IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
See this test: http://samba.org/~tridge/junkcode/aio_uid.c
The problem is that setresuid() tries to be clever about threads, and
tries to change the euid of any threads that are running. If a AIO read
or write completes while this is going on then the signal from the thread
where the IO completed is lost, as it gets -1/EPERM from rt_sigqueueinfo()
The simplest fix is to try to use setreuid() instead of setresuid(),
as setreuid() doesn't try to be clever. Unfortunately this also means
we must use become_root()/unbecome_root() in the aio code.
(This used to be commit 56c5a6f024)
When building on linux, default to building cifs.upcall. Throw a
warning if ADS support is disabled or keyutils isn't installed.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
(This used to be commit d8018d15f0)
This reverts commit fc9b30bed2.
Sorry, this got pushed by accident:
"This can not go upstream yet because it uses the non-GPL libgpfs."
Michael
(This used to be commit 26a3cf0be9)
This can not go upstream yet because it uses the non-GPL libgpfs. So it will
not be compiled by default and will not be included in the SOFS RPMs. But upon
Sven's request, we include it in the git tree and the source RPMs, so that it
can be built for in-house tests.
(This used to be commit fc9b30bed2)
Add a configure test for the availability of f_frsize in struct statvfs
(for broken platforms that define statvfs but still have
f_bsize/f_iosize). Also add sys/types.h to the other struct statvfs test
(again, some platforms need it).
(This used to be commit 591bf1d15f)
This adds an --enable-picky-developer option that will halt compilation
on warnings. Yes, this could be handled by a direct Makefile change, but
people should be encourage to do it!
(This used to be commit 10a2ab4077)
binary. The biggest change is that it renames it from cifs.spnego
to cifs.upcall since the cifs.spnego name really isn't applicable
anymore.
It also fixes a segfault when the program is run without any args
and adds a manpage. Comments and/or suggestions appreciated.
This set should apply cleanly to the 3.3 test branch.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Jeremy.
(This used to be commit c633f10d9e)
This sets the library versions (previously gotten from library-versions
file during make) already at configure time.
For library "foo", the version is stored in the LIBFOO_SOVER configure
variable. For all libraries except libwbclient, this is done through the
call of SMB_LIBRARY().
NOTE: Currently the versions are hard-coded in the SMB_LIBRARY() calls in
configure.in, but this change makes it easier to retrieve the version from
the libraries source dir at configure time, e.g. for talloc and tdb.
Michael
(This used to be commit 3c300f9f8c)
This test fails if GCC emits any warnings (presumably to detect the
function propertly), but unless we include this message then free()
fail. Why we need to call free in a configure test is probably
something to blame on valgrind...
Andrew Bartlett
(This used to be commit d013f6fadc)
Remaining problem is that not all *.syms linker scripts are generated
so the LD command needs to look in a single place. Out of tree
builds now work but report write access to the source tree in order
to create the $(srcdir)/expotrs/libtdb.sysms file.
(This used to be commit 6e9287c5c3)
This is used for two purposes:
First, I'm using it to test the async I/O code. In the forked process it is
pretty easy to delay a reply for a random amount of time. See the
BUILD_FARM_HACKS snippet.
Second, there are systems around that claim to have Posix AIO but which is
broken. This might be some help for those systems.
Also add tests how to pass file descriptors
(This used to be commit b0e52cecf2)
Use -bexpfull instead of -bexpall for LDSHFLAGS to also export
symbols starting with "_". This exports all necessary symbols
from libtalloc.so.
Michael
(This used to be commit 213ba3a719)
To disable internal use of shared libraries altogether (as opposed to
disabling use of single shared libs by --with-static-libs=LIBS), use
this new configure parameter --disable-shared-libs.
Michael
(This used to be commit 63bff18f3f)
I have to investigate the effect of these and evaluate the need of
their presence. For now establish more reasonable grouping.
Michael
(This used to be commit 3ba8fa1c4b)
This should work with m4_include(lib/tdb/libtdb.m4) like already
done with talloc, but currently this does not work (yet) - tdb
code has to be merged from s4 before that.
Michael
(This used to be commit e92fdf6b20)
Golly gosh, this one took me quite some time to figure out!
Hopefully, I did not introduce any typos copying...
