mbox series

[v5,00/12] netfs, cifs: Delegate high-level I/O to netfslib

Message ID 20240205225726.3104808-1-dhowells@redhat.com (mailing list archive)
Headers show
Series netfs, cifs: Delegate high-level I/O to netfslib | expand

Message

David Howells Feb. 5, 2024, 10:57 p.m. UTC
Hi Steve,

Here are patches to convert cifs to use the netfslib library.  With this I
can run a certain amount of xfstests on CIFS, though not all the tests work
correctly because of fallocate issues.

The patches remove around 2000 lines from CIFS

To do:

 (*) Implement write-retry.  Currently, netfslib errors out on a failed DIO
     write and relies on the VM to drive writepages again on a failed
     buffered write.  This needs some retry logic adding into
     fs/netfs/output.c.

     I'm not sure what the best way to handle this is.  One way is to
     resend each failing subreq as it fails, offloading this to a kernel
     thread that re-splits the subreq, calling out to a rreq->op to do the
     splitting, thereby allowing cifs to renegotiate credits.  If a subreq
     is split, the two parts need to be adjacent in the rreq->subrequests
     list.

     An alternative way might be to try and combine failing tests and then
     split them.

     Yet a third way might be to try each failing subreq a smaller bit at a
     time and keep track of what has been sent in
     wdata->subreq.transferred.

     Whichever way is chosen, NETFS_SREQ_RETRYING should be set in
     wdata->subreq.flags instead of setting wdata->replay.

Notes:

 (1) CIFS is made to use unbuffered I/O for unbuffered caching modes and
     write-through caching for cache=strict.

 (2) Various cifs fallocate() function implementations have issues that
     aren't easily fixed without enhanced protocol support.

 (3) It should be possible to turn on multipage folio support in CIFS now.

 (4) The then-unused CIFS code is removed in three patches, not one, to
     avoid the git patch generator from producing confusing patches in
     which it thinks code is being moved around rather than just being
     removed.

The patches can be found here also:

	https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=cifs-netfs

Changes
=======
ver #5)
 - Rebased to -rc3 plus SteveF's for-next branch as netfslib is now
   upstream, as are a couple of patches from this series.
 - Replace the ->replay bool Shyam added with a flag on the netfs
   subrequest.  This is tested by the code, but not currently set (see
   above).

ver #4)
 - Slimmed down the branch:
   - Split the cifs-related patches off to a separate branch (cifs-netfs)
   - Deferred the content-encryption to the in-progress ceph changes.
   - Deferred the use-PG_writeback rather than PG_fscache patch
 - Rebased on a later linux-next with afs-rotation patches.

ver #3)
 - Moved the fscache module into netfslib to avoid export cycles.
 - Fixed a bunch of bugs.
 - Got CIFS to pass as much of xfstests as possible.
 - Added a patch to make 9P use all the helpers.
 - Added a patch to stop using PG_fscache, but rather dirty pages on
   reading and have writepages write to the cache.

ver #2)
 - Folded the addition of NETFS_RREQ_NONBLOCK/BLOCKED into first patch that
   uses them.
 - Folded addition of rsize member into first user.
 - Don't set rsize in ceph (yet) and set it in kafs to 256KiB.  cifs sets
   it dynamically.
 - Moved direct_bv next to direct_bv_count in struct netfs_io_request and
   labelled it with a __counted_by().
 - Passed flags into netfs_xa_store_and_mark() rather than two bools.
 - Removed netfs_set_up_buffer() as it wasn't used.

David

Link: https://lore.kernel.org/r/20231213152350.431591-1-dhowells@redhat.com/ [1]
Link: https://lore.kernel.org/r/20231013160423.2218093-1-dhowells@redhat.com/ # v1
Link: https://lore.kernel.org/r/20231117211544.1740466-1-dhowells@redhat.com/ # v2
Link: https://lore.kernel.org/r/20231207212206.1379128-1-dhowells@redhat.com/ # v3
Link: https://lore.kernel.org/r/20231213154139.432922-1-dhowells@redhat.com/ # v4

