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.
With the removal of the call to flock LOCK_MAND, the only remaining use
of this VFS path is to register sharemodes with specific file systems.
Rename the VFS call to reflect that this is no longer related to flock.
Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The function kernel_flock will be deleted, drop the reference to it.
Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14700
RN: File owner not available when file unreadable
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Mon Aug 2 18:05:04 UTC 2021 on sn-devel-184
This was needed before we had pathref fsps, with pathref fsps we can do
operation requiring WRITE_OWNER_ACCESS, WRITE_DAC_ACCESS and READ_CONTROL_ACCESS
on the pathref fsp.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14700
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This helps code inside any module implementing fstat() looking at
fsp->fsp_name->st instead of the passed in stat buf.
I only ran afoul of this in a DEBUG message I added while debugging some inode
related problem.
No change in behaviour.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14756
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Currently in master when we call into openat() in the VFS
we violate the invarient:
fsp->base_fsp->fsp_name->fsp == fsp->base_fsp.
The reason for this is subtle. Inside open.c:non_widelink_open()
we change the fsp->base_fsp to be relative to the new $cwd.
We do this by the following code in open.c:non_widelink_open():
/* Also setup base_fsp to be relative to the new cwd */
if (fsp->base_fsp != NULL) {
base_smb_fname_rel = (struct smb_filename) {
.base_name = smb_fname_rel->base_name,
};
orig_base_fsp_name = fsp->base_fsp->fsp_name;
fsp->base_fsp->fsp_name = &base_smb_fname_rel;
}
Note that fsp->base_fsp->fsp_name now points at a
stack variable struct smb_filename, with smb_fname->fsp == NULL.
This fixes that problem by removing the horrid
stack based smb_filename and changing to use a
talloc'ed fsp->base_fsp->fsp_name, with
correctly linked fsp->base_fsp->fsp_name-> pointer.
Remove the selftest/knownfail.d/fruit_vfs_invariant
file as all vfs_fruit tests now pass again.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Save several system calls if we're operating at the root of the share.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Fri Jun 18 17:21:31 UTC 2021 on sn-devel-184
Optimization now becomes clear. We already have a valid stat of the parent
directory so we don't need to re-do a system call.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Optimization now becomes clear. We already have a valid stat of the parent
directory so we don't need to re-do a system call.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Caller is still using this !
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14736
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Noel Power<npower@samba.org>
Autobuild-User(master): Noel Power <npower@samba.org>
Autobuild-Date(master): Fri Jun 11 10:17:46 UTC 2021 on sn-devel-184
This is an inconsequential cosmetic change, it just caught my eye
as looking a bit out of place compared to the surrounding code style.
Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Pass down to smbd_check_access_rights_sd().
Always pass conn->cwd_fsp for now.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Pass down to parent_override_delete().
Always pass fsp->conn->cwd_fsp for now.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Pass down to can_delete_file_in_directory().
Always pass conn->cwd_fsp for now.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Pass this down into smbd_calculate_maximum_allowed_access_fsp().
Currently pass fsp->conn->cwd_fsp everywhere.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Pass the same fsp->conn->cwd_fsp parameter to can_delete_file_in_directory()
dirfsp for now.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Comment out smbd_calculate_maximum_allowed_access() as it is
a static function and we just removed the only caller.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Not yet used but this now uses smbd_calculate_maximum_allowed_access_fsp(),
so uncomment it.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Commented out as not yet used.
Signed-off-by: Noel Power <noel.power@suse.com>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
open_file() can now use parent_dir_fname->fsp for handle-based
access calls.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This will allow us to change directory_has_default_acl() to
directory_has_default_acl_fsp() later.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
atname is a better name, as base_name refers to the name without streams
and this is nothing to do with streams.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We don't need the parent_pathref() and more, and as we don't
own parent_dir_fname anymore, don't free it on exit.
Next step will be to remove the local variables.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Not yet used. Next step will be to use them to replace the
internal parent_dir_fname,base_name variables inside mkdir_internal().
Annotate them so we know what these extra params are.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Not yet used.