Michael
(This used to be commit 85ca2c9edb)
Formerly this was only needed for libwbclient, but now that we start
using more shared libs internally, this is needed more globally
to support linking internal libs dynamically on systemy without winbindd.
Michael
(This used to be commit ec79357218)
This also establishes a general configure mechanism to control static vs
dynamic linking of internal subsystems built as libraries:
This first simple approach is as follows.
* It applies only to "subsystems" that we build as libraries and for
that linking samba against the libraries (as opposed to linking in
the plain object files) has been configured in Makefile.in.
* If we do build the shared library, then we link dynamically by default.
* We only link statically if we don't build shared or if the library
appears in the new --with-static-libs configure option
(comma-separated list).
Example (currently only one):
--with-static-libs=talloc makes use of libtalloc.a instead
of linking the dynamic variant with -ltalloc.
A possilble way to setup linking against libraries in Makefile.in is this:
For a subsystem, "mylib" say, we build bin/libmylib.a and bin/libmylib.so.
The subsystem usually has a MYLIB_OBJ definition in Makefile.in. Define
LIBMYLIB_STATIC=bin/libmylib.a and and LIBMYLIB_LIBS=-lmylib in configure.in
as controlled by presence of "mylib" in the list given to --with-static-libs
and change uses of $(MYLIB_OBJ) to @LIBMYLIB_STATIC@ in Makefile.in and
add @LIBMYLIB_LIBS@ to the link targets as needed.
In the example of talloc, which is needed everywhere, I have simply
added @LIBTALLOC_LIBS@ to the definition of "LIBS" in Makefile.in.
For other subsystems, one will have to be more careful.
Michael
(This used to be commit 71b990d9d6)
Use the resulting HAVE_LBER_LOG_PRINT_FN to determine whether we can
use it in init_ldap_debugging to intercept LDAP debug output and print
it out in the samba logs (controlled with "ldap debug level").
Michael
(This used to be commit 8dfc389ff6)
-bexpall only exports the names without leading _, so we don't get
_talloc_reallo_array for example.
Thanks a lot to Heinrich Mislik for the hint!
(This used to be commit 74a71f1988)
Store streams in posix xattrs. A kludge, as xattrs are limited in many ways,
but it might be a help for some situations.
(This used to be commit 83a805220e)
This reverts commit e33286f4a6.
393bab185e0e9f02aac5740becc2aba2762133c8 is the v4-0-test fix
for the same problem.
metze
(This used to be commit b8a57a24c3)
This is an option for file systems that do not implement xattrs: in
lockdir/eas.tdb an array of xatts per inode is stored.
It can not solve the problem that xattrs might reappear if a posix-level
process deletes a file and happens to re-create it under the same name. On file
systems with birthtime we might have a chance to detect this, but not with
standard posix. A future version might put relief on file systems that do have
xattrs but where these are severely limited in size/speed/whatever: We can put
a simple marker as a native xattr, but the xattrs proper are stored in the tdb.
Volker
(This used to be commit 2036b4c5ad)
This patch adds the configure option --enable-libwbclient-shared that defaults
to yes. If you set --enable-developer=yes, then you can say
--enable-libwbclient-shared=no to avoid problems you might have with the wrong
shared libaries.
Jerry, is this acceptable to you? If yes, please push.
Thanks!
Volker
(This used to be commit 7a88cd61e0)
(based on the winbind_struct_protocol.h).
The API in incomplete, but sufficient to merge. See wbclienbt.h for the i
interface functions.
(This used to be commit 83d274b460)
* Added checks for spnego prereq keyutils.h and kerberos in configure.in
Signed-off-by: Igor Mammedov <niallain@gmail.com>
(This used to be commit f802db70b8)
Add "commit on eof" and "commit past eof" modes. Patch from David
Losada Carballo with some modifications during the merge.
(This used to be commit 12eee4f7fe)
The /usr/include/sys/capability.h defines _LINUX_TYPES_H
which prevents /usr/include/linux/types.h from being parsed
(when included afterwards). Thus certain types are undefined
that are for instance needed in /usr/include/linux/dqblk_xfs.h.
This breaks the build of lib/sysquotas_xfs.c.
This commit adds a configure check and a workaround for this.
Michael
(This used to be commit 11bcdf780e)
