clvm.h"
+#include "clvmd.h"
+
+extern debug_t debug;
+extern struct cluster_ops *clops;
+
+/* This is where all the real work happens:
+   NOTE: client will be NULL when this is executed on a remote node */
+int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
+	       char **buf, int buflen, int *retlen)
+{
+	char *args = msg->node + strlen(msg->node) + 1;
+	int arglen = msglen - sizeof(struct clvm_header) - strlen(msg->node);
+	int status = 0;
+	char *lockname;
+	const char *locktype;
+	struct utsname nodeinfo;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+
+	/* Do the command */
+	switch (msg->cmd) {
+		/* Just a test message */
+	case CLVMD_CMD_TEST:
+		if (arglen > buflen) {
+			char *new_buf;
+			buflen = arglen + 200;
+			new_buf = realloc(*buf, buflen);
+			if (new_buf == NULL) {
+				status = errno;
+				free (*buf);
+			}
+			*buf = new_buf;
+		}
+		if (*buf) {
+			uname(&nodeinfo);
+			*retlen = 1 + snprintf(*buf, buflen,
+					       "TEST from %s: %s v%s",
+					       nodeinfo.nodename, args,
+					       nodeinfo.release);
+		}
+		break;
+
+	case CLVMD_CMD_LOCK_VG:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		/* Check to see if the VG is in use by LVM1 */
+		status = do_check_lvm1(lockname);
+		do_lock_vg(lock_cmd, lock_flags, lockname);
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		/* This is the biggie */
+		lock_cmd = args[0] & (LCK_NONBLOCK | LCK_HOLD | LCK_SCOPE_MASK | LCK_TYPE_MASK);
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = do_lock_lv(lock_cmd, lock_flags, lockname);
+		/* Replace EIO with something less scary */
+		if (status == EIO) {
+			*retlen =
+			    1 + snprintf(*buf, buflen, "%s",
+					 get_last_lvm_error());
+			return EIO;
+		}
+		break;
+
+	case CLVMD_CMD_LOCK_QUERY:
+		lockname = &args[2];
+		if (buflen < 3)
+			return EIO;
+		if ((locktype = do_lock_query(lockname)))
+			*retlen = 1 + snprintf(*buf, buflen, "%s", locktype);
+		break;
+
+	case CLVMD_CMD_REFRESH:
+		do_refresh_cache();
+		break;
+
+	case CLVMD_CMD_SET_DEBUG:
+		debug = args[0];
+		break;
+
+	case CLVMD_CMD_GET_CLUSTERNAME:
+		status = clops->get_cluster_name(*buf, buflen);
+		if (!status)
+			*retlen = strlen(*buf)+1;
+		break;
+
+	case CLVMD_CMD_VG_BACKUP:
+		/*
+		 * Do not run backup on local node, caller should do that.
+		 */
+		if (!client)
+			lvm_do_backup(&args[2]);
+		break;
+
+	default:
+		/* Won't get here because command is validated in pre_command */
+		break;
+	}
+
+	/* Check the status of the command and return the error text */
+	if (status) {
+		*retlen = 1 + snprintf(*buf, buflen, "%s", strerror(status));
+	}
+
+	return status;
+
+}
+
+static int lock_vg(struct local_client *client)
+{
+    struct dm_hash_table *lock_hash;
+    struct clvm_header *header =
+	(struct clvm_header *) client->bits.localsock.cmd;
+    unsigned char lock_cmd;
+    unsigned char lock_flags;
+    int lock_mode;
+    char *args = header->node + strlen(header->node) + 1;
+    int lkid;
+    int status = 0;
+    char *lockname;
+
+    /* Keep a track of VG locks in our own hash table. In current
+       practice there should only ever be more than two VGs locked
+       if a user tries to merge lots of them at once */
+    if (client->bits.localsock.private) {
+	lock_hash = (struct dm_hash_table *)client->bits.localsock.private;
+    }
+    else {
+	lock_hash = dm_hash_create(3);
+	if (!lock_hash)
+	    return ENOMEM;
+	client->bits.localsock.private = (void *)lock_hash;
+    }
+
+    lock_cmd = args[0] & (LCK_NONBLOCK | LCK_HOLD | LCK_SCOPE_MASK | LCK_TYPE_MASK);
+    lock_mode = ((int)lock_cmd & LCK_TYPE_MASK);
+    lock_flags = args[1];
+    lockname = &args[2];
+    DEBUGLOG("doing PRE command LOCK_VG '%s' at %x (client=%p)\n", lockname, lock_cmd, client);
+
+    if (lock_mode == LCK_UNLOCK) {
+
+	lkid = (int)(long)dm_hash_lookup(lock_hash, lockname);
+	if (lkid == 0)
+	    return EINVAL;
+
+	status = sync_unlock(lockname, lkid);
+	if (status)
+	    status = errno;
+	else
+	    dm_hash_remove(lock_hash, lockname);
+    }
+    else {
+	/* Read locks need to be PR; other modes get passed through */
+	if (lock_mode == LCK_READ)
+	    lock_mode = LCK_PREAD;
+	status = sync_lock(lockname, lock_mode, (lock_cmd & LCK_NONBLOCK) ? LKF_NOQUEUE : 0, &lkid);
+	if (status)
+	    status = errno;
+	else
+	    dm_hash_insert(lock_hash, lockname, (void *)(long)lkid);
+    }
+
+    return status;
+}
+
+
+/* Pre-command is a good place to get locks that are needed only for the duration
+   of the commands around the cluster (don't forget to free them in post-command),
+   and to sanity check the command arguments */
+int do_pre_command(struct local_client *client)
+{
+	struct clvm_header *header =
+	    (struct clvm_header *) client->bits.localsock.cmd;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+	char *args = header->node + strlen(header->node) + 1;
+	int lockid;
+	int status = 0;
+	char *lockname;
+
+	switch (header->cmd) {
+	case CLVMD_CMD_TEST:
+		status = sync_lock("CLVMD_TEST", LKM_EXMODE, 0, &lockid);
+		client->bits.localsock.private = (void *)(long)lockid;
+		break;
+
+	case CLVMD_CMD_LOCK_VG:
+		lockname = &args[2];
+		/* We take out a real lock unless LCK_CACHE was set */
+		if (!strncmp(lockname, "V_", 2) ||
+		    !strncmp(lockname, "P_#", 3))
+			status = lock_vg(client);
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = pre_lock_lv(lock_cmd, lock_flags, lockname);
+		break;
+
+	case CLVMD_CMD_REFRESH:
+	case CLVMD_CMD_GET_CLUSTERNAME:
+	case CLVMD_CMD_SET_DEBUG:
+	case CLVMD_CMD_VG_BACKUP:
+	case CLVMD_CMD_LOCK_QUERY:
+		break;
+
+	default:
+		log_error("Unknown command %d received\n", header->cmd);
+		status = EINVAL;
+	}
+	return status;
+}
+
+/* Note that the post-command routine is called even if the pre-command or the real command
+   failed */
+int do_post_command(struct local_client *client)
+{
+	struct clvm_header *header =
+	    (struct clvm_header *) client->bits.localsock.cmd;
+	int status = 0;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+	char *args = header->node + strlen(header->node) + 1;
+	char *lockname;
+
+	switch (header->cmd) {
+	case CLVMD_CMD_TEST:
+		status =
+		    sync_unlock("CLVMD_TEST", (int) (long) client->bits.localsock.private);
+		client->bits.localsock.private = 0;
+		break;
+
+	case CLVMD_CMD_LOCK_VG:
+	case CLVMD_CMD_VG_BACKUP:
+	case CLVMD_CMD_LOCK_QUERY:
+		/* Nothing to do here */
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = post_lock_lv(lock_cmd, lock_flags, lockname);
+		break;
+	}
+	return status;
+}
+
+
+/* Called when the client is about to be deleted */
+void cmd_client_cleanup(struct local_client *client)
+{
+    if (client->bits.localsock.private) {
+
+	struct dm_hash_node *v;
+	struct dm_hash_table *lock_hash =
+	    (struct dm_hash_table *)client->bits.localsock.private;
+
+	dm_hash_iterate(v, lock_hash) {
+		int lkid = (int)(long)dm_hash_get_data(lock_hash, v);
+		char *lockname = dm_hash_get_key(lock_hash, v);
+
+		DEBUGLOG("cleanup: Unlocking lock %s %x\n", lockname, lkid);
+		sync_unlock(lockname, lkid);
+	}
+
+	dm_hash_destroy(lock_hash);
+	client->bits.localsock.private = 0;
+    }
+}

daemons/clvmd/clvmd-comms.h

@@ -0,0 +1,114 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Abstraction layer for clvmd cluster communications
+ */
+
+#ifndef _CLVMD_COMMS_H
+#define _CLVMD_COMMS_H
+
+struct local_client;
+
+struct cluster_ops {
+	void (*cluster_init_completed) (void);
+
+	int (*cluster_send_message) (const void *buf, int msglen,
+				     const char *csid,
+				     const char *errtext);
+	int (*name_from_csid) (const char *csid, char *name);
+	int (*csid_from_name) (char *csid, const char *name);
+	int (*get_num_nodes) (void);
+	int (*cluster_fd_callback) (struct local_client *fd, char *buf, int len,
+				    const char *csid,
+				    struct local_client **new_client);
+	int (*get_main_cluster_fd) (void);	/* gets accept FD or cman cluster socket */
+	int (*cluster_do_node_callback) (struct local_client *client,
+					 void (*callback) (struct local_client *,
+							   const char *csid,
+							   int node_up));
+	int (*is_quorate) (void);
+
+	void (*get_our_csid) (char *csid);
+	void (*add_up_node) (const char *csid);
+	void (*reread_config) (void);
+	void (*cluster_closedown) (void);
+
+	int (*get_cluster_name)(char *buf, int buflen);
+
+	int (*sync_lock) (const char *resource, int mode,
+			  int flags, int *lockid);
+	int (*sync_unlock) (const char *resource, int lockid);
+
+};
+
+#ifdef USE_GULM
+#  include "tcp-comms.h"
+struct cluster_ops *init_gulm_cluster(void);
+#define MAX_CSID_LEN 			GULM_MAX_CSID_LEN
+#define MAX_CLUSTER_MEMBER_NAME_LEN	GULM_MAX_CLUSTER_MEMBER_NAME_LEN
+#endif
+
+#ifdef USE_CMAN
+#  include <netinet/in.h>
+#  include "libcman.h"
+#  define CMAN_MAX_CSID_LEN 4
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN CMAN_MAX_CSID_LEN
+#  endif
+#  undef MAX_CLUSTER_MEMBER_NAME_LEN
+#  define MAX_CLUSTER_MEMBER_NAME_LEN   CMAN_MAX_NODENAME_LEN
+#  define CMAN_MAX_CLUSTER_MESSAGE 1500
+#  define CLUSTER_PORT_CLVMD 11
+struct cluster_ops *init_cman_cluster(void);
+#endif
+
+#ifdef USE_OPENAIS
+#  include <openais/saAis.h>
+#  include <corosync/totem/totem.h>
+#  define OPENAIS_CSID_LEN (sizeof(int))
+#  define OPENAIS_MAX_CLUSTER_MESSAGE         MESSAGE_SIZE_MAX
+#  define OPENAIS_MAX_CLUSTER_MEMBER_NAME_LEN SA_MAX_NAME_LENGTH
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       SA_MAX_NAME_LENGTH
+#  endif
+#  ifndef CMAN_MAX_CLUSTER_MESSAGE
+#    define CMAN_MAX_CLUSTER_MESSAGE          MESSAGE_SIZE_MAX
+#  endif
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_openais_cluster(void);
+#endif
+
+#ifdef USE_COROSYNC
+#  include <corosync/corotypes.h>
+#  define COROSYNC_CSID_LEN (sizeof(int))
+#  define COROSYNC_MAX_CLUSTER_MESSAGE         65535
+#  define COROSYNC_MAX_CLUSTER_MEMBER_NAME_LEN CS_MAX_NAME_LENGTH
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       CS_MAX_NAME_LENGTH
+#  endif
+#  ifndef CMAN_MAX_CLUSTER_MESSAGE
+#    define CMAN_MAX_CLUSTER_MESSAGE          65535
+#  endif
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_corosync_cluster(void);
+#endif
+
+
+#endif

daemons/clvmd/clvmd-corosync.c

@@ -0,0 +1,649 @@
+/*
+ * Copyright (C) 2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * This provides the interface between clvmd and corosync/DLM as the cluster
+ * and lock manager.
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <sys/file.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <signal.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <utmpx.h>
+#include <syslog.h>
+#include <assert.h>
+#include <libdevmapper.h>
+
+#include <corosync/corotypes.h>
+#include <corosync/cpg.h>
+#include <corosync/quorum.h>
+#include <corosync/confdb.h>
+#include <libdlm.h>
+
+#include "locking.h"
+#include "lvm-logging.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "lvm-functions.h"
+#include "clvmd.h"
+
+/* Timeout value for several corosync calls */
+#define LOCKSPACE_NAME "clvmd"
+
+static void corosync_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len);
+static void corosync_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries);
+static void _cluster_closedown(void);
+
+/* Hash list of nodes in the cluster */
+static struct dm_hash_table *node_hash;
+
+/* Number of active nodes */
+static int num_nodes;
+static unsigned int our_nodeid;
+
+static struct local_client *cluster_client;
+
+/* Corosync handles */
+static cpg_handle_t cpg_handle;
+static quorum_handle_t quorum_handle;
+
+/* DLM Handle */
+static dlm_lshandle_t *lockspace;
+
+static struct cpg_name cpg_group_name;
+
+/* Corosync callback structs */
+cpg_callbacks_t corosync_cpg_callbacks = {
+	.cpg_deliver_fn =            corosync_cpg_deliver_callback,
+	.cpg_confchg_fn =            corosync_cpg_confchg_callback,
+};
+
+quorum_callbacks_t quorum_callbacks = {
+	.quorum_notify_fn = NULL,
+};
+
+struct node_info
+{
+	enum {NODE_UNKNOWN, NODE_DOWN, NODE_UP, NODE_CLVMD} state;
+	int nodeid;
+};
+
+
+/* Set errno to something approximating the right value and return 0 or -1 */
+static int cs_to_errno(cs_error_t err)
+{
+	switch(err)
+	{
+	case CS_OK:
+		return 0;
+        case CS_ERR_LIBRARY:
+		errno = EINVAL;
+		break;
+        case CS_ERR_VERSION:
+		errno = EINVAL;
+		break;
+        case CS_ERR_INIT:
+		errno = EINVAL;
+		break;
+        case CS_ERR_TIMEOUT:
+		errno = ETIME;
+		break;
+        case CS_ERR_TRY_AGAIN:
+		errno = EAGAIN;
+		break;
+        case CS_ERR_INVALID_PARAM:
+		errno = EINVAL;
+		break;
+        case CS_ERR_NO_MEMORY:
+		errno = ENOMEM;
+		break;
+        case CS_ERR_BAD_HANDLE:
+		errno = EINVAL;
+		break;
+        case CS_ERR_BUSY:
+		errno = EBUSY;
+		break;
+        case CS_ERR_ACCESS:
+		errno = EPERM;
+		break;
+        case CS_ERR_NOT_EXIST:
+		errno = ENOENT;
+		break;
+        case CS_ERR_NAME_TOO_LONG:
+		errno = ENAMETOOLONG;
+		break;
+        case CS_ERR_EXIST:
+		errno = EEXIST;
+		break;
+        case CS_ERR_NO_SPACE:
+		errno = ENOSPC;
+		break;
+        case CS_ERR_INTERRUPT:
+		errno = EINTR;
+		break;
+	case CS_ERR_NAME_NOT_FOUND:
+		errno = ENOENT;
+		break;
+        case CS_ERR_NO_RESOURCES:
+		errno = ENOMEM;
+		break;
+        case CS_ERR_NOT_SUPPORTED:
+		errno = EOPNOTSUPP;
+		break;
+        case CS_ERR_BAD_OPERATION:
+		errno = EINVAL;
+		break;
+        case CS_ERR_FAILED_OPERATION:
+		errno = EIO;
+		break;
+        case CS_ERR_MESSAGE_ERROR:
+		errno = EIO;
+		break;
+        case CS_ERR_QUEUE_FULL:
+		errno = EXFULL;
+		break;
+        case CS_ERR_QUEUE_NOT_AVAILABLE:
+		errno = EINVAL;
+		break;
+        case CS_ERR_BAD_FLAGS:
+		errno = EINVAL;
+		break;
+        case CS_ERR_TOO_BIG:
+		errno = E2BIG;
+		break;
+        case CS_ERR_NO_SECTIONS:
+		errno = ENOMEM;
+		break;
+	default:
+		errno = EINVAL;
+		break;
+	}
+	return -1;
+}
+
+static char *print_corosync_csid(const char *csid)
+{
+	static char buf[128];
+	int id;
+
+	memcpy(&id, csid, sizeof(int));
+	sprintf(buf, "%d", id);
+	return buf;
+}
+
+static void corosync_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len)
+{
+	int target_nodeid;
+
+	memcpy(&target_nodeid, msg, COROSYNC_CSID_LEN);
+
+	DEBUGLOG("%u got message from nodeid %d for %d. len %zd\n",
+		 our_nodeid, nodeid, target_nodeid, msg_len-4);
+
+	if (nodeid != our_nodeid)
+		if (target_nodeid == our_nodeid || target_nodeid == 0)
+			process_message(cluster_client, (char *)msg+COROSYNC_CSID_LEN,
+					msg_len-COROSYNC_CSID_LEN, (char*)&nodeid);
+}
+
+static void corosync_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries)
+{
+	int i;
+	struct node_info *ninfo;
+
+	DEBUGLOG("confchg callback. %zd joined, %zd left, %zd members\n",
+		 joined_list_entries, left_list_entries, member_list_entries);
+
+	for (i=0; i<joined_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&joined_list[i].nodeid,
+					      COROSYNC_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = joined_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						      (char *)&ninfo->nodeid,
+						      COROSYNC_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	for (i=0; i<left_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&left_list[i].nodeid,
+					      COROSYNC_CSID_LEN);
+		if (ninfo)
+			ninfo->state = NODE_DOWN;
+	}
+
+	for (i=0; i<member_list_entries; i++) {
+		if (member_list[i].nodeid == 0) continue;
+		ninfo = dm_hash_lookup_binary(node_hash,
+				(char *)&member_list[i].nodeid,
+				COROSYNC_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = member_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						(char *)&ninfo->nodeid,
+						COROSYNC_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	num_nodes = member_list_entries;
+}
+
+static int _init_cluster(void)
+{
+	cs_error_t err;
+
+	node_hash = dm_hash_create(100);
+
+	err = cpg_initialize(&cpg_handle,
+			     &corosync_cpg_callbacks);
+	if (err != CS_OK) {
+		syslog(LOG_ERR, "Cannot initialise Corosync CPG service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise Corosync CPG service: %d", err);
+		return cs_to_errno(err);
+	}
+
+	err = quorum_initialize(&quorum_handle,
+				&quorum_callbacks);
+	if (err != CS_OK) {
+		syslog(LOG_ERR, "Cannot initialise Corosync quorum service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise Corosync quorum service: %d", err);
+		return cs_to_errno(err);
+	}
+
+
+	/* Create a lockspace for LV & VG locks to live in */
+	lockspace = dlm_create_lockspace(LOCKSPACE_NAME, 0600);
+	if (!lockspace) {
+		if (errno == EEXIST) {
+			lockspace = dlm_open_lockspace(LOCKSPACE_NAME);
+		}
+		if (!lockspace) {
+			syslog(LOG_ERR, "Unable to create lockspace for CLVM: %m");
+			quorum_finalize(quorum_handle);
+			return -1;
+		}
+	}
+	dlm_ls_pthread_init(lockspace);
+	DEBUGLOG("DLM initialisation complete\n");
+
+	/* Connect to the clvmd group */
+	strcpy((char *)cpg_group_name.value, "clvmd");
+	cpg_group_name.length = strlen((char *)cpg_group_name.value);
+	err = cpg_join(cpg_handle, &cpg_group_name);
+	if (err != CS_OK) {
+		cpg_finalize(cpg_handle);
+		quorum_finalize(quorum_handle);
+		dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+		syslog(LOG_ERR, "Cannot join clvmd process group");
+		DEBUGLOG("Cannot join clvmd process group: %d\n", err);
+		return cs_to_errno(err);
+	}
+
+	err = cpg_local_get(cpg_handle,
+			    &our_nodeid);
+	if (err != CS_OK) {
+		cpg_finalize(cpg_handle);
+		quorum_finalize(quorum_handle);
+		dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+		syslog(LOG_ERR, "Cannot get local node id\n");
+		return cs_to_errno(err);
+	}
+	DEBUGLOG("Our local node id is %d\n", our_nodeid);
+
+	DEBUGLOG("Connected to Corosync\n");
+
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	DEBUGLOG("cluster_closedown\n");
+	destroy_lvhash();
+
+	dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+	cpg_finalize(cpg_handle);
+	quorum_finalize(quorum_handle);
+}
+
+static void _get_our_csid(char *csid)
+{
+	memcpy(csid, &our_nodeid, sizeof(int));
+}
+
+/* Corosync doesn't really have nmode names so we
+   just use the node ID in hex instead */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int nodeid;
+	struct node_info *ninfo;
+
+	if (sscanf(name, "%x", &nodeid) == 1) {
+		ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+		if (ninfo)
+			return nodeid;
+	}
+	return -1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+	if (!ninfo)
+	{
+		sprintf(name, "UNKNOWN %s", print_corosync_csid(csid));
+		return -1;
+	}
+
+	sprintf(name, "%x", ninfo->nodeid);
+	return 0;
+}
+
+static int _get_num_nodes()
+{
+	DEBUGLOG("num_nodes = %d\n", num_nodes);
+	return num_nodes;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+	if (!ninfo) {
+		DEBUGLOG("corosync_add_up_node no node_hash entry for csid %s\n",
+			 print_corosync_csid(csid));
+		return;
+	}
+
+	DEBUGLOG("corosync_add_up_node %d\n", ninfo->nodeid);
+
+	ninfo->state = NODE_CLVMD;
+
+	return;
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+						      const char *csid, int node_up))
+{
+	struct dm_hash_node *hn;
+	struct node_info *ninfo;
+	int somedown = 0;
+
+	dm_hash_iterate(hn, node_hash)
+	{
+		char csid[COROSYNC_CSID_LEN];
+
+		ninfo = dm_hash_get_data(node_hash, hn);
+		memcpy(csid, dm_hash_get_key(node_hash, hn), COROSYNC_CSID_LEN);
+
+		DEBUGLOG("down_callback. node %d, state = %d\n", ninfo->nodeid,
+			 ninfo->state);
+
+		if (ninfo->state != NODE_DOWN)
+			callback(master_client, csid, ninfo->state == NODE_CLVMD);
+		if (ninfo->state != NODE_CLVMD)
+			somedown = -1;
+	}
+	return somedown;
+}
+
+/* Real locking */
+static int _lock_resource(const char *resource, int mode, int flags, int *lockid)
+{
+	struct dlm_lksb lksb;
+	int err;
+
+	DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
+
+	if (flags & LKF_CONVERT)
+		lksb.sb_lkid = *lockid;
+
+	err = dlm_ls_lock_wait(lockspace,
+			       mode,
+			       &lksb,
+			       flags,
+			       resource,
+			       strlen(resource),
+			       0,
+			       NULL, NULL, NULL);
+
+	if (err != 0)
+	{
+		DEBUGLOG("dlm_ls_lock returned %d\n", errno);
+		return err;
+	}
+	if (lksb.sb_status != 0)
+	{
+		DEBUGLOG("dlm_ls_lock returns lksb.sb_status %d\n", lksb.sb_status);
+		errno = lksb.sb_status;
+		return -1;
+	}
+
+	DEBUGLOG("lock_resource returning %d, lock_id=%x\n", err, lksb.sb_lkid);
+
+	*lockid = lksb.sb_lkid;
+
+	return 0;
+}
+
+
+static int _unlock_resource(const char *resource, int lockid)
+{
+	struct dlm_lksb lksb;
+	int err;
+
+	DEBUGLOG("unlock_resource: %s lockid: %x\n", resource, lockid);
+	lksb.sb_lkid = lockid;
+
+	err = dlm_ls_unlock_wait(lockspace,
+				 lockid,
+				 0,
+				 &lksb);
+	if (err != 0)
+	{
+		DEBUGLOG("Unlock returned %d\n", err);
+		return err;
+	}
+	if (lksb.sb_status != EUNLOCK)
+	{
+		DEBUGLOG("dlm_ls_unlock_wait returns lksb.sb_status: %d\n", lksb.sb_status);
+		errno = lksb.sb_status;
+		return -1;
+	}   
+
+
+	return 0;
+}
+
+static int _is_quorate()
+{
+	int quorate;
+	if (quorum_getquorate(quorum_handle, &quorate) == CS_OK)
+		return quorate;
+	else
+		return 0;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	int select_fd;
+
+	cpg_fd_get(cpg_handle, &select_fd);
+	return select_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	cluster_client = fd;
+	*new_client = NULL;
+	cpg_dispatch(cpg_handle, CS_DISPATCH_ONE);
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	struct iovec iov[2];
+	cs_error_t err;
+	int target_node;
+
+	if (csid)
+		memcpy(&target_node, csid, COROSYNC_CSID_LEN);
+	else
+		target_node = 0;
+
+	iov[0].iov_base = &target_node;
+	iov[0].iov_len = sizeof(int);
+	iov[1].iov_base = (char *)buf;
+	iov[1].iov_len = msglen;
+
+	err = cpg_mcast_joined(cpg_handle, CPG_TYPE_AGREED, iov, 2);
+	return cs_to_errno(err);
+}
+
+/*
+ * We are not necessarily connected to a Red Hat Cluster system,
+ * but if we are, this returns the cluster name from cluster.conf.
+ * I've used confdb rather than ccs to reduce the inter-package
+ * dependancies as well as to allow people to set a cluster name
+ * for themselves even if they are not running on RH cluster.
+ */
+static int _get_cluster_name(char *buf, int buflen)
+{
+	confdb_handle_t handle;
+	int result;
+	size_t namelen = buflen;
+	hdb_handle_t cluster_handle;
+	confdb_callbacks_t callbacks = {
+		.confdb_key_change_notify_fn = NULL,
+		.confdb_object_create_change_notify_fn = NULL,
+		.confdb_object_delete_change_notify_fn = NULL
+	};
+
+	/* This is a default in case everything else fails */
+	strncpy(buf, "Corosync", buflen);
+
+	/* Look for a cluster name in confdb */
+	result = confdb_initialize (&handle, &callbacks);
+        if (result != CS_OK)
+		return 0;
+
+        result = confdb_object_find_start(handle, OBJECT_PARENT_HANDLE);
+	if (result != CS_OK)
+		goto out;
+
+        result = confdb_object_find(handle, OBJECT_PARENT_HANDLE, (void *)"cluster", strlen("cluster"), &cluster_handle);
+        if (result != CS_OK)
+		goto out;
+
+        result = confdb_key_get(handle, cluster_handle, (void *)"name", strlen("name"), buf, &namelen);
+        if (result != CS_OK)
+		goto out;
+
+	buf[namelen] = '\0';
+
+out:
+	confdb_finalize(handle);
+	return 0;
+}
+
+static struct cluster_ops _cluster_corosync_ops = {
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _lock_resource,
+	.sync_unlock              = _unlock_resource,
+};
+
+struct cluster_ops *init_corosync_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_corosync_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-gulm.h

@@ -0,0 +1,13 @@
+
+/* DLM constant that clvmd uses as a generic NONBLOCK lock flag */
+#define LKF_NOQUEUE 1
+
+extern int get_next_node_csid(void **context, char *csid);
+extern void add_down_node(char *csid);
+extern int gulm_fd(void);
+extern int get_ip_address(const char *node, char *addr);
+extern void tcp_remove_client(const char *csid);
+extern int alloc_client(int fd, const char *csid, struct local_client **new_client);
+
+void gulm_add_up_node(const char *csid);
+int gulm_name_from_csid(const char *csid, char *name);

daemons/clvmd/clvmd-openais.c

@@ -0,0 +1,716 @@
+/*
+ * Copyright (C) 2007-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * This provides the interface between clvmd and OpenAIS as the cluster
+ * and lock manager.
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <sys/file.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <signal.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <utmpx.h>
+#include <syslog.h>
+#include <assert.h>
+#include <libdevmapper.h>
+
+#include <openais/saAis.h>
+#include <openais/saLck.h>
+
+#include <corosync/corotypes.h>
+#include <corosync/cpg.h>
+
+#include "locking.h"
+#include "lvm-logging.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "lvm-functions.h"
+#include "clvmd.h"
+
+/* Timeout value for several openais calls */
+#define TIMEOUT 10
+
+static void openais_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len);
+static void openais_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries);
+
+static void _cluster_closedown(void);
+
+/* Hash list of nodes in the cluster */
+static struct dm_hash_table *node_hash;
+
+/* For associating lock IDs & resource handles */
+static struct dm_hash_table *lock_hash;
+
+/* Number of active nodes */
+static int num_nodes;
+static unsigned int our_nodeid;
+
+static struct local_client *cluster_client;
+
+/* OpenAIS handles */
+static cpg_handle_t cpg_handle;
+static SaLckHandleT lck_handle;
+
+static struct cpg_name cpg_group_name;
+
+/* Openais callback structs */
+cpg_callbacks_t openais_cpg_callbacks = {
+	.cpg_deliver_fn =            openais_cpg_deliver_callback,
+	.cpg_confchg_fn =            openais_cpg_confchg_callback,
+};
+
+struct node_info
+{
+	enum {NODE_UNKNOWN, NODE_DOWN, NODE_UP, NODE_CLVMD} state;
+	int nodeid;
+};
+
+struct lock_info
+{
+	SaLckResourceHandleT res_handle;
+	SaLckLockIdT         lock_id;
+	SaNameT              lock_name;
+};
+
+/* Set errno to something approximating the right value and return 0 or -1 */
+static int ais_to_errno(SaAisErrorT err)
+{
+	switch(err)
+	{
+	case SA_AIS_OK:
+		return 0;
+        case SA_AIS_ERR_LIBRARY:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_VERSION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_INIT:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TIMEOUT:
+		errno = ETIME;
+		break;
+        case SA_AIS_ERR_TRY_AGAIN:
+		errno = EAGAIN;
+		break;
+        case SA_AIS_ERR_INVALID_PARAM:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_NO_MEMORY:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_BAD_HANDLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BUSY:
+		errno = EBUSY;
+		break;
+        case SA_AIS_ERR_ACCESS:
+		errno = EPERM;
+		break;
+        case SA_AIS_ERR_NOT_EXIST:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NAME_TOO_LONG:
+		errno = ENAMETOOLONG;
+		break;
+        case SA_AIS_ERR_EXIST:
+		errno = EEXIST;
+		break;
+        case SA_AIS_ERR_NO_SPACE:
+		errno = ENOSPC;
+		break;
+        case SA_AIS_ERR_INTERRUPT:
+		errno = EINTR;
+		break;
+	case SA_AIS_ERR_NAME_NOT_FOUND:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NO_RESOURCES:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_NOT_SUPPORTED:
+		errno = EOPNOTSUPP;
+		break;
+        case SA_AIS_ERR_BAD_OPERATION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_FAILED_OPERATION:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_MESSAGE_ERROR:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_QUEUE_FULL:
+		errno = EXFULL;
+		break;
+        case SA_AIS_ERR_QUEUE_NOT_AVAILABLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BAD_FLAGS:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TOO_BIG:
+		errno = E2BIG;
+		break;
+        case SA_AIS_ERR_NO_SECTIONS:
+		errno = ENOMEM;
+		break;
+	default:
+		errno = EINVAL;
+		break;
+	}
+	return -1;
+}
+
+static char *print_openais_csid(const char *csid)
+{
+	static char buf[128];
+	int id;
+
+	memcpy(&id, csid, sizeof(int));
+	sprintf(buf, "%d", id);
+	return buf;
+}
+
+static int add_internal_client(int fd, fd_callback_t callback)
+{
+	struct local_client *client;
+
+	DEBUGLOG("Add_internal_client, fd = %d\n", fd);
+
+	client = malloc(sizeof(struct local_client));
+	if (!client)
+	{
+		DEBUGLOG("malloc failed\n");
+		return -1;
+	}
+
+	memset(client, 0, sizeof(struct local_client));
+	client->fd = fd;
+	client->type = CLUSTER_INTERNAL;
+	client->callback = callback;
+	add_client(client);
+
+	/* Set Close-on-exec */
+	fcntl(fd, F_SETFD, 1);
+
+	return 0;
+}
+
+static void openais_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len)
+{
+	int target_nodeid;
+
+	memcpy(&target_nodeid, msg, OPENAIS_CSID_LEN);
+
+	DEBUGLOG("%u got message from nodeid %d for %d. len %d\n",
+		 our_nodeid, nodeid, target_nodeid, msg_len-4);
+
+	if (nodeid != our_nodeid)
+		if (target_nodeid == our_nodeid || target_nodeid == 0)
+			process_message(cluster_client, (char *)msg+OPENAIS_CSID_LEN,
+					msg_len-OPENAIS_CSID_LEN, (char*)&nodeid);
+}
+
+static void openais_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries)
+{
+	int i;
+	struct node_info *ninfo;
+
+	DEBUGLOG("confchg callback. %d joined, %d left, %d members\n",
+		 joined_list_entries, left_list_entries, member_list_entries);
+
+	for (i=0; i<joined_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&joined_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = joined_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						      (char *)&ninfo->nodeid,
+						      OPENAIS_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	for (i=0; i<left_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&left_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (ninfo)
+			ninfo->state = NODE_DOWN;
+	}
+
+	for (i=0; i<member_list_entries; i++) {
+		if (member_list[i].nodeid == 0) continue;
+		ninfo = dm_hash_lookup_binary(node_hash,
+				(char *)&member_list[i].nodeid,
+				OPENAIS_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = member_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						(char *)&ninfo->nodeid,
+						OPENAIS_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	num_nodes = member_list_entries;
+}
+
+static int lck_dispatch(struct local_client *client, char *buf, int len,
+			const char *csid, struct local_client **new_client)
+{
+	*new_client = NULL;
+	saLckDispatch(lck_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _init_cluster(void)
+{
+	SaAisErrorT err;
+	SaVersionT  ver = { 'B', 1, 1 };
+	int select_fd;
+
+	node_hash = dm_hash_create(100);
+	lock_hash = dm_hash_create(10);
+
+	err = cpg_initialize(&cpg_handle,
+			     &openais_cpg_callbacks);
+	if (err != SA_AIS_OK) {
+		syslog(LOG_ERR, "Cannot initialise OpenAIS CPG service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS CPG service: %d", err);
+		return ais_to_errno(err);
+	}
+
+	err = saLckInitialize(&lck_handle,
+					NULL,
+			      &ver);
+	if (err != SA_AIS_OK) {
+		cpg_initialize(&cpg_handle, &openais_cpg_callbacks);
+		syslog(LOG_ERR, "Cannot initialise OpenAIS lock service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS lock service: %d\n\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Connect to the clvmd group */
+	strcpy((char *)cpg_group_name.value, "clvmd");
+	cpg_group_name.length = strlen((char *)cpg_group_name.value);
+	err = cpg_join(cpg_handle, &cpg_group_name);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot join clvmd process group");
+		DEBUGLOG("Cannot join clvmd process group: %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	err = cpg_local_get(cpg_handle,
+			    &our_nodeid);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot get local node id\n");
+		return ais_to_errno(err);
+	}
+	DEBUGLOG("Our local node id is %d\n", our_nodeid);
+
+	saLckSelectionObjectGet(lck_handle, (SaSelectionObjectT *)&select_fd);
+	add_internal_client(select_fd, lck_dispatch);
+
+	DEBUGLOG("Connected to OpenAIS\n");
+
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	DEBUGLOG("cluster_closedown\n");
+	destroy_lvhash();
+
+	saLckFinalize(lck_handle);
+	cpg_finalize(cpg_handle);
+}
+
+static void _get_our_csid(char *csid)
+{
+	memcpy(csid, &our_nodeid, sizeof(int));
+}
+
+/* OpenAIS doesn't really have nmode names so we
+   just use the node ID in hex instead */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int nodeid;
+	struct node_info *ninfo;
+
+	if (sscanf(name, "%x", &nodeid) == 1) {
+		ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+		if (ninfo)
+			return nodeid;
+	}
+	return -1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo)
+	{
+		sprintf(name, "UNKNOWN %s", print_openais_csid(csid));
+		return -1;
+	}
+
+	sprintf(name, "%x", ninfo->nodeid);
+	return 0;
+}
+
+static int _get_num_nodes()
+{
+	DEBUGLOG("num_nodes = %d\n", num_nodes);
+	return num_nodes;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo) {
+		DEBUGLOG("openais_add_up_node no node_hash entry for csid %s\n",
+			 print_openais_csid(csid));
+		return;
+	}
+
+	DEBUGLOG("openais_add_up_node %d\n", ninfo->nodeid);
+
+	ninfo->state = NODE_CLVMD;
+
+	return;
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+						      const char *csid, int node_up))
+{
+	struct dm_hash_node *hn;
+	struct node_info *ninfo;
+	int somedown = 0;
+
+	dm_hash_iterate(hn, node_hash)
+	{
+		char csid[OPENAIS_CSID_LEN];
+
+		ninfo = dm_hash_get_data(node_hash, hn);
+		memcpy(csid, dm_hash_get_key(node_hash, hn), OPENAIS_CSID_LEN);
+
+		DEBUGLOG("down_callback. node %d, state = %d\n", ninfo->nodeid,
+			 ninfo->state);
+
+		if (ninfo->state != NODE_DOWN)
+			callback(master_client, csid, ninfo->state == NODE_CLVMD);
+		if (ninfo->state != NODE_CLVMD)
+			somedown = -1;
+	}
+	return somedown;
+}
+
+/* Real locking */
+static int _lock_resource(char *resource, int mode, int flags, int *lockid)
+{
+	struct lock_info *linfo;
+	SaLckResourceHandleT res_handle;
+	SaAisErrorT err;
+	SaLckLockIdT lock_id;
+	SaLckLockStatusT lockStatus;
+
+	/* This needs to be converted from DLM/LVM2 value for OpenAIS LCK */
+	if (flags & LCK_NONBLOCK) flags = SA_LCK_LOCK_NO_QUEUE;
+
+	linfo = malloc(sizeof(struct lock_info));
+	if (!linfo)
+		return -1;
+
+	DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
+
+	linfo->lock_name.length = strlen(resource)+1;
+	strcpy((char *)linfo->lock_name.value, resource);
+
+	err = saLckResourceOpen(lck_handle, &linfo->lock_name,
+				SA_LCK_RESOURCE_CREATE, TIMEOUT, &res_handle);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("ResourceOpen returned %d\n", err);
+		free(linfo);
+		return ais_to_errno(err);
+	}
+
+	err = saLckResourceLock(
+			res_handle,
+			&lock_id,
+			mode,
+			flags,
+			0,
+			SA_TIME_END,
+			&lockStatus);
+	if (err != SA_AIS_OK && lockStatus != SA_LCK_LOCK_GRANTED)
+	{
+		free(linfo);
+		saLckResourceClose(res_handle);
+		return ais_to_errno(err);
+	}
+			
+	/* Wait for it to complete */
+
+	DEBUGLOG("lock_resource returning %d, lock_id=%llx\n", err,
+		 lock_id);
+
+	linfo->lock_id = lock_id;
+	linfo->res_handle = res_handle;
+
+	dm_hash_insert(lock_hash, resource, linfo);
+
+	return ais_to_errno(err);
+}
+
+
+static int _unlock_resource(char *resource, int lockid)
+{
+	SaAisErrorT err;
+	struct lock_info *linfo;
+
+	DEBUGLOG("unlock_resource %s\n", resource);
+	linfo = dm_hash_lookup(lock_hash, resource);
+	if (!linfo)
+		return 0;
+
+	DEBUGLOG("unlock_resource: lockid: %llx\n", linfo->lock_id);
+	err = saLckResourceUnlock(linfo->lock_id, SA_TIME_END);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("Unlock returned %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Release the resource */
+	dm_hash_remove(lock_hash, resource);
+	saLckResourceClose(linfo->res_handle);
+	free(linfo);
+
+	return ais_to_errno(err);
+}
+
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
+{
+	int status;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	switch (mode)
+	{
+	case LCK_EXCL:
+		status = _lock_resource(lock1, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+
+		/* If we can't get this lock too then bail out */
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, LCK_NONBLOCK,
+					lockid);
+		if (status == SA_LCK_LOCK_NOT_QUEUED)
+		{
+			_unlock_resource(lock1, *lockid);
+			status = -1;
+			errno = EAGAIN;
+		}
+		break;
+
+	case LCK_PREAD:
+	case LCK_READ:
+		status = _lock_resource(lock1, SA_LCK_PR_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock2, *lockid);
+		break;
+
+	case LCK_WRITE:
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock1, *lockid);
+		break;
+
+	default:
+		status = -1;
+		errno = EINVAL;
+		break;
+	}
+out:
+	*lockid = mode;
+	return status;
+}
+
+static int _sync_unlock(const char *resource, int lockid)
+{
+	int status = 0;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	_unlock_resource(lock1, lockid);
+	_unlock_resource(lock2, lockid);
+
+	return status;
+}
+
+/* We are always quorate ! */
+static int _is_quorate()
+{
+	return 1;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	int select_fd;
+
+	cpg_fd_get(cpg_handle, &select_fd);
+	return select_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	cluster_client = fd;
+	*new_client = NULL;
+	cpg_dispatch(cpg_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	struct iovec iov[2];
+	SaAisErrorT err;
+	int target_node;
+
+	if (csid)
+		memcpy(&target_node, csid, OPENAIS_CSID_LEN);
+	else
+		target_node = 0;
+
+	iov[0].iov_base = &target_node;
+	iov[0].iov_len = sizeof(int);
+	iov[1].iov_base = (char *)buf;
+	iov[1].iov_len = msglen;
+
+	err = cpg_mcast_joined(cpg_handle, CPG_TYPE_AGREED, iov, 2);
+	return ais_to_errno(err);
+}
+
+/* We don't have a cluster name to report here */
+static int _get_cluster_name(char *buf, int buflen)
+{
+	strncpy(buf, "OpenAIS", buflen);
+	return 0;
+}
+
+static struct cluster_ops _cluster_openais_ops = {
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_openais_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_openais_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd.h

