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.
openat_pathref_fsp() does not need them anymore
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Fri Mar 11 19:19:21 UTC 2022 on sn-devel-184
With the simplifications in non_widelink_open() (don't depend on the
is_directory fsp flag) the main reason for requiring a valid stat
struct in openat_pathref_fsp() is gone. With this change
openat_pathref_fsp() is now capable of being the very first (and
authoritative) name-referencing operation with openat(O_PATH) for a
name.
Without having the stat information around before calling
openat_pathref_fsp(), the call to check_same_dev_ino() becomes
obsolete here.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
If I read Linux' man 2 open right (and susv4 agrees), O_DIRECTORY is
around to make sure opendir() is not raced against non-directory
files. opendir() needs to make sure the underlying object is actually
a directory. O_DIRECTORY is not required for opening directories in
RDONLY mode, regardless of having O_PATH or not.
At this point in openat_pathref_fsp() we don't care about the type of
the underlying object, we do fstat() and distinguish between files and
directories later according to the mode returned from fstat().
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
There's no reason why we would ever want to block on open(O_PATH). The
only cases that to me right now seem relevant is oplock breaks and
FIFOs, which can block forever. Oplock breaks don't happen for
O_PATH (hopefully...) but for the non-O_PATH case we don't want to
block either but we do handle this higher up.
We're handling EWOULDBLOCK for the oplock case correctly in
open_file_ntcreate() by setting up polling. So far we haven't done
this for the implicit openat_pathref_fsp() from filename_convert()
yet. But as our kernel oplock implementation lacks in functionality
big time anyway I would rather fail an open with NETWORK_BUSY than to
sit waiting for an oplock break for 30 seconds.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We don't need to look at the full share-relative path to figure out if
we have a stream name, the original smb_fname is sufficient for this.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Centralize the pattern
if (fsp->base_fsp != NULL) {
fsp = fsp->base_fsp;
}
with a descriptive name.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
To me this is more descriptive than "fsp->base_fsp != NULL". If this
turns out to be a performance problem, I would go and make this a
static inline in smbd/proto.h.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
To be reused in file_close_user(). Deliberately a separate commit to
make the previous commit easier to understand.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
close_file_free() needs to handle base fsps specially. This can be
simplified a lot if we pass the the open files a second time in case
we encountered base_fsps that we could not immediately delete.
file_close_conn() is not our hot code path, and also we don't expect
many thousand open files that we need to walk a second time.
A subsequent patch will simplify close_file_free(), the complicated
logic is now in files.c, where it IMHO belongs because
file_set_base_fsp() are here as well.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Quite a few places already had this in the caller, but not all. Rename
close_file() to close_file_free() appropriately. We'll factor out
close_file_smb() doing only parts of close_file_free() later.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Opening a stream base file only worked if "dirfsp == conn->cwd_fsp":
We have replaced fsp->fsp_name with the full dirfsp->relative pathname
at the point where open_pathref_base_fsp() is called. In case dirfsp
is already a subdirectory in a share, this breaks because the
open_pathref_base_fsp() uses fsp->fsp_name, not the original
dirfsp-relative one.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Remove skip test for the DISABLE_OPATH case.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14816
RN: Fix pathref open of a filesystem fifo in the DISABLE_OPATH build
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): Mon Sep 6 09:51:54 UTC 2021 on sn-devel-184
fd_openat() has done an FSTAT on the handle so update the smb_fname stat info
with "truth". from the handle.
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>
This reverts commit a6df051dd5:
"s3: smbd: In openat_pathref_fsp(), just check we're opening the same file type, not dev and inode."
The prior changes mean we can go back to checking dev/ino
matches.
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>
This occurs on first CHDIR to the root of the share.
Ensure we don't add conn->connectpath twice when doing
creating the file name hash.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
As this is an internal open and we don't have any
locks around, we don't have to mandate the dev and ino
pair are the same (and in fact not doing so fixes bugs
when this is called by VFS modules that like to play tricks
with ino number on stream paths (fruit, and streams_xattr
are the two that currently do this).
There's no security advantage to checking that, as the
fd_openat() ensures this is safe. As fd_openat() does an
FSTAT on the handle, update the smb_fname stat info with
the "correct" values from the handle.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We know that the parent name must
exist, and the name has been canonicalized
even if this was a POSIX pathname.
Ensure that we follow symlinks for
the parent. See the torture test
POSIX-SYMLINK-PARENT for details.
Remove knownfail entry.
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 Apr 7 15:39:45 UTC 2021 on sn-devel-184
The errno can be overwritten by the calls made by the fsp destructor.
This can cause problems if the original errno was required by subsequent
calls.
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Sachin Prabhu <sprabhu@redhat.com>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Wed Mar 10 22:55:17 UTC 2021 on sn-devel-184
NT_STATUS_STOPPED_ON_SYMLINK is returned when trying to open a symlink, most
callers are not interested in this.
Some callers that would want to know whether openat_pathref_fsp() failed
specifically on a symlink are setup_close_full_information(),
smbd_dirptr_get_entry(), unlink_internals() and filename_convert_internal(), so
we fix those callers to handle the symlink case themselves.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
No change in behaviour: the cleanup code at the fail label does the same as the
cleanup this patch removes. It has an extra fd_close() that is not existing in
the removed cleanup, but as fsp->fd is -1, that's a noop.
And when previously the
return NT_STATUS_OBJECT_NAME_NOT_FOUND;
returns an an explicit status code, when now doing goto fail status will also be
set to NT_STATUS_OBJECT_NAME_NOT_FOUND.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We're never creating files here, so instead of waiting for the underlying open()
to return ENOENT, just check that we have valid stat info, expecting all callers
to have called SMB_VFS_[L]STAT() on the smb_fname.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Similar to synthetic_smb_fname(), but also opens a pathref fsp.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Before we had open_pathref_fsp() we had the stream fsp before the base
fsp in the linked list we traverse for SHUTDOWN_CLOSE.
Now the order has changed. I could have used some DLIST_PROMOTE()
hacks, but that's still fragile.
Now we reference both fsp's via ->base_fsp and ->stream_fsp.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
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>
That will be useful in other places as well.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Otherwise we'll always keep the current talloc_stackframe arround.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
It's useful to watch this using: git show --histogram
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>
We also need to be sure both sides were not linked before.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We only modify smb_fname_src on success.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
