[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241213135013.2964079-1-dhowells@redhat.com>
Date: Fri, 13 Dec 2024 13:50:00 +0000
From: David Howells <dhowells@...hat.com>
To: Christian Brauner <christian@...uner.io>
Cc: David Howells <dhowells@...hat.com>,
Max Kellermann <max.kellermann@...os.com>,
Ilya Dryomov <idryomov@...il.com>,
Xiubo Li <xiubli@...hat.com>,
Trond Myklebust <trondmy@...nel.org>,
Jeff Layton <jlayton@...nel.org>,
Matthew Wilcox <willy@...radead.org>,
netfs@...ts.linux.dev,
linux-afs@...ts.infradead.org,
linux-cifs@...r.kernel.org,
linux-nfs@...r.kernel.org,
ceph-devel@...r.kernel.org,
v9fs@...ts.linux.dev,
linux-erofs@...ts.ozlabs.org,
linux-fsdevel@...r.kernel.org,
linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: [PATCH 00/10] netfs, ceph, nfs, cachefiles: Miscellaneous fixes/changes
Hi Christian,
Here are some miscellaneous fixes and changes for netfslib and the ceph and
nfs filesystems:
(1) Ignore silly-rename files from afs and nfs when building the header
archive in a kernel build.
(2) netfs: Fix the way read result collection applies results to folios
when each folio is being read by multiple subrequests and the results
come out of order.
(3) netfs: Fix ENOMEM handling in buffered reads.
(4) nfs: Fix an oops in nfs_netfs_init_request() when copying to the cache.
(5) cachefiles: Parse the "secctx" command immediately to get the correct
error rather than leaving it to the "bind" command.
(6) netfs: Remove a redundant smp_rmb(). This isn't a bug per se and
could be deferred.
(7) netfs: Fix missing barriers by using clear_and_wake_up_bit().
(8) netfs: Work around recursion in read retry by failing and abandoning
the retried subrequest if no I/O is performed.
[!] NOTE: This only works around the recursion problem if the
recursion keeps returning no data. If the server manages, say, to
repeatedly return a single byte of data faster than the retry
algorithm can complete, it will still recurse and the stack
overrun may still occur. Actually fixing this requires quite an
intrusive change which will hopefully make the next merge window.
(9) netfs: Fix the clearance of a folio_queue when unlocking the page if
we're going to want to subsequently send the queue for copying to the
cache (if, for example, we're using ceph).
(10) netfs: Fix the lack of cancellation of copy-to-cache when the cache
for a file is temporarily disabled (for example when a DIO write is
done to the file). This patch and (9) fix hangs with ceph.
With these patches, I can run xfstest -g quick to completion on ceph with a
local cache.
The patches can also be found here with a bonus cifs patch:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=netfs-fixes
Thanks,
David
David Howells (8):
kheaders: Ignore silly-rename files
netfs: Fix non-contiguous donation between completed reads
netfs: Fix enomem handling in buffered reads
nfs: Fix oops in nfs_netfs_init_request() when copying to cache
netfs: Fix missing barriers by using clear_and_wake_up_bit()
netfs: Work around recursion by abandoning retry if nothing read
netfs: Fix ceph copy to cache on write-begin
netfs: Fix the (non-)cancellation of copy when cache is temporarily
disabled
Max Kellermann (1):
cachefiles: Parse the "secctx" immediately
Zilin Guan (1):
netfs: Remove redundant use of smp_rmb()
fs/9p/vfs_addr.c | 6 +++++-
fs/afs/write.c | 5 ++++-
fs/cachefiles/daemon.c | 14 +++++++-------
fs/cachefiles/internal.h | 3 ++-
fs/cachefiles/security.c | 6 +++---
fs/netfs/buffered_read.c | 28 ++++++++++++++++------------
fs/netfs/direct_write.c | 1 -
fs/netfs/read_collect.c | 33 +++++++++++++++++++--------------
fs/netfs/read_pgpriv2.c | 4 ++++
fs/netfs/read_retry.c | 6 ++++--
fs/netfs/write_collect.c | 14 +++++---------
fs/netfs/write_issue.c | 2 ++
fs/nfs/fscache.c | 9 ++++++++-
fs/smb/client/cifssmb.c | 13 +++++++++----
fs/smb/client/smb2pdu.c | 9 ++++++---
include/linux/netfs.h | 6 +++---
kernel/gen_kheaders.sh | 1 +
17 files changed, 98 insertions(+), 62 deletions(-)
Powered by blists - more mailing lists