@@ -0,0 +1,126 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _CLVMD_H
+#define _CLVMD_H
+
+#define CLVMD_MAJOR_VERSION 0
+#define CLVMD_MINOR_VERSION 2
+#define CLVMD_PATCH_VERSION 1
+
+/* Name of the cluster LVM admin lock */
+#define ADMIN_LOCK_NAME "CLVMD_ADMIN"
+
+/* Default time (in seconds) we will wait for all remote commands to execute
+   before declaring them dead */
+#define DEFAULT_CMD_TIMEOUT 60
+
+/* One of these for each reply we get from command execution on a node */
+struct node_reply {
+	char node[MAX_CLUSTER_MEMBER_NAME_LEN];
+	char *replymsg;
+	int status;
+	struct node_reply *next;
+};
+
+typedef enum {DEBUG_OFF, DEBUG_STDERR, DEBUG_SYSLOG} debug_t;
+
+/*
+ * These exist for the use of local sockets only when we are
+ * collecting responses from all cluster nodes
+ */
+struct localsock_bits {
+	struct node_reply *replies;
+	int num_replies;
+	int expected_replies;
+	time_t sent_time;	/* So we can check for timeouts */
+	int in_progress;	/* Only execute one cmd at a time per client */
+	int sent_out;		/* Flag to indicate that a command was sent
+				   to remote nodes */
+	void *private;		/* Private area for command processor use */
+	void *cmd;		/* Whole command as passed down local socket */
+	int cmd_len;		/* Length of above */
+	int pipe;		/* Pipe to send PRE completion status down */
+	int finished;		/* Flag to tell subthread to exit */
+	int all_success;	/* Set to 0 if any node (or the pre_command)
+				   failed */
+	struct local_client *pipe_client;
+	pthread_t threadid;
+	enum { PRE_COMMAND, POST_COMMAND, QUIT } state;
+	pthread_mutex_t mutex;	/* Main thread and worker synchronisation */
+	pthread_cond_t cond;
+
+	pthread_mutex_t reply_mutex;	/* Protect reply structure */
+};
+
+/* Entries for PIPE clients */
+struct pipe_bits {
+	struct local_client *client;	/* Actual (localsock) client */
+	pthread_t threadid;		/* Our own copy of the thread id */
+};
+
+/* Entries for Network socket clients */
+struct netsock_bits {
+	void *private;
+	int flags;
+};
+
+typedef int (*fd_callback_t) (struct local_client * fd, char *buf, int len,
+			      const char *csid,
+			      struct local_client ** new_client);
+
+/* One of these for each fd we are listening on */
+struct local_client {
+	int fd;
+	enum { CLUSTER_MAIN_SOCK, CLUSTER_DATA_SOCK, LOCAL_RENDEZVOUS,
+		    LOCAL_SOCK, THREAD_PIPE, CLUSTER_INTERNAL } type;
+	struct local_client *next;
+	unsigned short xid;
+	fd_callback_t callback;
+	uint8_t removeme;
+
+	union {
+		struct localsock_bits localsock;
+		struct pipe_bits pipe;
+		struct netsock_bits net;
+	} bits;
+};
+
+#define DEBUGLOG(fmt, args...) debuglog(fmt, ## args);
+
+#ifndef max
+#define max(a,b) ((a)>(b)?(a):(b))
+#endif
+
+/* The real command processor is in clvmd-command.c */
+extern int do_command(struct local_client *client, struct clvm_header *msg,
+		      int msglen, char **buf, int buflen, int *retlen);
+
+/* Pre and post command routines are called only on the local node */
+extern int do_pre_command(struct local_client *client);
+extern int do_post_command(struct local_client *client);
+extern void cmd_client_cleanup(struct local_client *client);
+extern int add_client(struct local_client *new_client);
+
+extern void clvmd_cluster_init_completed(void);
+extern void process_message(struct local_client *client, const char *buf,
+			    int len, const char *csid);
+extern void debuglog(const char *fmt, ... )
+  __attribute__ ((format(printf, 1, 2)));
+
+int sync_lock(const char *resource, int mode, int flags, int *lockid);
+int sync_unlock(const char *resource, int lockid);
+
+#endif

daemons/clvmd/lvm-functions.c

@@ -0,0 +1,886 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <syslog.h>
+#include <assert.h>
+#include <libdevmapper.h>
+#include <libdlm.h>
+
+#include "lvm-types.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "clvmd.h"
+#include "lvm-functions.h"
+
+/* LVM2 headers */
+#include "toolcontext.h"
+#include "lvmcache.h"
+#include "lvm-logging.h"
+#include "lvm-globals.h"
+#include "activate.h"
+#include "locking.h"
+#include "archiver.h"
+#include "defaults.h"
+#include "memlock.h"
+
+static struct cmd_context *cmd = NULL;
+static struct dm_hash_table *lv_hash = NULL;
+static pthread_mutex_t lv_hash_lock;
+static pthread_mutex_t lvm_lock;
+static char last_error[1024];
+static int suspended = 0;
+
+struct lv_info {
+	int lock_id;
+	int lock_mode;
+};
+
+static const char *decode_locking_cmd(unsigned char cmdl)
+{
+	static char buf[128];
+	const char *type;
+	const char *scope;
+	const char *command;
+
+	switch (cmdl & LCK_TYPE_MASK) {
+	case LCK_NULL:
+		type = "NULL";
+		break;
+	case LCK_READ:
+		type = "READ";
+		break;
+	case LCK_PREAD:
+		type = "PREAD";
+		break;
+	case LCK_WRITE:
+		type = "WRITE";
+		break;
+	case LCK_EXCL:
+		type = "EXCL";
+		break;
+	case LCK_UNLOCK:
+		type = "UNLOCK";
+		break;
+	default:
+		type = "unknown";
+		break;
+	}
+
+	switch (cmdl & LCK_SCOPE_MASK) {
+	case LCK_VG:
+		scope = "VG";
+		command = "LCK_VG";
+		break;
+	case LCK_LV:
+		scope = "LV"; 
+		switch (cmdl & LCK_MASK) {
+		case LCK_LV_EXCLUSIVE & LCK_MASK:
+			command = "LCK_LV_EXCLUSIVE";
+			break;
+		case LCK_LV_SUSPEND & LCK_MASK:
+			command = "LCK_LV_SUSPEND";
+			break;
+		case LCK_LV_RESUME & LCK_MASK:
+			command = "LCK_LV_RESUME";
+			break;
+		case LCK_LV_ACTIVATE & LCK_MASK:
+			command = "LCK_LV_ACTIVATE";
+			break;
+		case LCK_LV_DEACTIVATE & LCK_MASK:
+			command = "LCK_LV_DEACTIVATE";
+			break;
+		default:
+			command = "unknown";
+			break;
+		}
+		break;
+	default:
+		scope = "unknown";
+		command = "unknown";
+		break;
+	}
+
+	sprintf(buf, "0x%x %s (%s|%s%s%s%s%s%s)", cmdl, command, type, scope,
+		cmdl & LCK_NONBLOCK   ? "|NONBLOCK" : "",
+		cmdl & LCK_HOLD       ? "|HOLD" : "",
+		cmdl & LCK_LOCAL      ? "|LOCAL" : "",
+		cmdl & LCK_CLUSTER_VG ? "|CLUSTER_VG" : "",
+		cmdl & LCK_CACHE      ? "|CACHE" : "");
+
+	return buf;
+}
+
+static const char *decode_flags(unsigned char flags)
+{
+	static char buf[128];
+
+	sprintf(buf, "0x%x (%s%s%s%s)", flags,
+		flags & LCK_PARTIAL_MODE	  ? "PARTIAL_MODE " : "",
+		flags & LCK_MIRROR_NOSYNC_MODE	  ? "MIRROR_NOSYNC " : "",
+		flags & LCK_DMEVENTD_MONITOR_MODE ? "DMEVENTD_MONITOR " : "",
+		flags & LCK_CONVERT ? "CONVERT " : "");
+
+	return buf;
+}
+
+char *get_last_lvm_error()
+{
+	return last_error;
+}
+
+/*
+ * Hash lock info helpers
+ */
+static struct lv_info *lookup_info(const char *resource)
+{
+	struct lv_info *lvi;
+
+	pthread_mutex_lock(&lv_hash_lock);
+	lvi = dm_hash_lookup(lv_hash, resource);
+	pthread_mutex_unlock(&lv_hash_lock);
+
+	return lvi;
+}
+
+static void insert_info(const char *resource, struct lv_info *lvi)
+{
+	pthread_mutex_lock(&lv_hash_lock);
+	dm_hash_insert(lv_hash, resource, lvi);
+	pthread_mutex_unlock(&lv_hash_lock);
+}
+
+static void remove_info(const char *resource)
+{
+	pthread_mutex_lock(&lv_hash_lock);
+	dm_hash_remove(lv_hash, resource);
+	pthread_mutex_unlock(&lv_hash_lock);
+}
+
+/*
+ * Return the mode a lock is currently held at (or -1 if not held)
+ */
+static int get_current_lock(char *resource)
+{
+	struct lv_info *lvi;
+
+	if ((lvi = lookup_info(resource)))
+		return lvi->lock_mode;
+
+	return -1;
+}
+
+
+void init_lvhash()
+{
+	/* Create hash table for keeping LV locks & status */
+	lv_hash = dm_hash_create(100);
+	pthread_mutex_init(&lv_hash_lock, NULL);
+	pthread_mutex_init(&lvm_lock, NULL);
+}
+
+/* Called at shutdown to tidy the lockspace */
+void destroy_lvhash()
+{
+	struct dm_hash_node *v;
+	struct lv_info *lvi;
+	char *resource;
+	int status;
+
+	pthread_mutex_lock(&lv_hash_lock);
+
+	dm_hash_iterate(v, lv_hash) {
+		lvi = dm_hash_get_data(lv_hash, v);
+		resource = dm_hash_get_key(lv_hash, v);
+
+		if ((status = sync_unlock(resource, lvi->lock_id)))
+			DEBUGLOG("unlock_all. unlock failed(%d): %s\n",
+				 status,  strerror(errno));
+		free(lvi);
+	}
+
+	dm_hash_destroy(lv_hash);
+	lv_hash = NULL;
+
+	pthread_mutex_unlock(&lv_hash_lock);
+}
+
+/* Gets a real lock and keeps the info in the hash table */
+int hold_lock(char *resource, int mode, int flags)
+{
+	int status;
+	int saved_errno;
+	struct lv_info *lvi;
+
+	/* Mask off invalid options */
+	flags &= LKF_NOQUEUE | LKF_CONVERT;
+
+	lvi = lookup_info(resource);
+
+	if (lvi && lvi->lock_mode == mode) {
+		DEBUGLOG("hold_lock, lock mode %d already held\n", mode);
+		return 0;
+	}
+
+	/* Only allow explicit conversions */
+	if (lvi && !(flags & LKF_CONVERT)) {
+		errno = EBUSY;
+		return -1;
+	}
+	if (lvi) {
+		/* Already exists - convert it */
+		status =
+		    sync_lock(resource, mode, flags, &lvi->lock_id);
+		saved_errno = errno;
+		if (!status)
+			lvi->lock_mode = mode;
+
+		if (status) {
+			DEBUGLOG("hold_lock. convert to %d failed: %s\n", mode,
+				 strerror(errno));
+		}
+		errno = saved_errno;
+	} else {
+		lvi = malloc(sizeof(struct lv_info));
+		if (!lvi)
+			return -1;
+
+		lvi->lock_mode = mode;
+		status = sync_lock(resource, mode, flags & ~LKF_CONVERT, &lvi->lock_id);
+		saved_errno = errno;
+		if (status) {
+			free(lvi);
+			DEBUGLOG("hold_lock. lock at %d failed: %s\n", mode,
+				 strerror(errno));
+		} else
+			insert_info(resource, lvi);
+
+		errno = saved_errno;
+	}
+	return status;
+}
+
+/* Unlock and remove it from the hash table */
+int hold_unlock(char *resource)
+{
+	struct lv_info *lvi;
+	int status;
+	int saved_errno;
+
+	if (!(lvi = lookup_info(resource))) {
+		DEBUGLOG("hold_unlock, lock not already held\n");
+		return 0;
+	}
+
+	status = sync_unlock(resource, lvi->lock_id);
+	saved_errno = errno;
+	if (!status) {
+		remove_info(resource);
+		free(lvi);
+	} else {
+		DEBUGLOG("hold_unlock. unlock failed(%d): %s\n", status,
+			 strerror(errno));
+	}
+
+	errno = saved_errno;
+	return status;
+}
+
+/* Watch the return codes here.
+   liblvm API functions return 1(true) for success, 0(false) for failure and don't set errno.
+   libdlm API functions return 0 for success, -1 for failure and do set errno.
+   These functions here return 0 for success or >0 for failure (where the retcode is errno)
+*/
+
+/* Activate LV exclusive or non-exclusive */
+static int do_activate_lv(char *resource, unsigned char lock_flags, int mode)
+{
+	int oldmode;
+	int status;
+	int activate_lv;
+	int exclusive = 0;
+	struct lvinfo lvi;
+
+	/* Is it already open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == mode && (lock_flags & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_activate_lv, lock already held at %d\n", oldmode);
+		return 0;	/* Nothing to do */
+	}
+
+	/* Does the config file want us to activate this LV ? */
+	if (!lv_activation_filter(cmd, resource, &activate_lv))
+		return EIO;
+
+	if (!activate_lv)
+		return 0;	/* Success, we did nothing! */
+
+	/* Do we need to activate exclusively? */
+	if ((activate_lv == 2) || (mode == LKM_EXMODE)) {
+		exclusive = 1;
+		mode = LKM_EXMODE;
+	}
+
+	/*
+	 * Try to get the lock if it's a clustered volume group.
+	 * Use lock conversion only if requested, to prevent implicit conversion
+	 * of exclusive lock to shared one during activation.
+	 */
+	if (lock_flags & LCK_CLUSTER_VG) {
+		status = hold_lock(resource, mode, LKF_NOQUEUE | (lock_flags & LCK_CONVERT?LKF_CONVERT:0));
+		if (status) {
+			/* Return an LVM-sensible error for this.
+			 * Forcing EIO makes the upper level return this text
+			 * rather than the strerror text for EAGAIN.
+			 */
+			if (errno == EAGAIN) {
+				sprintf(last_error, "Volume is busy on another node");
+				errno = EIO;
+			}
+			return errno;
+		}
+	}
+
+	/* If it's suspended then resume it */
+	if (!lv_info_by_lvid(cmd, resource, &lvi, 0, 0))
+		return EIO;
+
+	if (lvi.suspended)
+		if (!lv_resume(cmd, resource))
+			return EIO;
+
+	/* Now activate it */
+	if (!lv_activate(cmd, resource, exclusive))
+		return EIO;
+
+	return 0;
+}
+
+/* Resume the LV if it was active */
+static int do_resume_lv(char *resource, unsigned char lock_flags)
+{
+	int oldmode;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (lock_flags & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_resume_lv, lock not already held\n");
+		return 0;	/* We don't need to do anything */
+	}
+
+	if (!lv_resume_if_active(cmd, resource))
+		return EIO;
+
+	return 0;
+}
+
+/* Suspend the device if active */
+static int do_suspend_lv(char *resource, unsigned char lock_flags)
+{
+	int oldmode;
+	struct lvinfo lvi;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (lock_flags & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_suspend_lv, lock not already held\n");
+		return 0; /* Not active, so it's OK */
+	}
+
+	/* Only suspend it if it exists */
+	if (!lv_info_by_lvid(cmd, resource, &lvi, 0, 0))
+		return EIO;
+
+	if (lvi.exists) {
+		if (!lv_suspend_if_active(cmd, resource)) {
+			return EIO;
+		}
+	}
+	return 0;
+}
+
+static int do_deactivate_lv(char *resource, unsigned char lock_flags)
+{
+	int oldmode;
+	int status;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (lock_flags & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_deactivate_lock, lock not already held\n");
+		return 0;	/* We don't need to do anything */
+	}
+
+	if (!lv_deactivate(cmd, resource))
+		return EIO;
+
+	if (lock_flags & LCK_CLUSTER_VG) {
+		status = hold_unlock(resource);
+		if (status)
+			return errno;
+	}
+
+	return 0;
+}
+
+const char *do_lock_query(char *resource)
+{
+	int mode;
+	const char *type = NULL;
+
+	mode = get_current_lock(resource);
+	switch (mode) {
+		case LKM_NLMODE: type = "NL"; break;
+		case LKM_CRMODE: type = "CR"; break;
+		case LKM_CWMODE: type = "CW"; break;
+		case LKM_PRMODE: type = "PR"; break;
+		case LKM_PWMODE: type = "PW"; break;
+		case LKM_EXMODE: type = "EX"; break;
+	}
+
+	DEBUGLOG("do_lock_query: resource '%s', mode %i (%s)\n", resource, mode, type ?: "?");
+
+	return type;
+}
+
+/* This is the LOCK_LV part that happens on all nodes in the cluster -
+   it is responsible for the interaction with device-mapper and LVM */
+int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	int status = 0;
+
+	DEBUGLOG("do_lock_lv: resource '%s', cmd = %s, flags = %s, memlock = %d\n",
+		 resource, decode_locking_cmd(command), decode_flags(lock_flags), memlock());
+
+	if (!cmd->config_valid || config_files_changed(cmd)) {
+		/* Reinitialise various settings inc. logging, filters */
+		if (do_refresh_cache()) {
+			log_error("Updated config file invalid. Aborting.");
+			return EINVAL;
+		}
+	}
+
+	pthread_mutex_lock(&lvm_lock);
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(1);
+
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(0);
+
+	cmd->partial_activation = (lock_flags & LCK_PARTIAL_MODE) ? 1 : 0;
+
+	/* clvmd should never try to read suspended device */
+	init_ignore_suspended_devices(1);
+
+	switch (command & LCK_MASK) {
+	case LCK_LV_EXCLUSIVE:
+		status = do_activate_lv(resource, lock_flags, LKM_EXMODE);
+		break;
+
+	case LCK_LV_SUSPEND:
+		status = do_suspend_lv(resource, lock_flags);
+		if (!status)
+			suspended++;
+		break;
+
+	case LCK_UNLOCK:
+	case LCK_LV_RESUME:	/* if active */
+		status = do_resume_lv(resource, lock_flags);
+		if (!status)
+			suspended--;
+		break;
+
+	case LCK_LV_ACTIVATE:
+		status = do_activate_lv(resource, lock_flags, LKM_CRMODE);
+		break;
+
+	case LCK_LV_DEACTIVATE:
+		status = do_deactivate_lv(resource, lock_flags);
+		break;
+
+	default:
+		DEBUGLOG("Invalid LV command 0x%x\n", command);
+		status = EINVAL;
+		break;
+	}
+
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(0);
+
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(DEFAULT_DMEVENTD_MONITOR);
+
+	cmd->partial_activation = 0;
+
+	/* clean the pool for another command */
+	dm_pool_empty(cmd->mem);
+	pthread_mutex_unlock(&lvm_lock);
+
+	DEBUGLOG("Command return is %d, memlock is %d\n", status, memlock());
+	return status;
+}
+
+/* Functions to do on the local node only BEFORE the cluster-wide stuff above happens */
+int pre_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	/* Nearly all the stuff happens cluster-wide. Apart from SUSPEND. Here we get the
+	   lock out on this node (because we are the node modifying the metadata)
+	   before suspending cluster-wide.
+	   LKF_CONVERT is used always, local node is going to modify metadata
+	 */
+	if ((command & (LCK_SCOPE_MASK | LCK_TYPE_MASK)) == LCK_LV_SUSPEND &&
+	    (lock_flags & LCK_CLUSTER_VG)) {
+		DEBUGLOG("pre_lock_lv: resource '%s', cmd = %s, flags = %s\n",
+			 resource, decode_locking_cmd(command), decode_flags(lock_flags));
+
+		if (hold_lock(resource, LKM_PWMODE, LKF_NOQUEUE | LKF_CONVERT))
+			return errno;
+	}
+	return 0;
+}
+
+/* Functions to do on the local node only AFTER the cluster-wide stuff above happens */
+int post_lock_lv(unsigned char command, unsigned char lock_flags,
+		 char *resource)
+{
+	int status;
+
+	/* Opposite of above, done on resume after a metadata update */
+	if ((command & (LCK_SCOPE_MASK | LCK_TYPE_MASK)) == LCK_LV_RESUME &&
+	    (lock_flags & LCK_CLUSTER_VG)) {
+		int oldmode;
+
+		DEBUGLOG
+		    ("post_lock_lv: resource '%s', cmd = %s, flags = %s\n",
+		     resource, decode_locking_cmd(command), decode_flags(lock_flags));
+
+		/* If the lock state is PW then restore it to what it was */
+		oldmode = get_current_lock(resource);
+		if (oldmode == LKM_PWMODE) {
+			struct lvinfo lvi;
+
+			pthread_mutex_lock(&lvm_lock);
+			status = lv_info_by_lvid(cmd, resource, &lvi, 0, 0);
+			pthread_mutex_unlock(&lvm_lock);
+			if (!status)
+				return EIO;
+
+			if (lvi.exists) {
+				if (hold_lock(resource, LKM_CRMODE, LKF_CONVERT))
+					return errno;
+			} else {
+				if (hold_unlock(resource))
+					return errno;
+			}
+		}
+	}
+	return 0;
+}
+
+/* Check if a VG is in use by LVM1 so we don't stomp on it */
+int do_check_lvm1(const char *vgname)
+{
+	int status;
+
+	status = check_lvm1_vg_inactive(cmd, vgname);
+
+	return status == 1 ? 0 : EBUSY;
+}
+
+int do_refresh_cache()
+{
+	DEBUGLOG("Refreshing context\n");
+	log_notice("Refreshing context");
+
+	pthread_mutex_lock(&lvm_lock);
+
+	if (!refresh_toolcontext(cmd)) {
+		pthread_mutex_unlock(&lvm_lock);
+		return -1;
+	}
+
+	init_full_scan_done(0);
+	init_ignore_suspended_devices(1);
+	lvmcache_label_scan(cmd, 2);
+	dm_pool_empty(cmd->mem);
+
+	pthread_mutex_unlock(&lvm_lock);
+
+	return 0;
+}
+
+
+/* Only called at gulm startup. Drop any leftover VG or P_orphan locks
+   that might be hanging around if we died for any reason
+*/
+static void drop_vg_locks()
+{
+	char vg[128];
+	char line[255];
+	FILE *vgs =
+	    popen
+	    ("lvm pvs  --config 'log{command_names=0 prefix=\"\"}' --nolocking --noheadings -o vg_name", "r");
+
+	sync_unlock("P_" VG_ORPHANS, LCK_EXCL);
+	sync_unlock("P_" VG_GLOBAL, LCK_EXCL);
+
+	if (!vgs)
+		return;
+
+	while (fgets(line, sizeof(line), vgs)) {
+		char *vgend;
+		char *vgstart;
+
+		if (line[strlen(line)-1] == '\n')
+			line[strlen(line)-1] = '\0';
+
+		vgstart = line + strspn(line, " ");
+		vgend = vgstart + strcspn(vgstart, " ");
+		*vgend = '\0';
+
+		if (strncmp(vgstart, "WARNING:", 8) == 0)
+			continue;
+
+		sprintf(vg, "V_%s", vgstart);
+		sync_unlock(vg, LCK_EXCL);
+
+	}
+	if (fclose(vgs))
+		DEBUGLOG("vgs fclose failed: %s\n", strerror(errno));
+}
+
+/*
+ * Handle VG lock - drop metadata or update lvmcache state
+ */
+void do_lock_vg(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	uint32_t lock_cmd = command;
+	char *vgname = resource + 2;
+
+	DEBUGLOG("do_lock_vg: resource '%s', cmd = %s, flags = %s, memlock = %d\n",
+		 resource, decode_locking_cmd(command), decode_flags(lock_flags), memlock());
+
+	/* P_#global causes a full cache refresh */
+	if (!strcmp(resource, "P_" VG_GLOBAL)) {
+		do_refresh_cache();
+		return;
+	}
+
+	lock_cmd &= (LCK_SCOPE_MASK | LCK_TYPE_MASK | LCK_HOLD);
+
+	/*
+	 * Check if LCK_CACHE should be set. All P_ locks except # are cache related.
+	 */
+	if (strncmp(resource, "P_#", 3) && !strncmp(resource, "P_", 2))
+		lock_cmd |= LCK_CACHE;
+
+	pthread_mutex_lock(&lvm_lock);
+	switch (lock_cmd) {
+		case LCK_VG_COMMIT:
+			DEBUGLOG("vg_commit notification for VG %s\n", vgname);
+			lvmcache_commit_metadata(vgname);
+			break;
+		case LCK_VG_REVERT:
+			DEBUGLOG("vg_revert notification for VG %s\n", vgname);
+			lvmcache_drop_metadata(vgname, 1);
+			break;
+		case LCK_VG_DROP_CACHE:
+		default:
+			DEBUGLOG("Invalidating cached metadata for VG %s\n", vgname);
+			lvmcache_drop_metadata(vgname, 0);
+	}
+	pthread_mutex_unlock(&lvm_lock);
+}
+
+/*
+ * Ideally, clvmd should be started before any LVs are active
+ * but this may not be the case...
+ * I suppose this also comes in handy if clvmd crashes, not that it would!
+ */
+static void *get_initial_state()
+{
+	char lv[64], vg[64], flags[25], vg_flags[25];
+	char uuid[65];
+	char line[255];
+	FILE *lvs =
+	    popen
+	    ("lvm lvs  --config 'log{command_names=0 prefix=\"\"}' --nolocking --noheadings -o vg_uuid,lv_uuid,lv_attr,vg_attr",
+	     "r");
+
+	if (!lvs)
+		return NULL;
+
+	while (fgets(line, sizeof(line), lvs)) {
+	        if (sscanf(line, "%s %s %s %s\n", vg, lv, flags, vg_flags) == 4) {
+
+			/* States: s:suspended a:active S:dropped snapshot I:invalid snapshot */
+		        if (strlen(vg) == 38 &&                         /* is is a valid UUID ? */
+			    (flags[4] == 'a' || flags[4] == 's') &&	/* is it active or suspended? */
+			    vg_flags[5] == 'c') {			/* is it clustered ? */
+				/* Convert hyphen-separated UUIDs into one */
+				memcpy(&uuid[0], &vg[0], 6);
+				memcpy(&uuid[6], &vg[7], 4);
+				memcpy(&uuid[10], &vg[12], 4);
+				memcpy(&uuid[14], &vg[17], 4);
+				memcpy(&uuid[18], &vg[22], 4);
+				memcpy(&uuid[22], &vg[27], 4);
+				memcpy(&uuid[26], &vg[32], 6);
+				memcpy(&uuid[32], &lv[0], 6);
+				memcpy(&uuid[38], &lv[7], 4);
+				memcpy(&uuid[42], &lv[12], 4);
+				memcpy(&uuid[46], &lv[17], 4);
+				memcpy(&uuid[50], &lv[22], 4);
+				memcpy(&uuid[54], &lv[27], 4);
+				memcpy(&uuid[58], &lv[32], 6);
+				uuid[64] = '\0';
+
+				DEBUGLOG("getting initial lock for %s\n", uuid);
+				hold_lock(uuid, LKM_CRMODE, LKF_NOQUEUE);
+			}
+		}
+	}
+	if (fclose(lvs))
+		DEBUGLOG("lvs fclose failed: %s\n", strerror(errno));
+	return NULL;
+}
+
+static void lvm2_log_fn(int level, const char *file, int line, int dm_errno,
+			const char *message)
+{
+
+	/* Send messages to the normal LVM2 logging system too,
+	   so we get debug output when it's asked for.
+ 	   We need to NULL the function ptr otherwise it will just call
+	   back into here! */
+	init_log_fn(NULL);
+	print_log(level, file, line, dm_errno, "%s", message);
+	init_log_fn(lvm2_log_fn);
+
+	/*
+	 * Ignore non-error messages, but store the latest one for returning
+	 * to the user.
+	 */
+	if (level != _LOG_ERR && level != _LOG_FATAL)
+		return;
+
+	strncpy(last_error, message, sizeof(last_error));
+	last_error[sizeof(last_error)-1] = '\0';
+}
+
+/* This checks some basic cluster-LVM configuration stuff */
+static void check_config()
+{
+	int locking_type;
+
+	locking_type = find_config_tree_int(cmd, "global/locking_type", 1);
+
+	if (locking_type == 3) /* compiled-in cluster support */
+		return;
+
+	if (locking_type == 2) { /* External library, check name */
+		const char *libname;
+
+		libname = find_config_tree_str(cmd, "global/locking_library",
+					  "");
+		if (strstr(libname, "liblvm2clusterlock.so"))
+			return;
+
+		log_error("Incorrect LVM locking library specified in lvm.conf, cluster operations may not work.");
+		return;
+	}
+	log_error("locking_type not set correctly in lvm.conf, cluster operations will not work.");
+}
+
+/* Backups up the LVM metadata if it's changed */
+void lvm_do_backup(const char *vgname)
+{
+	struct volume_group * vg;
+	int consistent = 0;
+
+	DEBUGLOG("Triggering backup of VG metadata for %s. suspended=%d\n", vgname, suspended);
+
+	pthread_mutex_lock(&lvm_lock);
+
+	vg = vg_read_internal(cmd, vgname, NULL /*vgid*/, &consistent);
+
+	if (vg && consistent)
+		check_current_backup(vg);
+	else
+		log_error("Error backing up metadata, can't find VG for group %s", vgname);
+
+	vg_release(vg);
+	dm_pool_empty(cmd->mem);
+
+	pthread_mutex_unlock(&lvm_lock);
+}
+
+/* Called to initialise the LVM context of the daemon */
+int init_lvm(int using_gulm)
+{
+	if (!(cmd = create_toolcontext(1, NULL))) {
+		log_error("Failed to allocate command context");
+		return 0;
+	}
+
+	if (stored_errno()) {
+		destroy_toolcontext(cmd);
+		return 0;
+	}
+
+	/* Use LOG_DAEMON for syslog messages instead of LOG_USER */
+	init_syslog(LOG_DAEMON);
+	openlog("clvmd", LOG_PID, LOG_DAEMON);
+	cmd->cmd_line = "clvmd";
+
+	/* Check lvm.conf is setup for cluster-LVM */
+	check_config();
+	init_ignore_suspended_devices(1);
+
+	/* Remove any non-LV locks that may have been left around */
+	if (using_gulm)
+		drop_vg_locks();
+
+	get_initial_state();
+
+	/* Trap log messages so we can pass them back to the user */
+	init_log_fn(lvm2_log_fn);
+
+	return 1;
+}
+
+void destroy_lvm(void)
+{
+	if (cmd)
+		destroy_toolcontext(cmd);
+	cmd = NULL;
+}

daemons/clvmd/lvm-functions.h

@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/* Functions in lvm-functions.c */
+
+#ifndef _LVM_FUNCTIONS_H
+#define _LVM_FUNCTIONS_H
+
+extern int pre_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+		       char *resource);
+extern int do_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+		      char *resource);
+extern const char *do_lock_query(char *resource);
+extern int post_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+			char *resource);
+extern int do_check_lvm1(const char *vgname);
+extern int do_refresh_cache(void);
+extern int init_lvm(int using_gulm);
+extern void destroy_lvm(void);
+extern void init_lvhash(void);
+extern void destroy_lvhash(void);
+extern void lvm_do_backup(const char *vgname);
+extern int hold_unlock(char *resource);
+extern int hold_lock(char *resource, int mode, int flags);
+extern char *get_last_lvm_error(void);
+extern void do_lock_vg(unsigned char command, unsigned char lock_flags,
+		      char *resource);
+
+#endif

daemons/clvmd/refresh_clvmd.c

@@ -0,0 +1,367 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Tell all clvmds in a cluster to refresh their toolcontext
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <stddef.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <errno.h>
+#include <unistd.h>
+#include <libdevmapper.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <limits.h>
+
+#include "clvm.h"
+#include "refresh_clvmd.h"
+
+typedef struct lvm_response {
+	char node[255];
+	char *response;
+	int status;
+	int len;
+} lvm_response_t;
+
+/*
+ * This gets stuck at the start of memory we allocate so we
+ * can sanity-check it at deallocation time
+ */
+#define LVM_SIGNATURE 0x434C564D
+
+static int _clvmd_sock = -1;
+
+/* Open connection to the clvm daemon */
+static int _open_local_sock(void)
+{
+	int local_socket;
+	struct sockaddr_un sockaddr;
+
+	/* Open local socket */
+	if ((local_socket = socket(PF_UNIX, SOCK_STREAM, 0)) < 0) {
+		fprintf(stderr, "Local socket creation failed: %s", strerror(errno));
+		return -1;
+	}
+
+	memset(&sockaddr, 0, sizeof(sockaddr));
+	memcpy(sockaddr.sun_path, CLVMD_SOCKNAME, sizeof(CLVMD_SOCKNAME));
+
+	sockaddr.sun_family = AF_UNIX;
+
+	if (connect(local_socket,(struct sockaddr *) &sockaddr,
+		    sizeof(sockaddr))) {
+		int saved_errno = errno;
+
+		fprintf(stderr, "connect() failed on local socket: %s\n",
+			  strerror(errno));
+		if (close(local_socket))
+			return -1;
+
+		errno = saved_errno;
+		return -1;
+	}
+
+	return local_socket;
+}
+
+/* Send a request and return the status */
+static int _send_request(const char *inbuf, int inlen, char **retbuf)
+{
+	char outbuf[PIPE_BUF];
+	struct clvm_header *outheader = (struct clvm_header *) outbuf;
+	int len;
+	int off;
+	int buflen;
+	int err;
+
+	/* Send it to CLVMD */
+ rewrite:
+	if ( (err = write(_clvmd_sock, inbuf, inlen)) != inlen) {
+	        if (err == -1 && errno == EINTR)
+		        goto rewrite;
+		fprintf(stderr, "Error writing data to clvmd: %s", strerror(errno));
+		return 0;
+	}
+
+	/* Get the response */
+ reread:
+	if ((len = read(_clvmd_sock, outbuf, sizeof(struct clvm_header))) < 0) {
+	        if (errno == EINTR)
+		        goto reread;
+		fprintf(stderr, "Error reading data from clvmd: %s", strerror(errno));
+		return 0;
+	}
+
+	if (len == 0) {
+		fprintf(stderr, "EOF reading CLVMD");
+		errno = ENOTCONN;
+		return 0;
+	}
+
+	/* Allocate buffer */
+	buflen = len + outheader->arglen;
+	*retbuf = dm_malloc(buflen);
+	if (!*retbuf) {
+		errno = ENOMEM;
+		return 0;
+	}
+
+	/* Copy the header */
+	memcpy(*retbuf, outbuf, len);
+	outheader = (struct clvm_header *) *retbuf;
+
+	/* Read the returned values */
+	off = 1;		/* we've already read the first byte */
+	while (off <= outheader->arglen && len > 0) {
+		len = read(_clvmd_sock, outheader->args + off,
+			   buflen - off - offsetof(struct clvm_header, args));
+		if (len > 0)
+			off += len;
+	}
+
+	/* Was it an error ? */
+	if (outheader->status != 0) {
+		errno = outheader->status;
+
+		/* Only return an error here if there are no node-specific
+		   errors present in the message that might have more detail */
+		if (!(outheader->flags & CLVMD_FLAG_NODEERRS)) {
+			fprintf(stderr, "cluster request failed: %s\n", strerror(errno));
+			return 0;
+		}
+
+	}
+
+	return 1;
+}
+
+/* Build the structure header and parse-out wildcard node names */
+static void _build_header(struct clvm_header *head, int cmd, const char *node,
+			  int len)
+{
+	head->cmd = cmd;
+	head->status = 0;
+	head->flags = 0;
+	head->clientid = 0;
+	head->arglen = len;
+
+	if (node) {
+		/*
+		 * Allow a couple of special node names:
+		 * "*" for all nodes,
+		 * "." for the local node only
+		 */
+		if (strcmp(node, "*") == 0) {
+			head->node[0] = '\0';
+		} else if (strcmp(node, ".") == 0) {
+			head->node[0] = '\0';
+			head->flags = CLVMD_FLAG_LOCAL;
+		} else
+			strcpy(head->node, node);
+	} else
+		head->node[0] = '\0';
+}
+
+/*
+ * Send a message to a(or all) node(s) in the cluster and wait for replies
+ */
+static int _cluster_request(char cmd, const char *node, void *data, int len,
+			   lvm_response_t ** response, int *num)
+{
+	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1];
+	char *inptr;
+	char *retbuf = NULL;
+	int status;
+	int i;
+	int num_responses = 0;
+	struct clvm_header *head = (struct clvm_header *) outbuf;
+	lvm_response_t *rarray;
+
+	*num = 0;
+
+	if (_clvmd_sock == -1)
+		_clvmd_sock = _open_local_sock();
+
+	if (_clvmd_sock == -1)
+		return 0;
+
+	_build_header(head, cmd, node, len);
+	memcpy(head->node + strlen(head->node) + 1, data, len);
+
+	status = _send_request(outbuf, sizeof(struct clvm_header) +
+			      strlen(head->node) + len, &retbuf);
+	if (!status)
+		goto out;
+
+	/* Count the number of responses we got */
+	head = (struct clvm_header *) retbuf;
+	inptr = head->args;
+	while (inptr[0]) {
+		num_responses++;
+		inptr += strlen(inptr) + 1;
+		inptr += sizeof(int);
+		inptr += strlen(inptr) + 1;
+	}
+
+	/*
+	 * Allocate response array.
+	 * With an extra pair of INTs on the front to sanity
+	 * check the pointer when we are given it back to free
+	 */
+	*response = dm_malloc(sizeof(lvm_response_t) * num_responses +
+			    sizeof(int) * 2);
+	if (!*response) {
+		errno = ENOMEM;
+		status = 0;
+		goto out;
+	}
+
+	rarray = *response;
+
+	/* Unpack the response into an lvm_response_t array */
+	inptr = head->args;
+	i = 0;
+	while (inptr[0]) {
+		strcpy(rarray[i].node, inptr);
+		inptr += strlen(inptr) + 1;
+
+		memcpy(&rarray[i].status, inptr, sizeof(int));
+		inptr += sizeof(int);
+
+		rarray[i].response = dm_malloc(strlen(inptr) + 1);
+		if (rarray[i].response == NULL) {
+			/* Free up everything else and return error */
+			int j;
+			for (j = 0; j < i; j++)
+				dm_free(rarray[i].response);
+			free(*response);
+			errno = ENOMEM;
+			status = -1;
+			goto out;
+		}
+
+		strcpy(rarray[i].response, inptr);
+		rarray[i].len = strlen(inptr);
+		inptr += strlen(inptr) + 1;
+		i++;
+	}
+	*num = num_responses;
+	*response = rarray;
+
+      out:
+	if (retbuf)
+		dm_free(retbuf);
+
+	return status;
+}
+
+/* Free reply array */
+static int _cluster_free_request(lvm_response_t * response, int num)
+{
+	int i;
+
+	for (i = 0; i < num; i++) {
+		dm_free(response[i].response);
+	}
+
+	dm_free(response);
+
+	return 1;
+}
+
+int refresh_clvmd()
+{
+	int num_responses;
+	char args[1]; // No args really.
+	lvm_response_t *response;
+	int saved_errno;
+	int status;
+	int i;
+
+	status = _cluster_request(CLVMD_CMD_REFRESH, "*", args, 0, &response, &num_responses);
+
+	/* If any nodes were down then display them and return an error */
+	for (i = 0; i < num_responses; i++) {
+		if (response[i].status == EHOSTDOWN) {
+			fprintf(stderr, "clvmd not running on node %s",
+				  response[i].node);
+			status = 0;
+			errno = response[i].status;
+		} else if (response[i].status) {
+			fprintf(stderr, "Error resetting node %s: %s",
+				  response[i].node,
+				  response[i].response[0] ?
+				  	response[i].response :
+				  	strerror(response[i].status));
+			status = 0;
+			errno = response[i].status;
+		}
+	}
+
+	saved_errno = errno;
+	_cluster_free_request(response, num_responses);
+	errno = saved_errno;
+
+	return status;
+}
+
+int debug_clvmd(int level, int clusterwide)
+{
+	int num_responses;
+	char args[1];
+	const char *nodes;
+	lvm_response_t *response;
+	int saved_errno;
+	int status;
+	int i;
+
+	args[0] = level;
+	if (clusterwide)
+		nodes = "*";
+	else
+		nodes = ".";
+
+	status = _cluster_request(CLVMD_CMD_SET_DEBUG, nodes, args, 1, &response, &num_responses);
+
+	/* If any nodes were down then display them and return an error */
+	for (i = 0; i < num_responses; i++) {
+		if (response[i].status == EHOSTDOWN) {
+			fprintf(stderr, "clvmd not running on node %s",
+				  response[i].node);
+			status = 0;
+			errno = response[i].status;
+		} else if (response[i].status) {
+			fprintf(stderr, "Error setting debug on node %s: %s",
+				  response[i].node,
+				  response[i].response[0] ?
+				  	response[i].response :
+				  	strerror(response[i].status));
+			status = 0;
+			errno = response[i].status;
+		}
+	}
+
+	saved_errno = errno;
+	_cluster_free_request(response, num_responses);
+	errno = saved_errno;
+
+	return status;
+}

