Pull 9p update from Dominique Martinet:
- fix long-standing limitation on open-unlink-fop pattern
- add refcount to p9_fid (fixes the above and will allow for more
cleanups and simplifications in the future)
* tag '9p-for-5.11-rc1' of git://github.com/martinetd/linux:
9p: Remove unnecessary IS_ERR() check
9p: Uninitialized variable in v9fs_writeback_fid()
9p: Fix writeback fid incorrectly being attached to dentry
9p: apply review requests for fid refcounting
9p: add refcount to p9_fid struct
fs/9p: search open fids first
fs/9p: track open fids
fs/9p: fix create-unlink-getattr idiom
The v9fs file operations were missing the splice_read operations, which
breaks sendfile() of files on such a filesystem. I discovered this while
trying to load an eBPF program using iproute2 inside a 'virtme' environment
which uses 9pfs for the virtual file system. iproute2 relies on sendfile()
with an AF_ALG socket to hash files, which was erroring out in the virtual
environment.
Since generic_file_splice_read() seems to just implement splice_read in
terms of the read_iter operation, I simply added the generic implementation
to the file operations, which fixed the error I was seeing. A quick grep
indicates that this is what most other file systems do as well.
Link: http://lkml.kernel.org/r/20201201135409.55510-1-toke@redhat.com
Fixes: 36e2c7421f ("fs: don't allow splice read/write without explicit ops")
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
The "fid" variable can't be an error pointer so there is no need to
check. The code is slightly cleaner if we move the increment before
the break and remove the NULL check as well.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
If v9fs_fid_lookup_with_uid() fails then "fid" is not initialized.
The v9fs_fid_lookup_with_uid() can't return NULL. If it returns an
error pointer then we can still pass that to clone_fid() and it will
return the error pointer back again.
Fixes: 6636b6dcc3 ("9p: add refcount to p9_fid struct")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
Fix race issue in fid contention.
Eric's and Greg's patch offer a mechanism to fix open-unlink-f*syscall
bug in 9p. But there is race issue in fid parallel accesses.
As Greg's patch stores all of fids from opened files into according inode,
so all the lookup fid ops can retrieve fid from inode preferentially. But
there is no mechanism to handle the fid contention issue. For example,
there are two threads get the same fid in the same time and one of them
clunk the fid before the other thread ready to discard the fid. In this
scenario, it will lead to some fatal problems, even kernel core dump.
I introduce a mechanism to fix this race issue. A counter field introduced
into p9_fid struct to store the reference counter to the fid. When a fid
is allocated from the inode or dentry, the counter will increase, and
will decrease at the end of its occupation. It is guaranteed that the
fid won't be clunked before the reference counter go down to 0, then
we can avoid the clunked fid to be used.
tests:
race issue test from the old test case:
for file in {01..50}; do touch f.${file}; done
seq 1 1000 | xargs -n 1 -P 50 -I{} cat f.* > /dev/null
open-unlink-f*syscall test:
I have tested for f*syscall include: ftruncate fstat fchown fchmod faccessat.
Link: http://lkml.kernel.org/r/20200923141146.90046-5-jianyong.wu@arm.com
Fixes: 478ba09edc ("fs/9p: search open fids first")
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
A previous patch fixed the "create-unlink-getattr" idiom: if getattr is
called on an unlinked file, we try to find an open fid attached to the
corresponding inode.
We have a similar issue with file permissions and setattr:
open("./test.txt", O_RDWR|O_CREAT, 0666) = 4
chmod("./test.txt", 0) = 0
truncate("./test.txt", 0) = -1 EACCES (Permission denied)
ftruncate(4, 0) = -1 EACCES (Permission denied)
The failure is expected with truncate() but not with ftruncate().
This happens because the lookup code does find a matching fid in the
dentry list. Unfortunately, this is not an open fid and the server
will be forced to rely on the path name, rather than on an open file
descriptor. This is the case in QEMU: the setattr operation will use
truncate() and fail because of bad write permissions.