We will be passing this down to open_directory() and
open_file_ntcreate() and using it within create_file_unixpath()
as all of these functions need a parent pathref to check parent
ACLs etc.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
For the pathname verison of this function smbd_check_access_rights()
we return the st_mode bits turned into an NT ACL for a symlink.
For a symlink the mode bits are always 'lrwxrwxrwx' which means
smbd_check_access_rights() version always returned NT_STATUS_OK
for any access rights requested on a symlink.
For smbd_check_access_rights_fsp() to a symlink use the handle
access bits as this is a better representation of the access
allowed.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
smbd_check_access_rights() is checking permissions on a file *before*
open, so getting ACCESS_DENIED and mapping to NT_STATUS_OK when
reading the security descriptor is fine, as if we really don't have
access the open will fail.
smbd_check_access_rights_fsp() takes place *after* the open (pathref
or otherwise), so being unable to get the security decriptor should
be reported back to the caller and not mapped to NT_STATUS_OK.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This doesn't matter now, but later we will move to a handle-based call to
check access rights, so we will need the full handle setup.
Add a fd_close(fsp) in the error path now this is done after the fd open.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Now we do an early check for opening a directory, this code
could never have been triggered.
The only case we need to consider now is when the filesystem
object exists as a file when we enter open_file() and another
smbd removes and then re-creates the object as a directory before we
call reopen_from_fsp(). In that case, we will open the object,
and come back out to open_file_ntcreate(), where the race
condition detection code that calls check_same_dev_ino()
will catch the case and error out the client request.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Makes sure we keep the pathref fd open for the NT_FILE_IS_A_DIRECTORY case.
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Ralph Boehme <slow@samba.org>
This is really subtle. If someone passes in an smb_fname where smb_fname
actually is taken from fsp->fsp_name, then the lifetime of these objects is
meant to be the same.
This is commonly the case from an SMB1 path-based call
(eg call_trans2qfilepathinfo()) where we use the pathref fsp
(smb_fname->fsp) as the handle. In this case we must not unlink smb_fname->fsp
from it's owner.
The asserts below:
SMB_ASSERT(fsp->fsp_name->fsp != NULL);
SMB_ASSERT(fsp->fsp_name->fsp == fsp);
ensure the required invarients are met.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14732
Pair-Programmed-With: Ralph Boehme <slow@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Tue Jun 8 20:44:41 UTC 2021 on sn-devel-184
Valgrind trace follows.
==3627798== Invalid read of size 1
==3627798== at 0x483FF46: strlen (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==3627798== by 0x55DE412: strdup (strdup.c:41)
==3627798== by 0x4F4657E: smb_xstrdup (util.c:660)
==3627798== by 0x4C62C2E: vfs_ChDir (vfs.c:988)
==3627798== by 0x4C4A51C: process_symlink_open (open.c:656)
==3627798== by 0x4C4ADE7: non_widelink_open (open.c:862)
==3627798== by 0x4C4AFB7: fd_openat (open.c:918)
==3627798== by 0x4BBE895: openat_pathref_fsp (files.c:506)
==3627798== by 0x4C48A00: filename_convert_internal (filename.c:2027)
==3627798== by 0x4C48B77: filename_convert (filename.c:2067)
==3627798== by 0x4C32408: call_trans2qfilepathinfo (trans2.c:6173)
==3627798== by 0x4C3C5DA: handle_trans2 (trans2.c:10143)
==3627798== Address 0xda8bc90 is 96 bytes inside a block of size 217 free'd
==3627798== at 0x483DA3F: free (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==3627798== by 0x4FCA3C9: _tc_free_internal (talloc.c:1222)
==3627798== by 0x4FCA481: _talloc_free_internal (talloc.c:1248)
==3627798== by 0x4FCB825: _talloc_free (talloc.c:1792)
==3627798== by 0xDB248DD: store_cwd_data (vfs_shadow_copy2.c:1473)
==3627798== by 0xDB24BEF: shadow_copy2_chdir (vfs_shadow_copy2.c:1542)
==3627798== by 0x4C662A4: smb_vfs_call_chdir (vfs.c:2257)