daemons/clvmd/refresh_clvmd.h

@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+
+int refresh_clvmd(void);
+int debug_clvmd(int level, int clusterwide);
+

daemons/clvmd/tcp-comms.c

@@ -0,0 +1,511 @@
+/*
+ *  Copyright (C) 2002-2003 Sistina Software, Inc. All rights reserved.
+ *  Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * This provides the inter-clvmd communications for a system without CMAN.
+ * There is a listening TCP socket which accepts new connections in the
+ * normal way.
+ * It can also make outgoing connnections to the other clvmd nodes.
+ */
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <syslog.h>
+#include <netdb.h>
+#include <assert.h>
+#include <libdevmapper.h>
+
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "clvmd.h"
+#include "clvmd-gulm.h"
+
+#define DEFAULT_TCP_PORT 21064
+
+static int listen_fd = -1;
+static int tcp_port;
+struct dm_hash_table *sock_hash;
+
+static int get_our_ip_address(char *addr, int *family);
+static int read_from_tcpsock(struct local_client *fd, char *buf, int len, char *csid,
+			     struct local_client **new_client);
+
+/* Called by init_cluster() to open up the listening socket */
+int init_comms(unsigned short port)
+{
+    struct sockaddr_in6 addr;
+
+    sock_hash = dm_hash_create(100);
+    tcp_port = port ? : DEFAULT_TCP_PORT;
+
+    listen_fd = socket(AF_INET6, SOCK_STREAM, 0);
+
+    if (listen_fd < 0)
+    {
+	return -1;
+    }
+    else
+    {
+	int one = 1;
+	setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &one, sizeof(int));
+	setsockopt(listen_fd, SOL_SOCKET, SO_KEEPALIVE, &one, sizeof(int));
+    }
+
+    memset(&addr, 0, sizeof(addr)); // Bind to INADDR_ANY
+    addr.sin6_family = AF_INET6;
+    addr.sin6_port = htons(tcp_port);
+
+    if (bind(listen_fd, (struct sockaddr *)&addr, sizeof(addr)) < 0)
+    {
+	DEBUGLOG("Can't bind to port: %s\n", strerror(errno));
+	syslog(LOG_ERR, "Can't bind to port %d, is clvmd already running ?", tcp_port);
+	close(listen_fd);
+	return -1;
+    }
+
+    listen(listen_fd, 5);
+
+    /* Set Close-on-exec */
+    fcntl(listen_fd, F_SETFD, 1);
+
+    return 0;
+}
+
+void tcp_remove_client(const char *c_csid)
+{
+    struct local_client *client;
+    char csid[GULM_MAX_CSID_LEN];
+    unsigned int i;
+    memcpy(csid, c_csid, sizeof csid);
+    DEBUGLOG("tcp_remove_client\n");
+
+    /* Don't actually close the socket here - that's the
+       job of clvmd.c whch will do the job when it notices the
+       other end has gone. We just need to remove the client(s) from
+       the hash table so we don't try to use it for sending any more */
+    for (i = 0; i < 2; i++)
+    {
+	client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+	if (client)
+	{
+	    dm_hash_remove_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+	    client->removeme = 1;
+	    close(client->fd);
+	}
+	/* Look for a mangled one too, on the 2nd iteration. */
+	csid[0] ^= 0x80;
+    }
+}
+
+int alloc_client(int fd, const char *c_csid, struct local_client **new_client)
+{
+    struct local_client *client;
+    char csid[GULM_MAX_CSID_LEN];
+    memcpy(csid, c_csid, sizeof csid);
+
+    DEBUGLOG("alloc_client %d csid = %s\n", fd, print_csid(csid));
+
+    /* Create a local_client and return it */
+    client = malloc(sizeof(struct local_client));
+    if (!client)
+    {
+	DEBUGLOG("malloc failed\n");
+	return -1;
+    }
+
+    memset(client, 0, sizeof(struct local_client));
+    client->fd = fd;
+    client->type = CLUSTER_DATA_SOCK;
+    client->callback = read_from_tcpsock;
+    if (new_client)
+	*new_client = client;
+
+    /* Add to our list of node sockets */
+    if (dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN))
+    {
+	DEBUGLOG("alloc_client mangling CSID for second connection\n");
+	/* This is a duplicate connection but we can't close it because
+	   the other end may already have started sending.
+	   So, we mangle the IP address and keep it, all sending will
+	   go out of the main FD
+	*/
+	csid[0] ^= 0x80;
+	client->bits.net.flags = 1; /* indicate mangled CSID */
+
+        /* If it still exists then kill the connection as we should only
+           ever have one incoming connection from each node */
+        if (dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN))
+        {
+	    DEBUGLOG("Multiple incoming connections from node\n");
+            syslog(LOG_ERR, " Bogus incoming connection from %d.%d.%d.%d\n", csid[0],csid[1],csid[2],csid[3]);
+
+	    free(client);
+            errno = ECONNREFUSED;
+            return -1;
+        }
+    }
+    dm_hash_insert_binary(sock_hash, csid, GULM_MAX_CSID_LEN, client);
+
+    return 0;
+}
+
+int get_main_gulm_cluster_fd()
+{
+    return listen_fd;
+}
+
+
+/* Read on main comms (listen) socket, accept it */
+int cluster_fd_gulm_callback(struct local_client *fd, char *buf, int len, const char *csid,
+			struct local_client **new_client)
+{
+    int newfd;
+    struct sockaddr_in6 addr;
+    socklen_t addrlen = sizeof(addr);
+    int status;
+    char name[GULM_MAX_CLUSTER_MEMBER_NAME_LEN];
+
+    DEBUGLOG("cluster_fd_callback\n");
+    *new_client = NULL;
+    newfd = accept(listen_fd, (struct sockaddr *)&addr, &addrlen);
+
+    DEBUGLOG("cluster_fd_callback, newfd=%d (errno=%d)\n", newfd, errno);
+    if (!newfd)
+    {
+	syslog(LOG_ERR, "error in accept: %m");
+	errno = EAGAIN;
+	return -1; /* Don't return an error or clvmd will close the listening FD */
+    }
+
+    /* Check that the client is a member of the cluster
+       and reject if not.
+    */
+    if (gulm_name_from_csid((char *)&addr.sin6_addr, name) < 0)
+    {
+	syslog(LOG_ERR, "Got connect from non-cluster node %s\n",
+	       print_csid((char *)&addr.sin6_addr));
+	DEBUGLOG("Got connect from non-cluster node %s\n",
+		 print_csid((char *)&addr.sin6_addr));
+	close(newfd);
+
+	errno = EAGAIN;
+	return -1;
+    }
+
+    status = alloc_client(newfd, (char *)&addr.sin6_addr, new_client);
+    if (status)
+    {
+	DEBUGLOG("cluster_fd_callback, alloc_client failed, status = %d\n", status);
+	close(newfd);
+	/* See above... */
+	errno = EAGAIN;
+	return -1;
+    }
+    DEBUGLOG("cluster_fd_callback, returning %d, %p\n", newfd, *new_client);
+    return newfd;
+}
+
+/* Try to get at least 'len' bytes from the socket */
+static int really_read(int fd, char *buf, int len)
+{
+	int got, offset;
+
+	got = offset = 0;
+
+	do {
+		got = read(fd, buf+offset, len-offset);
+		DEBUGLOG("really_read. got %d bytes\n", got);
+		offset += got;
+	} while (got > 0 && offset < len);
+
+	if (got < 0)
+		return got;
+	else
+		return offset;
+}
+
+
+static int read_from_tcpsock(struct local_client *client, char *buf, int len, char *csid,
+			     struct local_client **new_client)
+{
+    struct sockaddr_in6 addr;
+    socklen_t slen = sizeof(addr);
+    struct clvm_header *header = (struct clvm_header *)buf;
+    int status;
+    uint32_t arglen;
+
+    DEBUGLOG("read_from_tcpsock fd %d\n", client->fd);
+    *new_client = NULL;
+
+    /* Get "csid" */
+    getpeername(client->fd, (struct sockaddr *)&addr, &slen);
+    memcpy(csid, &addr.sin6_addr, GULM_MAX_CSID_LEN);
+
+    /* Read just the header first, then get the rest if there is any.
+     * Stream sockets, sigh.
+     */
+    status = really_read(client->fd, buf, sizeof(struct clvm_header));
+    if (status > 0)
+    {
+	    int status2;
+
+	    arglen = ntohl(header->arglen);
+
+	    /* Get the rest */
+	    if (arglen && arglen < GULM_MAX_CLUSTER_MESSAGE)
+	    {
+		    status2 = really_read(client->fd, buf+status, arglen);
+		    if (status2 > 0)
+			    status += status2;
+		    else
+			    status = status2;
+	    }
+    }
+
+    DEBUGLOG("read_from_tcpsock, status = %d(errno = %d)\n", status, errno);
+
+    /* Remove it from the hash table if there's an error, clvmd will
+       remove the socket from its lists and free the client struct */
+    if (status == 0 ||
+	(status < 0 && errno != EAGAIN && errno != EINTR))
+    {
+	char remcsid[GULM_MAX_CSID_LEN];
+
+	memcpy(remcsid, csid, GULM_MAX_CSID_LEN);
+	close(client->fd);
+
+	/* If the csid was mangled, then make sure we remove the right entry */
+	if (client->bits.net.flags)
+	    remcsid[0] ^= 0x80;
+	dm_hash_remove_binary(sock_hash, remcsid, GULM_MAX_CSID_LEN);
+
+	/* Tell cluster manager layer */
+	add_down_node(remcsid);
+    }
+    else {
+	    gulm_add_up_node(csid);
+	    /* Send it back to clvmd */
+	    process_message(client, buf, status, csid);
+    }
+    return status;
+}
+
+int gulm_connect_csid(const char *csid, struct local_client **newclient)
+{
+    int fd;
+    struct sockaddr_in6 addr;
+    int status;
+    int one = 1;
+
+    DEBUGLOG("Connecting socket\n");
+    fd = socket(PF_INET6, SOCK_STREAM, 0);
+
+    if (fd < 0)
+    {
+	syslog(LOG_ERR, "Unable to create new socket: %m");
+	return -1;
+    }
+
+    addr.sin6_family = AF_INET6;
+    memcpy(&addr.sin6_addr, csid, GULM_MAX_CSID_LEN);
+    addr.sin6_port = htons(tcp_port);
+
+    DEBUGLOG("Connecting socket %d\n", fd);
+    if (connect(fd, (struct sockaddr *)&addr, sizeof(struct sockaddr_in6)) < 0)
+    {
+	/* "Connection refused" is "normal" because clvmd may not yet be running
+	 * on that node.
+	 */
+	if (errno != ECONNREFUSED)
+	{
+	    syslog(LOG_ERR, "Unable to connect to remote node: %m");
+	}
+	DEBUGLOG("Unable to connect to remote node: %s\n", strerror(errno));
+	close(fd);
+	return -1;
+    }
+
+    /* Set Close-on-exec */
+    fcntl(fd, F_SETFD, 1);
+    setsockopt(fd, SOL_SOCKET, SO_KEEPALIVE, &one, sizeof(int));
+
+    status = alloc_client(fd, csid, newclient);
+    if (status)
+	close(fd);
+    else
+	add_client(*newclient);
+
+    /* If we can connect to it, it must be running a clvmd */
+    gulm_add_up_node(csid);
+    return status;
+}
+
+/* Send a message to a known CSID */
+static int tcp_send_message(void *buf, int msglen, const char *csid, const char *errtext)
+{
+    int status;
+    struct local_client *client;
+    char ourcsid[GULM_MAX_CSID_LEN];
+
+    assert(csid);
+
+    DEBUGLOG("tcp_send_message, csid = %s, msglen = %d\n", print_csid(csid), msglen);
+
+    /* Don't connect to ourself */
+    get_our_gulm_csid(ourcsid);
+    if (memcmp(csid, ourcsid, GULM_MAX_CSID_LEN) == 0)
+	return msglen;
+
+    client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+    if (!client)
+    {
+	status = gulm_connect_csid(csid, &client);
+	if (status)
+	    return -1;
+    }
+    DEBUGLOG("tcp_send_message, fd = %d\n", client->fd);
+
+    return write(client->fd, buf, msglen);
+}
+
+
+int gulm_cluster_send_message(void *buf, int msglen, const char *csid, const char *errtext)
+{
+    int status=0;
+
+    DEBUGLOG("cluster send message, csid = %p, msglen = %d\n", csid, msglen);
+
+    /* If csid is NULL then send to all known (not just connected) nodes */
+    if (!csid)
+    {
+	void *context = NULL;
+	char loop_csid[GULM_MAX_CSID_LEN];
+
+	/* Loop round all gulm-known nodes */
+	while (get_next_node_csid(&context, loop_csid))
+	{
+	    status = tcp_send_message(buf, msglen, loop_csid, errtext);
+	    if (status == 0 ||
+		(status < 0 && (errno == EAGAIN || errno == EINTR)))
+		break;
+	}
+    }
+    else
+    {
+
+	status = tcp_send_message(buf, msglen, csid, errtext);
+    }
+    return status;
+}
+
+/* To get our own IP address we get the locally bound address of the
+   socket that's talking to GULM in the assumption(eek) that it will
+   be on the "right" network in a multi-homed system */
+static int get_our_ip_address(char *addr, int *family)
+{
+	struct utsname info;
+
+	uname(&info);
+	get_ip_address(info.nodename, addr);
+
+	return 0;
+}
+
+/* Public version of above for those that don't care what protocol
+   we're using */
+void get_our_gulm_csid(char *csid)
+{
+    static char our_csid[GULM_MAX_CSID_LEN];
+    static int got_csid = 0;
+
+    if (!got_csid)
+    {
+	int family;
+
+	memset(our_csid, 0, sizeof(our_csid));
+	if (get_our_ip_address(our_csid, &family))
+	{
+	    got_csid = 1;
+	}
+    }
+    memcpy(csid, our_csid, GULM_MAX_CSID_LEN);
+}
+
+static void map_v4_to_v6(struct in_addr *ip4, struct in6_addr *ip6)
+{
+   ip6->s6_addr32[0] = 0;
+   ip6->s6_addr32[1] = 0;
+   ip6->s6_addr32[2] = htonl(0xffff);
+   ip6->s6_addr32[3] = ip4->s_addr;
+}
+
+/* Get someone else's IP address from DNS */
+int get_ip_address(const char *node, char *addr)
+{
+    struct hostent *he;
+
+    memset(addr, 0, GULM_MAX_CSID_LEN);
+
+    // TODO: what do we do about multi-homed hosts ???
+    // CCSs ip_interfaces solved this but some bugger removed it.
+
+    /* Try IPv6 first. The man page for gethostbyname implies that
+       it will lookup ip6 & ip4 names, but it seems not to */
+    he = gethostbyname2(node, AF_INET6);
+    if (he)
+    {
+	memcpy(addr, he->h_addr_list[0],
+	       he->h_length);
+    }
+    else
+    {
+	he = gethostbyname2(node, AF_INET);
+	if (!he)
+	    return -1;
+	map_v4_to_v6((struct in_addr *)he->h_addr_list[0], (struct in6_addr *)addr);
+    }
+
+    return 0;
+}
+
+char *print_csid(const char *csid)
+{
+    static char buf[128];
+    int *icsid = (int *)csid;
+
+    sprintf(buf, "[%x.%x.%x.%x]",
+	    icsid[0],icsid[1],icsid[2],icsid[3]);
+
+    return buf;
+}

daemons/clvmd/tcp-comms.h

@@ -0,0 +1,13 @@
+#include <netinet/in.h>
+
+#define GULM_MAX_CLUSTER_MESSAGE 1600
+#define GULM_MAX_CSID_LEN sizeof(struct in6_addr)
+#define GULM_MAX_CLUSTER_MEMBER_NAME_LEN 128
+
+extern int init_comms(unsigned short);
+extern char *print_csid(const char *);
+int get_main_gulm_cluster_fd(void);
+int cluster_fd_gulm_callback(struct local_client *fd, char *buf, int len, const char *csid, struct local_client **new_client);
+int gulm_cluster_send_message(void *buf, int msglen, const char *csid, const char *errtext);
+void get_our_gulm_csid(char *csid);
+int gulm_connect_csid(const char *csid, struct local_client **newclient);

daemons/cmirrord/Makefile.in

@@ -0,0 +1,40 @@
+#
+# Copyright (C) 2009 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+CPG_LIBS = @CPG_LIBS@
+CPG_CFLAGS = @CPG_CFLAGS@
+SACKPT_LIBS = @SACKPT_LIBS@
+SACKPT_CFLAGS = @SACKPT_CFLAGS@
+
+SOURCES = clogd.c cluster.c compat.c functions.c link_mon.c local.c logging.c
+
+TARGETS = cmirrord
+
+include ../../make.tmpl
+
+LIBS += -ldevmapper
+LMLIBS += $(CPG_LIBS) $(SACKPT_LIBS)
+CFLAGS += $(CPG_CFLAGS) $(SACKPT_CFLAGS)
+
+cmirrord: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
+	$(CC) $(CFLAGS) $(LDFLAGS) -o cmirrord $(OBJECTS) \
+		$(LVMLIBS) $(LMLIBS) $(LIBS)
+
+install: $(TARGETS)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) cmirrord \
+		$(usrsbindir)/cmirrord

daemons/cmirrord/clogd.c

@@ -0,0 +1,267 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#include "logging.h"
+#include "common.h"
+#include "functions.h"
+#include "link_mon.h"
+#include "local.h"
+
+#include <errno.h>
+#include <fcntl.h>
+#include <sys/socket.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+static volatile sig_atomic_t exit_now = 0;
+/* FIXME Review signal handling.  Should be volatile sig_atomic_t */
+static sigset_t signal_mask;
+static volatile sig_atomic_t signal_received;
+
+static void process_signals(void);
+static void daemonize(void);
+static void init_all(void);
+static void cleanup_all(void);
+
+int main(int argc __attribute((unused)), char *argv[] __attribute((unused)))
+{
+	daemonize();
+
+	init_all();
+
+	/* Parent can now exit, we're ready to handle requests */
+	kill(getppid(), SIGTERM);
+
+	LOG_PRINT("Starting cmirrord:");
+	LOG_PRINT(" Built: "__DATE__" "__TIME__"\n");
+	LOG_DBG(" Compiled with debugging.");
+
+	while (!exit_now) {
+		links_monitor();
+
+		links_issue_callbacks();
+
+		process_signals();
+	}
+	exit(EXIT_SUCCESS);
+}
+
+/*
+ * parent_exit_handler: exit the parent
+ * @sig: the signal
+ *
+ */
+static void parent_exit_handler(int sig __attribute((unused)))
+{
+	exit_now = 1;
+}
+
+/*
+ * create_lockfile - create and lock a lock file
+ * @lockfile: location of lock file
+ *
+ * Returns: 0 on success, -1 otherwise
+ */
+static int create_lockfile(const char *lockfile)
+{
+	int fd;
+	struct flock lock;
+	char buffer[50];
+
+	if((fd = open(lockfile, O_CREAT | O_WRONLY,
+		      (S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH))) < 0)
+		return -errno;
+
+	lock.l_type = F_WRLCK;
+	lock.l_start = 0;
+	lock.l_whence = SEEK_SET;
+	lock.l_len = 0;
+
+	if (fcntl(fd, F_SETLK, &lock) < 0) {
+		close(fd);
+		return -errno;
+	}
+
+	if (ftruncate(fd, 0) < 0) {
+		close(fd);
+		return -errno;
+	}
+
+	sprintf(buffer, "%d\n", getpid());
+
+	/* FIXME Handle other non-error returns without aborting */
+	if (write(fd, buffer, strlen(buffer)) < strlen(buffer)){
+		close(fd);
+		unlink(lockfile);
+		return -errno;
+	}
+
+	return 0;
+}
+
+static void sig_handler(int sig)
+{
+	/* FIXME Races - don't touch signal_mask here. */
+	sigaddset(&signal_mask, sig);
+	signal_received = 1;
+}
+
+static void process_signal(int sig){
+	int r = 0;
+
+	switch(sig) {
+	case SIGINT:
+	case SIGQUIT:
+	case SIGTERM:
+	case SIGHUP:
+		r += log_status();
+		break;
+	case SIGUSR1:
+	case SIGUSR2:
+		log_debug();
+		/*local_debug();*/
+		cluster_debug();
+		return;
+	default:
+		LOG_PRINT("Unknown signal received... ignoring");
+		return;
+	}
+
+	if (!r) {
+		LOG_DBG("No current cluster logs... safe to exit.");
+		cleanup_all();
+		exit(EXIT_SUCCESS);
+	}
+
+	LOG_ERROR("Cluster logs exist.  Refusing to exit.");
+}
+
+static void process_signals(void)
+{
+	int x;
+
+	if (!signal_received)
+		return;
+
+	signal_received = 0;
+
+	for (x = 1; x < _NSIG; x++) {
+		if (sigismember(&signal_mask, x)) {
+			sigdelset(&signal_mask, x);
+			process_signal(x);
+		}
+	}
+}
+
+/*
+ * daemonize
+ *
+ * Performs the steps necessary to become a daemon.
+ */
+static void daemonize(void)
+{
+	int pid;
+	int status;
+
+	signal(SIGTERM, &parent_exit_handler);
+
+	pid = fork();
+
+	if (pid < 0) {
+		LOG_ERROR("Unable to fork()");
+		exit(EXIT_FAILURE);
+	}
+
+	if (pid) {
+		/* Parent waits here for child to get going */
+		while (!waitpid(pid, &status, WNOHANG) && !exit_now);
+		if (exit_now)
+			exit(EXIT_SUCCESS);
+
+		switch (WEXITSTATUS(status)) {
+		case EXIT_LOCKFILE:
+			LOG_ERROR("Failed to create lockfile");
+			LOG_ERROR("Process already running?");
+			break;
+		case EXIT_KERNEL_SOCKET:
+			LOG_ERROR("Unable to create netlink socket");
+			break;
+		case EXIT_KERNEL_BIND:
+			LOG_ERROR("Unable to bind to netlink socket");
+			break;
+		case EXIT_KERNEL_SETSOCKOPT:
+			LOG_ERROR("Unable to setsockopt on netlink socket");
+			break;
+		case EXIT_CLUSTER_CKPT_INIT:
+			LOG_ERROR("Unable to initialize checkpoint service");
+			LOG_ERROR("Has the cluster infrastructure been started?");
+			break;
+		case EXIT_FAILURE:
+			LOG_ERROR("Failed to start: Generic error");
+			break;
+		default:
+			LOG_ERROR("Failed to start: Unknown error");
+			break;
+		}
+		exit(EXIT_FAILURE);
+	}
+
+	setsid();
+	chdir("/");
+	umask(0);
+
+	close(0); close(1); close(2);
+	open("/dev/null", O_RDONLY); /* reopen stdin */
+	open("/dev/null", O_WRONLY); /* reopen stdout */
+	open("/dev/null", O_WRONLY); /* reopen stderr */
+
+	LOG_OPEN("cmirrord", LOG_PID, LOG_DAEMON);
+
+	if (create_lockfile(CMIRRORD_PIDFILE))
+		exit(EXIT_LOCKFILE);
+
+	/* FIXME Replace with sigaction. (deprecated) */
+	signal(SIGINT, &sig_handler);
+	signal(SIGQUIT, &sig_handler);
+	signal(SIGTERM, &sig_handler);
+	signal(SIGHUP, &sig_handler);
+	signal(SIGPIPE, SIG_IGN);
+	signal(SIGUSR1, &sig_handler);
+	signal(SIGUSR2, &sig_handler);
+	sigemptyset(&signal_mask);
+	signal_received = 0;
+}
+
+/*
+ * init_all
+ *
+ * Initialize modules.  Exit on failure.
+ */
+static void init_all(void)
+{
+	int r;
+
+	if ((r = init_local()) ||
+	    (r = init_cluster())) {
+		exit(r);
+	}
+}
+
+/*
+ * cleanup_all
+ *
+ * Clean up before exiting
+ */
+static void cleanup_all(void)
+{
+	cleanup_local();
+	cleanup_cluster();
+}

daemons/cmirrord/cluster.h

@@ -0,0 +1,76 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#ifndef _LVM_CLOG_CLUSTER_H
+#define _LVM_CLOG_CLUSTER_H
+
+#include "dm-log-userspace.h"
+#include "libdevmapper.h"
+
+#define DM_ULOG_RESPONSE 0x1000U /* in last byte of 32-bit value */
+#define DM_ULOG_CHECKPOINT_READY 21
+#define DM_ULOG_MEMBER_JOIN      22
+
+/*
+ * There is other information in addition to what can
+ * be found in the dm_ulog_request structure that we
+ * need for processing.  'clog_request' is the wrapping
+ * structure we use to make the additional fields
+ * available.
+ */
+struct clog_request {
+	/*
+	 * If we don't use a union, the structure size will
+	 * vary between 32-bit and 64-bit machines.  So, we
+	 * pack two 64-bit version numbers in there to force
+	 * the size of the structure to be the same.
+	 *
+	 * The two version numbers also help us with endian
+	 * issues.  The first is always little endian, while
+	 * the second is in native format of the sending
+	 * machine.  If the two are equal, there is no need
+	 * to do endian conversions.
+	 */
+	union {
+		uint64_t version[2]; /* LE version and native version */
+		struct dm_list list;
+	} u;
+
+	/*
+	 * 'originator' is the machine from which the requests
+	 * was made.
+	 */
+	uint32_t originator;
+
+	/*
+	 * 'pit_server' is the "point-in-time" server for the
+	 * request.  (I.e.  The machine that was the server at
+	 * the time the request was issued - only important during
+	 * startup.
+	 */
+	uint32_t pit_server;
+
+	/*
+	 * The request from the kernel that is being processed
+	 */
+	struct dm_ulog_request u_rq;
+};
+
+int init_cluster(void);
+void cleanup_cluster(void);
+void cluster_debug(void);
+
+int create_cluster_cpg(char *uuid, uint64_t luid);
+int destroy_cluster_cpg(char *uuid);
+
+int cluster_send(struct clog_request *rq);
+
+#endif /* _LVM_CLOG_CLUSTER_H */

daemons/cmirrord/common.h

@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#ifndef _LVM_CLOG_COMMON_H
+#define _LVM_CLOG_COMMON_H
+
+/*
+ * If there are problems when forking off to become a daemon,
+ * the child will exist with one of these codes.  This allows
+ * the parent to know the reason for the failure and print it
+ * to the launching terminal.
+ *
+ * #define EXIT_SUCCESS 0 (from stdlib.h)
+ * #define EXIT_FAILURE 1 (from stdlib.h)
+ */
+#define EXIT_LOCKFILE              2
+#define EXIT_KERNEL_SOCKET         3 /* Failed netlink socket create */
+#define EXIT_KERNEL_BIND           4
+#define EXIT_KERNEL_SETSOCKOPT     5
+#define EXIT_CLUSTER_CKPT_INIT     6 /* Failed to init checkpoint */
+#define EXIT_QUEUE_NOMEM           7
+
+#define DM_ULOG_REQUEST_SIZE 1024
+
+#endif /* _LVM_CLOG_COMMON_H */

daemons/cmirrord/compat.c

@@ -0,0 +1,210 @@
+/*
+ * Copyright (C) 2010 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ */
+#include "logging.h"
+#include "cluster.h"
+#include "compat.h"
+#include "xlate.h"
+
+#include <errno.h>
+
+/*
+ * Older versions of the log daemon communicate with different
+ * versions of the inter-machine communication structure, which
+ * varies in size and fields.  The older versions append the
+ * standard upstream version of the structure to every request.
+ * COMPAT_OFFSET is where the upstream structure starts.
+ */
+#define COMPAT_OFFSET 256
+
+static void v5_data_endian_switch(struct clog_request *rq, int to_network __attribute((unused)))
+{
+	int i, end;
+	int64_t *pi64;
+	uint64_t *pu64;
+	uint32_t rq_type = rq->u_rq.request_type & ~DM_ULOG_RESPONSE;
+
+	if (rq->u_rq.request_type & DM_ULOG_RESPONSE) {
+		switch (rq_type) {
+		case DM_ULOG_CTR:
+		case DM_ULOG_DTR:
+			LOG_ERROR("Invalid response type in endian switch");
+			exit(EXIT_FAILURE);
+
+		case DM_ULOG_PRESUSPEND:
+		case DM_ULOG_POSTSUSPEND:
+		case DM_ULOG_RESUME:
+		case DM_ULOG_FLUSH:
+		case DM_ULOG_MARK_REGION:
+		case DM_ULOG_CLEAR_REGION:
+		case DM_ULOG_SET_REGION_SYNC:
+		case DM_ULOG_CHECKPOINT_READY:
+		case DM_ULOG_MEMBER_JOIN:
+		case DM_ULOG_STATUS_INFO:
+		case DM_ULOG_STATUS_TABLE:
+			/* No outbound data */
+			break;
+
+		case DM_ULOG_GET_REGION_SIZE:
+		case DM_ULOG_GET_SYNC_COUNT:
+			pu64 = (uint64_t *)rq->u_rq.data;
+			*pu64 = xlate64(*pu64);
+			break;
+		case DM_ULOG_IS_CLEAN:
+		case DM_ULOG_IN_SYNC:
+			pi64 = (int64_t *)rq->u_rq.data;
+			*pi64 = xlate64(*pi64);
+			break;
+		case DM_ULOG_GET_RESYNC_WORK:
+		case DM_ULOG_IS_REMOTE_RECOVERING:
+			pi64 = (int64_t *)rq->u_rq.data;
+			pu64 = ((uint64_t *)rq->u_rq.data) + 1;
+			*pi64 = xlate64(*pi64);
+			*pu64 = xlate64(*pu64);
+			break;
+		default:
+			LOG_ERROR("Unknown request type, %u", rq_type);
+			return;
+		}
+	} else {
+		switch (rq_type) {
+		case DM_ULOG_CTR:
+		case DM_ULOG_DTR:
+			LOG_ERROR("Invalid request type in endian switch");
+			exit(EXIT_FAILURE);
+
+		case DM_ULOG_PRESUSPEND:
+		case DM_ULOG_POSTSUSPEND:
+		case DM_ULOG_RESUME:
+		case DM_ULOG_GET_REGION_SIZE:
+		case DM_ULOG_FLUSH:
+		case DM_ULOG_GET_RESYNC_WORK:
+		case DM_ULOG_GET_SYNC_COUNT:
+		case DM_ULOG_STATUS_INFO:
+		case DM_ULOG_STATUS_TABLE:
+		case DM_ULOG_CHECKPOINT_READY:
+		case DM_ULOG_MEMBER_JOIN:
+			/* No incoming data */
+			break;
+		case DM_ULOG_IS_CLEAN:
+		case DM_ULOG_IN_SYNC:
+		case DM_ULOG_IS_REMOTE_RECOVERING:
+			pu64 = (uint64_t *)rq->u_rq.data;
+			*pu64 = xlate64(*pu64);
+			break;
+		case DM_ULOG_MARK_REGION:
+		case DM_ULOG_CLEAR_REGION:
+			end = rq->u_rq.data_size/sizeof(uint64_t);
+
+			pu64 = (uint64_t *)rq->u_rq.data;
+			for (i = 0; i < end; i++)
+				pu64[i] = xlate64(pu64[i]);
+			break;
+		case DM_ULOG_SET_REGION_SYNC:
+			pu64 = (uint64_t *)rq->u_rq.data;
+			pi64 = ((int64_t *)rq->u_rq.data) + 1;
+			*pu64 = xlate64(*pu64);
+			*pi64 = xlate64(*pi64);
+			break;
+		default:
+			LOG_ERROR("Unknown request type, %u", rq_type);
+			exit(EXIT_FAILURE);
+		}
+	}
+}
+
+static int v5_endian_to_network(struct clog_request *rq)
+{
+	int size;
+	struct dm_ulog_request *u_rq = &rq->u_rq;
+
+	size = sizeof(*rq) + u_rq->data_size;
+
+	u_rq->error = xlate32(u_rq->error);
+	u_rq->seq = xlate32(u_rq->seq);
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate64(u_rq->data_size);
+
+	rq->originator = xlate32(rq->originator);
+
+	v5_data_endian_switch(rq, 1);
+
+	return size;
+}
+
+int clog_request_to_network(struct clog_request *rq)
+{
+	int r;
+
+	/* FIXME: Remove this safety check */
+	if (rq->u.version[0] != xlate64(rq->u.version[1])) {
+		LOG_ERROR("Programmer error:  version[0] must be LE");
+		exit(EXIT_FAILURE);
+	}
+
+	/*
+	 * Are we already running in the endian mode we send
+	 * over the wire?
+	 */
+	if (rq->u.version[0] == rq->u.version[1])
+		return 0;
+
+	r = v5_endian_to_network(rq);
+	if (r < 0)
+		return r;
+	return 0;
+}
+
+static int v5_endian_from_network(struct clog_request *rq)
+{
+	int size;
+	struct dm_ulog_request *u_rq = &rq->u_rq;
+
+	u_rq->error = xlate32(u_rq->error);
+	u_rq->seq = xlate32(u_rq->seq);
+	u_rq->request_type = xlate32(u_rq->request_type);
+	u_rq->data_size = xlate64(u_rq->data_size);
+
+	rq->originator = xlate32(rq->originator);
+
+	size = sizeof(*rq) + u_rq->data_size;
+
+	v5_data_endian_switch(rq, 0);
+
+	return size;
+}
+
+int clog_request_from_network(void *data, size_t data_len)
+{
+	uint64_t *vp = data;
+	uint64_t version = xlate64(vp[0]);
+	uint64_t unconverted_version = vp[1];
+	struct clog_request *rq = data;
+
+	switch (version) {
+	case 5: /* Upstream */
+		if (version == unconverted_version)
+			return 0;
+		break;
+	case 4: /* RHEL 5.[45] */
+	case 3: /* RHEL 5.3 */
+	case 2: /* RHEL 5.2 */
+		/* FIXME: still need to account for payload */
+		if (data_len < (COMPAT_OFFSET + sizeof(*rq)))
+			return -ENOSPC;
+
+		rq = (struct clog_request *)((char *)data + COMPAT_OFFSET);
+		break;
+	default:
+		LOG_ERROR("Unable to process cluster message: "
+			  "Incompatible version");
+		return -EINVAL;
+	}
+
+	v5_endian_from_network(rq);
+	return 0;
+}

daemons/cmirrord/compat.h

@@ -0,0 +1,25 @@
+/*
+ * Copyright (C) 2010 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ */
+#ifndef _LVM_CLOG_COMPAT_H
+#define _LVM_CLOG_COMPAT_H
+
+/*
+ * The intermachine communication structure version are:
+ *	0: Unused
+ *	1: Never in the wild
+ *	2: RHEL 5.2
+ *	3: RHEL 5.3
+ *	4: RHEL 5.4, RHEL 5.5
+ *	5: RHEL 6, Current Upstream Format
+ */
+#define CLOG_TFR_VERSION 5
+
+int clog_request_to_network(struct clog_request *rq);
+int clog_request_from_network(void *data, size_t data_len);
+
+#endif /* _LVM_CLOG_COMPAT_H */

daemons/cmirrord/functions.h

@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#ifndef _LVM_CLOG_FUNCTIONS_H
+#define _LVM_CLOG_FUNCTIONS_H
+
+#include "dm-log-userspace.h"
+#include "cluster.h"
+
+#define LOG_RESUMED   1
+#define LOG_SUSPENDED 2
+
+int local_resume(struct dm_ulog_request *rq);
+int cluster_postsuspend(char *, uint64_t);
+
+int do_request(struct clog_request *rq, int server);
+int push_state(const char *uuid, uint64_t luid,
+	       const char *which, char **buf, uint32_t debug_who);
+int pull_state(const char *uuid, uint64_t luid,
+	       const char *which, char *buf, int size);
+
+int log_get_state(struct dm_ulog_request *rq);
+int log_status(void);
+void log_debug(void);
+
+#endif /* _LVM_CLOG_FUNCTIONS_H */