This patch changes the logic in the lookup code, so that we consider
open fids first. It gives a chance to the server to match this open
fid to an open file descriptor and use ftruncate() instead of truncate().
This does not change the current behaviour for truncate() and other
path name based syscalls, since file permissions are checked earlier
in the VFS layer.
With this patch, we get:
open("./test.txt", O_RDWR|O_CREAT, 0666) = 4
chmod("./test.txt", 0) = 0
truncate("./test.txt", 0) = -1 EACCES (Permission denied)
ftruncate(4, 0) = 0
Link: http://lkml.kernel.org/r/20200923141146.90046-4-jianyong.wu@arm.com
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
Fixes several outstanding bug reports of not being able to getattr from an
open file after an unlink. This patch cleans up transient fids on an unlink
and will search open fids on a client if it detects a dentry that appears to
have been unlinked. This search is necessary because fstat does not pass fd
information through the VFS API to the filesystem, only the dentry which for
9p has an imperfect match to fids.
Inherent in this patch is also a fix for the qid handling on create/open
which apparently wasn't being set correctly and was necessary for the search
to succeed.
A possible optimization over this fix is to include accounting of open fids
with the inode in the private data (in a similar fashion to the way we track
transient fids with dentries). This would allow a much quicker search for
a matching open fid.
(changed v9fs_fid_find_global to v9fs_fid_find_inode in comment)
Link: http://lkml.kernel.org/r/20200923141146.90046-2-jianyong.wu@arm.com
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
Pull misc vfs updates from Al Viro:
"Assorted stuff all over the place (the largest group here is
Christoph's stat cleanups)"
* 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fs: remove KSTAT_QUERY_FLAGS
fs: remove vfs_stat_set_lookup_flags
fs: move vfs_fstatat out of line
fs: implement vfs_stat and vfs_lstat in terms of vfs_fstatat
fs: remove vfs_statx_fd
fs: omfs: use kmemdup() rather than kmalloc+memcpy
[PATCH] reduce boilerplate in fsid handling
fs: Remove duplicated flag O_NDELAY occurring twice in VALID_OPEN_FLAGS
selftests: mount: add nosymfollow tests
Add a "nosymfollow" mount option.
Pull 9p updates from Dominique Martinet:
"A couple of small fixes (loff_t overflow on 32bit, syzbot
uninitialized variable warning) and code cleanup (xen)"
* tag '9p-for-5.10-rc1' of git://github.com/martinetd/linux:
net: 9p: initialize sun_server.sun_path to have addr's value only when addr is valid
9p/xen: Fix format argument warning
9P: Cast to loff_t before multiplying
Replace the two negative flags that are always used together with a
single positive flag that indicates the writeback capability instead
of two related non-capabilities. Also remove the pointless wrappers
to just check the flag.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Set up a readahead size by default, as very few users have a good
reason to change it. This means code, ecryptfs, and orangefs now
set up the values while they were previously missing it, while ubifs,
mtd and vboxsf manually set it to 0 to avoid readahead.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: David Sterba <dsterba@suse.com> [btrfs]
Acked-by: Richard Weinberger <richard@nod.at> [ubifs, mtd]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In the current setattr implementation in 9p, fid is always retrieved
from dentry no matter file instance exists or not. If so, there may be
some info related to opened file instance dropped. So it's better
to retrieve fid from file instance when it is passed to setattr.
for example:
fd=open("tmp", O_RDWR);
ftruncate(fd, 10);
The file context related with the fd will be lost as fid is always
retrieved from dentry, then the backend can't get the info of
file context. It is against the original intention of user and
may lead to bug.
Link: http://lkml.kernel.org/r/20200710101548.10108-1-jianyong.wu@arm.com
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
Linus Torvalds
Dominique Martinet
Toke Høiland-Jørgensen
Dan Carpenter
Jianyong Wu
Greg Kurz
Greg Kurz
Eric Van Hensbergen
Matthew Wilcox (Oracle)
Christoph Hellwig
Al Viro
Gustavo A. R. Silva
Li Heng
Zheng Bin
Sergey Alirzaev
Krzysztof Kozlowski