==3627798== by 0x4C62B48: vfs_ChDir (vfs.c:940)
==3627798== by 0x4C4A51C: process_symlink_open (open.c:656)
==3627798== by 0x4C4ADE7: non_widelink_open (open.c:862)
==3627798== by 0x4C4AFB7: fd_openat (open.c:918)
==3627798== by 0x4BBE895: openat_pathref_fsp (files.c:506)
==3627798== Block was alloc'd at
==3627798== at 0x483C7F3: malloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==3627798== by 0x4FC9365: __talloc_with_prefix (talloc.c:783)
==3627798== by 0x4FC94FF: __talloc (talloc.c:825)
==3627798== by 0x4FCCFDC: __talloc_strlendup (talloc.c:2454)
==3627798== by 0x4FCD096: talloc_strdup (talloc.c:2470)
==3627798== by 0xDB24977: store_cwd_data (vfs_shadow_copy2.c:1476)
==3627798== by 0xDB24BEF: shadow_copy2_chdir (vfs_shadow_copy2.c:1542)
==3627798== by 0x4C662A4: smb_vfs_call_chdir (vfs.c:2257)
==3627798== by 0x4C62B48: vfs_ChDir (vfs.c:940)
==3627798== by 0x4C4A92D: non_widelink_open (open.c:755)
==3627798== by 0x4C4AFB7: fd_openat (open.c:918)
==3627798== by 0x4BBE895: openat_pathref_fsp (files.c:506)
==3627798==
Even though SMB_VFS_CONNECTPATH() returns a const char,
vfs_shadow_copy2() can free and reallocate this whilst
in use inside process_symlink_open().
Take a copy to make sure we don't reference free'd memory.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14721
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu May 27 17:25:43 UTC 2021 on sn-devel-184
We just freed subreq and a pointer to subreq is stored in open_rec->watch_req,
so we must invalidate the pointer.
Otherwise if the poll open timer fires it will do a
TALLOC_FREE(open_rec->watch_req);
on the dangling pointer which may crash or do something worse like freeing some
other random talloc memory.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14672
CI: https://gitlab.com/samba-team/samba/-/merge_requests/1843
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
the smb_filename/smb_filename->fsp passed to vfs_streaminfo
eventually is passed to SMB_VFS_FLISTXATTR, we need to ensure this is
properly setup and not NULL
Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Mon Feb 1 18:38:23 UTC 2021 on sn-devel-184
This ensures smb_fname_base knows if it's a file or directory.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We can just call close_file(req, fsp, ERROR_CLOSE), as it handles
non-fsa fsp's and base_fsp's just fine.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Thu Jan 14 12:40:56 UTC 2021 on sn-devel-184
This allows us to add some more logic for bi-directional linking between
base and stream fsp in the next commits.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This has already been set a few lines above. The duplicate was the result of
restructuring create_file_unixpath() a few months ago, allocating fsp in
create_file_unixpath() instead of in the callees open_file_ntcreate() or
open_directory() respectively.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
In future we may move the reopen logic to the VFS,
but for now we just keep it in one place.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Both callers pass in a helper variable that points to
fsp->fsp_name and it was only used for a debug message,
so we can simply use fsp_str_dgb() instead.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Calling fd_open_atomic() without O_CREAT is the same as calling
fd_openat() directly, so we can also use it to open an existing
directory.
In the next step we'll move the reopen_from_procfd() fallback logic to
a single helper function.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This was only done if fsp->fsp_name already existed, but not the first time.
This also makes sure we modify fsp->fsp_name and fsp->name_hash only on success.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Now that all VFS modules support real dirfsps in SMB_VFS_MKDIRAT(), pass the
pathref fsp from the parent directory and the basename of the new directory.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
After the directory has been created by SMB_VFS_MKDIRAT(), open the fsp on the
new directory as pathref fsp so we can use handle based VFS functions.
open_directory() will reopen the fsp as a full fsp, but that doesn't really hurt
thanks to the reopen_from_procfd() optimisation.