daemons/cmirrord/link_mon.c

@@ -0,0 +1,151 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#include "logging.h"
+#include "link_mon.h"
+
+#include <errno.h>
+#include <poll.h>
+#include <stdlib.h>
+
+struct link_callback {
+	int fd;
+	const char *name;
+	void *data;
+	int (*callback)(void *data);
+
+	struct link_callback *next;
+};
+
+static unsigned used_pfds = 0;
+static unsigned free_pfds = 0;
+static struct pollfd *pfds = NULL;
+static struct link_callback *callbacks = NULL;
+
+int links_register(int fd, const char *name, int (*callback)(void *data), void *data)
+{
+	unsigned i;
+	struct link_callback *lc;
+
+	for (i = 0; i < used_pfds; i++) {
+		if (fd == pfds[i].fd) {
+			LOG_ERROR("links_register: Duplicate file descriptor");
+			return -EINVAL;
+		}
+	}
+
+	lc = malloc(sizeof(*lc));
+	if (!lc)
+		return -ENOMEM;
+
+	lc->fd = fd;
+	lc->name = name;
+	lc->data = data;
+	lc->callback = callback;
+
+	if (!free_pfds) {
+		struct pollfd *tmp;
+		tmp = realloc(pfds, sizeof(struct pollfd) * ((used_pfds*2) + 1));
+		if (!tmp) {
+			free(lc);
+			return -ENOMEM;
+		}
+		
+		pfds = tmp;
+		free_pfds = used_pfds + 1;
+	}
+
+	free_pfds--;
+	pfds[used_pfds].fd = fd;
+	pfds[used_pfds].events = POLLIN;
+	pfds[used_pfds].revents = 0;
+	used_pfds++;
+
+	lc->next = callbacks;
+	callbacks = lc;
+	LOG_DBG("Adding %s/%d", lc->name, lc->fd);
+	LOG_DBG(" used_pfds = %u, free_pfds = %u",
+		used_pfds, free_pfds);
+
+	return 0;
+}
+
+int links_unregister(int fd)
+{
+	unsigned i;
+	struct link_callback *p, *c;
+
+	for (i = 0; i < used_pfds; i++)
+		if (fd == pfds[i].fd) {
+			/* entire struct is copied (overwritten) */
+			pfds[i] = pfds[used_pfds - 1];
+			used_pfds--;
+			free_pfds++;
+		}
+
+	for (p = NULL, c = callbacks; c; p = c, c = c->next)
+		if (fd == c->fd) {
+			LOG_DBG("Freeing up %s/%d", c->name, c->fd);
+			LOG_DBG(" used_pfds = %u, free_pfds = %u",
+				used_pfds, free_pfds);
+			if (p)
+				p->next = c->next;
+			else
+				callbacks = c->next;
+			free(c);
+			break;
+		}
+
+	return 0;
+}
+
+int links_monitor(void)
+{
+	unsigned i;
+	int r;
+
+	for (i = 0; i < used_pfds; i++) {
+		pfds[i].revents = 0;
+	}
+
+	r = poll(pfds, used_pfds, -1);
+	if (r <= 0)
+		return r;
+
+	r = 0;
+	/* FIXME: handle POLLHUP */
+	for (i = 0; i < used_pfds; i++)
+		if (pfds[i].revents & POLLIN) {
+			LOG_DBG("Data ready on %d", pfds[i].fd);
+				
+			/* FIXME: Add this back return 1;*/
+			r++;
+		}
+
+	return r;
+}
+
+int links_issue_callbacks(void)
+{
+	unsigned i;
+	struct link_callback *lc;
+
+	for (i = 0; i < used_pfds; i++)
+		if (pfds[i].revents & POLLIN)
+			for (lc = callbacks; lc; lc = lc->next)
+				if (pfds[i].fd == lc->fd) {
+					LOG_DBG("Issuing callback on %s/%d",
+						lc->name, lc->fd);
+					lc->callback(lc->data);
+					break;
+				}
+	return 0;
+}

daemons/cmirrord/link_mon.h

@@ -0,0 +1,20 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#ifndef _LVM_CLOG_LINK_MON_H
+#define _LVM_CLOG_LINK_MON_H
+
+int links_register(int fd, const char *name, int (*callback)(void *data), void *data);
+int links_unregister(int fd);
+int links_monitor(void);
+int links_issue_callbacks(void);
+
+#endif /* _LVM_CLOG_LINK_MON_H */

daemons/cmirrord/local.c

@@ -0,0 +1,416 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#include "logging.h"
+#include "common.h"
+#include "functions.h"
+#include "link_mon.h"
+#include "local.h"
+
+#include <errno.h>
+#include <sys/socket.h>
+#include <linux/connector.h>
+#include <linux/netlink.h>
+#include <unistd.h>
+
+#ifndef CN_IDX_DM
+/* Kernel 2.6.31 is required to run this code */
+#define CN_IDX_DM                       0x7     /* Device Mapper */
+#define CN_VAL_DM_USERSPACE_LOG         0x1
+#endif
+
+static int cn_fd;  /* Connector (netlink) socket fd */
+static char recv_buf[2048];
+static char send_buf[2048];
+
+
+/* FIXME: merge this function with kernel_send_helper */
+static int kernel_ack(uint32_t seq, int error)
+{
+	int r;
+	struct nlmsghdr *nlh = (struct nlmsghdr *)send_buf;
+	struct cn_msg *msg = NLMSG_DATA(nlh);
+
+	if (error < 0) {
+		LOG_ERROR("Programmer error: error codes must be positive");
+		return -EINVAL;
+	}
+
+	memset(send_buf, 0, sizeof(send_buf));
+
+	nlh->nlmsg_seq = 0;
+	nlh->nlmsg_pid = getpid();
+	nlh->nlmsg_type = NLMSG_DONE;
+	nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct cn_msg));
+	nlh->nlmsg_flags = 0;
+
+	msg->len = 0;
+	msg->id.idx = CN_IDX_DM;
+	msg->id.val = CN_VAL_DM_USERSPACE_LOG;
+	msg->seq = seq;
+	msg->ack = error;
+
+	r = send(cn_fd, nlh, NLMSG_LENGTH(sizeof(struct cn_msg)), 0);
+	/* FIXME: do better error processing */
+	if (r <= 0)
+		return -EBADE;
+
+	return 0;
+}
+
+
+/*
+ * kernel_recv
+ * @rq: the newly allocated request from kernel
+ *
+ * Read requests from the kernel and allocate space for the new request.
+ * If there is no request from the kernel, *rq is NULL.
+ *
+ * This function is not thread safe due to returned stack pointer.  In fact,
+ * the returned pointer must not be in-use when this function is called again.
+ *
+ * Returns: 0 on success, -EXXX on error
+ */
+static int kernel_recv(struct clog_request **rq)
+{
+	int r = 0;
+	ssize_t len;
+	char *foo;
+	struct cn_msg *msg;
+	struct dm_ulog_request *u_rq;
+	struct nlmsghdr *nlmsg_h;
+
+	*rq = NULL;
+	memset(recv_buf, 0, sizeof(recv_buf));
+
+	len = recv(cn_fd, recv_buf, sizeof(recv_buf), 0);
+	if (len < 0) {
+		LOG_ERROR("Failed to recv message from kernel");
+		r = -errno;
+		goto fail;
+	}
+
+	nlmsg_h = (struct nlmsghdr *)recv_buf;
+	switch (nlmsg_h->nlmsg_type) {
+	case NLMSG_ERROR:
+		LOG_ERROR("Unable to recv message from kernel: NLMSG_ERROR");
+		r = -EBADE;
+		goto fail;
+	case NLMSG_DONE:
+		msg = (struct cn_msg *)NLMSG_DATA((struct nlmsghdr *)recv_buf);
+		len -= (ssize_t)sizeof(struct nlmsghdr);
+
+		if (len < (ssize_t)sizeof(struct cn_msg)) {
+			LOG_ERROR("Incomplete request from kernel received");
+			r = -EBADE;
+			goto fail;
+		}
+
+		if (msg->len > DM_ULOG_REQUEST_SIZE) {
+			LOG_ERROR("Not enough space to receive kernel request (%d/%d)",
+				  msg->len, DM_ULOG_REQUEST_SIZE);
+			r = -EBADE;
+			goto fail;
+		}
+
+		if (!msg->len)
+			LOG_ERROR("Zero length message received");
+
+		len -= (ssize_t)sizeof(struct cn_msg);
+
+		if (len < msg->len)
+			LOG_ERROR("len = %zd, msg->len = %" PRIu16, len, msg->len);
+
+		msg->data[msg->len] = '\0'; /* Cleaner way to ensure this? */
+		u_rq = (struct dm_ulog_request *)msg->data;
+
+		if (!u_rq->request_type) {
+			LOG_DBG("Bad transmission, requesting resend [%u]",
+				msg->seq);
+			r = -EAGAIN;
+
+			if (kernel_ack(msg->seq, EAGAIN)) {
+				LOG_ERROR("Failed to NACK kernel transmission [%u]",
+					  msg->seq);
+				r = -EBADE;
+			}
+		}
+
+		/*
+		 * Now we've got sizeof(struct cn_msg) + sizeof(struct nlmsghdr)
+		 * worth of space that precede the request structure from the
+		 * kernel.  Since that space isn't going to be used again, we
+		 * can take it for our purposes; rather than allocating a whole
+		 * new structure and doing a memcpy.
+		 *
+		 * We should really make sure 'clog_request' doesn't grow
+		 * beyond what is available to us, but we need only check it
+		 * once... perhaps at compile time?
+		 */
+		foo = (char *)u_rq;
+		foo -= (sizeof(struct clog_request) - sizeof(struct dm_ulog_request));
+		*rq = (struct clog_request *) foo;
+
+		/* Clear the wrapper container fields */
+		memset(*rq, 0, (size_t)((char *)u_rq - (char *)(*rq)));
+		break;
+	default:
+		LOG_ERROR("Unknown nlmsg_type");
+		r = -EBADE;
+	}
+
+fail:
+	if (r)
+		*rq = NULL;
+
+	return (r == -EAGAIN) ? 0 : r;
+}
+
+static int kernel_send_helper(void *data, uint16_t out_size)
+{
+	int r;
+	struct nlmsghdr *nlh;
+	struct cn_msg *msg;
+
+	memset(send_buf, 0, sizeof(send_buf));
+
+	nlh = (struct nlmsghdr *)send_buf;
+	nlh->nlmsg_seq = 0;  /* FIXME: Is this used? */
+	nlh->nlmsg_pid = getpid();
+	nlh->nlmsg_type = NLMSG_DONE;
+	nlh->nlmsg_len = NLMSG_LENGTH(out_size + sizeof(struct cn_msg));
+	nlh->nlmsg_flags = 0;
+
+	msg = NLMSG_DATA(nlh);
+	memcpy(msg->data, data, out_size);
+	msg->len = out_size;
+	msg->id.idx = CN_IDX_DM;
+	msg->id.val = CN_VAL_DM_USERSPACE_LOG;
+	msg->seq = 0;
+
+	r = send(cn_fd, nlh, NLMSG_LENGTH(out_size + sizeof(struct cn_msg)), 0);
+	/* FIXME: do better error processing */
+	if (r <= 0)
+		return -EBADE;
+
+	return 0;
+}
+
+/*
+ * do_local_work
+ *
+ * Any processing errors are placed in the 'rq'
+ * structure to be reported back to the kernel.
+ * It may be pointless for this function to
+ * return an int.
+ *
+ * Returns: 0 on success, -EXXX on failure
+ */
+static int do_local_work(void *data __attribute((unused)))
+{
+	int r;
+	struct clog_request *rq;
+	struct dm_ulog_request *u_rq = NULL;
+
+	r = kernel_recv(&rq);
+	if (r)
+		return r;
+
+	if (!rq)
+		return 0;
+
+	u_rq = &rq->u_rq;
+	LOG_DBG("[%s]  Request from kernel received: [%s/%u]",
+		SHORT_UUID(u_rq->uuid), RQ_TYPE(u_rq->request_type),
+		u_rq->seq);
+	switch (u_rq->request_type) {
+	case DM_ULOG_CTR:
+	case DM_ULOG_DTR:
+	case DM_ULOG_GET_REGION_SIZE:
+	case DM_ULOG_IN_SYNC:
+	case DM_ULOG_GET_SYNC_COUNT:
+	case DM_ULOG_STATUS_INFO:
+	case DM_ULOG_STATUS_TABLE:
+	case DM_ULOG_PRESUSPEND:
+		/* We do not specify ourselves as server here */
+		r = do_request(rq, 0);
+		if (r)
+			LOG_DBG("Returning failed request to kernel [%s]",
+				RQ_TYPE(u_rq->request_type));
+		r = kernel_send(u_rq);
+		if (r)
+			LOG_ERROR("Failed to respond to kernel [%s]",
+				  RQ_TYPE(u_rq->request_type));
+			
+		break;
+	case DM_ULOG_RESUME:
+		/*
+		 * Resume is a special case that requires a local
+		 * component to join the CPG, and a cluster component
+		 * to handle the request.
+		 */
+		r = local_resume(u_rq);
+		if (r) {
+			LOG_DBG("Returning failed request to kernel [%s]",
+				RQ_TYPE(u_rq->request_type));
+			r = kernel_send(u_rq);
+			if (r)
+				LOG_ERROR("Failed to respond to kernel [%s]",
+					  RQ_TYPE(u_rq->request_type));
+			break;
+		}
+		/* ELSE, fall through */
+	case DM_ULOG_IS_CLEAN:
+	case DM_ULOG_FLUSH:
+	case DM_ULOG_MARK_REGION:
+	case DM_ULOG_GET_RESYNC_WORK:
+	case DM_ULOG_SET_REGION_SYNC:
+	case DM_ULOG_IS_REMOTE_RECOVERING:
+	case DM_ULOG_POSTSUSPEND:
+		r = cluster_send(rq);
+		if (r) {
+			u_rq->data_size = 0;
+			u_rq->error = r;
+			kernel_send(u_rq);
+		}
+
+		break;
+	case DM_ULOG_CLEAR_REGION:
+		r = kernel_ack(u_rq->seq, 0);
+
+		r = cluster_send(rq);
+		if (r) {
+			/*
+			 * FIXME: store error for delivery on flush
+			 *        This would allow us to optimize MARK_REGION
+			 *        too.
+			 */
+		}
+
+		break;
+	default:
+		LOG_ERROR("Invalid log request received (%u), ignoring.",
+			  u_rq->request_type);
+
+		return 0;
+	}
+
+	if (r && !u_rq->error)
+		u_rq->error = r;
+
+	return r;
+}
+
+/*
+ * kernel_send
+ * @u_rq: result to pass back to kernel
+ *
+ * This function returns the u_rq structure
+ * (containing the results) to the kernel.
+ * It then frees the structure.
+ *
+ * WARNING: should the structure be freed if
+ * there is an error?  I vote 'yes'.  If the
+ * kernel doesn't get the response, it should
+ * resend the request.
+ *
+ * Returns: 0 on success, -EXXX on failure
+ */
+int kernel_send(struct dm_ulog_request *u_rq)
+{
+	int r;
+	uint16_t size;
+
+	if (!u_rq)
+		return -EINVAL;
+
+	size = (uint16_t)(sizeof(struct dm_ulog_request) + u_rq->data_size);
+
+	if (!u_rq->data_size && !u_rq->error) {
+		/* An ACK is all that is needed */
+
+		/* FIXME: add ACK code */
+	} else if (size > DM_ULOG_REQUEST_SIZE) {
+		/*
+		 * If we gotten here, we've already overrun
+		 * our allotted space somewhere.
+		 *
+		 * We must do something, because the kernel
+		 * is waiting for a response.
+		 */
+		LOG_ERROR("Not enough space to respond to server");
+		u_rq->error = -ENOSPC;
+		size = sizeof(struct dm_ulog_request);
+	}
+
+	r = kernel_send_helper(u_rq, size);
+	if (r)
+		LOG_ERROR("Failed to send msg to kernel.");
+
+	return r;
+}
+
+/*
+ * init_local
+ *
+ * Initialize kernel communication socket (netlink)
+ *
+ * Returns: 0 on success, values from common.h on failure
+ */
+int init_local(void)
+{
+	int r = 0;
+	unsigned opt;
+	struct sockaddr_nl addr;
+
+	cn_fd = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_CONNECTOR);
+	if (cn_fd < 0)
+		return EXIT_KERNEL_SOCKET;
+
+	/* memset to fix valgrind complaint */
+	memset(&addr, 0, sizeof(struct sockaddr_nl));
+
+	addr.nl_family = AF_NETLINK;
+	addr.nl_groups = CN_IDX_DM;
+	addr.nl_pid = 0;
+
+	r = bind(cn_fd, (struct sockaddr *) &addr, sizeof(addr));
+	if (r < 0) {
+		close(cn_fd);
+		return EXIT_KERNEL_BIND;
+	}
+
+	opt = addr.nl_groups;
+	r = setsockopt(cn_fd, 270, NETLINK_ADD_MEMBERSHIP, &opt, sizeof(opt));
+	if (r) {
+		close(cn_fd);
+		return EXIT_KERNEL_SETSOCKOPT;
+	}
+
+	/*
+	r = fcntl(cn_fd, F_SETFL, FNDELAY);
+	*/
+
+	links_register(cn_fd, "local", do_local_work, NULL);
+
+	return 0;
+}
+
+/*
+ * cleanup_local
+ *
+ * Clean up before exiting
+ */
+void cleanup_local(void)
+{
+	links_unregister(cn_fd);
+	close(cn_fd);
+}

daemons/cmirrord/local.h

@@ -0,0 +1,20 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#ifndef _LVM_CLOG_LOCAL_H
+#define _LVM_CLOG_LOCAL_H
+
+int init_local(void);
+void cleanup_local(void);
+
+int kernel_send(struct dm_ulog_request *rq);
+
+#endif /* _LVM_CLOG_LOCAL_H */

daemons/cmirrord/logging.c

@@ -0,0 +1,57 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+#include "logging.h"
+
+const char *__rq_types_off_by_one[] = {
+	"DM_ULOG_CTR",
+	"DM_ULOG_DTR",
+	"DM_ULOG_PRESUSPEND",
+	"DM_ULOG_POSTSUSPEND",
+	"DM_ULOG_RESUME",
+	"DM_ULOG_GET_REGION_SIZE",
+	"DM_ULOG_IS_CLEAN",
+	"DM_ULOG_IN_SYNC",
+	"DM_ULOG_FLUSH",
+	"DM_ULOG_MARK_REGION",
+	"DM_ULOG_CLEAR_REGION",
+	"DM_ULOG_GET_RESYNC_WORK",
+	"DM_ULOG_SET_REGION_SYNC",
+	"DM_ULOG_GET_SYNC_COUNT",
+	"DM_ULOG_STATUS_INFO",
+	"DM_ULOG_STATUS_TABLE",
+	"DM_ULOG_IS_REMOTE_RECOVERING",
+	NULL
+};
+
+int log_tabbing = 0;
+int log_is_open = 0;
+
+/*
+ * Variables for various conditional logging
+ */
+#ifdef MEMB
+int log_membership_change = 1;
+#else
+int log_membership_change = 0;
+#endif
+
+#ifdef CKPT
+int log_checkpoint = 1;
+#else
+int log_checkpoint = 0;
+#endif
+
+#ifdef RESEND
+int log_resend_requests = 1;
+#else
+int log_resend_requests = 0;
+#endif

daemons/cmirrord/logging.h

@@ -0,0 +1,77 @@
+/*
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_CLOG_LOGGING_H
+#define _LVM_CLOG_LOGGING_H
+
+#define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
+
+#include <configure.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <syslog.h>
+
+/* SHORT_UUID - print last 8 chars of a string */
+#define SHORT_UUID(x) (strlen(x) > 8) ? ((x) + (strlen(x) - 8)) : (x)
+
+extern const char *__rq_types_off_by_one[];
+#define RQ_TYPE(x) __rq_types_off_by_one[(x) - 1]
+
+extern int log_tabbing;
+extern int log_is_open;
+extern int log_membership_change;
+extern int log_checkpoint;
+extern int log_resend_requests;
+
+#define LOG_OPEN(ident, option, facility) do { \
+		openlog(ident, option, facility); \
+		log_is_open = 1;		  \
+	} while (0)
+
+#define LOG_CLOSE(void) do { \
+		log_is_open = 0; \
+		closelog();	 \
+	} while (0)
+
+#define LOG_OUTPUT(level, f, arg...) do {				\
+		int __i;						\
+		char __buffer[16];					\
+		FILE *fp = (level > LOG_NOTICE) ? stderr : stdout;	\
+		if (log_is_open) {					\
+			for (__i = 0; (__i < log_tabbing) && (__i < 15); __i++) \
+				__buffer[__i] = '\t';			\
+			__buffer[__i] = '\0';				\
+			syslog(level, "%s" f "\n", __buffer, ## arg);	\
+		} else {						\
+			for (__i = 0; __i < log_tabbing; __i++)		\
+				fprintf(fp, "\t");			\
+			fprintf(fp, f "\n", ## arg);			\
+		}							\
+	} while (0)
+
+
+#ifdef DEBUG
+#define LOG_DBG(f, arg...) LOG_OUTPUT(LOG_DEBUG, f, ## arg)
+#else /* DEBUG */
+#define LOG_DBG(f, arg...)
+#endif /* DEBUG */
+
+#define LOG_COND(__X, f, arg...) do {\
+		if (__X) { 	     \
+			LOG_OUTPUT(LOG_NOTICE, f, ## arg); \
+		} \
+	} while (0)
+#define LOG_PRINT(f, arg...) LOG_OUTPUT(LOG_NOTICE, f, ## arg)
+#define LOG_ERROR(f, arg...) LOG_OUTPUT(LOG_ERR, f, ## arg)
+
+#endif /* _LVM_CLOG_LOGGING_H */

daemons/dmeventd/.exported_symbols

@@ -0,0 +1,19 @@
+dm_event_handler_create
+dm_event_handler_destroy
+dm_event_handler_set_dso
+dm_event_handler_set_dev_name
+dm_event_handler_set_uuid
+dm_event_handler_set_major
+dm_event_handler_set_minor
+dm_event_handler_set_event_mask
+dm_event_handler_get_dso
+dm_event_handler_get_devname
+dm_event_handler_get_uuid
+dm_event_handler_get_major
+dm_event_handler_get_minor
+dm_event_handler_get_event_mask
+dm_event_register_handler
+dm_event_unregister_handler
+dm_event_get_registered_device
+dm_event_handler_set_timeout
+dm_event_handler_get_timeout

daemons/dmeventd/Makefile.in

@@ -0,0 +1,114 @@
+#
+# Copyright (C) 2005-2007 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the device-mapper userspace tools.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU Lesser General Public License v.2.1.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+SOURCES = libdevmapper-event.c
+TARGETS = dmeventd
+
+.PHONY: install_lib_dynamic install_lib_static install_include \
+	install_pkgconfig install_dmeventd_dynamic install_dmeventd_static \
+	install_lib install_dmeventd
+
+INSTALL_DMEVENTD_TARGETS = install_dmeventd_dynamic
+INSTALL_LIB_TARGETS = install_lib_dynamic
+
+ifeq ("@STATIC_LINK@", "yes")
+  LIB_STATIC = libdevmapper-event.a
+  TARGETS += dmeventd.static
+  INSTALL_DMEVENTD_TARGETS += install_dmeventd_static
+  INSTALL_LIB_TARGETS += install_lib_static
+endif
+
+LIB_VERSION = $(LIB_VERSION_DM)
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event.dylib
+else
+  LIB_SHARED = libdevmapper-event.so
+  VERSIONED_SHLIB = $(LIB_SHARED).$(LIB_VERSION)
+endif
+
+CLEAN_TARGETS = dmeventd.o dmeventd.static
+
+ifneq ($(MAKECMDGOALS),device-mapper)
+  SUBDIRS+=plugins
+endif
+
+include ../../make.tmpl
+
+all: device-mapper
+device-mapper: $(TARGETS)
+
+LIBS += -ldl
+LVMLIBS += -ldevmapper-event -lpthread -ldevmapper
+
+$(VERSIONED_SHLIB): $(LIB_SHARED)
+	$(RM) -f $@
+	$(LN_S) $(LIB_SHARED) $@
+
+dmeventd: $(LIB_SHARED) $(VERSIONED_SHLIB) dmeventd.o
+	$(CC) $(CFLAGS) $(LDFLAGS) -L. -o $@ dmeventd.o \
+	$(LVMLIBS) $(LIBS) -rdynamic
+
+dmeventd.static: $(LIB_STATIC) dmeventd.o $(interfacebuilddir)/libdevmapper.a
+	$(CC) $(CFLAGS) $(LDFLAGS) -static -L. -L$(interfacebuilddir) -o $@ \
+	dmeventd.o $(LVMLIBS) $(LIBS) -rdynamic
+
+ifeq ("@PKGCONFIG@", "yes")
+  INSTALL_LIB_TARGETS += install_pkgconfig
+endif
+
+install_include:
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/libdevmapper-event.h \
+		$(includedir)/libdevmapper-event.h
+
+install_pkgconfig:
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper-event.pc \
+		$(usrlibdir)/pkgconfig/devmapper-event.pc
+
+install_lib_dynamic: libdevmapper-event.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION)
+	$(LN_S) -f libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION) \
+		$(libdir)/libdevmapper-event.$(LIB_SUFFIX)
+
+install_lib_static: libdevmapper-event.a
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/libdevmapper-event.a.$(LIB_VERSION)
+	$(LN_S) -f libdevmapper-event.a.$(LIB_VERSION) $(libdir)/libdevmapper-event.a
+
+install_lib: $(INSTALL_LIB_TARGETS)
+
+install_dmeventd_dynamic: dmeventd
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+
+install_dmeventd_static: dmeventd.static
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+
+install_dmeventd: $(INSTALL_DMEVENTD_TARGETS)
+
+install: install_include install_lib install_dmeventd
+
+install_device-mapper: install_include install_lib install_dmeventd
+
+.PHONY: distclean_lib distclean
+
+distclean_lib: 
+	$(RM) libdevmapper-event.pc
+
+distclean: distclean_lib
+

daemons/dmeventd/dmeventd.h

@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2005-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of the device-mapper userspace tools.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef __DMEVENTD_DOT_H__
+#define __DMEVENTD_DOT_H__
+
+/* FIXME This stuff must be configurable. */
+
+#define	DM_EVENT_DAEMON		"/sbin/dmeventd"
+#define DM_EVENT_LOCKFILE	"/var/lock/dmeventd"
+#define	DM_EVENT_FIFO_CLIENT	"/var/run/dmeventd-client"
+#define	DM_EVENT_FIFO_SERVER	"/var/run/dmeventd-server"
+#define DM_EVENT_PIDFILE	"/var/run/dmeventd.pid"
+
+#define DM_EVENT_DEFAULT_TIMEOUT 10
+
+/* Commands for the daemon passed in the message below. */
+enum dm_event_command {
+	DM_EVENT_CMD_ACTIVE = 1,
+	DM_EVENT_CMD_REGISTER_FOR_EVENT,
+	DM_EVENT_CMD_UNREGISTER_FOR_EVENT,
+	DM_EVENT_CMD_GET_REGISTERED_DEVICE,
+	DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE,
+	DM_EVENT_CMD_SET_TIMEOUT,
+	DM_EVENT_CMD_GET_TIMEOUT,
+	DM_EVENT_CMD_HELLO,
+};
+
+/* Message passed between client and daemon. */
+struct dm_event_daemon_message {
+	uint32_t cmd;
+	uint32_t size;
+	char *data;
+};
+
+/* FIXME Is this meant to be exported?  I can't see where the
+   interface uses it. */
+/* Fifos for client/daemon communication. */
+struct dm_event_fifos {
+	int client;
+	int server;
+	const char *client_path;
+	const char *server_path;
+};
+
+/*      EXIT_SUCCESS             0 -- stdlib.h */
+/*      EXIT_FAILURE             1 -- stdlib.h */
+#define EXIT_LOCKFILE_INUSE      2
+#define EXIT_DESC_CLOSE_FAILURE  3
+#define EXIT_DESC_OPEN_FAILURE   4
+#define EXIT_OPEN_PID_FAILURE    5
+#define EXIT_FIFO_FAILURE        6
+#define EXIT_CHDIR_FAILURE       7
+
+#endif /* __DMEVENTD_DOT_H__ */

daemons/dmeventd/libdevmapper-event.c

@@ -0,0 +1,817 @@
+/*
+ * Copyright (C) 2005-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of the device-mapper userspace tools.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "dmlib.h"
+#include "libdevmapper-event.h"
+//#include "libmultilog.h"
+#include "dmeventd.h"
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/file.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <sys/wait.h>
+#include <arpa/inet.h>		/* for htonl, ntohl */
+
+static int _sequence_nr = 0;
+
+struct dm_event_handler {
+	char *dso;
+
+	char *dev_name;
+
+	char *uuid;
+	int major;
+	int minor;
+	uint32_t timeout;
+
+	enum dm_event_mask mask;
+};
+
+static void _dm_event_handler_clear_dev_info(struct dm_event_handler *dmevh)
+{
+	if (dmevh->dev_name)
+		dm_free(dmevh->dev_name);
+	if (dmevh->uuid)
+		dm_free(dmevh->uuid);
+	dmevh->dev_name = dmevh->uuid = NULL;
+	dmevh->major = dmevh->minor = 0;
+}
+
+struct dm_event_handler *dm_event_handler_create(void)
+{
+	struct dm_event_handler *dmevh = NULL;
+
+	if (!(dmevh = dm_malloc(sizeof(*dmevh))))
+		return NULL;
+
+	dmevh->dso = dmevh->dev_name = dmevh->uuid = NULL;
+	dmevh->major = dmevh->minor = 0;
+	dmevh->mask = 0;
+	dmevh->timeout = 0;
+
+	return dmevh;
+}
+
+void dm_event_handler_destroy(struct dm_event_handler *dmevh)
+{
+	_dm_event_handler_clear_dev_info(dmevh);
+	if (dmevh->dso)
+		dm_free(dmevh->dso);
+	dm_free(dmevh);
+}
+
+int dm_event_handler_set_dso(struct dm_event_handler *dmevh, const char *path)
+{
+	if (!path) /* noop */
+		return 0;
+	if (dmevh->dso)
+		dm_free(dmevh->dso);
+
+	dmevh->dso = dm_strdup(path);
+	if (!dmevh->dso)
+		return -ENOMEM;
+
+	return 0;
+}
+
+int dm_event_handler_set_dev_name(struct dm_event_handler *dmevh, const char *dev_name)
+{
+	if (!dev_name)
+		return 0;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->dev_name = dm_strdup(dev_name);
+	if (!dmevh->dev_name)
+		return -ENOMEM;
+	return 0;
+}
+
+int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid)
+{
+	if (!uuid)
+		return 0;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->uuid = dm_strdup(uuid);
+	if (!dmevh->dev_name)
+		return -ENOMEM;
+	return 0;
+}
+
+void dm_event_handler_set_major(struct dm_event_handler *dmevh, int major)
+{
+	int minor = dmevh->minor;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->major = major;
+	dmevh->minor = minor;
+}
+
+void dm_event_handler_set_minor(struct dm_event_handler *dmevh, int minor)
+{
+	int major = dmevh->major;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->major = major;
+	dmevh->minor = minor;
+}
+
+void dm_event_handler_set_event_mask(struct dm_event_handler *dmevh,
+				     enum dm_event_mask evmask)
+{
+	dmevh->mask = evmask;
+}
+
+void dm_event_handler_set_timeout(struct dm_event_handler *dmevh, int timeout)
+{
+	dmevh->timeout = timeout;
+}
+
+const char *dm_event_handler_get_dso(const struct dm_event_handler *dmevh)
+{
+	return dmevh->dso;
+}
+
+const char *dm_event_handler_get_dev_name(const struct dm_event_handler *dmevh)
+{
+	return dmevh->dev_name;
+}
+
+const char *dm_event_handler_get_uuid(const struct dm_event_handler *dmevh)
+{
+	return dmevh->uuid;
+}
+
+int dm_event_handler_get_major(const struct dm_event_handler *dmevh)
+{
+	return dmevh->major;
+}
+
+int dm_event_handler_get_minor(const struct dm_event_handler *dmevh)
+{
+	return dmevh->minor;
+}
+
+int dm_event_handler_get_timeout(const struct dm_event_handler *dmevh)
+{
+	return dmevh->timeout;
+}
+
+enum dm_event_mask dm_event_handler_get_event_mask(const struct dm_event_handler *dmevh)
+{
+	return dmevh->mask;
+}
+
+static int _check_message_id(struct dm_event_daemon_message *msg)
+{
+	int pid, seq_nr;
+
+	if ((sscanf(msg->data, "%d:%d", &pid, &seq_nr) != 2) ||
+	    (pid != getpid()) || (seq_nr != _sequence_nr)) {
+		log_error("Ignoring out-of-sequence reply from dmeventd. "
+			  "Expected %d:%d but received %s", getpid(),
+			  _sequence_nr, msg->data);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * daemon_read
+ * @fifos
+ * @msg
+ *
+ * Read message from daemon.
+ *
+ * Returns: 0 on failure, 1 on success
+ */
+static int _daemon_read(struct dm_event_fifos *fifos,
+			struct dm_event_daemon_message *msg)
+{
+	unsigned bytes = 0;
+	int ret, i;
+	fd_set fds;
+	struct timeval tval = { 0, 0 };
+	size_t size = 2 * sizeof(uint32_t);	/* status + size */
+	char *buf = alloca(size);
+	int header = 1;
+
+	while (bytes < size) {
+		for (i = 0, ret = 0; (i < 20) && (ret < 1); i++) {
+			/* Watch daemon read FIFO for input. */
+			FD_ZERO(&fds);
+			FD_SET(fifos->server, &fds);
+			tval.tv_sec = 1;
+			ret = select(fifos->server + 1, &fds, NULL, NULL,
+				     &tval);
+			if (ret < 0 && errno != EINTR) {
+				log_error("Unable to read from event server");
+				return 0;
+			}
+		}
+		if (ret < 1) {
+			log_error("Unable to read from event server.");
+			return 0;
+		}
+
+		ret = read(fifos->server, buf + bytes, size);
+		if (ret < 0) {
+			if ((errno == EINTR) || (errno == EAGAIN))
+				continue;
+			else {
+				log_error("Unable to read from event server.");
+				return 0;
+			}
+		}
+
+		bytes += ret;
+		if (bytes == 2 * sizeof(uint32_t) && header) {
+			msg->cmd = ntohl(*((uint32_t *)buf));
+			msg->size = ntohl(*((uint32_t *)buf + 1));
+			buf = msg->data = dm_malloc(msg->size);
+			size = msg->size;
+			bytes = 0;
+			header = 0;
+		}
+	}
+
+	if (bytes != size) {
+		if (msg->data)
+			dm_free(msg->data);
+		msg->data = NULL;
+	}
+
+	return bytes == size;
+}
+
+/* Write message to daemon. */
+static int _daemon_write(struct dm_event_fifos *fifos,
+			 struct dm_event_daemon_message *msg)
+{
+	unsigned bytes = 0;
+	int ret = 0;
+	fd_set fds;
+
+	size_t size = 2 * sizeof(uint32_t) + msg->size;
+	char *buf = alloca(size);
+	char drainbuf[128];
+	struct timeval tval = { 0, 0 };
+
+	*((uint32_t *)buf) = htonl(msg->cmd);
+	*((uint32_t *)buf + 1) = htonl(msg->size);
+	memcpy(buf + 2 * sizeof(uint32_t), msg->data, msg->size);
+
+	/* drain the answer fifo */
+	while (1) {
+		FD_ZERO(&fds);
+		FD_SET(fifos->server, &fds);
+		tval.tv_usec = 100;
+		ret = select(fifos->server + 1, &fds, NULL, NULL, &tval);
+		if ((ret < 0) && (errno != EINTR)) {
+			log_error("Unable to talk to event daemon");
+			return 0;
+		}
+		if (ret == 0)
+			break;
+		read(fifos->server, drainbuf, 127);
+	}
+
+	while (bytes < size) {
+		do {
+			/* Watch daemon write FIFO to be ready for output. */
+			FD_ZERO(&fds);
+			FD_SET(fifos->client, &fds);
+			ret = select(fifos->client + 1, NULL, &fds, NULL, NULL);
+			if ((ret < 0) && (errno != EINTR)) {
+				log_error("Unable to talk to event daemon");
+				return 0;
+			}
+		} while (ret < 1);
+
+		ret = write(fifos->client, ((char *) buf) + bytes,
+			    size - bytes);
+		if (ret < 0) {
+			if ((errno == EINTR) || (errno == EAGAIN))
+				continue;
+			else {
+				log_error("Unable to talk to event daemon");
+				return 0;
+			}
+		}
+
+		bytes += ret;
+	}
+
+	return bytes == size;
+}
+
+static int _daemon_talk(struct dm_event_fifos *fifos,
+			struct dm_event_daemon_message *msg, int cmd,
+			const char *dso_name, const char *dev_name,
+			enum dm_event_mask evmask, uint32_t timeout)
+{
+	const char *dso = dso_name ? dso_name : "";
+	const char *dev = dev_name ? dev_name : "";
+	const char *fmt = "%d:%d %s %s %u %" PRIu32;
+	int msg_size;
+	memset(msg, 0, sizeof(*msg));
+
+	/*
+	 * Set command and pack the arguments
+	 * into ASCII message string.
+	 */
+	msg->cmd = cmd;
+	if (cmd == DM_EVENT_CMD_HELLO)
+		fmt = "%d:%d HELLO";
+	if ((msg_size = dm_asprintf(&(msg->data), fmt, getpid(), _sequence_nr,
+				    dso, dev, evmask, timeout)) < 0) {
+		log_error("_daemon_talk: message allocation failed");
+		return -ENOMEM;
+	}
+	msg->size = msg_size;
+
+	/*
+	 * Write command and message to and
+	 * read status return code from daemon.
+	 */
+	if (!_daemon_write(fifos, msg)) {
+		stack;
+		dm_free(msg->data);
+		msg->data = 0;
+		return -EIO;
+	}
+
+	do {
+
+		if (msg->data)
+			dm_free(msg->data);
+		msg->data = 0;
+
+		if (!_daemon_read(fifos, msg)) {
+			stack;
+			return -EIO;
+		}
+	} while (!_check_message_id(msg));
+
+	_sequence_nr++;
+
+	return (int32_t) msg->cmd;
+}
+
+/*
+ * start_daemon
+ *
+ * This function forks off a process (dmeventd) that will handle
+ * the events.  I am currently test opening one of the fifos to
+ * ensure that the daemon is running and listening...  I thought
+ * this would be less expensive than fork/exec'ing every time.
+ * Perhaps there is an even quicker/better way (no, checking the
+ * lock file is _not_ a better way).
+ *
+ * Returns: 1 on success, 0 otherwise
+ */
+static int _start_daemon(struct dm_event_fifos *fifos)
+{
+	int pid, ret = 0;
+	int status;
+	struct stat statbuf;
+	char dmeventdpath[] = DMEVENTD_PATH; /* const type for execvp */
+	char * const args[] = { dmeventdpath, NULL };
+
+	if (stat(fifos->client_path, &statbuf))
+		goto start_server;
+
+	if (!S_ISFIFO(statbuf.st_mode)) {
+		log_error("%s is not a fifo.", fifos->client_path);
+		return 0;
+	}
+
+	/* Anyone listening?  If not, errno will be ENXIO */
+	fifos->client = open(fifos->client_path, O_WRONLY | O_NONBLOCK);
+	if (fifos->client >= 0) {
+		/* server is running and listening */
+
+		close(fifos->client);
+		return 1;
+	} else if (errno != ENXIO) {
+		/* problem */
+
+		log_error("%s: Can't open client fifo %s: %s",
+			  __func__, fifos->client_path, strerror(errno));
+		stack;
+		return 0;
+	}
+
+      start_server:
+	/* server is not running */
+
+	if (!strncmp(DMEVENTD_PATH, "/", 1) && stat(DMEVENTD_PATH, &statbuf)) {
+		log_error("Unable to find dmeventd.");
+		return_0;
+	}
+
+	pid = fork();
+
+	if (pid < 0)
+		log_error("Unable to fork.");
+
+	else if (!pid) {
+		execvp(args[0], args);
+		_exit(EXIT_FAILURE);
+	} else {
+		if (waitpid(pid, &status, 0) < 0)
+			log_error("Unable to start dmeventd: %s",
+				  strerror(errno));
+		else if (WEXITSTATUS(status))
+			log_error("Unable to start dmeventd.");
+		else
+			ret = 1;
+	}
+
+	return ret;
+}
+
+/* Initialize client. */
+static int _init_client(struct dm_event_fifos *fifos)
+{
+	/* FIXME? Is fifo the most suitable method? Why not share
+	   comms/daemon code with something else e.g. multipath? */
+
+	/* init fifos */
+	memset(fifos, 0, sizeof(*fifos));
+	fifos->client_path = DM_EVENT_FIFO_CLIENT;
+	fifos->server_path = DM_EVENT_FIFO_SERVER;
+
+	if (!_start_daemon(fifos)) {
+		stack;
+		return 0;
+	}
+
+	/* Open the fifo used to read from the daemon. */
+	if ((fifos->server = open(fifos->server_path, O_RDWR)) < 0) {
+		log_error("%s: open server fifo %s",
+			  __func__, fifos->server_path);
+		stack;
+		return 0;
+	}
+
+	/* Lock out anyone else trying to do communication with the daemon. */
+	if (flock(fifos->server, LOCK_EX) < 0) {
+		log_error("%s: flock %s", __func__, fifos->server_path);
+		close(fifos->server);
+		return 0;
+	}
+
+/*	if ((fifos->client = open(fifos->client_path, O_WRONLY | O_NONBLOCK)) < 0) {*/
+	if ((fifos->client = open(fifos->client_path, O_RDWR | O_NONBLOCK)) < 0) {
+		log_error("%s: Can't open client fifo %s: %s",
+			  __func__, fifos->client_path, strerror(errno));
+		close(fifos->server);
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+static void _dtr_client(struct dm_event_fifos *fifos)
+{
+	if (flock(fifos->server, LOCK_UN))
+		log_error("flock unlock %s", fifos->server_path);
+
+	close(fifos->client);
+	close(fifos->server);
+}
+
+/* Get uuid of a device */
+static struct dm_task *_get_device_info(const struct dm_event_handler *dmevh)
+{
+	struct dm_task *dmt;
+	struct dm_info info;
+
+	if (!(dmt = dm_task_create(DM_DEVICE_INFO))) {
+		log_error("_get_device_info: dm_task creation for info failed");
+		return NULL;
+	}
+
+	if (dmevh->uuid)
+		dm_task_set_uuid(dmt, dmevh->uuid);
+	else if (dmevh->dev_name)
+		dm_task_set_name(dmt, dmevh->dev_name);
+	else if (dmevh->major && dmevh->minor) {
+		dm_task_set_major(dmt, dmevh->major);
+		dm_task_set_minor(dmt, dmevh->minor);
+        }
+
+	/* FIXME Add name or uuid or devno to messages */
+	if (!dm_task_run(dmt)) {
+		log_error("_get_device_info: dm_task_run() failed");
+		goto failed;
+	}
+
+	if (!dm_task_get_info(dmt, &info)) {
+		log_error("_get_device_info: failed to get info for device");
+		goto failed;
+	}
+
+	if (!info.exists) {
+		log_error("_get_device_info: device not found");
+		goto failed;
+	}
+
+	return dmt;
+
+failed:
+	dm_task_destroy(dmt);
+	return NULL;
+}
+
+/* Handle the event (de)registration call and return negative error codes. */
+static int _do_event(int cmd, struct dm_event_daemon_message *msg,
+		     const char *dso_name, const char *dev_name,
+		     enum dm_event_mask evmask, uint32_t timeout)
+{
+	int ret;
+	struct dm_event_fifos fifos;
+
+	if (!_init_client(&fifos)) {
+		stack;
+		return -ESRCH;
+	}
+
+	ret = _daemon_talk(&fifos, msg, DM_EVENT_CMD_HELLO, 0, 0, 0, 0);
+
+	if (msg->data)
+		dm_free(msg->data);
+	msg->data = 0;
+
+	if (!ret)
+		ret = _daemon_talk(&fifos, msg, cmd, dso_name, dev_name, evmask, timeout);
+
+	/* what is the opposite of init? */
+	_dtr_client(&fifos);
+
+	return ret;
+}
+
+/* External library interface. */
+int dm_event_register_handler(const struct dm_event_handler *dmevh)
+{
+	int ret = 1, err;
+	const char *uuid;
+	struct dm_task *dmt;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if ((err = _do_event(DM_EVENT_CMD_REGISTER_FOR_EVENT, &msg,
+			     dmevh->dso, uuid, dmevh->mask, dmevh->timeout)) < 0) {
+		log_error("%s: event registration failed: %s",
+			  dm_task_get_name(dmt),
+			  msg.data ? msg.data : strerror(-err));
+		ret = 0;
+	}
+
+	if (msg.data)
+		dm_free(msg.data);
+
+	dm_task_destroy(dmt);
+
+	return ret;
+}
+
+int dm_event_unregister_handler(const struct dm_event_handler *dmevh)
+{
+	int ret = 1, err;
+	const char *uuid;
+	struct dm_task *dmt;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if ((err = _do_event(DM_EVENT_CMD_UNREGISTER_FOR_EVENT, &msg,
+			    dmevh->dso, uuid, dmevh->mask, dmevh->timeout)) < 0) {
+		log_error("%s: event deregistration failed: %s",
+			  dm_task_get_name(dmt),
+			  msg.data ? msg.data : strerror(-err));
+		ret = 0;
+	}
+
+	if (msg.data)
+		dm_free(msg.data);
+
+	dm_task_destroy(dmt);
+
+	return ret;
+}
+
+/* Fetch a string off src and duplicate it into *dest. */
+/* FIXME: move to separate module to share with the daemon. */
+static char *_fetch_string(char **src, const int delimiter)
+{
+	char *p, *ret;
+
+	if ((p = strchr(*src, delimiter)))
+		*p = 0;
+
+	if ((ret = dm_strdup(*src)))
+		*src += strlen(ret) + 1;
+
+	if (p)
+		*p = delimiter;
+
+	return ret;
+}
+
+/* Parse a device message from the daemon. */
+static int _parse_message(struct dm_event_daemon_message *msg, char **dso_name,
+			 char **uuid, enum dm_event_mask *evmask)
+{
+	char *id = NULL;
+	char *p = msg->data;
+
+	if ((id = _fetch_string(&p, ' ')) &&
+	    (*dso_name = _fetch_string(&p, ' ')) &&
+	    (*uuid = _fetch_string(&p, ' '))) {
+		*evmask = atoi(p);
+
+		dm_free(id);
+		return 0;
+	}
+
+	if (id)
+		dm_free(id);
+	return -ENOMEM;
+}
+
+/*
+ * Returns 0 if handler found; error (-ENOMEM, -ENOENT) otherwise.
+ */
+int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
+{
+	int ret = 0;
+	const char *uuid = NULL;
+	char *reply_dso = NULL, *reply_uuid = NULL;
+	enum dm_event_mask reply_mask = 0;
+	struct dm_task *dmt = NULL;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if (!(ret = _do_event(next ? DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE :
+			     DM_EVENT_CMD_GET_REGISTERED_DEVICE,
+			      &msg, dmevh->dso, uuid, dmevh->mask, 0))) {
+		/* FIXME this will probably horribly break if we get
+		   ill-formatted reply */
+		ret = _parse_message(&msg, &reply_dso, &reply_uuid, &reply_mask);
+	} else {
+		ret = -ENOENT;
+		goto fail;
+	}
+
+	dm_task_destroy(dmt);
+	dmt = NULL;
+
+	if (msg.data) {
+		dm_free(msg.data);
+		msg.data = NULL;
+	}
+
+	_dm_event_handler_clear_dev_info(dmevh);
+	dmevh->uuid = dm_strdup(reply_uuid);
+	if (!dmevh->uuid) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		ret = -ENXIO; /* dmeventd probably gave us bogus uuid back */
+		goto fail;
+	}
+
+	dm_event_handler_set_dso(dmevh, reply_dso);
+	dm_event_handler_set_event_mask(dmevh, reply_mask);
+
+	if (reply_dso) {
+		dm_free(reply_dso);
+		reply_dso = NULL;
+	}
+
+	if (reply_uuid) {
+		dm_free(reply_uuid);
+		reply_uuid = NULL;
+	}
+
+	dmevh->dev_name = dm_strdup(dm_task_get_name(dmt));
+	if (!dmevh->dev_name) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	struct dm_info info;
+	if (!dm_task_get_info(dmt, &info)) {
+		ret = -1;
+		goto fail;
+	}
+
+	dmevh->major = info.major;
+	dmevh->minor = info.minor;
+
+	dm_task_destroy(dmt);
+
+	return ret;
+
+ fail:
+	if (msg.data)
+		dm_free(msg.data);
+	if (reply_dso)
+		dm_free(reply_dso);
+	if (reply_uuid)
+		dm_free(reply_uuid);
+	_dm_event_handler_clear_dev_info(dmevh);
+	if (dmt)
+		dm_task_destroy(dmt);
+	return ret;
+}
+
+#if 0				/* left out for now */
+
+static char *_skip_string(char *src, const int delimiter)
+{
+	src = srtchr(src, delimiter);
+	if (src && *(src + 1))
+		return src + 1;
+	return NULL;
+}
+
+int dm_event_set_timeout(const char *device_path, uint32_t timeout)
+{
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!device_exists(device_path))
+		return -ENODEV;
+
+	return _do_event(DM_EVENT_CMD_SET_TIMEOUT, &msg,
+			 NULL, device_path, 0, timeout);
+}
+
+int dm_event_get_timeout(const char *device_path, uint32_t *timeout)
+{
+	int ret;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!device_exists(device_path))
+		return -ENODEV;
+	if (!(ret = _do_event(DM_EVENT_CMD_GET_TIMEOUT, &msg, NULL, device_path,
+			     0, 0))) {
+		char *p = _skip_string(msg.data, ' ');
+		if (!p) {
+			log_error("malformed reply from dmeventd '%s'\n",
+				  msg.data);
+			return -EIO;
+		}
+		*timeout = atoi(p);
+	}
+	if (msg.data)
+		dm_free(msg.data);
+	return ret;
+}
+#endif