David Howells (12):
  cifs: Replace cifs_readdata with a wrapper around netfs_io_subrequest
  cifs: Set zero_point in the copy_file_range() and remap_file_range()
  cifs: Replace cifs_writedata with a wrapper around netfs_io_subrequest
  cifs: Use more fields from netfs_io_subrequest
  cifs: Make wait_mtu_credits take size_t args
  cifs: Implement netfslib hooks
  cifs: Replace the writedata replay bool with a netfs sreq flag
  cifs: Move cifs_loose_read_iter() and cifs_file_write_iter() to file.c
  cifs: Cut over to using netfslib
  cifs: Remove some code that's no longer used, part 1
  cifs: Remove some code that's no longer used, part 2
  cifs: Remove some code that's no longer used, part 3

 fs/netfs/buffered_write.c    |    3 +
 fs/netfs/io.c                |    7 +-
 fs/smb/client/Kconfig        |    1 +
 fs/smb/client/cifsfs.c       |   69 +-
 fs/smb/client/cifsfs.h       |   10 +-
 fs/smb/client/cifsglob.h     |   59 +-
 fs/smb/client/cifsproto.h    |   14 +-
 fs/smb/client/cifssmb.c      |  111 +-
 fs/smb/client/file.c         | 2911 ++++++----------------------------
 fs/smb/client/fscache.c      |  109 --
 fs/smb/client/fscache.h      |   54 -
 fs/smb/client/inode.c        |   19 +-
 fs/smb/client/smb2ops.c      |   10 +-
 fs/smb/client/smb2pdu.c      |  169 +-
 fs/smb/client/smb2proto.h    |    5 +-
 fs/smb/client/trace.h        |  144 +-
 fs/smb/client/transport.c    |   17 +-
 include/linux/netfs.h        |    2 +
 include/trace/events/netfs.h |    1 +
 19 files changed, 852 insertions(+), 2863 deletions(-)

Comments

Steve French Feb. 8, 2024, 3:55 a.m. UTC | #1
This looks promising but I do hit an oops fairly early in xfstest runs
with the 12 patches in your series (on 6.8-rc3)