Note that smb_dname == fsp->fsp_name.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Prepares for calling SMB_VFS_MKDIRAT() below with a real dirfsp/atname. As
parent_dir_fname now has a pathref fsp in parent_dir_fname->fsp, make sure to
talloc_free() the parent_dir_fname before leaving the function, so the pathref
fsp is closed right there and not left around until the talloc tos is
destroyed (parent_dir_fname is a child of talloc-tos).
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is the correct place where this code should be. It also means opaque VFS
modules that implement their own mkdirat() like glusterfs now use this logic.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
fd_close() was accidentally added twice by
d00d09fdcf. If it would have been removed by
5770cdd763 that adds the reopen_from_procfd()
optimisation, all would be fine. But fd_close() still being called before
reopen_from_procfd() means we're closing the pathref fd prevening the
optimisation.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We can't call directly into close_file(), as that cannot deal
with regular file and directory opens where fsp->fsp_flags.is_fsa
hasn't been set to true (uncompleted opens).
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We know we must have a valid fsp.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Wed Dec 16 10:15:11 UTC 2020 on sn-devel-184
This replaces the code in a bunch of places where we choose between stat() and
lstat() based on req->posix_pathname. The new code inside vfs_stat() is based on
checking the smb_fname flag SMB_FILENAME_POSIX_PATH.
req->posix_pathname is inherited from the global POSIX pathnames state and the
smb_fname flags is also inherited from that indirectly via the UCF flags.
Tl;dr: no change in behaviour. :)
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is an optimisation that avoids going through the expensive
non_widelink_open() logic a second time. It depends on a usable /proc/%d/fd/%d
filesystem and this is checked and set as "can_reopen" flag by the VFS in the
openat() function in the fsp.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is the big bang for the internal pathref fsps: up to this point the pathref
fsps were lingering around unused inside smb_fname->fsp.
With this change, the internal fsp will be the one that is going to be returned
from SMB_VFS_CREATE_FILE() if the client requested access mask matches the
criteria in open_file():
uint32_t need_fd_mask =
FILE_READ_DATA |
FILE_WRITE_DATA |
FILE_APPEND_DATA |
FILE_EXECUTE |
WRITE_DAC_ACCESS |
WRITE_OWNER_ACCESS |
SEC_FLAG_SYSTEM_SECURITY |
READ_CONTROL_ACCESS;
As long as the client doesn't request any of the access rights listed above, we
reuse the smb_fname->fsp, otherwise we close the smb_fname->fsp and call
fd_open() to open a new fsp.
In the future we can remove the four non-IO related access rights from the list:
WRITE_DAC_ACCESS |
WRITE_OWNER_ACCESS |
SEC_FLAG_SYSTEM_SECURITY |
READ_CONTROL_ACCESS
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
In order to make everything handle based, we will need the basefile handle when
eg the client requests setting any of the filemetadata that is common across all
streams, eg the file's timestamps.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Ensures we have a pathref handle in the smb_fname we pass to
SMB_VFS_CREATE_FILE().
As the create_disposition is FILE_OPEN we just return the error if
openat_pathref_fsp() fails
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
All callers except rename_open_files() can ignore non FSA fsps.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This way we can avoid stating twice: once here and possibly a second time in
the caller open_file().
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
non_widelink_open() now also returns NT_STATUS_STOPPED_ON_SYMLINK in case an
attempt was made to either
1. open a symlink from a POSIX client, or
2. open a symlink from a Windows client but any of the symlink behaviour
configuring options "follow symlink", "wide links" or "allow insecure wide
links" prevents access to the symlink target
Caller open_file() has already been updated to map NT_STATUS_STOPPED_ON_SYMLINK
to NT_STATUS_NT_STATUS_OBJECT_PATH_NOT_FOUND.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
If we get a real dirfsp, skip the parent-directory logic. Just pass the dirfsp
to SMB_VFS_OPENAT() which by now supports real dirfsps.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Instead of setting and resetting the name to the relative name every time we
call into the VFS, just set it once and reset it at the end and when recursing
via process_symlink_open().