daemons/dmeventd/libdevmapper-event.h

@@ -0,0 +1,106 @@
+/*
+ * Copyright (C) 2005-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of the device-mapper userspace tools.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Note that this file is released only as part of a technology preview
+ * and its contents may change in future updates in ways that do not
+ * preserve compatibility.
+ */
+
+#ifndef LIB_DMEVENT_H
+#define LIB_DMEVENT_H
+
+#include <stdint.h>
+
+/*
+ * Event library interface.
+ */
+
+enum dm_event_mask {
+	DM_EVENT_SETTINGS_MASK  = 0x0000FF,
+	DM_EVENT_SINGLE		= 0x000001, /* Report multiple errors just once. */
+	DM_EVENT_MULTI		= 0x000002, /* Report all of them. */
+
+	DM_EVENT_ERROR_MASK     = 0x00FF00,
+	DM_EVENT_SECTOR_ERROR	= 0x000100, /* Failure on a particular sector. */
+	DM_EVENT_DEVICE_ERROR	= 0x000200, /* Device failure. */
+	DM_EVENT_PATH_ERROR	= 0x000400, /* Failure on an io path. */
+	DM_EVENT_ADAPTOR_ERROR	= 0x000800, /* Failure of a host adaptor. */
+
+	DM_EVENT_STATUS_MASK    = 0xFF0000,
+	DM_EVENT_SYNC_STATUS	= 0x010000, /* Mirror synchronization completed/failed. */
+	DM_EVENT_TIMEOUT	= 0x020000, /* Timeout has occured */
+
+	DM_EVENT_REGISTRATION_PENDING = 0x1000000, /* Monitor thread is setting-up/shutting-down */
+};
+
+#define DM_EVENT_ALL_ERRORS DM_EVENT_ERROR_MASK
+
+struct dm_event_handler;
+
+struct dm_event_handler *dm_event_handler_create(void);
+void dm_event_handler_destroy(struct dm_event_handler *dmevh);
+
+/*
+ * Path of shared library to handle events.
+ *
+ * All of dso, device_name and uuid strings are duplicated, you do not
+ * need to keep the pointers valid after the call succeeds. Thes may
+ * return -ENOMEM though.
+ */
+int dm_event_handler_set_dso(struct dm_event_handler *dmevh, const char *path);
+
+/*
+ * Identify the device to monitor by exactly one of device_name, uuid or
+ * device number. String arguments are duplicated, see above.
+ */
+int dm_event_handler_set_dev_name(struct dm_event_handler *dmevh, const char *device_name);
+
+int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid);
+
+void dm_event_handler_set_major(struct dm_event_handler *dmevh, int major);
+void dm_event_handler_set_minor(struct dm_event_handler *dmevh, int minor);
+void dm_event_handler_set_timeout(struct dm_event_handler *dmevh, int timeout);
+
+/*
+ * Specify mask for events to monitor.
+ */
+void dm_event_handler_set_event_mask(struct dm_event_handler *dmevh,
+				     enum dm_event_mask evmask);
+
+const char *dm_event_handler_get_dso(const struct dm_event_handler *dmevh);
+const char *dm_event_handler_get_dev_name(const struct dm_event_handler *dmevh);
+const char *dm_event_handler_get_uuid(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_major(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_minor(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_timeout(const struct dm_event_handler *dmevh);
+enum dm_event_mask dm_event_handler_get_event_mask(const struct dm_event_handler *dmevh);
+
+/* FIXME Review interface (what about this next thing?) */
+int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next);
+
+/*
+ * Initiate monitoring using dmeventd.
+ */
+int dm_event_register_handler(const struct dm_event_handler *dmevh);
+int dm_event_unregister_handler(const struct dm_event_handler *dmevh);
+
+/* Prototypes for DSO interface, see dmeventd.c, struct dso_data for
+   detailed descriptions. */
+void process_event(struct dm_task *dmt, enum dm_event_mask evmask, void **user);
+int register_device(const char *device_name, const char *uuid, int major, int minor, void **user);
+int unregister_device(const char *device_name, const char *uuid, int major,
+		      int minor, void **user);
+
+#endif

daemons/dmeventd/libdevmapper-event.pc.in

@@ -0,0 +1,12 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: devmapper-event
+Description: device-mapper event library
+Version: @DM_LIB_VERSION@
+Requires: devmapper
+Cflags: -I${includedir}
+Libs: -L${libdir} -ldevmapper-event
+Libs.private: -lpthread -ldl

daemons/dmeventd/plugins/Makefile.in

@@ -0,0 +1,26 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+SUBDIRS += lvm2 mirror snapshot
+
+mirror: lvm2
+snapshot: lvm2
+
+include ../../../make.tmpl
+

daemons/dmeventd/plugins/lvm2/.exported_symbols

@@ -0,0 +1,6 @@
+dmeventd_lvm2_init
+dmeventd_lvm2_exit
+dmeventd_lvm2_lock
+dmeventd_lvm2_unlock
+dmeventd_lvm2_pool
+dmeventd_lvm2_run

daemons/dmeventd/plugins/lvm2/Makefile.in

@@ -0,0 +1,39 @@
+#
+# Copyright (C) 2010 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+CLDFLAGS += -L${top_builddir}/tools
+
+SOURCES = dmeventd_lvm.c
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event-lvm2.dylib
+else
+  LIB_SHARED = libdevmapper-event-lvm2.so
+endif
+LIB_VERSION = $(LIB_VERSION_LVM)
+
+include $(top_builddir)/make.tmpl
+
+LIBS += -ldevmapper @LIB_PTHREAD@ @LVM2CMD_LIB@
+
+install_lvm2: libdevmapper-event-lvm2.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/$<.$(LIB_VERSION)
+	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
+
+install: install_lvm2

daemons/dmeventd/plugins/lvm2/dmeventd_lvm.c

@@ -0,0 +1,127 @@
+/*
+ * Copyright (C) 2010 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+
+#include "lvm2cmd.h"
+#include "errors.h"
+#include "libdevmapper-event.h"
+#include "dmeventd_lvm.h"
+
+#include <pthread.h>
+#include <syslog.h>
+
+/*
+ * register_device() is called first and performs initialisation.
+ * Only one device may be registered or unregistered at a time.
+ */
+static pthread_mutex_t _register_mutex = PTHREAD_MUTEX_INITIALIZER;
+
+/*
+ * Number of active registrations.
+ */
+static int _register_count = 0;
+static struct dm_pool *_mem_pool = NULL;
+static void *_lvm_handle = NULL;
+
+/*
+ * Currently only one event can be processed at a time.
+ */
+static pthread_mutex_t _event_mutex = PTHREAD_MUTEX_INITIALIZER;
+
+/* FIXME Remove this: Pass messages back to dmeventd core for processing. */
+static void _temporary_log_fn(int level,
+			      const char *file __attribute((unused)),
+			      int line __attribute((unused)),
+			      int dm_errno __attribute((unused)),
+			      const char *format)
+{
+	level &= ~_LOG_STDERR;
+
+	if (!strncmp(format, "WARNING: ", 9) && (level < 5))
+		syslog(LOG_CRIT, "%s", format);
+	else
+		syslog(LOG_DEBUG, "%s", format);
+}
+
+void dmeventd_lvm2_lock(void)
+{
+	if (pthread_mutex_trylock(&_event_mutex)) {
+		syslog(LOG_NOTICE, "Another thread is handling an event. Waiting...");
+		pthread_mutex_lock(&_event_mutex);
+	}
+}
+
+void dmeventd_lvm2_unlock(void)
+{
+	pthread_mutex_unlock(&_event_mutex);
+}
+
+int dmeventd_lvm2_init(void)
+{
+	int r = 0;
+
+	pthread_mutex_lock(&_register_mutex);
+
+	/*
+	 * Need some space for allocations.  1024 should be more
+	 * than enough for what we need (device mapper name splitting)
+	 */
+	if (!_mem_pool && !(_mem_pool = dm_pool_create("mirror_dso", 1024)))
+		goto out;
+
+	if (!_lvm_handle) {
+		lvm2_log_fn(_temporary_log_fn);
+		if (!(_lvm_handle = lvm2_init())) {
+			dm_pool_destroy(_mem_pool);
+			_mem_pool = NULL;
+			goto out;
+		}
+		/* FIXME Temporary: move to dmeventd core */
+		lvm2_run(_lvm_handle, "_memlock_inc");
+	}
+
+	_register_count++;
+	r = 1;
+
+out:
+	pthread_mutex_unlock(&_register_mutex);
+	return r;
+}
+
+void dmeventd_lvm2_exit(void)
+{
+	pthread_mutex_lock(&_register_mutex);
+
+	if (!--_register_count) {
+		dm_pool_destroy(_mem_pool);
+		_mem_pool = NULL;
+		lvm2_run(_lvm_handle, "_memlock_dec");
+		lvm2_exit(_lvm_handle);
+		_lvm_handle = NULL;
+	}
+
+	pthread_mutex_unlock(&_register_mutex);
+}
+
+struct dm_pool *dmeventd_lvm2_pool(void)
+{
+	return _mem_pool;
+}
+
+int dmeventd_lvm2_run(const char *cmdline)
+{
+	return lvm2_run(_lvm_handle, cmdline);
+}
+

daemons/dmeventd/plugins/lvm2/dmeventd_lvm.h

@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2010 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Wrappers around liblvm2cmd functions for dmeventd plug-ins.
+ *
+ * liblvm2cmd is not thread-safe so the locking in this library helps dmeventd
+ * threads to co-operate in sharing a single instance.
+ *
+ * FIXME Either support this properly as a generic liblvm2cmd wrapper or make
+ * liblvm2cmd thread-safe so this can go away.
+ */
+
+#include "libdevmapper.h"
+
+#ifndef _DMEVENTD_LVMWRAP_H
+#define _DMEVENTD_LVMWRAP_H
+
+int dmeventd_lvm2_init(void);
+void dmeventd_lvm2_exit(void);
+int dmeventd_lvm2_run(const char *cmdline);
+
+void dmeventd_lvm2_lock(void);
+void dmeventd_lvm2_unlock(void);
+
+struct dm_pool *dmeventd_lvm2_pool(void);
+
+#endif /* _DMEVENTD_LVMWRAP_H */

daemons/dmeventd/plugins/mirror/.exported_symbols

@@ -0,0 +1,3 @@
+process_event
+register_device
+unregister_device

daemons/dmeventd/plugins/mirror/Makefile.in

@@ -0,0 +1,42 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005, 2008-2010 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+INCLUDES += -I${top_srcdir}/tools -I$(top_srcdir)/daemons/dmeventd/plugins/lvm2
+CLDFLAGS += -L${top_builddir}/tools -L${top_builddir}/daemons/dmeventd/plugins/lvm2
+
+SOURCES = dmeventd_mirror.c
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event-lvm2mirror.dylib
+else
+  LIB_SHARED = libdevmapper-event-lvm2mirror.so
+endif
+
+LIB_VERSION = $(LIB_VERSION_LVM)
+
+include $(top_builddir)/make.tmpl
+
+LIBS += -ldevmapper @LIB_PTHREAD@ @LVM2CMD_LIB@ -ldevmapper-event-lvm2
+
+install_lvm2: libdevmapper-event-lvm2mirror.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/$<.$(LIB_VERSION)
+	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
+
+install: install_lvm2

daemons/dmeventd/plugins/mirror/dmeventd_mirror.c

@@ -0,0 +1,243 @@
+/*
+ * Copyright (C) 2005-2010 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+
+#include "lvm2cmd.h"
+#include "errors.h"
+#include "libdevmapper-event.h"
+#include "dmeventd_lvm.h"
+
+#include <pthread.h>
+
+#include <syslog.h> /* FIXME Replace syslog with multilog */
+/* FIXME Missing openlog? */
+/* FIXME Replace most syslogs with log_error() style messages and add complete context. */
+/* FIXME Reformat to 80 char lines. */
+
+#define ME_IGNORE    0
+#define ME_INSYNC    1
+#define ME_FAILURE   2
+
+static int _process_status_code(const char status_code, const char *dev_name,
+				const char *dev_type, int r)
+{
+	/*
+	 *    A => Alive - No failures
+	 *    D => Dead - A write failure occurred leaving mirror out-of-sync
+	 *    F => Flush failed.
+	 *    S => Sync - A sychronization failure occurred, mirror out-of-sync
+	 *    R => Read - A read failure occurred, mirror data unaffected
+	 *    U => Unclassified failure (bug)
+	 */ 
+	if (status_code == 'F') {
+		syslog(LOG_ERR, "%s device %s flush failed.\n",
+		       dev_type, dev_name);
+		r = ME_FAILURE;
+	} else if (status_code == 'S')
+		syslog(LOG_ERR, "%s device %s sync failed.\n",
+		       dev_type, dev_name);
+	else if (status_code == 'R')
+		syslog(LOG_ERR, "%s device %s read failed.\n",
+		       dev_type, dev_name);
+	else if (status_code != 'A') {
+		syslog(LOG_ERR, "%s device %s has failed (%c).\n",
+		       dev_type, dev_name, status_code);
+		r = ME_FAILURE;
+	}
+
+	return r;
+}
+
+static int _get_mirror_event(char *params)
+{
+	int i, r = ME_INSYNC;
+	char **args = NULL;
+	char *dev_status_str;
+	char *log_status_str;
+	char *sync_str;
+	char *p = NULL;
+	int log_argc, num_devs;
+
+	/*
+	 * dm core parms:	     0 409600 mirror
+	 * Mirror core parms:	     2 253:4 253:5 400/400
+	 * New-style failure params: 1 AA
+	 * New-style log params:     3 cluster 253:3 A
+	 *			 or  3 disk 253:3 A
+	 *			 or  1 core
+	 */
+
+	/* number of devices */
+	if (!dm_split_words(params, 1, 0, &p))
+		goto out_parse;
+
+	if (!(num_devs = atoi(p)))
+		goto out_parse;
+	p += strlen(p) + 1;
+
+	/* devices names + "400/400" + "1 AA" + 1 or 3 log parms + NULL */
+	args = dm_malloc((num_devs + 7) * sizeof(char *));
+	if (!args || dm_split_words(p, num_devs + 7, 0, args) < num_devs + 5)
+		goto out_parse;
+
+	dev_status_str = args[2 + num_devs];
+	log_argc = atoi(args[3 + num_devs]);
+	log_status_str = args[3 + num_devs + log_argc];
+	sync_str = args[num_devs];
+
+	/* Check for bad mirror devices */
+	for (i = 0; i < num_devs; i++)
+		r = _process_status_code(dev_status_str[i], args[i],
+			i ? "Secondary mirror" : "Primary mirror", r);
+
+	/* Check for bad disk log device */
+	if (log_argc > 1)
+		r = _process_status_code(log_status_str[0],
+					 args[2 + num_devs + log_argc],
+					 "Log", r);
+
+	if (r == ME_FAILURE)
+		goto out;
+
+	p = strstr(sync_str, "/");
+	if (p) {
+		p[0] = '\0';
+		if (strcmp(sync_str, p+1))
+			r = ME_IGNORE;
+		p[0] = '/';
+	} else
+		goto out_parse;
+
+out:
+	if (args)
+		dm_free(args);
+	return r;
+	
+out_parse:
+	if (args)
+		dm_free(args);
+	syslog(LOG_ERR, "Unable to parse mirror status string.");
+	return ME_IGNORE;
+}
+
+static int _remove_failed_devices(const char *device)
+{
+	int r;
+#define CMD_SIZE 256	/* FIXME Use system restriction */
+	char cmd_str[CMD_SIZE];
+	char *vg = NULL, *lv = NULL, *layer = NULL;
+
+	if (strlen(device) > 200)  /* FIXME Use real restriction */
+		return -ENAMETOOLONG;	/* FIXME These return code distinctions are not used so remove them! */
+
+	if (!dm_split_lvm_name(dmeventd_lvm2_pool(), device, &vg, &lv, &layer)) {
+		syslog(LOG_ERR, "Unable to determine VG name from %s",
+		       device);
+		return -ENOMEM;	/* FIXME Replace with generic error return - reason for failure has already got logged */
+	}
+
+	/* FIXME Is any sanity-checking required on %s? */
+	if (CMD_SIZE <= snprintf(cmd_str, CMD_SIZE, "lvconvert --config devices{ignore_suspended_devices=1} --repair --use-policies %s/%s", vg, lv)) {
+		/* this error should be caught above, but doesn't hurt to check again */
+		syslog(LOG_ERR, "Unable to form LVM command: Device name too long");
+		return -ENAMETOOLONG; /* FIXME Replace with generic error return - reason for failure has already got logged */
+	}
+
+	r = dmeventd_lvm2_run(cmd_str);
+
+	syslog(LOG_INFO, "Repair of mirrored LV %s/%s %s.", vg, lv, (r == ECMD_PROCESSED) ? "finished successfully" : "failed");
+
+	return (r == ECMD_PROCESSED) ? 0 : -1;
+}
+
+void process_event(struct dm_task *dmt,
+		   enum dm_event_mask event __attribute((unused)),
+		   void **unused __attribute((unused)))
+{
+	void *next = NULL;
+	uint64_t start, length;
+	char *target_type = NULL;
+	char *params;
+	const char *device = dm_task_get_name(dmt);
+
+	dmeventd_lvm2_lock();
+
+	do {
+		next = dm_get_next_target(dmt, next, &start, &length,
+					  &target_type, &params);
+
+		if (!target_type) {
+			syslog(LOG_INFO, "%s mapping lost.\n", device);
+			continue;
+		}
+
+		if (strcmp(target_type, "mirror")) {
+			syslog(LOG_INFO, "%s has unmirrored portion.\n", device);
+			continue;
+		}
+
+		switch(_get_mirror_event(params)) {
+		case ME_INSYNC:
+			/* FIXME: all we really know is that this
+			   _part_ of the device is in sync
+			   Also, this is not an error
+			*/
+			syslog(LOG_NOTICE, "%s is now in-sync\n", device);
+			break;
+		case ME_FAILURE:
+			syslog(LOG_ERR, "Device failure in %s\n", device);
+			if (_remove_failed_devices(device))
+				/* FIXME Why are all the error return codes unused? Get rid of them? */
+				syslog(LOG_ERR, "Failed to remove faulty devices in %s\n",
+				       device);
+			/* Should check before warning user that device is now linear
+			else
+				syslog(LOG_NOTICE, "%s is now a linear device.\n",
+					device);
+			*/
+			break;
+		case ME_IGNORE:
+			break;
+		default:
+			/* FIXME Provide value then! */
+			syslog(LOG_INFO, "Unknown event received.\n");
+		}
+	} while (next);
+
+	dmeventd_lvm2_unlock();
+}
+
+int register_device(const char *device,
+		    const char *uuid __attribute((unused)),
+		    int major __attribute((unused)),
+		    int minor __attribute((unused)),
+		    void **unused __attribute((unused)))
+{
+	int r = dmeventd_lvm2_init();
+	syslog(LOG_INFO, "Monitoring mirror device %s for events", device);
+	return r;
+}
+
+int unregister_device(const char *device,
+		      const char *uuid __attribute((unused)),
+		      int major __attribute((unused)),
+		      int minor __attribute((unused)),
+		      void **unused __attribute((unused)))
+{
+	syslog(LOG_INFO, "No longer monitoring mirror device %s for events\n",
+	       device);
+	dmeventd_lvm2_exit();
+	return 1;
+}

daemons/dmeventd/plugins/snapshot/.exported_symbols

@@ -0,0 +1,3 @@
+process_event
+register_device
+unregister_device

daemons/dmeventd/plugins/snapshot/Makefile.in

@@ -0,0 +1,42 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2010 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+INCLUDES += -I${top_srcdir}/tools -I$(top_srcdir)/daemons/dmeventd/plugins/lvm2
+CLDFLAGS += -L${top_builddir}/tools -L${top_builddir}/daemons/dmeventd/plugins/lvm2
+
+SOURCES = dmeventd_snapshot.c
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event-lvm2snapshot.dylib
+else
+  LIB_SHARED = libdevmapper-event-lvm2snapshot.so
+endif
+
+LIB_VERSION = $(LIB_VERSION_LVM)
+
+include $(top_builddir)/make.tmpl
+
+LIBS += -ldevmapper @LIB_PTHREAD@ @LVM2CMD_LIB@ -ldevmapper-event-lvm2
+
+install_lvm2: libdevmapper-event-lvm2snapshot.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/$<.$(LIB_VERSION)
+	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
+
+install: install_lvm2

daemons/dmeventd/plugins/snapshot/dmeventd_snapshot.c

@@ -0,0 +1,141 @@
+/*
+ * Copyright (C) 2007-2008 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+
+#include "lvm2cmd.h"
+#include "errors.h"
+#include "libdevmapper-event.h"
+#include "dmeventd_lvm.h"
+
+#include "lvm-string.h"
+
+#include <pthread.h>
+#include <syslog.h> /* FIXME Replace syslog with multilog */
+/* FIXME Missing openlog? */
+
+/* First warning when snapshot is 80% full. */
+#define WARNING_THRESH 80
+/* Further warnings at 85%, 90% and 95% fullness. */
+#define WARNING_STEP 5
+
+struct snap_status {
+	int invalid;
+	int used;
+	int max;
+};
+
+/* FIXME possibly reconcile this with target_percent when we gain
+   access to regular LVM library here. */
+static void _parse_snapshot_params(char *params, struct snap_status *stat)
+{
+	char *p;
+	/*
+	 * xx/xx	-- fractions used/max
+	 * Invalid	-- snapshot invalidated
+	 * Unknown	-- status unknown
+	 */
+	stat->used = stat->max = 0;
+
+	if (!strncmp(params, "Invalid", 7)) {
+		stat->invalid = 1;
+		return;
+	}
+
+	/*
+	 * When we return without setting non-zero max, the parent is
+	 * responsible for reporting errors.
+	 */
+	if (!strncmp(params, "Unknown", 7))
+		return;
+
+	if (!(p = strstr(params, "/")))
+		return;
+
+	*p = '\0';
+	p++;
+
+	stat->used = atoi(params);
+	stat->max = atoi(p);
+}
+
+void process_event(struct dm_task *dmt,
+		   enum dm_event_mask event __attribute((unused)),
+		   void **private)
+{
+	void *next = NULL;
+	uint64_t start, length;
+	char *target_type = NULL;
+	char *params;
+	struct snap_status stat = { 0 };
+	const char *device = dm_task_get_name(dmt);
+	int percent, *percent_warning = (int*)private;
+
+	/* No longer monitoring, waiting for remove */
+	if (!*percent_warning)
+		return;
+
+	dmeventd_lvm2_lock();
+
+	dm_get_next_target(dmt, next, &start, &length, &target_type, &params);
+	if (!target_type)
+		goto out;
+
+	_parse_snapshot_params(params, &stat);
+	/*
+	 * If the snapshot has been invalidated or we failed to parse
+	 * the status string. Report the full status string to syslog.
+	 */
+	if (stat.invalid || !stat.max) {
+		syslog(LOG_ERR, "Snapshot %s changed state to: %s\n", device, params);
+		*percent_warning = 0;
+		goto out;
+	}
+
+	percent = 100 * stat.used / stat.max;
+	if (percent >= *percent_warning) {
+		syslog(LOG_WARNING, "Snapshot %s is now %i%% full.\n", device, percent);
+		/* Print warning on the next multiple of WARNING_STEP. */
+		*percent_warning = (percent / WARNING_STEP) * WARNING_STEP + WARNING_STEP;
+	}
+out:
+	dmeventd_lvm2_unlock();
+}
+
+int register_device(const char *device,
+		    const char *uuid __attribute((unused)),
+		    int major __attribute((unused)),
+		    int minor __attribute((unused)),
+		    void **private)
+{
+	int *percent_warning = (int*)private;
+	int r = dmeventd_lvm2_init();
+
+	*percent_warning = WARNING_THRESH; /* Print warning if snapshot is full */
+
+	syslog(LOG_INFO, "Monitoring snapshot %s\n", device);
+	return r;
+}
+
+int unregister_device(const char *device,
+		      const char *uuid __attribute((unused)),
+		      int major __attribute((unused)),
+		      int minor __attribute((unused)),
+		      void **unused __attribute((unused)))
+{
+	syslog(LOG_INFO, "No longer monitoring snapshot %s\n",
+	       device);
+	dmeventd_lvm2_exit();
+	return 1;
+}

doc/Makefile.in

@@ -0,0 +1,31 @@
+#
+# Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+CONFSRC=example.conf
+CONFDEST=lvm.conf
+
+include ../make.tmpl
+
+install_lvm2: $(CONFSRC)
+	@if [ ! -e $(confdir)/$(CONFDEST) ]; then \
+		echo "Installing $(CONFSRC) as $(confdir)/$(CONFDEST)"; \
+		@INSTALL@ -D $(OWNER) $(GROUP) -m 644 $< \
+			$(confdir)/$(CONFDEST); \
+	fi
+
+install: install_lvm2

doc/README

@@ -1 +0,0 @@
-Wow!  This is really incredible documentation!

doc/example.conf

@@ -0,0 +1,474 @@
+# This is an example configuration file for the LVM2 system.
+# It contains the default settings that would be used if there was no
+# /etc/lvm/lvm.conf file.
+#
+# Refer to 'man lvm.conf' for further information including the file layout.
+#
+# To put this file in a different directory and override /etc/lvm set
+# the environment variable LVM_SYSTEM_DIR before running the tools.
+
+
+# This section allows you to configure which block devices should
+# be used by the LVM system.
+devices {
+
+    # Where do you want your volume groups to appear ?
+    dir = "/dev"
+
+    # An array of directories that contain the device nodes you wish
+    # to use with LVM2.
+    scan = [ "/dev" ]
+
+    # If several entries in the scanned directories correspond to the
+    # same block device and the tools need to display a name for device,
+    # all the pathnames are matched against each item in the following
+    # list of regular expressions in turn and the first match is used.
+    preferred_names = [ ]
+
+    # Try to avoid using undescriptive /dev/dm-N names, if present.
+    # preferred_names = [ "^/dev/mpath/", "^/dev/mapper/mpath", "^/dev/[hs]d" ]
+
+    # A filter that tells LVM2 to only use a restricted set of devices.
+    # The filter consists of an array of regular expressions.  These
+    # expressions can be delimited by a character of your choice, and
+    # prefixed with either an 'a' (for accept) or 'r' (for reject).
+    # The first expression found to match a device name determines if
+    # the device will be accepted or rejected (ignored).  Devices that
+    # don't match any patterns are accepted.
+
+    # Be careful if there there are symbolic links or multiple filesystem 
+    # entries for the same device as each name is checked separately against
+    # the list of patterns.  The effect is that if any name matches any 'a'
+    # pattern, the device is accepted; otherwise if any name matches any 'r'
+    # pattern it is rejected; otherwise it is accepted.
+
+    # Don't have more than one filter line active at once: only one gets used.
+
+    # Run vgscan after you change this parameter to ensure that
+    # the cache file gets regenerated (see below).
+    # If it doesn't do what you expect, check the output of 'vgscan -vvvv'.
+
+
+    # By default we accept every block device:
+    filter = [ "a/.*/" ]
+
+    # Exclude the cdrom drive
+    # filter = [ "r|/dev/cdrom|" ]
+
+    # When testing I like to work with just loopback devices:
+    # filter = [ "a/loop/", "r/.*/" ]
+
+    # Or maybe all loops and ide drives except hdc:
+    # filter =[ "a|loop|", "r|/dev/hdc|", "a|/dev/ide|", "r|.*|" ]
+
+    # Use anchors if you want to be really specific
+    # filter = [ "a|^/dev/hda8$|", "r/.*/" ]
+
+    # The results of the filtering are cached on disk to avoid
+    # rescanning dud devices (which can take a very long time).
+    # By default this cache is stored in the /etc/lvm/cache directory
+    # in a file called '.cache'.
+    # It is safe to delete the contents: the tools regenerate it.
+    # (The old setting 'cache' is still respected if neither of
+    # these new ones is present.)
+    cache_dir = "/etc/lvm/cache"
+    cache_file_prefix = ""
+
+    # You can turn off writing this cache file by setting this to 0.
+    write_cache_state = 1
+
+    # Advanced settings.
+
+    # List of pairs of additional acceptable block device types found 
+    # in /proc/devices with maximum (non-zero) number of partitions.
+    # types = [ "fd", 16 ]
+
+    # If sysfs is mounted (2.6 kernels) restrict device scanning to 
+    # the block devices it believes are valid.
+    # 1 enables; 0 disables.
+    sysfs_scan = 1
+
+    # By default, LVM2 will ignore devices used as components of
+    # software RAID (md) devices by looking for md superblocks.
+    # 1 enables; 0 disables.
+    md_component_detection = 1
+
+    # By default, if a PV is placed directly upon an md device, LVM2
+    # will align its data blocks with the md device's stripe-width.
+    # 1 enables; 0 disables.
+    md_chunk_alignment = 1
+
+    # By default, the start of a PV's data area will be a multiple of
+    # the 'minimum_io_size' or 'optimal_io_size' exposed in sysfs.
+    # - minimum_io_size - the smallest request the device can perform
+    #   w/o incurring a read-modify-write penalty (e.g. MD's chunk size)
+    # - optimal_io_size - the device's preferred unit of receiving I/O
+    #   (e.g. MD's stripe width)
+    # minimum_io_size is used if optimal_io_size is undefined (0).
+    # If md_chunk_alignment is enabled, that detects the optimal_io_size.
+    # This setting takes precedence over md_chunk_alignment.
+    # 1 enables; 0 disables.
+    data_alignment_detection = 1
+
+    # Alignment (in KB) of start of data area when creating a new PV.
+    # If a PV is placed directly upon an md device and md_chunk_alignment or
+    # data_alignment_detection is enabled this parameter is ignored.
+    # Set to 0 for the default alignment of 64KB or page size, if larger.
+    data_alignment = 0
+
+    # By default, the start of the PV's aligned data area will be shifted by
+    # the 'alignment_offset' exposed in sysfs.  This offset is often 0 but
+    # may be non-zero; e.g.: certain 4KB sector drives that compensate for
+    # windows partitioning will have an alignment_offset of 3584 bytes
+    # (sector 7 is the lowest aligned logical block, the 4KB sectors start
+    # at LBA -1, and consequently sector 63 is aligned on a 4KB boundary).
+    # 1 enables; 0 disables.
+    data_alignment_offset_detection = 1
+
+    # If, while scanning the system for PVs, LVM2 encounters a device-mapper
+    # device that has its I/O suspended, it waits for it to become accessible.
+    # Set this to 1 to skip such devices.  This should only be needed
+    # in recovery situations.
+    ignore_suspended_devices = 0
+}
+
+# This section that allows you to configure the nature of the
+# information that LVM2 reports.
+log {
+
+    # Controls the messages sent to stdout or stderr.
+    # There are three levels of verbosity, 3 being the most verbose.
+    verbose = 0
+
+    # Should we send log messages through syslog?
+    # 1 is yes; 0 is no.
+    syslog = 1
+
+    # Should we log error and debug messages to a file?
+    # By default there is no log file.
+    #file = "/var/log/lvm2.log"
+
+    # Should we overwrite the log file each time the program is run?
+    # By default we append.
+    overwrite = 0
+
+    # What level of log messages should we send to the log file and/or syslog?
+    # There are 6 syslog-like log levels currently in use - 2 to 7 inclusive.
+    # 7 is the most verbose (LOG_DEBUG).
+    level = 0
+
+    # Format of output messages
+    # Whether or not (1 or 0) to indent messages according to their severity
+    indent = 1
+
+    # Whether or not (1 or 0) to display the command name on each line output
+    command_names = 0
+
+    # A prefix to use before the message text (but after the command name,
+    # if selected).  Default is two spaces, so you can see/grep the severity
+    # of each message.
+    prefix = "  "
+
+    # To make the messages look similar to the original LVM tools use:
+    #   indent = 0
+    #   command_names = 1
+    #   prefix = " -- "
+
+    # Set this if you want log messages during activation.
+    # Don't use this in low memory situations (can deadlock).
+    # activation = 0
+}
+
+# Configuration of metadata backups and archiving.  In LVM2 when we
+# talk about a 'backup' we mean making a copy of the metadata for the
+# *current* system.  The 'archive' contains old metadata configurations.
+# Backups are stored in a human readeable text format.
+backup {
+
+    # Should we maintain a backup of the current metadata configuration ?
+    # Use 1 for Yes; 0 for No.
+    # Think very hard before turning this off!
+    backup = 1
+
+    # Where shall we keep it ?
+    # Remember to back up this directory regularly!
+    backup_dir = "/etc/lvm/backup"
+
+    # Should we maintain an archive of old metadata configurations.
+    # Use 1 for Yes; 0 for No.
+    # On by default.  Think very hard before turning this off.
+    archive = 1
+
+    # Where should archived files go ?
+    # Remember to back up this directory regularly!
+    archive_dir = "/etc/lvm/archive"
+
+    # What is the minimum number of archive files you wish to keep ?
+    retain_min = 10
+
+    # What is the minimum time you wish to keep an archive file for ?
+    retain_days = 30
+}
+
+# Settings for the running LVM2 in shell (readline) mode.
+shell {
+
+    # Number of lines of history to store in ~/.lvm_history
+    history_size = 100
+}
+
+
+# Miscellaneous global LVM2 settings
+global {
+
+    # The file creation mask for any files and directories created.
+    # Interpreted as octal if the first digit is zero.
+    umask = 077
+
+    # Allow other users to read the files
+    #umask = 022
+
+    # Enabling test mode means that no changes to the on disk metadata
+    # will be made.  Equivalent to having the -t option on every
+    # command.  Defaults to off.
+    test = 0
+
+    # Default value for --units argument
+    units = "h"
+
+    # Since version 2.02.54, the tools distinguish between powers of
+    # 1024 bytes (e.g. KiB, MiB, GiB) and powers of 1000 bytes (e.g.
+    # KB, MB, GB).
+    # If you have scripts that depend on the old behaviour, set this to 0
+    # temporarily until you update them.
+    si_unit_consistency = 1
+
+    # Whether or not to communicate with the kernel device-mapper.
+    # Set to 0 if you want to use the tools to manipulate LVM metadata 
+    # without activating any logical volumes.
+    # If the device-mapper kernel driver is not present in your kernel
+    # setting this to 0 should suppress the error messages.
+    activation = 1
+
+    # If we can't communicate with device-mapper, should we try running 
+    # the LVM1 tools?
+    # This option only applies to 2.4 kernels and is provided to help you
+    # switch between device-mapper kernels and LVM1 kernels.
+    # The LVM1 tools need to be installed with .lvm1 suffices
+    # e.g. vgscan.lvm1 and they will stop working after you start using
+    # the new lvm2 on-disk metadata format.
+    # The default value is set when the tools are built.
+    # fallback_to_lvm1 = 0
+
+    # The default metadata format that commands should use - "lvm1" or "lvm2".
+    # The command line override is -M1 or -M2.
+    # Defaults to "lvm2".
+    # format = "lvm2"
+
+    # Location of proc filesystem
+    proc = "/proc"
+
+    # Type of locking to use. Defaults to local file-based locking (1).
+    # Turn locking off by setting to 0 (dangerous: risks metadata corruption
+    # if LVM2 commands get run concurrently).
+    # Type 2 uses the external shared library locking_library.
+    # Type 3 uses built-in clustered locking.
+    # Type 4 uses read-only locking which forbids any operations that might 
+    # change metadata.
+    locking_type = 1
+
+    # Set to 0 to fail when a lock request cannot be satisfied immediately.
+    wait_for_locks = 1
+
+    # If using external locking (type 2) and initialisation fails,
+    # with this set to 1 an attempt will be made to use the built-in
+    # clustered locking.
+    # If you are using a customised locking_library you should set this to 0.
+    fallback_to_clustered_locking = 1
+
+    # If an attempt to initialise type 2 or type 3 locking failed, perhaps
+    # because cluster components such as clvmd are not running, with this set
+    # to 1 an attempt will be made to use local file-based locking (type 1).
+    # If this succeeds, only commands against local volume groups will proceed.
+    # Volume Groups marked as clustered will be ignored.
+    fallback_to_local_locking = 1
+
+    # Local non-LV directory that holds file-based locks while commands are
+    # in progress.  A directory like /tmp that may get wiped on reboot is OK.
+    locking_dir = "/var/lock/lvm"
+
+    # Whenever there are competing read-only and read-write access requests for
+    # a volume group's metadata, instead of always granting the read-only
+    # requests immediately, delay them to allow the read-write requests to be
+    # serviced.  Without this setting, write access may be stalled by a high
+    # volume of read-only requests.
+    # NB. This option only affects locking_type = 1 viz. local file-based
+    # locking.
+    prioritise_write_locks = 1
+
+    # Other entries can go here to allow you to load shared libraries
+    # e.g. if support for LVM1 metadata was compiled as a shared library use
+    #   format_libraries = "liblvm2format1.so" 
+    # Full pathnames can be given.
+
+    # Search this directory first for shared libraries.
+    #   library_dir = "/lib"
+
+    # The external locking library to load if locking_type is set to 2.
+    #   locking_library = "liblvm2clusterlock.so"
+
+    # Treat any internal errors as fatal errors, aborting the process that
+    # encountered the internal error. Please only enable for debugging.
+    abort_on_internal_errors = 0
+}
+
+activation {
+    # Set to 0 to disable udev synchronisation (if compiled into the binaries).
+    # Processes will not wait for notification from udev.
+    # They will continue irrespective of any possible udev processing
+    # in the background.  You should only use this if udev is not running
+    # or has rules that ignore the devices LVM2 creates.
+    # The command line argument --nodevsync takes precedence over this setting.
+    # If set to 1 when udev is not running, and there are LVM2 processes
+    # waiting for udev, run 'dmsetup udevcomplete_all' manually to wake them up.
+    udev_sync = 1
+
+    # Set to 0 to disable the udev rules installed by LVM2 (if built with
+    # --enable-udev_rules). LVM2 will then manage the /dev nodes and symlinks
+    # for active logical volumes directly itself.
+    # N.B. Manual intervention may be required if this setting is changed
+    # while any logical volumes are active.
+    udev_rules = 1
+
+    # How to fill in missing stripes if activating an incomplete volume.
+    # Using "error" will make inaccessible parts of the device return
+    # I/O errors on access.  You can instead use a device path, in which 
+    # case, that device will be used to in place of missing stripes.
+    # But note that using anything other than "error" with mirrored 
+    # or snapshotted volumes is likely to result in data corruption.
+    missing_stripe_filler = "error"
+
+    # How much stack (in KB) to reserve for use while devices suspended
+    reserved_stack = 256
+
+    # How much memory (in KB) to reserve for use while devices suspended
+    reserved_memory = 8192
+
+    # Nice value used while devices suspended
+    process_priority = -18
+
+    # If volume_list is defined, each LV is only activated if there is a
+    # match against the list.
+    #   "vgname" and "vgname/lvname" are matched exactly.
+    #   "@tag" matches any tag set in the LV or VG.
+    #   "@*" matches if any tag defined on the host is also set in the LV or VG
+    #
+    # volume_list = [ "vg1", "vg2/lvol1", "@tag1", "@*" ]
+
+    # Size (in KB) of each copy operation when mirroring
+    mirror_region_size = 512
+
+    # Setting to use when there is no readahead value stored in the metadata.
+    #
+    # "none" - Disable readahead.
+    # "auto" - Use default value chosen by kernel.
+    readahead = "auto"
+
+    # 'mirror_image_fault_policy' and 'mirror_log_fault_policy' define
+    # how a device failure affecting a mirror is handled.
+    # A mirror is composed of mirror images (copies) and a log.
+    # A disk log ensures that a mirror does not need to be re-synced
+    # (all copies made the same) every time a machine reboots or crashes.
+    #
+    # In the event of a failure, the specified policy will be used to determine
+    # what happens. This applies to automatic repairs (when the mirror is being
+    # monitored by dmeventd) and to manual lvconvert --repair when
+    # --use-policies is given.
+    #
+    # "remove" - Simply remove the faulty device and run without it.  If
+    #            the log device fails, the mirror would convert to using
+    #            an in-memory log.  This means the mirror will not
+    #            remember its sync status across crashes/reboots and
+    #            the entire mirror will be re-synced.  If a
+    #            mirror image fails, the mirror will convert to a
+    #            non-mirrored device if there is only one remaining good
+    #            copy.
+    #
+    # "allocate" - Remove the faulty device and try to allocate space on
+    #            a new device to be a replacement for the failed device.
+    #            Using this policy for the log is fast and maintains the
+    #            ability to remember sync state through crashes/reboots.
+    #            Using this policy for a mirror device is slow, as it
+    #            requires the mirror to resynchronize the devices, but it
+    #            will preserve the mirror characteristic of the device.
+    #            This policy acts like "remove" if no suitable device and
+    #            space can be allocated for the replacement.
+    #
+    # "allocate_anywhere" - Not yet implemented. Useful to place the log device
+    #            temporarily on same physical volume as one of the mirror
+    #            images. This policy is not recommended for mirror devices
+    #            since it would break the redundant nature of the mirror. This
+    #            policy acts like "remove" if no suitable device and space can
+    #            be allocated for the replacement.
+
+    mirror_log_fault_policy = "allocate"
+    mirror_image_fault_policy = "remove"
+}
+
+
+####################
+# Advanced section #
+####################
+
+# Metadata settings
+#
+# metadata {
+    # Default number of copies of metadata to hold on each PV.  0, 1 or 2.
+    # You might want to override it from the command line with 0 
+    # when running pvcreate on new PVs which are to be added to large VGs.
+
+    # pvmetadatacopies = 1
+
+    # Approximate default size of on-disk metadata areas in sectors.
+    # You should increase this if you have large volume groups or
+    # you want to retain a large on-disk history of your metadata changes.
+
+    # pvmetadatasize = 255
+
+    # List of directories holding live copies of text format metadata.
+    # These directories must not be on logical volumes!
+    # It's possible to use LVM2 with a couple of directories here,
+    # preferably on different (non-LV) filesystems, and with no other 
+    # on-disk metadata (pvmetadatacopies = 0). Or this can be in
+    # addition to on-disk metadata areas.
+    # The feature was originally added to simplify testing and is not
+    # supported under low memory situations - the machine could lock up.
+    #
+    # Never edit any files in these directories by hand unless you
+    # you are absolutely sure you know what you are doing! Use
+    # the supplied toolset to make changes (e.g. vgcfgrestore).
+
+    # dirs = [ "/etc/lvm/metadata", "/mnt/disk2/lvm/metadata2" ]
+#}
+
+# Event daemon
+#
+dmeventd {
+    # mirror_library is the library used when monitoring a mirror device.
+    #
+    # "libdevmapper-event-lvm2mirror.so" attempts to recover from
+    # failures.  It removes failed devices from a volume group and
+    # reconfigures a mirror as necessary. If no mirror library is
+    # provided, mirrors are not monitored through dmeventd.
+
+    mirror_library = "libdevmapper-event-lvm2mirror.so"
+
+    # snapshot_library is the library used when monitoring a snapshot device.
+    #
+    # "libdevmapper-event-lvm2snapshot.so" monitors the filling of
+    # snapshots and emits a warning through syslog, when the use of
+    # snapshot exceedes 80%. The warning is repeated when 85%, 90% and
+    # 95% of the snapshot are filled.
+
+    snapshot_library = "libdevmapper-event-lvm2snapshot.so"
+}

doc/example_cmdlib.c

@@ -0,0 +1,48 @@
+/*
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lvm2cmd.h"
+
+/* All output gets passed to this function line-by-line */
+void test_log_fn(int level, int dm_errno, const char *file, int line,
+		 const char *format)
+{
+	/* Extract and process output here rather than printing it */
+
+	if (level != 4)
+		return;
+
+	printf("%s\n", format);
+	return;
+}
+
+int main(int argc, char **argv)
+{
+	void *handle;
+	int r;
+
+	lvm2_log_fn(test_log_fn);
+
+	handle = lvm2_init();
+
+	lvm2_log_level(handle, 1);
+	r = lvm2_run(handle, "vgs --noheadings vg1");
+
+	/* More commands here */
+
+	lvm2_exit(handle);
+
+	return r;
+}
+

doc/pvmove_outline.txt

@@ -0,0 +1,52 @@
+Let's say we have an LV, made up of three segments of different PV's,
+I've also added in the device major:minor as this will be useful
+later:
+
++-----------------------------+
+|  PV1     |   PV2   |   PV3  | 254:3
++----------+---------+--------+
+
+
+Now our hero decides to PV move PV2 to PV4:
+
+1. Suspend our LV (254:3), this starts queueing all io, and flushes
+   all pending io.  Once the suspend has completed we are free to change
+   the mapping table.
+
+2. Set up *another* (254:4) device with the mapping table of our LV.
+
+3. Load a new mapping table into (254:3) that has identity targets for
+   parts that aren't moving, and a mirror target for parts that are.
+
+4. Unsuspend (254:3)
+
+So now we have:
+                           destination of copy
+               +--------------------->--------------+
+               |                                    |
++-----------------------------+               + -----------+
+| Identity | mirror  | Ident. | 254:3         |    PV4     |
++----------+---------+--------+               +------------+
+     |         |         |
+     \/        \/        \/
++-----------------------------+
+|  PV1     |   PV2   |   PV3  | 254:4
++----------+---------+--------+
+
+Any writes to segment2 of the LV get intercepted by the mirror target
+who checks that that chunk has been copied to the new destination, if
+it hasn't it queues the initial copy and defers the current io until
+it has finished.  Then the current io is written to *both* PV2 and the
+PV4.
+
+5. When the copying has completed 254:3 is suspended/pending flushed.
+
+6. 254:4 is taken down
+
+7. metadata is updated on disk
+
+8. 254:3 has new mapping table loaded:
+
++-----------------------------+
+|  PV1     |   PV4   |   PV3  | 254:3
++----------+---------+--------+

doc/tagging.txt

@@ -0,0 +1,165 @@
+Tagging aims
+============
+  1) Ability to attach an unordered list of tags to LVM metadata objects.
+  2) Ability to add or remove tags easily.
+  3) Ability to select LVM objects for processing according to presence/absence
+     of specific tags.
+  4) Ability to control through the config file which VGs/LVs are activated 
+     on different machines using names or tags.
+  5) Ability to overlay settings from different config files e.g. override
+     some settings in a global config file locally.
+
+Clarifications
+==============
+  1) Tag character set: A-Za-z0-9_+.- 
+     Can't start with hyphen & max length is 128 (NAME_LEN).
+  2) LVM object types that can be tagged:
+       VG, LV, LV segment
+       PV - tags are stored in VG metadata so disappear when PV becomes orphaned
+     Snapshots can't be tagged, but their origin may be.
+  3) A tag can be used in place of any command line LVM object reference that
+     accepts (a) a list of objects; or (b) a single object as long as the
+     tag expands to a single object.  This is not supported everywhere yet.
+     Duplicate arguments in a list after argument expansion may get removed 
+     retaining the first copy of each argument.
+  4) Wherever there may be ambiguity of argument type, a tag must be prefixed 
+     by '@'; elsewhere an '@' prefix is optional.
+  5) LVM1 objects cannot be tagged, as the disk format doesn't support it.
+  6) Tags can be added or removed with --addtag or --deltag.
+
+Config file Extensions
+======================
+  To define host tags in config file:
+
+  tags {
+  	# Set a tag with the hostname
+	hosttags = 1
+
+	tag1 { }
+
+  	tag2 {
+		# If no exact match, tag is not set.
+		host_list = [ "hostname", "dbase" ]
+	}
+  }
+
+Activation config file example
+==============================
+  activation {
+      volume_list = [ "vg1/lvol0", "@database" ]
+  }
+
+  Matches against vgname, vgname/lvname or @tag set in *metadata*.
+  @* matches exactly against *any* tag set on the host.
+  The VG or LV only gets activated if a metadata tag matches.
+  The default if there is no match is not to activate.
+  If volume_list is not present and any tags are defined on the host 
+  then it only activates if a host tag matches a metadata tag.
+  If volume_list is not present and no tags are defined on the host 
+  then it does activate.
+
+Multiple config files
+=====================
+  (a) lvm.conf
+  (b) lvm_<host_tag>.conf
+
+  At startup, load lvm.conf.
+  Process tag settings.
+  If any host tags were defined, load lvm_tag.conf for each tag, if present.
+
+  When searching for a specific config file entry, search order is (b)
+  then (a), stopping at the first match.  
+  Within (b) use reverse order tags got set, so file for last tag set is
+  searched first.
+  New tags set in (b) *do* trigger additional config file loads. 
+
+Usage Examples
+==============
+  1) Simple activation control via metadata with static config files
+
+  lvm.conf:  (Identical on every machine - global settings)
+    tags {
+      hostname_tags = 1
+    }
+
+  From any machine in the cluster, add db1 to the list of machines that
+  activate vg1/lvol2:
+
+  lvchange --tag @db1 vg1/lvol2
+  (followed by lvchange -ay to actually activate it)
+
+
+  2) Multiple hosts.  
+
+    Activate vg1 only on the database hosts, db1 and db2.
+    Activate vg2 only on the fileserver host fs1.
+    Activate nothing initially on the fileserver backup host fsb1, but be
+    prepared for it to take over from fs1.
+
+  Option (i) - centralised admin, static configuration replicated between hosts  
+    # Add @database tag to vg1's metadata
+    vgchange --tag @database vg1
+
+    # Add @fileserver tag to vg2's metadata
+    vgchange --tag @fileserver vg2
+
+    lvm.conf:  (Identical on every machine)
+      tags {
+        database {
+          host_list = [ "db1", "db2" ]
+        }
+        fileserver {
+	  host_list = [ "fs1" ]
+        }
+        fileserverbackup {
+          host_list = [ "fsb1" ]
+        }
+      }
+
+      activation {
+        # Only activate if host has a tag that matches a metadata tag
+        volume_list = [ "@*" ]
+      }
+  
+  In the event of the fileserver host going down, vg2 can be brought up
+  on fsb1 by running *on any node* 'vgchange --tag @fileserverbackup vg2'
+  followed by 'vgchange -ay vg2'
+  
+  
+  Option (ii) - localised admin & configuation
+  (i.e. each host holds *locally* which classes of volumes to activate)
+    # Add @database tag to vg1's metadata
+    vgchange --tag @database vg1
+  
+    # Add @fileserver tag to vg2's metadata
+    vgchange --tag @fileserver vg2
+  
+    lvm.conf:  (Identical on every machine - global settings)
+      tags {
+        hosttags = 1
+      }
+  
+    lvm_db1.conf: (only needs to be on db1 - could be symlink to lvm_db.conf)
+      activation {
+        volume_list = [ "@database" ]
+      }
+  
+    lvm_db2.conf: (only needs to be on db2 - could be symlink to lvm_db.conf)
+      activation {
+        volume_list = [ "@database" ]
+      }
+  
+    lvm_fs1.conf: (only needs to be on fs1 - could be symlink to lvm_fs.conf)
+      activation {
+        volume_list = [ "@fileserver" ]
+      }
+  
+    If fileserver goes down, to bring a spare machine fsb1 in as fileserver,
+    create lvm_fsb1.conf on fsb1 (or symlink to lvm_fs.conf):
+
+      activation {
+        volume_list = [ "@fileserver" ]
+      }
+
+    and run 'vgchange -ay vg2' or 'vgchange -ay @fileserver'
+

doc/testing.txt

@@ -0,0 +1,41 @@
+Here's how I test new LVM2 builds without interfering with the stable
+LVM2 that is running the LV's on my development box.
+
+1) Create a set of loopback devices.
+
+2) Create a new directory to contain the LVM2 configuration files for
+   this setup.  (I use /etc/lvm_loops)
+
+3) Write a suitable lvm.conf file, this goes in the directory you just
+   created.  eg, my /etc/lvm_loops/lvm.conf looks like:
+
+   log {
+        file="/tmp/lvm2_loop.log"
+        level=9
+        verbose=0
+        overwrite=1
+   }
+
+   devices {
+        scan = "/dev"
+        filter = ["a/loop/", "r/.*/"]
+   }
+
+
+   The important thing to note is the devices section which makes sure
+   that only the loopback devices are considered for LVM2 operations.
+
+4) When you want to use this test setup just set the environment
+   variable LVM_SYSTEM_DIR to point to your config directory
+   (/etc/lvm_loops in my case).
+
+5) It's a good idea to do a vgscan to initialise the filters:
+
+   export LVM_SYSTEM_DIR=/etc/lvm_loops
+   ./lvm vgscan
+
+   where ./lvm is the new build of LVM2 that I'm trying out.
+
+7) Test away.  Make sure that you are explicit about which lvm
+   executable you want to execute (eg, ./lvm if you are in
+   LVM2/tools).

driver/README

@@ -1 +0,0 @@
-The driver directory

include/.symlinks.in

@@ -0,0 +1,62 @@
+@top_srcdir@/daemons/clvmd/clvm.h
+@top_srcdir@/daemons/dmeventd/libdevmapper-event.h
+@top_srcdir@/liblvm/lvm2app.h
+@top_srcdir@/lib/activate/activate.h
+@top_srcdir@/lib/activate/targets.h
+@top_srcdir@/lib/cache/lvmcache.h
+@top_srcdir@/lib/commands/errors.h
+@top_srcdir@/lib/commands/toolcontext.h
+@top_srcdir@/lib/config/config.h
+@top_srcdir@/lib/config/defaults.h
+@top_srcdir@/lib/datastruct/btree.h
+@top_srcdir@/lib/datastruct/lvm-types.h
+@top_srcdir@/lib/datastruct/str_list.h
+@top_srcdir@/lib/device/dev-cache.h
+@top_srcdir@/lib/device/device.h
+@top_srcdir@/lib/display/display.h
+@top_srcdir@/lib/filters/filter-composite.h
+@top_srcdir@/lib/filters/filter-md.h
+@top_srcdir@/lib/filters/filter-persistent.h
+@top_srcdir@/lib/filters/filter-regex.h
+@top_srcdir@/lib/filters/filter-sysfs.h
+@top_srcdir@/lib/filters/filter.h
+@top_srcdir@/lib/format1/format1.h
+@top_srcdir@/lib/format_pool/format_pool.h
+@top_srcdir@/lib/format_text/archiver.h
+@top_srcdir@/lib/format_text/format-text.h
+@top_srcdir@/lib/format_text/text_export.h
+@top_srcdir@/lib/format_text/text_import.h
+@top_srcdir@/lib/label/label.h
+@top_srcdir@/lib/locking/locking.h
+@top_srcdir@/lib/log/log.h
+@top_srcdir@/lib/log/lvm-logging.h
+@top_srcdir@/lib/metadata/lv_alloc.h
+@top_srcdir@/lib/metadata/metadata.h
+@top_srcdir@/lib/metadata/metadata-exported.h
+@top_srcdir@/lib/metadata/pv_alloc.h
+@top_srcdir@/lib/metadata/segtype.h
+@top_srcdir@/lib/mm/memlock.h
+@top_srcdir@/lib/mm/xlate.h
+@top_builddir@/lib/misc/configure.h
+@top_srcdir@/lib/misc/crc.h
+@top_srcdir@/lib/misc/intl.h
+@top_srcdir@/lib/misc/util.h
+@top_srcdir@/lib/misc/last-path-component.h
+@top_srcdir@/lib/misc/lib.h
+@top_srcdir@/lib/misc/lvm-exec.h
+@top_srcdir@/lib/misc/lvm-file.h
+@top_srcdir@/lib/misc/lvm-globals.h
+@top_srcdir@/lib/misc/lvm-string.h
+@top_builddir@/lib/misc/lvm-version.h
+@top_srcdir@/lib/misc/lvm-wrappers.h
+@top_srcdir@/lib/misc/sharedlib.h
+@top_srcdir@/lib/report/report.h
+@top_srcdir@/lib/uuid/uuid.h
+@top_srcdir@/libdm/libdevmapper.h
+@top_srcdir@/libdm/misc/dm-ioctl.h
+@top_srcdir@/libdm/misc/dm-logging.h
+@top_srcdir@/libdm/misc/dm-log-userspace.h
+@top_srcdir@/libdm/misc/dmlib.h
+@top_srcdir@/libdm/misc/kdev_t.h
+@top_srcdir@/po/pogen.h
+@top_srcdir@/tools/lvm2cmd.h

include/Makefile.in

@@ -0,0 +1,52 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+SHELL = /bin/sh
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+LN_S = @LN_S@
+
+.PHONY: clean distclean all install pofile install_cluster install_device-mapper
+
+all: .symlinks_created
+
+.symlinks_created: .symlinks 
+	find . -maxdepth 1 -type l -exec $(RM) \{\} \;
+	for i in `cat $<`; do $(LN_S) $$i ; done
+	touch $@
+
+distclean:
+	find . -maxdepth 1 -type l -exec $(RM) \{\} \;
+	$(RM) Makefile .include_symlinks .symlinks_created .symlinks
+
+pofile: all
+
+device-mapper: all
+
+clean:
+
+install:
+
+install_cluster:
+
+install_device-mapper:
+
+install_lvm2:
+
+cflow:
+

lib/Makefile.in

@@ -0,0 +1,165 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+VPATH = @srcdir@
+
+ifeq ("@LVM1@", "shared")
+  SUBDIRS = format1
+endif
+
+ifeq ("@POOL@", "shared")
+  SUBDIRS += format_pool
+endif
+
+ifeq ("@SNAPSHOTS@", "shared")
+  SUBDIRS += snapshot
+endif
+
+ifeq ("@MIRRORS@", "shared")
+  SUBDIRS += mirror
+endif
+
+SOURCES =\
+	activate/activate.c \
+	cache/lvmcache.c \
+	commands/toolcontext.c \
+	config/config.c \
+	datastruct/btree.c \
+	datastruct/str_list.c \
+	device/dev-cache.c \
+	device/dev-io.c \
+	device/dev-md.c \
+	device/dev-swap.c \
+	device/device.c \
+	display/display.c \
+	error/errseg.c \
+	unknown/unknown.c \
+	filters/filter-composite.c \
+	filters/filter-persistent.c \
+	filters/filter-regex.c \
+	filters/filter-sysfs.c \
+	filters/filter-md.c \
+	filters/filter.c \
+	format_text/archive.c \
+	format_text/archiver.c \
+	format_text/export.c \
+	format_text/flags.c \
+	format_text/format-text.c \
+	format_text/import.c \
+	format_text/import_vsn1.c \
+	format_text/tags.c \
+	format_text/text_label.c \
+	freeseg/freeseg.c \
+	label/label.c \
+	locking/file_locking.c \
+	locking/locking.c \
+	locking/no_locking.c \
+	log/log.c \
+	metadata/lv_manip.c \
+	metadata/merge.c \
+	metadata/metadata.c \
+	metadata/mirror.c \
+	metadata/pv_manip.c \
+	metadata/pv_map.c \
+	metadata/segtype.c \
+	metadata/snapshot_manip.c \
+	misc/crc.c \
+	misc/lvm-exec.c \
+	misc/lvm-file.c \
+	misc/lvm-globals.c \
+	misc/lvm-string.c \
+	misc/lvm-wrappers.c \
+	misc/util.c \
+	mm/memlock.c \
+	report/report.c \
+	striped/striped.c \
+	uuid/uuid.c \
+	zero/zero.c
+
+ifeq ("@HAVE_REALTIME@", "yes")
+  SOURCES +=\
+	misc/timestamp.c
+endif
+
+ifeq ("@LVM1@", "internal")
+  SOURCES +=\
+	format1/disk-rep.c \
+	format1/format1.c \
+	format1/import-export.c \
+	format1/import-extents.c \
+	format1/layout.c \
+	format1/lvm1-label.c \
+	format1/vg_number.c
+endif
+
+ifeq ("@POOL@", "internal")
+  SOURCES +=\
+	format_pool/disk_rep.c \
+	format_pool/format_pool.c \
+	format_pool/import_export.c \
+	format_pool/pool_label.c
+endif
+
+ifeq ("@CLUSTER@", "internal")
+  SOURCES += locking/cluster_locking.c
+endif
+
+ifeq ("@CLUSTER@", "shared")
+  SUBDIRS += locking
+endif
+
+ifeq ("@SNAPSHOTS@", "internal")
+  SOURCES += snapshot/snapshot.c
+endif
+
+ifeq ("@MIRRORS@", "internal")
+  SOURCES += mirror/mirrored.c
+endif
+
+ifeq ("@DEVMAPPER@", "yes")
+  SOURCES +=\
+	activate/dev_manager.c \
+	activate/fs.c
+endif
+
+ifeq ("@HAVE_LIBDL@", "yes")
+  SOURCES +=\
+	locking/external_locking.c \
+	misc/sharedlib.c
+endif
+
+ifeq ("@DMEVENTD@", "yes")
+  CLDFLAGS += -L../daemons/dmeventd
+  LIBS += -ldevmapper-event
+endif
+
+LIB_NAME = liblvm-internal
+LIB_STATIC = $(LIB_NAME).a
+
+CLEAN_TARGETS += $(LIB_NAME).cflow
+
+include ../make.tmpl
+
+$(SUBDIRS): $(LIB_STATIC)
+
+$(LIB_NAME).cflow: $(SOURCES)
+	set -e; (echo -n "SOURCES += "; \
+		 echo $(SOURCES) | \
+		 sed "s/^/ /;s/ / $(top_srcdir)\/lib\//g;s/$$//"; \
+		 ) > $@
+
+cflow: $(LIB_NAME).cflow

lib/README

@@ -1 +0,0 @@
-Base library directory

lib/activate/activate.h

@@ -0,0 +1,113 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef LVM_ACTIVATE_H
+#define LVM_ACTIVATE_H
+
+#include "metadata-exported.h"
+
+struct lvinfo {
+	int exists;
+	int suspended;
+	unsigned int open_count;
+	int major;
+	int minor;
+	int read_only;
+	int live_table;
+	int inactive_table;
+	uint32_t read_ahead;
+};
+
+/* target attribute flags */
+#define MIRROR_LOG_CLUSTERED	0x00000001U
+
+void set_activation(int activation);
+int activation(void);
+
+int driver_version(char *version, size_t size);
+int library_version(char *version, size_t size);
+int lvm1_present(struct cmd_context *cmd);
+
+int module_present(struct cmd_context *cmd, const char *target_name);
+int target_present(struct cmd_context *cmd, const char *target_name,
+		   int use_modprobe);
+int target_version(const char *target_name, uint32_t *maj,
+                   uint32_t *min, uint32_t *patchlevel);
+int list_segment_modules(struct dm_pool *mem, const struct lv_segment *seg,
+			 struct dm_list *modules);
+int list_lv_modules(struct dm_pool *mem, const struct logical_volume *lv,
+		    struct dm_list *modules);
+
+void activation_release(void);
+void activation_exit(void);
+
+int lv_suspend(struct cmd_context *cmd, const char *lvid_s);
+int lv_suspend_if_active(struct cmd_context *cmd, const char *lvid_s);
+int lv_resume(struct cmd_context *cmd, const char *lvid_s);
+int lv_resume_if_active(struct cmd_context *cmd, const char *lvid_s);
+int lv_activate(struct cmd_context *cmd, const char *lvid_s, int exclusive);
+int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s,
+			    int exclusive);
+int lv_deactivate(struct cmd_context *cmd, const char *lvid_s);
+
+int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv);
+
+/*
+ * Returns 1 if info structure has been populated, else 0.
+ */
+int lv_info(struct cmd_context *cmd, const struct logical_volume *lv, struct lvinfo *info,
+	    int with_open_count, int with_read_ahead);
+int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s,
+		    struct lvinfo *info, int with_open_count, int with_read_ahead);
+
+/*
+ * Returns 1 if activate_lv has been set: 1 = activate; 0 = don't.
+ */
+int lv_activation_filter(struct cmd_context *cmd, const char *lvid_s,
+			 int *activate_lv);
+
+/*
+ * Returns 1 if percent has been set, else 0.
+ */
+int lv_snapshot_percent(const struct logical_volume *lv, float *percent,
+			percent_range_t *percent_range);
+int lv_mirror_percent(struct cmd_context *cmd, struct logical_volume *lv,
+		      int wait, float *percent, percent_range_t *percent_range,
+		      uint32_t *event_nr);
+
+/*
+ * Return number of LVs in the VG that are active.
+ */
+int lvs_in_vg_activated(struct volume_group *vg);
+int lvs_in_vg_activated_by_uuid_only(struct volume_group *vg);
+int lvs_in_vg_opened(const struct volume_group *vg);
+
+int lv_is_active(struct logical_volume *lv);
+
+int monitor_dev_for_events(struct cmd_context *cmd,
+			    struct logical_volume *lv, int do_reg);
+
+/*
+ * Returns 1 if PV has a dependency tree that uses anything in VG.
+ */
+int pv_uses_vg(struct physical_volume *pv,
+	       struct volume_group *vg);
+
+/*
+ * Returns 1 if mapped device is not suspended.
+ */
+int device_is_usable(dev_t dev);
+
+#endif

lib/activate/dev_manager.h

@@ -0,0 +1,73 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_DEV_MANAGER_H
+#define _LVM_DEV_MANAGER_H
+
+#include "metadata-exported.h"
+
+struct logical_volume;
+struct volume_group;
+struct cmd_context;
+struct dev_manager;
+struct dm_info;
+struct device;
+
+/*
+ * Constructor and destructor.
+ */
+struct dev_manager *dev_manager_create(struct cmd_context *cmd,
+				       const char *vg_name);
+void dev_manager_destroy(struct dev_manager *dm);
+void dev_manager_release(void);
+void dev_manager_exit(void);
+
+/*
+ * The device handler is responsible for creating all the layered
+ * dm devices, and ensuring that all constraints are maintained
+ * (eg, an origin is created before its snapshot, but is not
+ * unsuspended until the snapshot is also created.)
+ */
+int dev_manager_info(struct dm_pool *mem, const char *name,
+		     const struct logical_volume *lv,
+		     int mknodes, int with_open_count, int with_read_ahead,
+		     struct dm_info *info, uint32_t *read_ahead);
+int dev_manager_snapshot_percent(struct dev_manager *dm,
+				 const struct logical_volume *lv,
+				 float *percent,
+				 percent_range_t *percent_range);
+int dev_manager_mirror_percent(struct dev_manager *dm,
+			       const struct logical_volume *lv, int wait,
+			       float *percent, percent_range_t *percent_range,
+			       uint32_t *event_nr);
+int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv,
+			int lockfs, int flush_required);
+int dev_manager_activate(struct dev_manager *dm, struct logical_volume *lv);
+int dev_manager_preload(struct dev_manager *dm, struct logical_volume *lv,
+			int *flush_required);
+int dev_manager_deactivate(struct dev_manager *dm, struct logical_volume *lv);
+
+int dev_manager_lv_mknodes(const struct logical_volume *lv);
+int dev_manager_lv_rmnodes(const struct logical_volume *lv);
+
+/*
+ * Put the desired changes into effect.
+ */
+int dev_manager_execute(struct dev_manager *dm);
+
+int dev_manager_device_uses_vg(struct device *dev,
+			       struct volume_group *vg);
+
+#endif