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Callers still all pass conn->cwd_fsp so no change in behaviour yet.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
For now no change in behaviour as all callers still pass conn->cwd_fsp. This
just prepared fd_openat() to deal with real dirfsp's pass by callers later on
when adding calls to fd_openat(dirfspm ...) in the directory enumeration loop.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Calling openat() with O_PATH|O_NOFOLLOW will open a handle on the symlink
itself. That would be a nice feature if it would be supported on more platforms,
but being a Linux only thing, we have to preserve the behaviour of failing to
open a handle on symlinks.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
A subsequent commit will add a consumer of the fd to non_widelink_open() (by
calling SMB_VFS_FSTAT()), so we need to set the fd already here. And it makes
more sense anyway. :)
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
In all places where we access or modify a file or it's associated metadata, we
use fsp_get_io_fd() to fetch the low-level fd from the fsp. This ensures we
don't accidentally use a pathref fsp where the fd would be opened as root on
systems lacking O_PATH.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The call to share_mode_have_entries() was put in before
fresh_share_mode_lock() initialized d->flags to be completely
permissive. With that correct initialization the call to
share_conflict() a few lines down will also make open_mode_check()
pass for any share_access/access_mask.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
smbd: Fix failure to check dstdir for delete on close
We're preventing ourselves from holding two locks here,
not protecting from waiting for a lock someone else
holds.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <Volker.Lendecke@SerNet.DE>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sun Nov 22 01:22:36 UTC 2020 on sn-devel-184
In smb2_setinfo.c the call to smbd_do_setfilepathinfo() to perform the
rename takes place while holding a share mode lock. The function
check_parent_access() called below tries to query the destination
directory's locking.tdb entry to check whether the delete on close
flag is set on the destination directory. This fails because the
file to be renamed already has the share mode entry locked, we can't
lock two share mode entries simultaneously.
Convert the check to use fetch_share_mode_unlocked(). This might
introduce races, but this whole check is racy anyway. It does not
really matter whether we do the check for delete_on_close under a lock
or not, fetch_share_mode_unlocked() retrieves a consistent status of
the locking.tdb entry at some point in time as well.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Nov 20 00:20:06 UTC 2020 on sn-devel-184
These routines parse and marshall the uint16 summary flag in
share_mode_data. open_file_ntcreate() and open_directory() are the
only real users of this. The user in oplock.c is just the lazy reset
of the "read lease exists somewhere" after asynchronously breaking
read oplocks after a write request.
This moves handling the flags into locking/ to consolidate data
structure handling of "share_mode_data" there.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
When calling open_mode_check() we know the file id. We can pass it to
the lower levels without dereferencing struct share_mode_lock.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
A new lease never triggers a retry. Setting d->modified to true just
triggered the watchers needlessly.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
To me this is then easier to figure out what is defined there, and
where it's exactly used.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This was supposed to be a shortcut to avoid passing dirfsp around as an explicit
function argument throughout the whole codebase when the new VFS design idea was
based on using *AT functions throughout the VFS.
Now that we've opted for basing the VFS on handles and *AT functions will only
be used in a much more limitted extent, it makes sense to remove this internal
dirfsp reference, otherwise the combination of internal fsp->dirfsp and
smb_fname->fsp is going to be a tough to wrap your head around.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Oct 2 21:00:05 UTC 2020 on sn-devel-184
This was supposed to be a shortcut to avoid passing dirfsp around as an explicit
function argument throughout the whole codebase when the new VFS design idea was
based on using *AT functions throughout the VFS.
Now that we've opted for basing the VFS on handles and *AT functions will only
be used in a much more limitted extent, it makes sense to remove this internal
dirfsp reference, otherwise the combination of internal fsp->dirfsp and
smb_fname->fsp is going to be a tough to wrap your head around.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Prepares for removing the dirfsp arg from SMB_VFS_CREATE_FILE() again. In the
future dirfsp has to opened as needed within create_file_unixpath() and below.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