lib/activate/fs.c

@@ -0,0 +1,395 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "fs.h"
+#include "toolcontext.h"
+#include "lvm-string.h"
+#include "lvm-file.h"
+#include "memlock.h"
+
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <limits.h>
+#include <dirent.h>
+
+static int _mk_dir(const char *dev_dir, const char *vg_name)
+{
+	char vg_path[PATH_MAX];
+	mode_t old_umask;
+
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+			 dev_dir, vg_name) == -1) {
+		log_error("Couldn't construct name of volume "
+			  "group directory.");
+		return 0;
+	}
+
+	if (dir_exists(vg_path))
+		return 1;
+
+	log_very_verbose("Creating directory %s", vg_path);
+
+	old_umask = umask(DM_DEV_DIR_UMASK);
+	if (mkdir(vg_path, 0777)) {
+		log_sys_error("mkdir", vg_path);
+		umask(old_umask);
+		return 0;
+	}
+	umask(old_umask);
+
+	return 1;
+}
+
+static int _rm_dir(const char *dev_dir, const char *vg_name)
+{
+	char vg_path[PATH_MAX];
+
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+			 dev_dir, vg_name) == -1) {
+		log_error("Couldn't construct name of volume "
+			  "group directory.");
+		return 0;
+	}
+
+	if (dir_exists(vg_path) && is_empty_dir(vg_path)) {
+		log_very_verbose("Removing directory %s", vg_path);
+		rmdir(vg_path);
+	}
+
+	return 1;
+}
+
+static void _rm_blks(const char *dir)
+{
+	const char *name;
+	char path[PATH_MAX];
+	struct dirent *dirent;
+	struct stat buf;
+	DIR *d;
+
+	if (!(d = opendir(dir))) {
+		log_sys_error("opendir", dir);
+		return;
+	}
+
+	while ((dirent = readdir(d))) {
+		name = dirent->d_name;
+
+		if (!strcmp(name, ".") || !strcmp(name, ".."))
+			continue;
+
+		if (dm_snprintf(path, sizeof(path), "%s/%s", dir, name) == -1) {
+			log_error("Couldn't create path for %s", name);
+			continue;
+		}
+
+		if (!lstat(path, &buf)) {
+			if (!S_ISBLK(buf.st_mode))
+				continue;
+			log_very_verbose("Removing %s", path);
+			if (unlink(path) < 0)
+				log_sys_error("unlink", path);
+		}
+	}
+}
+
+static int _mk_link(const char *dev_dir, const char *vg_name,
+		    const char *lv_name, const char *dev, int check_udev)
+{
+	char lv_path[PATH_MAX], link_path[PATH_MAX], lvm1_group_path[PATH_MAX];
+	char vg_path[PATH_MAX];
+	struct stat buf, buf_lp;
+
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+			 dev_dir, vg_name) == -1) {
+		log_error("Couldn't create path for volume group dir %s",
+			  vg_name);
+		return 0;
+	}
+
+	if (dm_snprintf(lv_path, sizeof(lv_path), "%s/%s", vg_path,
+			 lv_name) == -1) {
+		log_error("Couldn't create source pathname for "
+			  "logical volume link %s", lv_name);
+		return 0;
+	}
+
+	if (dm_snprintf(link_path, sizeof(link_path), "%s/%s",
+			 dm_dir(), dev) == -1) {
+		log_error("Couldn't create destination pathname for "
+			  "logical volume link for %s", lv_name);
+		return 0;
+	}
+
+	if (dm_snprintf(lvm1_group_path, sizeof(lvm1_group_path), "%s/group",
+			 vg_path) == -1) {
+		log_error("Couldn't create pathname for LVM1 group file for %s",
+			  vg_name);
+		return 0;
+	}
+
+	/* To reach this point, the VG must have been locked.
+	 * As locking fails if the VG is active under LVM1, it's
+	 * now safe to remove any LVM1 devices we find here
+	 * (as well as any existing LVM2 symlink). */
+	if (!lstat(lvm1_group_path, &buf)) {
+		if (!S_ISCHR(buf.st_mode)) {
+			log_error("Non-LVM1 character device found at %s",
+				  lvm1_group_path);
+		} else {
+			_rm_blks(vg_path);
+
+			log_very_verbose("Removing %s", lvm1_group_path);
+			if (unlink(lvm1_group_path) < 0)
+				log_sys_error("unlink", lvm1_group_path);
+		}
+	}
+
+	if (!lstat(lv_path, &buf)) {
+		if (!S_ISLNK(buf.st_mode) && !S_ISBLK(buf.st_mode)) {
+			log_error("Symbolic link %s not created: file exists",
+				  link_path);
+			return 0;
+		}
+
+		if (dm_udev_get_sync_support() && udev_checking() && check_udev) {
+			/* Check udev created the correct link. */
+			if (!stat(link_path, &buf_lp) &&
+			    !stat(lv_path, &buf)) {
+				if (buf_lp.st_rdev == buf.st_rdev)
+					return 1;
+				else
+					log_warn("Symlink %s that should have been "
+						 "created by udev does not have "
+						 "correct target. Falling back to "
+						 "direct link creation", lv_path);
+			} else
+				log_warn("Symlink %s that should have been "
+					 "created by udev could not be checked "
+					 "for its correctness. Falling back to "
+					 "direct link creation.", lv_path);
+
+		}
+
+		log_very_verbose("Removing %s", lv_path);
+		if (unlink(lv_path) < 0) {
+			log_sys_error("unlink", lv_path);
+			return 0;
+		}
+	} else if (dm_udev_get_sync_support() && udev_checking() && check_udev)
+		log_warn("The link %s should had been created by udev "
+			  "but it was not found. Falling back to "
+			  "direct link creation.", lv_path);
+
+	log_very_verbose("Linking %s -> %s", lv_path, link_path);
+	if (symlink(link_path, lv_path) < 0) {
+		log_sys_error("symlink", lv_path);
+		return 0;
+	}
+
+	if (!dm_set_selinux_context(lv_path, S_IFLNK))
+		return_0;
+
+	return 1;
+}
+
+static int _rm_link(const char *dev_dir, const char *vg_name,
+		    const char *lv_name, int check_udev)
+{
+	struct stat buf;
+	char lv_path[PATH_MAX];
+
+	if (dm_snprintf(lv_path, sizeof(lv_path), "%s%s/%s",
+			 dev_dir, vg_name, lv_name) == -1) {
+		log_error("Couldn't determine link pathname.");
+		return 0;
+	}
+
+	if (lstat(lv_path, &buf) && errno == ENOENT)
+		return 1;
+	else if (dm_udev_get_sync_support() && udev_checking() && check_udev)
+		log_warn("The link %s should have been removed by udev "
+			 "but it is still present. Falling back to "
+			 "direct link removal.", lv_path);
+
+	if (!S_ISLNK(buf.st_mode)) {
+		log_error("%s not symbolic link - not removing", lv_path);
+		return 0;
+	}
+
+	log_very_verbose("Removing link %s", lv_path);
+	if (unlink(lv_path) < 0) {
+		log_sys_error("unlink", lv_path);
+		return 0;
+	}
+
+	return 1;
+}
+
+typedef enum {
+	FS_ADD,
+	FS_DEL,
+	FS_RENAME
+} fs_op_t;
+
+static int _do_fs_op(fs_op_t type, const char *dev_dir, const char *vg_name,
+		     const char *lv_name, const char *dev,
+		     const char *old_lv_name, int check_udev)
+{
+	switch (type) {
+	case FS_ADD:
+		if (!_mk_dir(dev_dir, vg_name) ||
+		    !_mk_link(dev_dir, vg_name, lv_name, dev, check_udev))
+			return_0;
+		break;
+	case FS_DEL:
+		if (!_rm_link(dev_dir, vg_name, lv_name, check_udev) ||
+		    !_rm_dir(dev_dir, vg_name))
+			return_0;
+		break;
+		/* FIXME Use rename() */
+	case FS_RENAME:
+		if (old_lv_name && !_rm_link(dev_dir, vg_name, old_lv_name,
+					     check_udev))
+			stack;
+
+		if (!_mk_link(dev_dir, vg_name, lv_name, dev, check_udev))
+			stack;
+	}
+
+	return 1;
+}
+
+static DM_LIST_INIT(_fs_ops);
+
+struct fs_op_parms {
+	struct dm_list list;
+	fs_op_t type;
+	int check_udev;
+	char *dev_dir;
+	char *vg_name;
+	char *lv_name;
+	char *dev;
+	char *old_lv_name;
+	char names[0];
+};
+
+static void _store_str(char **pos, char **ptr, const char *str)
+{
+	strcpy(*pos, str);
+	*ptr = *pos;
+	*pos += strlen(*ptr) + 1;
+}
+
+static int _stack_fs_op(fs_op_t type, const char *dev_dir, const char *vg_name,
+			const char *lv_name, const char *dev,
+			const char *old_lv_name, int check_udev)
+{
+	struct fs_op_parms *fsp;
+	size_t len = strlen(dev_dir) + strlen(vg_name) + strlen(lv_name) +
+	    strlen(dev) + strlen(old_lv_name) + 5;
+	char *pos;
+
+	if (!(fsp = dm_malloc(sizeof(*fsp) + len))) {
+		log_error("No space to stack fs operation");
+		return 0;
+	}
+
+	pos = fsp->names;
+	fsp->type = type;
+	fsp->check_udev = check_udev;
+
+	_store_str(&pos, &fsp->dev_dir, dev_dir);
+	_store_str(&pos, &fsp->vg_name, vg_name);
+	_store_str(&pos, &fsp->lv_name, lv_name);
+	_store_str(&pos, &fsp->dev, dev);
+	_store_str(&pos, &fsp->old_lv_name, old_lv_name);
+
+	dm_list_add(&_fs_ops, &fsp->list);
+
+	return 1;
+}
+
+static void _pop_fs_ops(void)
+{
+	struct dm_list *fsph, *fspht;
+	struct fs_op_parms *fsp;
+
+	dm_list_iterate_safe(fsph, fspht, &_fs_ops) {
+		fsp = dm_list_item(fsph, struct fs_op_parms);
+		_do_fs_op(fsp->type, fsp->dev_dir, fsp->vg_name, fsp->lv_name,
+			  fsp->dev, fsp->old_lv_name, fsp->check_udev);
+		dm_list_del(&fsp->list);
+		dm_free(fsp);
+	}
+}
+
+static int _fs_op(fs_op_t type, const char *dev_dir, const char *vg_name,
+		  const char *lv_name, const char *dev, const char *old_lv_name,
+		  int check_udev)
+{
+	if (memlock()) {
+		if (!_stack_fs_op(type, dev_dir, vg_name, lv_name, dev,
+				  old_lv_name, check_udev))
+			return_0;
+		return 1;
+	}
+
+	return _do_fs_op(type, dev_dir, vg_name, lv_name, dev,
+			 old_lv_name, check_udev);
+}
+
+int fs_add_lv(const struct logical_volume *lv, const char *dev)
+{
+	return _fs_op(FS_ADD, lv->vg->cmd->dev_dir, lv->vg->name, lv->name,
+		      dev, "", lv->vg->cmd->current_settings.udev_rules);
+}
+
+int fs_del_lv(const struct logical_volume *lv)
+{
+	return _fs_op(FS_DEL, lv->vg->cmd->dev_dir, lv->vg->name, lv->name,
+		      "", "", lv->vg->cmd->current_settings.udev_rules);
+}
+
+int fs_del_lv_byname(const char *dev_dir, const char *vg_name,
+		     const char *lv_name, int check_udev)
+{
+	return _fs_op(FS_DEL, dev_dir, vg_name, lv_name, "", "", check_udev);
+}
+
+int fs_rename_lv(struct logical_volume *lv, const char *dev, 
+		const char *old_vgname, const char *old_lvname)
+{
+	if (strcmp(old_vgname, lv->vg->name)) {
+		return
+			(_fs_op(FS_DEL, lv->vg->cmd->dev_dir, old_vgname,
+				old_lvname, "", "", lv->vg->cmd->current_settings.udev_rules) &&
+			 _fs_op(FS_ADD, lv->vg->cmd->dev_dir, lv->vg->name,
+				lv->name, dev, "", lv->vg->cmd->current_settings.udev_rules));
+	}
+	else 
+		return _fs_op(FS_RENAME, lv->vg->cmd->dev_dir, lv->vg->name, lv->name,
+			      dev, old_lvname, lv->vg->cmd->current_settings.udev_rules);
+}
+
+void fs_unlock(void)
+{
+	if (!memlock()) {
+		dm_lib_release();
+		_pop_fs_ops();
+	}
+}

lib/activate/fs.h

@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_FS_H
+#define _LVM_FS_H
+
+#include "metadata.h"
+
+/*
+ * These calls, private to the activate unit, set
+ * up the volume group directory in /dev and the
+ * symbolic links to the dm device.
+ */
+int fs_add_lv(const struct logical_volume *lv, const char *dev);
+int fs_del_lv(const struct logical_volume *lv);
+int fs_del_lv_byname(const char *dev_dir, const char *vg_name,
+		     const char *lv_name, int check_udev);
+int fs_rename_lv(struct logical_volume *lv, const char *dev, 
+		 const char *old_vgname, const char *old_lvname);
+void fs_unlock(void);
+
+#endif

lib/activate/targets.h

@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2003-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_TARGETS_H
+#define _LVM_TARGETS_H
+
+struct dev_manager;
+struct lv_segment;
+
+int compose_areas_line(struct dev_manager *dm, struct lv_segment *seg,
+		       char *params, size_t paramsize, int *pos,
+		       int start_area, int areas);
+
+int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
+                   struct dm_tree_node *node, uint32_t start_area, uint32_t areas);
+
+int build_dev_string(struct dev_manager *dm, char *dlid, char *devbuf,
+		     size_t bufsize, const char *desc);
+
+char *build_dlid(struct dev_manager *dm, const char *lvid, const char *layer);
+
+#endif

lib/cache/lvmcache.h

@@ -0,0 +1,117 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_CACHE_H
+#define _LVM_CACHE_H
+
+#include "dev-cache.h"
+#include "uuid.h"
+#include "label.h"
+
+#define ORPHAN_PREFIX "#"
+#define ORPHAN_VG_NAME(fmt) ORPHAN_PREFIX "orphans_" fmt
+
+#define CACHE_INVALID	0x00000001
+#define CACHE_LOCKED	0x00000002
+
+/* LVM specific per-volume info */
+/* Eventual replacement for struct physical_volume perhaps? */
+
+struct cmd_context;
+struct format_type;
+struct volume_group;
+
+/* One per VG */
+struct lvmcache_vginfo {
+	struct dm_list list;	/* Join these vginfos together */
+	struct dm_list infos;	/* List head for lvmcache_infos */
+	const struct format_type *fmt;
+	char *vgname;		/* "" == orphan */
+	uint32_t status;
+	char vgid[ID_LEN + 1];
+	char _padding[7];
+	struct lvmcache_vginfo *next; /* Another VG with same name? */
+	char *creation_host;
+	char *vgmetadata;	/* Copy of VG metadata as format_text string */
+	unsigned precommitted;	/* Is vgmetadata live or precommitted? */
+};
+
+/* One per device */
+struct lvmcache_info {
+	struct dm_list list;	/* Join VG members together */
+	struct dm_list mdas;	/* list head for metadata areas */
+	struct dm_list das;	/* list head for data areas */
+	struct lvmcache_vginfo *vginfo;	/* NULL == unknown */
+	struct label *label;
+	const struct format_type *fmt;
+	struct device *dev;
+	uint64_t device_size;	/* Bytes */
+	uint32_t status;
+};
+
+int lvmcache_init(void);
+void lvmcache_destroy(struct cmd_context *cmd, int retain_orphans);
+
+/* Set full_scan to 1 to reread every filtered device label or
+ * 2 to rescan /dev for new devices */
+int lvmcache_label_scan(struct cmd_context *cmd, int full_scan);
+
+/* Add/delete a device */
+struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
+				   struct device *dev,
+				   const char *vgname, const char *vgid,
+				   uint32_t vgstatus);
+int lvmcache_add_orphan_vginfo(const char *vgname, struct format_type *fmt);
+void lvmcache_del(struct lvmcache_info *info);
+
+/* Update things */
+int lvmcache_update_vgname_and_id(struct lvmcache_info *info,
+				  const char *vgname, const char *vgid,
+				  uint32_t vgstatus, const char *hostname);
+int lvmcache_update_vg(struct volume_group *vg, unsigned precommitted);
+
+void lvmcache_lock_vgname(const char *vgname, int read_only);
+void lvmcache_unlock_vgname(const char *vgname);
+int lvmcache_verify_lock_order(const char *vgname);
+
+/* Queries */
+const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid);
+struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname,
+					   const char *vgid);
+struct lvmcache_vginfo *vginfo_from_vgid(const char *vgid);
+struct lvmcache_info *info_from_pvid(const char *pvid, int valid_only);
+const char *vgname_from_vgid(struct dm_pool *mem, const char *vgid);
+struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid);
+int vgs_locked(void);
+int vgname_is_locked(const char *vgname);
+
+/* Returns list of struct str_lists containing pool-allocated copy of vgnames */
+/* Set full_scan to 1 to reread every filtered device label */
+struct dm_list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan);
+
+/* Returns list of struct str_lists containing pool-allocated copy of vgids */
+/* Set full_scan to 1 to reread every filtered device label */
+struct dm_list *lvmcache_get_vgids(struct cmd_context *cmd, int full_scan);
+
+/* Returns list of struct str_lists containing pool-allocated copy of pvids */
+struct dm_list *lvmcache_get_pvids(struct cmd_context *cmd, const char *vgname,
+				const char *vgid);
+
+/* Returns cached volume group metadata. */
+struct volume_group *lvmcache_get_vg(const char *vgid, unsigned precommitted);
+void lvmcache_drop_metadata(const char *vgname, int drop_precommitted);
+void lvmcache_commit_metadata(const char *vgname);
+
+#endif

lib/commands/errors.h

@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_ERRORS_H
+#define _LVM_ERRORS_H
+
+#define ECMD_PROCESSED		1
+#define ENO_SUCH_CMD		2
+#define EINVALID_CMD_LINE	3
+#define ECMD_FAILED		5
+
+/* FIXME Also returned by cmdlib. */
+
+#endif

lib/commands/toolcontext.h

@@ -0,0 +1,113 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_TOOLCONTEXT_H
+#define _LVM_TOOLCONTEXT_H
+
+#include "dev-cache.h"
+
+#include <stdio.h>
+#include <limits.h>
+
+/*
+ * Config options that can be changed while commands are processed
+ */
+struct config_info {
+	int debug;
+	int verbose;
+	int test;
+	int syslog;
+	int activation;
+	int suffix;
+	int archive;		/* should we archive ? */
+	int backup;		/* should we backup ? */
+	int read_ahead;		/* DM_READ_AHEAD_NONE or _AUTO */
+	int udev_rules;
+	int udev_sync;
+	int cache_vgmetadata;
+	const char *msg_prefix;
+	struct format_type *fmt;
+	uint64_t unit_factor;
+	int cmd_name;		/* Show command name? */
+	mode_t umask;
+	char unit_type;
+	char _padding[1];
+};
+
+struct config_tree;
+struct archive_params;
+struct backup_params;
+
+/* FIXME Split into tool & library contexts */
+/* command-instance-related variables needed by library */
+struct cmd_context {
+	struct dm_pool *libmem;	/* For permanent config data */
+	struct dm_pool *mem;	/* Transient: Cleared between each command */
+
+	const struct format_type *fmt;	/* Current format to use by default */
+	struct format_type *fmt_backup;	/* Format to use for backups */
+
+	struct dm_list formats;	/* Available formats */
+	struct dm_list segtypes;	/* Available segment types */
+	const char *hostname;
+	const char *kernel_vsn;
+
+	unsigned rand_seed;
+	const char *cmd_line;
+	struct command *command;
+	char **argv;
+	unsigned is_long_lived:1;	/* Optimises persistent_filter handling */
+	unsigned handles_missing_pvs:1;
+	unsigned handles_unknown_segments:1;
+	unsigned partial_activation:1;
+	unsigned si_unit_consistency:1;
+
+	struct dev_filter *filter;
+	int dump_filter;	/* Dump filter when exiting? */
+
+	struct dm_list config_files;
+	int config_valid;
+	struct config_tree *cft;
+	struct config_tree *cft_override;
+	struct config_info default_settings;
+	struct config_info current_settings;
+
+	struct archive_params *archive_params;
+	struct backup_params *backup_params;
+	const char *stripe_filler;
+
+	/* List of defined tags */
+	struct dm_list tags;
+	int hosttags;
+
+	char system_dir[PATH_MAX];
+	char dev_dir[PATH_MAX];
+	char proc_dir[PATH_MAX];
+	char sysfs_dir[PATH_MAX];
+};
+
+/*
+ * system_dir may be NULL to use the default value.
+ * The environment variable LVM_SYSTEM_DIR always takes precedence.
+ */
+struct cmd_context *create_toolcontext(unsigned is_long_lived,
+				       const char *system_dir);
+void destroy_toolcontext(struct cmd_context *cmd);
+int refresh_toolcontext(struct cmd_context *cmd);
+int refresh_filters(struct cmd_context *cmd);
+int config_files_changed(struct cmd_context *cmd);
+int init_lvmcache_orphans(struct cmd_context *cmd);
+
+#endif

lib/config/config.h

@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_CONFIG_H
+#define _LVM_CONFIG_H
+
+#include "lvm-types.h"
+
+struct device;
+struct cmd_context;
+
+enum {
+	CFG_STRING,
+	CFG_FLOAT,
+	CFG_INT,
+	CFG_EMPTY_ARRAY
+};
+
+struct config_value {
+	int type;
+	union {
+		int64_t i;
+		float r;
+		char *str;
+	} v;
+	struct config_value *next;	/* for arrays */
+};
+
+struct config_node {
+	char *key;
+	struct config_node *parent, *sib, *child;
+	struct config_value *v;
+};
+
+struct config_tree {
+	struct config_node *root;
+};
+
+struct config_tree_list {
+	struct dm_list list;
+	struct config_tree *cft;
+};
+
+struct config_tree *create_config_tree(const char *filename, int keep_open);
+struct config_tree *create_config_tree_from_string(struct cmd_context *cmd,
+						   const char *config_settings);
+int override_config_tree_from_string(struct cmd_context *cmd,
+				     const char *config_settings);
+void destroy_config_tree(struct config_tree *cft);
+
+typedef uint32_t (*checksum_fn_t) (uint32_t initial, const void *buf, uint32_t size);
+
+int read_config_fd(struct config_tree *cft, struct device *dev,
+		   off_t offset, size_t size, off_t offset2, size_t size2,
+		   checksum_fn_t checksum_fn, uint32_t checksum);
+
+int read_config_file(struct config_tree *cft);
+int write_config_file(struct config_tree *cft, const char *file,
+		      int argc, char **argv);
+
+typedef int (*putline_fn)(const char *line, void *baton);
+int write_config_node(const struct config_node *cn, putline_fn putline, void *baton);
+
+time_t config_file_timestamp(struct config_tree *cft);
+int config_file_changed(struct config_tree *cft);
+int merge_config_tree(struct cmd_context *cmd, struct config_tree *cft,
+		      struct config_tree *newdata);
+
+struct config_node *find_config_node(const struct config_node *cn,
+				     const char *path);
+const char *find_config_str(const struct config_node *cn, const char *path,
+			    const char *fail);
+int find_config_int(const struct config_node *cn, const char *path, int fail);
+float find_config_float(const struct config_node *cn, const char *path,
+			float fail);
+
+/*
+ * These versions check an override tree, if present, first.
+ */
+struct config_node *find_config_tree_node(struct cmd_context *cmd,
+					  const char *path);
+const char *find_config_tree_str(struct cmd_context *cmd,
+				 const char *path, const char *fail);
+int find_config_tree_int(struct cmd_context *cmd, const char *path,
+			 int fail);
+float find_config_tree_float(struct cmd_context *cmd, const char *path,
+			     float fail);
+
+/*
+ * Understands (0, ~0), (y, n), (yes, no), (on,
+ * off), (true, false).
+ */
+int find_config_bool(const struct config_node *cn, const char *path, int fail);
+int find_config_tree_bool(struct cmd_context *cmd, const char *path, int fail);
+
+int get_config_uint32(const struct config_node *cn, const char *path,
+		      uint32_t *result);
+
+int get_config_uint64(const struct config_node *cn, const char *path,
+		      uint64_t *result);
+
+int get_config_str(const struct config_node *cn, const char *path,
+		   char **result);
+
+unsigned maybe_config_section(const char *str, unsigned len);
+
+const char *config_parent_name(const struct config_node *n);
+
+struct config_node *clone_config_node(struct dm_pool *mem, const struct config_node *cn,
+				      int siblings);
+#endif

lib/config/defaults.h

@@ -0,0 +1,145 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_DEFAULTS_H
+#define _LVM_DEFAULTS_H
+
+#define DEFAULT_ARCHIVE_ENABLED 1
+#define DEFAULT_BACKUP_ENABLED 1
+
+#define DEFAULT_ARCHIVE_SUBDIR "archive"
+#define DEFAULT_BACKUP_SUBDIR "backup"
+#define DEFAULT_CACHE_SUBDIR "cache"
+#define DEFAULT_CACHE_FILE_PREFIX ""
+
+#define DEFAULT_ARCHIVE_DAYS 30
+#define DEFAULT_ARCHIVE_NUMBER 10
+
+#define DEFAULT_SYS_DIR "/etc/lvm"
+#define DEFAULT_DEV_DIR "/dev"
+#define DEFAULT_PROC_DIR "/proc"
+#define DEFAULT_SYSFS_SCAN 1
+#define DEFAULT_MD_COMPONENT_DETECTION 1
+#define DEFAULT_MD_CHUNK_ALIGNMENT 1
+#define DEFAULT_IGNORE_SUSPENDED_DEVICES 1
+#define DEFAULT_DATA_ALIGNMENT_OFFSET_DETECTION 1
+#define DEFAULT_DATA_ALIGNMENT_DETECTION 1
+
+#define DEFAULT_LOCK_DIR "/var/lock/lvm"
+#define DEFAULT_LOCKING_LIB "liblvm2clusterlock.so"
+#define DEFAULT_FALLBACK_TO_LOCAL_LOCKING 1
+#define DEFAULT_FALLBACK_TO_CLUSTERED_LOCKING 1
+#define DEFAULT_WAIT_FOR_LOCKS 1
+#define DEFAULT_PRIORITISE_WRITE_LOCKS 1
+
+#define DEFAULT_MIRRORLOG "disk"
+#define DEFAULT_MIRROR_LOG_FAULT_POLICY "allocate"
+#define DEFAULT_MIRROR_IMAGE_FAULT_POLICY "remove"
+#define DEFAULT_MIRROR_MAX_IMAGES 8 /* limited by kernel DM_KCOPYD_MAX_REGIONS */
+#define DEFAULT_DMEVENTD_MIRROR_LIB "libdevmapper-event-lvm2mirror.so"
+#define DEFAULT_DMEVENTD_MONITOR 1
+#define DEFAULT_BACKGROUND_POLLING 1
+
+#define DEFAULT_UMASK 0077
+
+#ifdef LVM1_FALLBACK
+#  define DEFAULT_FALLBACK_TO_LVM1 1
+#else
+#  define DEFAULT_FALLBACK_TO_LVM1 0
+#endif
+
+#define DEFAULT_FORMAT "lvm2"
+
+#define DEFAULT_STRIPESIZE 64	/* KB */
+#define DEFAULT_PVMETADATASIZE 255
+#define DEFAULT_PVMETADATACOPIES 1
+#define DEFAULT_LABELSECTOR UINT64_C(1)
+#define DEFAULT_READ_AHEAD "auto"
+#define DEFAULT_UDEV_RULES 1
+#define DEFAULT_UDEV_SYNC 0
+#define DEFAULT_EXTENT_SIZE 4096	/* In KB */
+#define DEFAULT_MAX_PV 0
+#define DEFAULT_MAX_LV 0
+#define DEFAULT_ALLOC_POLICY ALLOC_NORMAL
+#define DEFAULT_CLUSTERED 0
+
+#define DEFAULT_MSG_PREFIX "  "
+#define DEFAULT_CMD_NAME 0
+#define DEFAULT_OVERWRITE 0
+
+#ifndef DEFAULT_LOG_FACILITY
+#  define DEFAULT_LOG_FACILITY LOG_USER
+#endif
+
+#define DEFAULT_SYSLOG 1
+#define DEFAULT_VERBOSE 0
+#define DEFAULT_LOGLEVEL 0
+#define DEFAULT_INDENT 1
+#define DEFAULT_ABORT_ON_INTERNAL_ERRORS 0
+#define DEFAULT_UNITS "h"
+#define DEFAULT_SUFFIX 1
+#define DEFAULT_HOSTTAGS 0
+
+#ifndef DEFAULT_SI_UNIT_CONSISTENCY
+#  define DEFAULT_SI_UNIT_CONSISTENCY 1
+#endif
+
+#ifdef DEVMAPPER_SUPPORT
+#  define DEFAULT_ACTIVATION 1
+#  define DEFAULT_RESERVED_MEMORY 8192
+#  define DEFAULT_RESERVED_STACK 256
+#  define DEFAULT_PROCESS_PRIORITY -18
+#else
+#  define DEFAULT_ACTIVATION 0
+#endif
+
+#define DEFAULT_STRIPE_FILLER "error"
+#define DEFAULT_MIRROR_REGION_SIZE 512	/* KB */
+#define DEFAULT_INTERVAL 15
+
+#ifdef READLINE_SUPPORT
+#  define DEFAULT_MAX_HISTORY 100
+#endif
+
+#define DEFAULT_REP_ALIGNED 1
+#define DEFAULT_REP_BUFFERED 1
+#define DEFAULT_REP_COLUMNS_AS_ROWS 0
+#define DEFAULT_REP_HEADINGS 1
+#define DEFAULT_REP_PREFIXES 0
+#define DEFAULT_REP_QUOTED 1
+#define DEFAULT_REP_SEPARATOR " "
+
+#define DEFAULT_LVS_COLS "lv_name,vg_name,lv_attr,lv_size,origin,snap_percent,move_pv,mirror_log,copy_percent,convert_lv"
+#define DEFAULT_VGS_COLS "vg_name,pv_count,lv_count,snap_count,vg_attr,vg_size,vg_free"
+#define DEFAULT_PVS_COLS "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free"
+#define DEFAULT_SEGS_COLS "lv_name,vg_name,lv_attr,stripes,segtype,seg_size"
+#define DEFAULT_PVSEGS_COLS "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,pvseg_start,pvseg_size"
+
+#define DEFAULT_LVS_COLS_VERB "lv_name,vg_name,seg_count,lv_attr,lv_size,lv_major,lv_minor,lv_kernel_major,lv_kernel_minor,origin,snap_percent,move_pv,copy_percent,mirror_log,convert_lv,lv_uuid"
+#define DEFAULT_VGS_COLS_VERB "vg_name,vg_attr,vg_extent_size,pv_count,lv_count,snap_count,vg_size,vg_free,vg_uuid"
+#define DEFAULT_PVS_COLS_VERB "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,dev_size,pv_uuid"
+#define DEFAULT_SEGS_COLS_VERB "lv_name,vg_name,lv_attr,seg_start,seg_size,stripes,segtype,stripesize,chunksize"
+#define DEFAULT_PVSEGS_COLS_VERB "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,pvseg_start,pvseg_size,lv_name,seg_start_pe,segtype,seg_pe_ranges"
+
+#define DEFAULT_LVS_SORT "vg_name,lv_name"
+#define DEFAULT_VGS_SORT "vg_name"
+#define DEFAULT_PVS_SORT "pv_name"
+#define DEFAULT_SEGS_SORT "vg_name,lv_name,seg_start"
+#define DEFAULT_PVSEGS_SORT "pv_name,pvseg_start"
+
+#define DEFAULT_MIRROR_DEVICE_FAULT_POLICY "remove"
+#define DEFAULT_MIRROR_LOG_FAULT_POLICY "allocate"
+
+#endif				/* _LVM_DEFAULTS_H */

lib/datastruct/btree.c

@@ -0,0 +1,137 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "btree.h"
+
+struct node {
+	uint32_t key;
+	struct node *l, *r, *p;
+
+	void *data;
+};
+
+struct btree {
+	struct dm_pool *mem;
+	struct node *root;
+};
+
+struct btree *btree_create(struct dm_pool *mem)
+{
+	struct btree *t = dm_pool_alloc(mem, sizeof(*t));
+
+	if (t) {
+		t->mem = mem;
+		t->root = NULL;
+	}
+
+	return t;
+}
+
+/*
+ * Shuffle the bits in a key, to try and remove
+ * any ordering.
+ */
+static uint32_t _shuffle(uint32_t k)
+{
+#if 1
+	return ((k & 0xff) << 24 |
+		(k & 0xff00) << 8 |
+		(k & 0xff0000) >> 8 | (k & 0xff000000) >> 24);
+#else
+	return k;
+#endif
+}
+
+static struct node **_lookup(struct node *const *c, uint32_t key,
+			     struct node **p)
+{
+	*p = NULL;
+	while (*c) {
+		*p = *c;
+		if ((*c)->key == key)
+			break;
+
+		if (key < (*c)->key)
+			c = &(*c)->l;
+
+		else
+			c = &(*c)->r;
+	}
+
+	return (struct node **)c;
+}
+
+void *btree_lookup(const struct btree *t, uint32_t k)
+{
+	uint32_t key = _shuffle(k);
+	struct node *p, **c = _lookup(&t->root, key, &p);
+	return (*c) ? (*c)->data : NULL;
+}
+
+int btree_insert(struct btree *t, uint32_t k, void *data)
+{
+	uint32_t key = _shuffle(k);
+	struct node *p, **c = _lookup(&t->root, key, &p), *n;
+
+	if (!*c) {
+		if (!(n = dm_pool_alloc(t->mem, sizeof(*n))))
+			return_0;
+
+		n->key = key;
+		n->data = data;
+		n->l = n->r = NULL;
+		n->p = p;
+
+		*c = n;
+	}
+
+	return 1;
+}
+
+void *btree_get_data(const struct btree_iter *it)
+{
+	return ((struct node *) it)->data;
+}
+
+static struct node *_left(struct node *n)
+{
+	while (n->l)
+		n = n->l;
+	return n;
+}
+
+struct btree_iter *btree_first(const struct btree *t)
+{
+	if (!t->root)
+		return NULL;
+
+	return (struct btree_iter *) _left(t->root);
+}
+
+struct btree_iter *btree_next(const struct btree_iter *it)
+{
+	struct node *n = (struct node *) it;
+	uint32_t k = n->key;
+
+	if (n->r)
+		return (struct btree_iter *) _left(n->r);
+
+	do
+		n = n->p;
+	while (n && k > n->key);
+
+	return (struct btree_iter *) n;
+}

lib/datastruct/btree.h

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_BTREE_H
+#define _LVM_BTREE_H
+
+struct btree;
+
+struct btree *btree_create(struct dm_pool *mem);
+
+void *btree_lookup(const struct btree *t, uint32_t k);
+int btree_insert(struct btree *t, uint32_t k, void *data);
+
+struct btree_iter;
+void *btree_get_data(const struct btree_iter *it);
+
+struct btree_iter *btree_first(const struct btree *t);
+struct btree_iter *btree_next(const struct btree_iter *it);
+
+#endif

lib/datastruct/lvm-types.h

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_TYPES_H
+#define _LVM_TYPES_H
+
+#include <sys/types.h>
+#include <inttypes.h>
+
+/* Define some portable printing types */
+#define PRIsize_t "zu"
+#define PRIptrdiff_t "td"
+#define PRIpid_t PRId32
+
+struct str_list {
+	struct dm_list list;
+	const char *str;
+};
+
+#endif

lib/datastruct/str_list.c

@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2003-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "str_list.h"
+
+struct dm_list *str_list_create(struct dm_pool *mem)
+{
+	struct dm_list *sl;
+
+	if (!(sl = dm_pool_alloc(mem, sizeof(struct dm_list)))) {
+		log_errno(ENOMEM, "str_list allocation failed");
+		return NULL;
+	}
+
+	dm_list_init(sl);
+
+	return sl;
+}
+
+int str_list_add(struct dm_pool *mem, struct dm_list *sll, const char *str)
+{
+	struct str_list *sln;
+
+	if (!str)
+		return_0;
+
+	/* Already in list? */
+	if (str_list_match_item(sll, str))
+		return 1;
+
+	if (!(sln = dm_pool_alloc(mem, sizeof(*sln))))
+		return_0;
+
+	sln->str = str;
+	dm_list_add(sll, &sln->list);
+
+	return 1;
+}
+
+int str_list_del(struct dm_list *sll, const char *str)
+{
+	struct dm_list *slh, *slht;
+
+	dm_list_iterate_safe(slh, slht, sll) {
+		if (!strcmp(str, dm_list_item(slh, struct str_list)->str))
+			 dm_list_del(slh);
+	}
+
+	return 1;
+}
+
+int str_list_dup(struct dm_pool *mem, struct dm_list *sllnew,
+		 const struct dm_list *sllold)
+{
+	struct str_list *sl;
+
+	dm_list_init(sllnew);
+
+	dm_list_iterate_items(sl, sllold) {
+		if (!str_list_add(mem, sllnew, dm_pool_strdup(mem, sl->str)))
+			return_0;
+	}
+
+	return 1;
+}
+
+/*
+ * Is item on list?
+ */
+int str_list_match_item(const struct dm_list *sll, const char *str)
+{
+	struct str_list *sl;
+
+	dm_list_iterate_items(sl, sll)
+	    if (!strcmp(str, sl->str))
+		return 1;
+
+	return 0;
+}
+
+/*
+ * Is at least one item on both lists?
+ */
+int str_list_match_list(const struct dm_list *sll, const struct dm_list *sll2)
+{
+	struct str_list *sl;
+
+	dm_list_iterate_items(sl, sll)
+	    if (str_list_match_item(sll2, sl->str))
+		return 1;
+
+	return 0;
+}
+
+/*
+ * Do both lists contain the same set of items?
+ */
+int str_list_lists_equal(const struct dm_list *sll, const struct dm_list *sll2)
+{
+	struct str_list *sl;
+
+	if (dm_list_size(sll) != dm_list_size(sll2))
+		return 0;
+
+	dm_list_iterate_items(sl, sll)
+	    if (!str_list_match_item(sll2, sl->str))
+		return 0;
+
+	return 1;
+}

lib/datastruct/str_list.h

@@ -0,0 +1,28 @@
+/*
+ * Copyright (C) 2003-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_STR_LIST_H
+#define _LVM_STR_LIST_H
+
+struct dm_list *str_list_create(struct dm_pool *mem);
+int str_list_add(struct dm_pool *mem, struct dm_list *sll, const char *str);
+int str_list_del(struct dm_list *sll, const char *str);
+int str_list_match_item(const struct dm_list *sll, const char *str);
+int str_list_match_list(const struct dm_list *sll, const struct dm_list *sll2);
+int str_list_lists_equal(const struct dm_list *sll, const struct dm_list *sll2);
+int str_list_dup(struct dm_pool *mem, struct dm_list *sllnew,
+		 const struct dm_list *sllold);
+
+#endif

lib/device/dev-cache.c

@@ -0,0 +1,783 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "dev-cache.h"
+#include "lvm-types.h"
+#include "btree.h"
+#include "filter.h"
+#include "filter-persistent.h"
+#include "toolcontext.h"
+
+#include <unistd.h>
+#include <sys/param.h>
+#include <dirent.h>
+
+struct dev_iter {
+	struct btree_iter *current;
+	struct dev_filter *filter;
+};
+
+struct dir_list {
+	struct dm_list list;
+	char dir[0];
+};
+
+static struct {
+	struct dm_pool *mem;
+	struct dm_hash_table *names;
+	struct btree *devices;
+	struct dm_regex *preferred_names_matcher;
+
+	int has_scanned;
+	struct dm_list dirs;
+	struct dm_list files;
+
+} _cache;
+
+#define _alloc(x) dm_pool_zalloc(_cache.mem, (x))
+#define _free(x) dm_pool_free(_cache.mem, (x))
+#define _strdup(x) dm_pool_strdup(_cache.mem, (x))
+
+static int _insert(const char *path, int rec);
+
+struct device *dev_create_file(const char *filename, struct device *dev,
+			       struct str_list *alias, int use_malloc)
+{
+	int allocate = !dev;
+
+	if (allocate) {
+		if (use_malloc) {
+			if (!(dev = dm_malloc(sizeof(*dev)))) {
+				log_error("struct device allocation failed");
+				return NULL;
+			}
+			if (!(alias = dm_malloc(sizeof(*alias)))) {
+				log_error("struct str_list allocation failed");
+				dm_free(dev);
+				return NULL;
+			}
+			if (!(alias->str = dm_strdup(filename))) {
+				log_error("filename strdup failed");
+				dm_free(dev);
+				dm_free(alias);
+				return NULL;
+			}
+			dev->flags = DEV_ALLOCED;
+		} else {
+			if (!(dev = _alloc(sizeof(*dev)))) {
+				log_error("struct device allocation failed");
+				return NULL;
+			}
+			if (!(alias = _alloc(sizeof(*alias)))) {
+				log_error("struct str_list allocation failed");
+				_free(dev);
+				return NULL;
+			}
+			if (!(alias->str = _strdup(filename))) {
+				log_error("filename strdup failed");
+				return NULL;
+			}
+		}
+	} else if (!(alias->str = dm_strdup(filename))) {
+		log_error("filename strdup failed");
+		return NULL;
+	}
+
+	dev->flags |= DEV_REGULAR;
+	dm_list_init(&dev->aliases);
+	dm_list_add(&dev->aliases, &alias->list);
+	dev->end = UINT64_C(0);
+	dev->dev = 0;
+	dev->fd = -1;
+	dev->open_count = 0;
+	dev->block_size = -1;
+	dev->read_ahead = -1;
+	memset(dev->pvid, 0, sizeof(dev->pvid));
+	dm_list_init(&dev->open_list);
+
+	return dev;
+}
+
+static struct device *_dev_create(dev_t d)
+{
+	struct device *dev;
+
+	if (!(dev = _alloc(sizeof(*dev)))) {
+		log_error("struct device allocation failed");
+		return NULL;
+	}
+	dev->flags = 0;
+	dm_list_init(&dev->aliases);
+	dev->dev = d;
+	dev->fd = -1;
+	dev->open_count = 0;
+	dev->block_size = -1;
+	dev->read_ahead = -1;
+	dev->end = UINT64_C(0);
+	memset(dev->pvid, 0, sizeof(dev->pvid));
+	dm_list_init(&dev->open_list);
+
+	return dev;
+}
+
+void dev_set_preferred_name(struct str_list *sl, struct device *dev)
+{
+	/*
+	 * Don't interfere with ordering specified in config file.
+	 */
+	if (_cache.preferred_names_matcher)
+		return;
+
+	log_debug("%s: New preferred name", sl->str);
+	dm_list_del(&sl->list);
+	dm_list_add_h(&dev->aliases, &sl->list);
+}
+
+/* Return 1 if we prefer path1 else return 0 */
+static int _compare_paths(const char *path0, const char *path1)
+{
+	int slash0 = 0, slash1 = 0;
+	int m0, m1;
+	const char *p;
+	char p0[PATH_MAX], p1[PATH_MAX];
+	char *s0, *s1;
+	struct stat stat0, stat1;
+
+	/*
+	 * FIXME Better to compare patterns one-at-a-time against all names.
+	 */
+	if (_cache.preferred_names_matcher) {
+		m0 = dm_regex_match(_cache.preferred_names_matcher, path0);
+		m1 = dm_regex_match(_cache.preferred_names_matcher, path1);
+
+		if (m0 != m1) {
+			if (m0 < 0)
+				return 1;
+			if (m1 < 0)
+				return 0;
+			if (m0 < m1)
+				return 1;
+			if (m1 < m0)
+				return 0;
+		}
+	}
+
+	/*
+	 * Built-in rules.
+	 */
+
+	/* Return the path with fewer slashes */
+	for (p = path0; p++; p = (const char *) strchr(p, '/'))
+		slash0++;
+
+	for (p = path1; p++; p = (const char *) strchr(p, '/'))
+		slash1++;
+
+	if (slash0 < slash1)
+		return 0;
+	if (slash1 < slash0)
+		return 1;
+
+	strncpy(p0, path0, PATH_MAX);
+	strncpy(p1, path1, PATH_MAX);
+	s0 = &p0[0] + 1;
+	s1 = &p1[0] + 1;
+
+	/* We prefer symlinks - they exist for a reason!
+	 * So we prefer a shorter path before the first symlink in the name.
+	 * FIXME Configuration option to invert this? */
+	while (s0) {
+		s0 = strchr(s0, '/');
+		s1 = strchr(s1, '/');
+		if (s0) {
+			*s0 = '\0';
+			*s1 = '\0';
+		}
+		if (lstat(p0, &stat0)) {
+			log_sys_very_verbose("lstat", p0);
+			return 1;
+		}
+		if (lstat(p1, &stat1)) {
+			log_sys_very_verbose("lstat", p1);
+			return 0;
+		}
+		if (S_ISLNK(stat0.st_mode) && !S_ISLNK(stat1.st_mode))
+			return 0;
+		if (!S_ISLNK(stat0.st_mode) && S_ISLNK(stat1.st_mode))
+			return 1;
+		if (s0) {
+			*s0++ = '/';
+			*s1++ = '/';
+		}
+	}
+
+	/* ASCII comparison */
+	if (strcmp(path0, path1) < 0)
+		return 0;
+	else
+		return 1;
+}
+
+static int _add_alias(struct device *dev, const char *path)
+{
+	struct str_list *sl = _alloc(sizeof(*sl));
+	struct str_list *strl;
+	const char *oldpath;
+	int prefer_old = 1;
+
+	if (!sl)
+		return_0;
+
+	/* Is name already there? */
+	dm_list_iterate_items(strl, &dev->aliases) {
+		if (!strcmp(strl->str, path)) {
+			log_debug("%s: Already in device cache", path);
+			return 1;
+		}
+	}
+
+	if (!(sl->str = dm_pool_strdup(_cache.mem, path)))
+		return_0;
+
+	if (!dm_list_empty(&dev->aliases)) {
+		oldpath = dm_list_item(dev->aliases.n, struct str_list)->str;
+		prefer_old = _compare_paths(path, oldpath);
+		log_debug("%s: Aliased to %s in device cache%s",
+			  path, oldpath, prefer_old ? "" : " (preferred name)");
+
+	} else
+		log_debug("%s: Added to device cache", path);
+
+	if (prefer_old)
+		dm_list_add(&dev->aliases, &sl->list);
+	else
+		dm_list_add_h(&dev->aliases, &sl->list);
+
+	return 1;
+}
+
+/*
+ * Either creates a new dev, or adds an alias to
+ * an existing dev.
+ */
+static int _insert_dev(const char *path, dev_t d)
+{
+	struct device *dev;
+	static dev_t loopfile_count = 0;
+	int loopfile = 0;
+
+	/* Generate pretend device numbers for loopfiles */
+	if (!d) {
+		if (dm_hash_lookup(_cache.names, path))
+			return 1;
+		d = ++loopfile_count;
+		loopfile = 1;
+	}
+
+	/* is this device already registered ? */
+	if (!(dev = (struct device *) btree_lookup(_cache.devices,
+						   (uint32_t) d))) {
+		/* create new device */
+		if (loopfile) {
+			if (!(dev = dev_create_file(path, NULL, NULL, 0)))
+				return_0;
+		} else if (!(dev = _dev_create(d)))
+			return_0;
+
+		if (!(btree_insert(_cache.devices, (uint32_t) d, dev))) {
+			log_error("Couldn't insert device into binary tree.");
+			_free(dev);
+			return 0;
+		}
+	}
+
+	if (!loopfile && !_add_alias(dev, path)) {
+		log_error("Couldn't add alias to dev cache.");
+		return 0;
+	}
+
+	if (!dm_hash_insert(_cache.names, path, dev)) {
+		log_error("Couldn't add name to hash in dev cache.");
+		return 0;
+	}
+
+	return 1;
+}
+
+static char *_join(const char *dir, const char *name)
+{
+	size_t len = strlen(dir) + strlen(name) + 2;
+	char *r = dm_malloc(len);
+	if (r)
+		snprintf(r, len, "%s/%s", dir, name);
+
+	return r;
+}
+
+/*
+ * Get rid of extra slashes in the path string.
+ */
+static void _collapse_slashes(char *str)
+{
+	char *ptr;
+	int was_slash = 0;
+
+	for (ptr = str; *ptr; ptr++) {
+		if (*ptr == '/') {
+			if (was_slash)
+				continue;
+
+			was_slash = 1;
+		} else
+			was_slash = 0;
+		*str++ = *ptr;
+	}
+
+	*str = *ptr;
+}
+
+static int _insert_dir(const char *dir)
+{
+	int n, dirent_count, r = 1;
+	struct dirent **dirent;
+	char *path;
+
+	dirent_count = scandir(dir, &dirent, NULL, alphasort);
+	if (dirent_count > 0) {
+		for (n = 0; n < dirent_count; n++) {
+			if (dirent[n]->d_name[0] == '.') {
+				free(dirent[n]);
+				continue;
+			}
+
+			if (!(path = _join(dir, dirent[n]->d_name)))
+				return_0;
+
+			_collapse_slashes(path);
+			r &= _insert(path, 1);
+			dm_free(path);
+
+			free(dirent[n]);
+		}
+		free(dirent);
+	}
+
+	return r;
+}
+
+static int _insert_file(const char *path)
+{
+	struct stat info;
+
+	if (stat(path, &info) < 0) {
+		log_sys_very_verbose("stat", path);
+		return 0;
+	}
+
+	if (!S_ISREG(info.st_mode)) {
+		log_debug("%s: Not a regular file", path);
+		return 0;
+	}
+
+	if (!_insert_dev(path, 0))
+		return_0;
+
+	return 1;
+}
+
+static int _insert(const char *path, int rec)
+{
+	struct stat info;
+	int r = 0;
+
+	if (stat(path, &info) < 0) {
+		log_sys_very_verbose("stat", path);
+		return 0;
+	}
+
+	if (S_ISDIR(info.st_mode)) {	/* add a directory */
+		/* check it's not a symbolic link */
+		if (lstat(path, &info) < 0) {
+			log_sys_very_verbose("lstat", path);
+			return 0;
+		}
+
+		if (S_ISLNK(info.st_mode)) {
+			log_debug("%s: Symbolic link to directory", path);
+			return 0;
+		}
+
+		if (rec)
+			r = _insert_dir(path);
+
+	} else {		/* add a device */
+		if (!S_ISBLK(info.st_mode)) {
+			log_debug("%s: Not a block device", path);
+			return 0;
+		}
+
+		if (!_insert_dev(path, info.st_rdev))
+			return_0;
+
+		r = 1;
+	}
+
+	return r;
+}
+
+static void _full_scan(int dev_scan)
+{
+	struct dir_list *dl;
+
+	if (_cache.has_scanned && !dev_scan)
+		return;
+
+	dm_list_iterate_items(dl, &_cache.dirs)
+		_insert_dir(dl->dir);
+
+	dm_list_iterate_items(dl, &_cache.files)
+		_insert_file(dl->dir);
+
+	_cache.has_scanned = 1;
+	init_full_scan_done(1);
+}
+
+int dev_cache_has_scanned(void)
+{
+	return _cache.has_scanned;
+}
+
+void dev_cache_scan(int do_scan)
+{
+	if (!do_scan)
+		_cache.has_scanned = 1;
+	else
+		_full_scan(1);
+}
+
+static int _init_preferred_names(struct cmd_context *cmd)
+{
+	const struct config_node *cn;
+	struct config_value *v;
+	struct dm_pool *scratch = NULL;
+	char **regex;
+	unsigned count = 0;
+	int i, r = 0;
+
+	_cache.preferred_names_matcher = NULL;
+
+	if (!(cn = find_config_tree_node(cmd, "devices/preferred_names")) ||
+	    cn->v->type == CFG_EMPTY_ARRAY) {
+		log_very_verbose("devices/preferred_names not found in config file: "
+				 "using built-in preferences");
+		return 1;
+	}
+
+	for (v = cn->v; v; v = v->next) {
+		if (v->type != CFG_STRING) {
+			log_error("preferred_names patterns must be enclosed in quotes");
+			return 0;
+		}
+
+		count++;
+	}
+
+	if (!(scratch = dm_pool_create("preferred device name matcher", 1024)))
+		return_0;
+
+	if (!(regex = dm_pool_alloc(scratch, sizeof(*regex) * count))) {
+		log_error("Failed to allocate preferred device name "
+			  "pattern list.");
+		goto out;
+	}
+
+	for (v = cn->v, i = count - 1; v; v = v->next, i--) {
+		if (!(regex[i] = dm_pool_strdup(scratch, v->v.str))) {
+			log_error("Failed to allocate a preferred device name "
+				  "pattern.");
+			goto out;
+		}
+	}
+
+	if (!(_cache.preferred_names_matcher =
+		dm_regex_create(_cache.mem,(const char **) regex, count))) {
+		log_error("Preferred device name pattern matcher creation failed.");
+		goto out;
+	}
+
+	r = 1;
+
+out:
+	dm_pool_destroy(scratch);
+
+	return r;
+}
+
+int dev_cache_init(struct cmd_context *cmd)
+{
+	_cache.names = NULL;
+	_cache.has_scanned = 0;
+
+	if (!(_cache.mem = dm_pool_create("dev_cache", 10 * 1024)))
+		return_0;
+
+	if (!(_cache.names = dm_hash_create(128))) {
+		dm_pool_destroy(_cache.mem);
+		_cache.mem = 0;
+		return_0;
+	}
+
+	if (!(_cache.devices = btree_create(_cache.mem))) {
+		log_error("Couldn't create binary tree for dev-cache.");
+		goto bad;
+	}
+
+	dm_list_init(&_cache.dirs);
+	dm_list_init(&_cache.files);
+
+	if (!_init_preferred_names(cmd))
+		goto_bad;
+
+	return 1;
+
+      bad:
+	dev_cache_exit();
+	return 0;
+}
+
+static void _check_closed(struct device *dev)
+{
+	if (dev->fd >= 0)
+		log_error("Device '%s' has been left open.", dev_name(dev));
+}
+
+static void _check_for_open_devices(void)
+{
+	dm_hash_iter(_cache.names, (dm_hash_iterate_fn) _check_closed);
+}
+
+void dev_cache_exit(void)
+{
+	if (_cache.names)
+		_check_for_open_devices();
+
+	if (_cache.preferred_names_matcher)
+		_cache.preferred_names_matcher = NULL;
+
+	if (_cache.mem) {
+		dm_pool_destroy(_cache.mem);
+		_cache.mem = NULL;
+	}
+
+	if (_cache.names) {
+		dm_hash_destroy(_cache.names);
+		_cache.names = NULL;
+	}
+
+	_cache.devices = NULL;
+	_cache.has_scanned = 0;
+	dm_list_init(&_cache.dirs);
+	dm_list_init(&_cache.files);
+}
+
+int dev_cache_add_dir(const char *path)
+{
+	struct dir_list *dl;
+	struct stat st;
+
+	if (stat(path, &st)) {
+		log_error("Ignoring %s: %s", path, strerror(errno));
+		/* But don't fail */
+		return 1;
+	}
+
+	if (!S_ISDIR(st.st_mode)) {
+		log_error("Ignoring %s: Not a directory", path);
+		return 1;
+	}
+
+	if (!(dl = _alloc(sizeof(*dl) + strlen(path) + 1))) {
+		log_error("dir_list allocation failed");
+		return 0;
+	}
+
+	strcpy(dl->dir, path);
+	dm_list_add(&_cache.dirs, &dl->list);
+	return 1;
+}
+
+int dev_cache_add_loopfile(const char *path)
+{
+	struct dir_list *dl;
+	struct stat st;
+
+	if (stat(path, &st)) {
+		log_error("Ignoring %s: %s", path, strerror(errno));
+		/* But don't fail */
+		return 1;
+	}
+
+	if (!S_ISREG(st.st_mode)) {
+		log_error("Ignoring %s: Not a regular file", path);
+		return 1;
+	}
+
+	if (!(dl = _alloc(sizeof(*dl) + strlen(path) + 1))) {
+		log_error("dir_list allocation failed for file");
+		return 0;
+	}
+
+	strcpy(dl->dir, path);
+	dm_list_add(&_cache.files, &dl->list);
+	return 1;
+}
+
+/* Check cached device name is still valid before returning it */
+/* This should be a rare occurrence */
+/* set quiet if the cache is expected to be out-of-date */
+/* FIXME Make rest of code pass/cache struct device instead of dev_name */
+const char *dev_name_confirmed(struct device *dev, int quiet)
+{
+	struct stat buf;
+	const char *name;
+	int r;
+
+	if ((dev->flags & DEV_REGULAR))
+		return dev_name(dev);
+
+	while ((r = stat(name = dm_list_item(dev->aliases.n,
+					  struct str_list)->str, &buf)) ||
+	       (buf.st_rdev != dev->dev)) {
+		if (r < 0) {
+			if (quiet)
+				log_sys_debug("stat", name);
+			else
+				log_sys_error("stat", name);
+		}
+		if (quiet)
+			log_debug("Path %s no longer valid for device(%d,%d)",
+				  name, (int) MAJOR(dev->dev),
+				  (int) MINOR(dev->dev));
+		else
+			log_error("Path %s no longer valid for device(%d,%d)",
+				  name, (int) MAJOR(dev->dev),
+				  (int) MINOR(dev->dev));
+
+		/* Remove the incorrect hash entry */
+		dm_hash_remove(_cache.names, name);
+
+		/* Leave list alone if there isn't an alternative name */
+		/* so dev_name will always find something to return. */
+		/* Otherwise add the name to the correct device. */
+		if (dm_list_size(&dev->aliases) > 1) {
+			dm_list_del(dev->aliases.n);
+			if (!r)
+				_insert(name, 0);
+			continue;
+		}
+
+		/* Scanning issues this inappropriately sometimes. */
+		log_debug("Aborting - please provide new pathname for what "
+			  "used to be %s", name);
+		return NULL;
+	}
+
+	return dev_name(dev);
+}
+
+struct device *dev_cache_get(const char *name, struct dev_filter *f)
+{
+	struct stat buf;
+	struct device *d = (struct device *) dm_hash_lookup(_cache.names, name);
+
+	if (d && (d->flags & DEV_REGULAR))
+		return d;
+
+	/* If the entry's wrong, remove it */
+	if (d && (stat(name, &buf) || (buf.st_rdev != d->dev))) {
+		dm_hash_remove(_cache.names, name);
+		d = NULL;
+	}
+
+	if (!d) {
+		_insert(name, 0);
+		d = (struct device *) dm_hash_lookup(_cache.names, name);
+		if (!d) {
+			_full_scan(0);
+			d = (struct device *) dm_hash_lookup(_cache.names, name);
+		}
+	}
+
+	return (d && (!f || (d->flags & DEV_REGULAR) ||
+		      f->passes_filter(f, d))) ? d : NULL;
+}
+
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan)
+{
+	struct dev_iter *di = dm_malloc(sizeof(*di));
+
+	if (!di) {
+		log_error("dev_iter allocation failed");
+		return NULL;
+	}
+
+	if (dev_scan && !trust_cache()) {
+		/* Flag gets reset between each command */
+		if (!full_scan_done())
+			persistent_filter_wipe(f); /* Calls _full_scan(1) */
+	} else
+		_full_scan(0);
+
+	di->current = btree_first(_cache.devices);
+	di->filter = f;
+
+	return di;
+}
+
+void dev_iter_destroy(struct dev_iter *iter)
+{
+	dm_free(iter);
+}
+
+static struct device *_iter_next(struct dev_iter *iter)
+{
+	struct device *d = btree_get_data(iter->current);
+	iter->current = btree_next(iter->current);
+	return d;
+}
+
+struct device *dev_iter_get(struct dev_iter *iter)
+{
+	while (iter->current) {
+		struct device *d = _iter_next(iter);
+		if (!iter->filter || (d->flags & DEV_REGULAR) ||
+		    iter->filter->passes_filter(iter->filter, d))
+			return d;
+	}
+
+	return NULL;
+}
+
+int dev_fd(struct device *dev)
+{
+	return dev->fd;
+}
+
+const char *dev_name(const struct device *dev)
+{
+	return (dev) ? dm_list_item(dev->aliases.n, struct str_list)->str :
+	    "unknown device";
+}

lib/device/dev-cache.h

@@ -0,0 +1,55 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_DEV_CACHE_H
+#define _LVM_DEV_CACHE_H
+
+#include "device.h"
+
+/*
+ * predicate for devices.
+ */
+struct dev_filter {
+	int (*passes_filter) (struct dev_filter * f, struct device * dev);
+	void (*destroy) (struct dev_filter * f);
+	void *private;
+};
+
+/*
+ * The global device cache.
+ */
+struct cmd_context;
+int dev_cache_init(struct cmd_context *cmd);
+void dev_cache_exit(void);
+
+/* Trigger(1) or avoid(0) a scan */
+void dev_cache_scan(int do_scan);
+int dev_cache_has_scanned(void);
+
+int dev_cache_add_dir(const char *path);
+int dev_cache_add_loopfile(const char *path);
+struct device *dev_cache_get(const char *name, struct dev_filter *f);
+
+void dev_set_preferred_name(struct str_list *sl, struct device *dev);
+
+/*
+ * Object for iterating through the cache.
+ */
+struct dev_iter;
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan);
+void dev_iter_destroy(struct dev_iter *iter);
+struct device *dev_iter_get(struct dev_iter *iter);
+
+#endif

lib/device/dev-io.c

@@ -0,0 +1,712 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "lvm-types.h"
+#include "device.h"
+#include "metadata.h"
+#include "lvmcache.h"
+#include "memlock.h"
+#include "locking.h"
+
+#include <limits.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+
+#ifdef linux
+#  define u64 uint64_t		/* Missing without __KERNEL__ */
+#  undef WNOHANG		/* Avoid redefinition */
+#  undef WUNTRACED		/* Avoid redefinition */
+#  include <linux/fs.h>		/* For block ioctl definitions */
+#  define BLKSIZE_SHIFT SECTOR_SHIFT
+#  ifndef BLKGETSIZE64		/* fs.h out-of-date */
+#    define BLKGETSIZE64 _IOR(0x12, 114, size_t)
+#  endif /* BLKGETSIZE64 */
+#else
+#  include <sys/disk.h>
+#  define BLKBSZGET DKIOCGETBLOCKSIZE
+#  define BLKSSZGET DKIOCGETBLOCKSIZE
+#  define BLKGETSIZE64 DKIOCGETBLOCKCOUNT
+#  define BLKFLSBUF DKIOCSYNCHRONIZECACHE
+#  define BLKSIZE_SHIFT 0
+#endif
+
+#ifdef O_DIRECT_SUPPORT
+#  ifndef O_DIRECT
+#    error O_DIRECT support configured but O_DIRECT definition not found in headers
+#  endif
+#endif
+
+static DM_LIST_INIT(_open_devices);
+
+/*-----------------------------------------------------------------
+ * The standard io loop that keeps submitting an io until it's
+ * all gone.
+ *---------------------------------------------------------------*/
+static int _io(struct device_area *where, void *buffer, int should_write)
+{
+	int fd = dev_fd(where->dev);
+	ssize_t n = 0;
+	size_t total = 0;
+
+	if (fd < 0) {
+		log_error("Attempt to read an unopened device (%s).",
+			  dev_name(where->dev));
+		return 0;
+	}
+
+	/*
+	 * Skip all writes in test mode.
+	 */
+	if (should_write && test_mode())
+		return 1;
+
+	if (where->size > SSIZE_MAX) {
+		log_error("Read size too large: %" PRIu64, where->size);
+		return 0;
+	}
+
+	if (lseek(fd, (off_t) where->start, SEEK_SET) < 0) {
+		log_error("%s: lseek %" PRIu64 " failed: %s",
+			  dev_name(where->dev), (uint64_t) where->start,
+			  strerror(errno));
+		return 0;
+	}
+
+	while (total < (size_t) where->size) {
+		do
+			n = should_write ?
+			    write(fd, buffer, (size_t) where->size - total) :
+			    read(fd, buffer, (size_t) where->size - total);
+		while ((n < 0) && ((errno == EINTR) || (errno == EAGAIN)));
+
+		if (n < 0)
+			log_error("%s: %s failed after %" PRIu64 " of %" PRIu64
+				  " at %" PRIu64 ": %s", dev_name(where->dev),
+				  should_write ? "write" : "read",
+				  (uint64_t) total,
+				  (uint64_t) where->size,
+				  (uint64_t) where->start, strerror(errno));
+
+		if (n <= 0)
+			break;
+
+		total += n;
+		buffer += n;
+	}
+
+	return (total == (size_t) where->size);
+}
+
+/*-----------------------------------------------------------------
+ * LVM2 uses O_DIRECT when performing metadata io, which requires
+ * block size aligned accesses.  If any io is not aligned we have
+ * to perform the io via a bounce buffer, obviously this is quite
+ * inefficient.
+ *---------------------------------------------------------------*/
+
+/*
+ * Get the sector size from an _open_ device.
+ */
+static int _get_block_size(struct device *dev, unsigned int *size)
+{
+	const char *name = dev_name(dev);
+
+	if ((dev->block_size == -1)) {
+		if (ioctl(dev_fd(dev), BLKBSZGET, &dev->block_size) < 0) {
+			log_sys_error("ioctl BLKBSZGET", name);
+			return 0;
+		}
+		log_debug("%s: block size is %u bytes", name, dev->block_size);
+	}
+
+	*size = (unsigned int) dev->block_size;
+
+	return 1;
+}
+
+/*
+ * Widens a region to be an aligned region.
+ */
+static void _widen_region(unsigned int block_size, struct device_area *region,
+			  struct device_area *result)
+{
+	uint64_t mask = block_size - 1, delta;
+	memcpy(result, region, sizeof(*result));
+
+	/* adjust the start */
+	delta = result->start & mask;
+	if (delta) {
+		result->start -= delta;
+		result->size += delta;
+	}
+
+	/* adjust the end */
+	delta = (result->start + result->size) & mask;
+	if (delta)
+		result->size += block_size - delta;
+}
+
+static int _aligned_io(struct device_area *where, void *buffer,
+		       int should_write)
+{
+	void *bounce;
+	unsigned int block_size = 0;
+	uintptr_t mask;
+	struct device_area widened;
+
+	if (!(where->dev->flags & DEV_REGULAR) &&
+	    !_get_block_size(where->dev, &block_size))
+		return_0;
+
+	if (!block_size)
+		block_size = lvm_getpagesize();
+
+	_widen_region(block_size, where, &widened);
+
+	/* Do we need to use a bounce buffer? */
+	mask = block_size - 1;
+	if (!memcmp(where, &widened, sizeof(widened)) &&
+	    !((uintptr_t) buffer & mask))
+		return _io(where, buffer, should_write);
+
+	/* Allocate a bounce buffer with an extra block */
+	if (!(bounce = alloca((size_t) widened.size + block_size))) {
+		log_error("Bounce buffer alloca failed");
+		return 0;
+	}
+
+	/*
+	 * Realign start of bounce buffer (using the extra sector)
+	 */
+	if (((uintptr_t) bounce) & mask)
+		bounce = (void *) ((((uintptr_t) bounce) + mask) & ~mask);
+
+	/* channel the io through the bounce buffer */
+	if (!_io(&widened, bounce, 0)) {
+		if (!should_write)
+			return_0;
+		/* FIXME pre-extend the file */
+		memset(bounce, '\n', widened.size);
+	}
+
+	if (should_write) {
+		memcpy(bounce + (where->start - widened.start), buffer,
+		       (size_t) where->size);
+
+		/* ... then we write */
+		return _io(&widened, bounce, 1);
+	}
+
+	memcpy(buffer, bounce + (where->start - widened.start),
+	       (size_t) where->size);
+
+	return 1;
+}
+
+static int _dev_get_size_file(const struct device *dev, uint64_t *size)
+{
+	const char *name = dev_name(dev);
+	struct stat info;
+
+	if (stat(name, &info)) {
+		log_sys_error("stat", name);
+		return 0;
+	}
+
+	*size = info.st_size;
+	*size >>= SECTOR_SHIFT;	/* Convert to sectors */
+
+	log_very_verbose("%s: size is %" PRIu64 " sectors", name, *size);
+
+	return 1;
+}
+
+static int _dev_get_size_dev(const struct device *dev, uint64_t *size)
+{
+	int fd;
+	const char *name = dev_name(dev);
+
+	if ((fd = open(name, O_RDONLY)) < 0) {
+		log_sys_error("open", name);
+		return 0;
+	}
+
+	if (ioctl(fd, BLKGETSIZE64, size) < 0) {
+		log_sys_error("ioctl BLKGETSIZE64", name);
+		if (close(fd))
+			log_sys_error("close", name);
+		return 0;
+	}
+
+	*size >>= BLKSIZE_SHIFT;	/* Convert to sectors */
+	if (close(fd))
+		log_sys_error("close", name);
+
+	log_very_verbose("%s: size is %" PRIu64 " sectors", name, *size);
+
+	return 1;
+}
+
+static int _dev_read_ahead_dev(struct device *dev, uint32_t *read_ahead)
+{
+	long read_ahead_long;
+
+	if (dev->read_ahead != -1) {
+		*read_ahead = (uint32_t) dev->read_ahead;
+		return 1;
+	}
+
+	if (!dev_open(dev))
+		return_0;
+
+	if (ioctl(dev->fd, BLKRAGET, &read_ahead_long) < 0) {
+		log_sys_error("ioctl BLKRAGET", dev_name(dev));
+		if (!dev_close(dev))
+			stack;
+		return 0;
+	}
+
+	if (!dev_close(dev))
+		stack;
+
+	*read_ahead = (uint32_t) read_ahead_long;
+	dev->read_ahead = read_ahead_long;
+
+	log_very_verbose("%s: read_ahead is %u sectors",
+			 dev_name(dev), *read_ahead);
+
+	return 1;
+}
+
+/*-----------------------------------------------------------------
+ * Public functions
+ *---------------------------------------------------------------*/
+
+int dev_get_size(const struct device *dev, uint64_t *size)
+{
+	if (!dev)
+		return 0;
+
+	if ((dev->flags & DEV_REGULAR))
+		return _dev_get_size_file(dev, size);
+	else
+		return _dev_get_size_dev(dev, size);
+}
+
+int dev_get_read_ahead(struct device *dev, uint32_t *read_ahead)
+{
+	if (!dev)
+		return 0;
+
+	if (dev->flags & DEV_REGULAR) {
+		*read_ahead = 0;
+		return 1;
+	}
+
+	return _dev_read_ahead_dev(dev, read_ahead);
+}
+
+/* FIXME Unused
+int dev_get_sectsize(struct device *dev, uint32_t *size)
+{
+	int fd;
+	int s;
+	const char *name = dev_name(dev);
+
+	if ((fd = open(name, O_RDONLY)) < 0) {
+		log_sys_error("open", name);
+		return 0;
+	}
+
+	if (ioctl(fd, BLKSSZGET, &s) < 0) {
+		log_sys_error("ioctl BLKSSZGET", name);
+		if (close(fd))
+			log_sys_error("close", name);
+		return 0;
+	}
+
+	if (close(fd))
+		log_sys_error("close", name);
+
+	*size = (uint32_t) s;
+
+	log_very_verbose("%s: sector size is %" PRIu32 " bytes", name, *size);
+
+	return 1;
+}
+*/
+
+void dev_flush(struct device *dev)
+{
+	if (!(dev->flags & DEV_REGULAR) && ioctl(dev->fd, BLKFLSBUF, 0) >= 0)
+		return;
+
+	if (fsync(dev->fd) >= 0)
+		return;
+
+	sync();
+}
+
+int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
+{
+	struct stat buf;
+	const char *name;
+	int need_excl = 0, need_rw = 0;
+
+	if ((flags & O_ACCMODE) == O_RDWR)
+		need_rw = 1;
+
+	if ((flags & O_EXCL))
+		need_excl = 1;
+
+	if (dev->fd >= 0) {
+		if (((dev->flags & DEV_OPENED_RW) || !need_rw) &&
+		    ((dev->flags & DEV_OPENED_EXCL) || !need_excl)) {
+			dev->open_count++;
+			return 1;
+		}
+
+		if (dev->open_count && !need_excl) {
+			/* FIXME Ensure we never get here */
+			log_debug("WARNING: %s already opened read-only",
+				  dev_name(dev));
+			dev->open_count++;
+		}
+
+		dev_close_immediate(dev);
+	}
+
+	if (memlock())
+		log_error("WARNING: dev_open(%s) called while suspended",
+			  dev_name(dev));
+
+	if (dev->flags & DEV_REGULAR)
+		name = dev_name(dev);
+	else if (!(name = dev_name_confirmed(dev, quiet)))
+		return_0;
+
+	if (!(dev->flags & DEV_REGULAR)) {
+		if (stat(name, &buf) < 0) {
+			log_sys_error("%s: stat failed", name);
+			return 0;
+		}
+		if (buf.st_rdev != dev->dev) {
+			log_error("%s: device changed", name);
+			return 0;
+		}
+	}
+
+#ifdef O_DIRECT_SUPPORT
+	if (direct) {
+		if (!(dev->flags & DEV_O_DIRECT_TESTED))
+			dev->flags |= DEV_O_DIRECT;
+
+		if ((dev->flags & DEV_O_DIRECT))
+			flags |= O_DIRECT;
+	}
+#endif
+
+#ifdef O_NOATIME
+	/* Don't update atime on device inodes */
+	if (!(dev->flags & DEV_REGULAR))
+		flags |= O_NOATIME;
+#endif
+
+	if ((dev->fd = open(name, flags, 0777)) < 0) {
+#ifdef O_DIRECT_SUPPORT
+		if (direct && !(dev->flags & DEV_O_DIRECT_TESTED)) {
+			flags &= ~O_DIRECT;
+			if ((dev->fd = open(name, flags, 0777)) >= 0) {
+				dev->flags &= ~DEV_O_DIRECT;
+				log_debug("%s: Not using O_DIRECT", name);
+				goto opened;
+			}
+		}
+#endif
+		if (quiet)
+			log_sys_debug("open", name);
+		else
+			log_sys_error("open", name);
+		return 0;
+	}
+
+#ifdef O_DIRECT_SUPPORT
+      opened:
+	if (direct)
+		dev->flags |= DEV_O_DIRECT_TESTED;
+#endif
+	dev->open_count++;
+	dev->flags &= ~DEV_ACCESSED_W;
+
+	if (need_rw)
+		dev->flags |= DEV_OPENED_RW;
+	else
+		dev->flags &= ~DEV_OPENED_RW;
+
+	if (need_excl)
+		dev->flags |= DEV_OPENED_EXCL;
+	else
+		dev->flags &= ~DEV_OPENED_EXCL;
+
+	if (!(dev->flags & DEV_REGULAR) &&
+	    ((fstat(dev->fd, &buf) < 0) || (buf.st_rdev != dev->dev))) {
+		log_error("%s: fstat failed: Has device name changed?", name);
+		dev_close_immediate(dev);
+		return 0;
+	}
+
+#ifndef O_DIRECT_SUPPORT
+	if (!(dev->flags & DEV_REGULAR))
+		dev_flush(dev);
+#endif
+
+	if ((flags & O_CREAT) && !(flags & O_TRUNC))
+		dev->end = lseek(dev->fd, (off_t) 0, SEEK_END);
+
+	dm_list_add(&_open_devices, &dev->open_list);
+
+	log_debug("Opened %s %s%s%s", dev_name(dev),
+		  dev->flags & DEV_OPENED_RW ? "RW" : "RO",
+		  dev->flags & DEV_OPENED_EXCL ? " O_EXCL" : "",
+		  dev->flags & DEV_O_DIRECT ? " O_DIRECT" : "");
+
+	return 1;
+}
+
+int dev_open_quiet(struct device *dev)
+{
+	int flags;
+
+	flags = vg_write_lock_held() ? O_RDWR : O_RDONLY;
+
+	return dev_open_flags(dev, flags, 1, 1);
+}
+
+int dev_open(struct device *dev)
+{
+	int flags;
+
+	flags = vg_write_lock_held() ? O_RDWR : O_RDONLY;
+
+	return dev_open_flags(dev, flags, 1, 0);
+}
+
+int dev_test_excl(struct device *dev)
+{
+	int flags;
+	int r;
+
+	flags = vg_write_lock_held() ? O_RDWR : O_RDONLY;
+	flags |= O_EXCL;
+
+	r = dev_open_flags(dev, flags, 1, 1);
+	if (r)
+		dev_close_immediate(dev);
+
+	return r;
+}
+
+static void _close(struct device *dev)
+{
+	if (close(dev->fd))
+		log_sys_error("close", dev_name(dev));
+	dev->fd = -1;
+	dev->block_size = -1;
+	dm_list_del(&dev->open_list);
+
+	log_debug("Closed %s", dev_name(dev));
+
+	if (dev->flags & DEV_ALLOCED) {
+		dm_free((void *) dm_list_item(dev->aliases.n, struct str_list)->
+			 str);
+		dm_free(dev->aliases.n);
+		dm_free(dev);
+	}
+}
+
+static int _dev_close(struct device *dev, int immediate)
+{
+	struct lvmcache_info *info;
+
+	if (dev->fd < 0) {
+		log_error("Attempt to close device '%s' "
+			  "which is not open.", dev_name(dev));
+		return 0;
+	}
+
+#ifndef O_DIRECT_SUPPORT
+	if (dev->flags & DEV_ACCESSED_W)
+		dev_flush(dev);
+#endif
+
+	if (dev->open_count > 0)
+		dev->open_count--;
+
+	if (immediate && dev->open_count)
+		log_debug("%s: Immediate close attempt while still referenced",
+			  dev_name(dev));
+
+	/* Close unless device is known to belong to a locked VG */
+	if (immediate ||
+	    (dev->open_count < 1 &&
+	     (!(info = info_from_pvid(dev->pvid, 0)) ||
+	      !info->vginfo ||
+	      !vgname_is_locked(info->vginfo->vgname))))
+		_close(dev);
+
+	return 1;
+}
+
+int dev_close(struct device *dev)
+{
+	return _dev_close(dev, 0);
+}
+
+int dev_close_immediate(struct device *dev)
+{
+	return _dev_close(dev, 1);
+}
+
+void dev_close_all(void)
+{
+	struct dm_list *doh, *doht;
+	struct device *dev;
+
+	dm_list_iterate_safe(doh, doht, &_open_devices) {
+		dev = dm_list_struct_base(doh, struct device, open_list);
+		if (dev->open_count < 1)
+			_close(dev);
+	}
+}
+
+int dev_read(struct device *dev, uint64_t offset, size_t len, void *buffer)
+{
+	struct device_area where;
+
+	if (!dev->open_count)
+		return_0;
+
+	where.dev = dev;
+	where.start = offset;
+	where.size = len;
+
+	return _aligned_io(&where, buffer, 0);
+}
+
+/*
+ * Read from 'dev' into 'buf', possibly in 2 distinct regions, denoted
+ * by (offset,len) and (offset2,len2).  Thus, the total size of
+ * 'buf' should be len+len2.
+ */
+int dev_read_circular(struct device *dev, uint64_t offset, size_t len,
+		      uint64_t offset2, size_t len2, void *buf)
+{
+	if (!dev_read(dev, offset, len, buf)) {
+		log_error("Read from %s failed", dev_name(dev));
+		return 0;
+	}
+
+	/*
+	 * The second region is optional, and allows for
+	 * a circular buffer on the device.
+	 */
+	if (!len2)
+		return 1;
+
+	if (!dev_read(dev, offset2, len2, buf + len)) {
+		log_error("Circular read from %s failed",
+			  dev_name(dev));
+		return 0;
+	}
+
+	return 1;
+}
+
+/* FIXME If O_DIRECT can't extend file, dev_extend first; dev_truncate after.
+ *       But fails if concurrent processes writing
+ */
+
+/* FIXME pre-extend the file */
+int dev_append(struct device *dev, size_t len, void *buffer)
+{
+	int r;
+
+	if (!dev->open_count)
+		return_0;
+
+	r = dev_write(dev, dev->end, len, buffer);
+	dev->end += (uint64_t) len;
+
+#ifndef O_DIRECT_SUPPORT
+	dev_flush(dev);
+#endif
+	return r;
+}
+
+int dev_write(struct device *dev, uint64_t offset, size_t len, void *buffer)
+{
+	struct device_area where;
+
+	if (!dev->open_count)
+		return_0;
+
+	where.dev = dev;
+	where.start = offset;
+	where.size = len;
+
+	dev->flags |= DEV_ACCESSED_W;
+
+	return _aligned_io(&where, buffer, 1);
+}
+
+int dev_set(struct device *dev, uint64_t offset, size_t len, int value)
+{
+	size_t s;
+	char buffer[4096] __attribute((aligned(8)));
+
+	if (!dev_open(dev))
+		return_0;
+
+	if ((offset % SECTOR_SIZE) || (len % SECTOR_SIZE))
+		log_debug("Wiping %s at %" PRIu64 " length %" PRIsize_t,
+			  dev_name(dev), offset, len);
+	else
+		log_debug("Wiping %s at sector %" PRIu64 " length %" PRIsize_t
+			  " sectors", dev_name(dev), offset >> SECTOR_SHIFT,
+			  len >> SECTOR_SHIFT);
+
+	memset(buffer, value, sizeof(buffer));
+	while (1) {
+		s = len > sizeof(buffer) ? sizeof(buffer) : len;
+		if (!dev_write(dev, offset, s, buffer))
+			break;
+
+		len -= s;
+		if (!len)
+			break;
+
+		offset += s;
+	}
+
+	dev->flags |= DEV_ACCESSED_W;
+
+	if (!dev_close(dev))
+		stack;
+
+	return (len == 0);
+}