[  228.136056] run fstests generic/306 at 2024-02-07 21:32:16

[  228.573734] ------------[ cut here ]------------
[  228.573737] kernel BUG at lib/iov_iter.c:582!
[  228.573744] invalid opcode: 0000 [#1] PREEMPT SMP PTI
[  228.573748] CPU: 2 PID: 4033 Comm: cifsd Tainted: G            E
  6.8.0-rc3+ #2
[  228.573751] Hardware name: Microsoft Corporation Virtual
Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 11/28/2023
[  228.573752] RIP: 0010:iov_iter_revert+0x114/0x120
[  228.573758] Code: 48 89 78 08 31 c0 31 d2 31 c9 31 f6 31 ff 45 31
c0 c3 cc cc cc cc 0f 0b 31 c0 31 d2 31 c9 31 f6 31 ff 45 31 c0 c3 cc
cc cc cc <0f> 0b 66 2e 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90 90
90 90
[  228.573760] RSP: 0018:ffffb8d005c27d68 EFLAGS: 00010246
[  228.573763] RAX: ffffb8d005c27d98 RBX: ffff99934d039b00 RCX: 0000000000000000
[  228.573764] RDX: 0000000000000000 RSI: 0000000000000200 RDI: 0000000000000004
[  228.573766] RBP: ffffb8d005c27e00 R08: 0000000000000000 R09: 0000000000000000
[  228.573767] R10: 0000000000000000 R11: 0000000000000000 R12: ffff999342736900
[  228.573769] R13: ffff9993a7fd4800 R14: ffff9993a7fd2000 R15: 0000000000000001
[  228.573770] FS:  0000000000000000(0000) GS:ffff9994f2b00000(0000)
knlGS:0000000000000000
[  228.573772] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  228.573774] CR2: 00007fe530dffe38 CR3: 0000000103fc8001 CR4: 00000000003706f0
[  228.573777] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  228.573778] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  228.573779] Call Trace:
[  228.573781]  <TASK>
[  228.573783]  ? show_regs+0x6d/0x80
[  228.573787]  ? die+0x37/0xa0
[  228.573789]  ? do_trap+0xd4/0xf0
[  228.573793]  ? do_error_trap+0x71/0xb0
[  228.573795]  ? iov_iter_revert+0x114/0x120
[  228.573798]  ? exc_invalid_op+0x52/0x80
[  228.573801]  ? iov_iter_revert+0x114/0x120
[  228.573803]  ? asm_exc_invalid_op+0x1b/0x20
[  228.573808]  ? iov_iter_revert+0x114/0x120
[  228.573813]  ? smb2_readv_callback+0x50f/0x5b0 [cifs]
[  228.573874]  cifs_demultiplex_thread+0x46e/0xe40 [cifs]
[  228.573920]  ? __pfx_cifs_demultiplex_thread+0x10/0x10 [cifs]
[  228.573962]  kthread+0xef/0x120
[  228.573966]  ? __pfx_kthread+0x10/0x10
[  228.573969]  ret_from_fork+0x44/0x70
[  228.573972]  ? __pfx_kthread+0x10/0x10
[  228.573975]  ret_from_fork_asm+0x1b/0x30
[  228.573979]  </TASK>
[  228.573980] Modules linked in: cmac(E) nls_utf8(E) cifs(E)
cifs_arc4(E) nls_ucs2_utils(E) cifs_md4(E) netfs(E) snd_seq_dummy(E)
snd_hrtimer(E) snd_seq_midi(E) snd_seq_midi_event(E) snd_rawmidi(E)
snd_seq(E) snd_seq_device(E) snd_timer(E) snd(E) soundcore(E)
sunrpc(E) binfmt_misc(E) intel_rapl_msr(E) intel_rapl_common(E)
intel_uncore_frequency_common(E) intel_pmc_core(E) intel_vsec(E)
pmt_telemetry(E) pmt_class(E) crct10dif_pclmul(E) polyval_clmulni(E)
polyval_generic(E) ghash_clmulni_intel(E) sha512_ssse3(E)
sha256_ssse3(E) sha1_ssse3(E) aesni_intel(E) crypto_simd(E) cryptd(E)
rapl(E) hyperv_drm(E) drm_shmem_helper(E) drm_kms_helper(E)
hv_balloon(E) hyperv_fb(E) vmgenid(E) joydev(E) mac_hid(E)
serio_raw(E) msr(E) parport_pc(E) ppdev(E) lp(E) drm(E) parport(E)
efi_pstore(E) dmi_sysfs(E) ip_tables(E) x_tables(E) autofs4(E)
btrfs(E) blake2b_generic(E) xor(E) raid6_pq(E) libcrc32c(E)
hid_generic(E) hid_hyperv(E) hv_storvsc(E) hv_netvsc(E) hid(E)
hyperv_keyboard(E) scsi_transport_fc(E) hv_utils(E) crc32_pclmul(E)
[  228.574031]  hv_vmbus(E)
[  228.574035] ---[ end trace 0000000000000000 ]---
[  228.636462] RIP: 0010:iov_iter_revert+0x114/0x120
[  228.636471] Code: 48 89 78 08 31 c0 31 d2 31 c9 31 f6 31 ff 45 31
c0 c3 cc cc cc cc 0f 0b 31 c0 31 d2 31 c9 31 f6 31 ff 45 31 c0 c3 cc
cc cc cc <0f> 0b 66 2e 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90 90
90 90
[  228.636474] RSP: 0018:ffffb8d005c27d68 EFLAGS: 00010246
[  228.636477] RAX: ffffb8d005c27d98 RBX: ffff99934d039b00 RCX: 0000000000000000
[  228.636479] RDX: 0000000000000000 RSI: 0000000000000200 RDI: 0000000000000004
[  228.636481] RBP: ffffb8d005c27e00 R08: 0000000000000000 R09: 0000000000000000
[  228.636482] R10: 0000000000000000 R11: 0000000000000000 R12: ffff999342736900
[  228.636484] R13: ffff9993a7fd4800 R14: ffff9993a7fd2000 R15: 0000000000000001
[  228.636485] FS:  0000000000000000(0000) GS:ffff9994f2b00000(0000)
knlGS:0000000000000000
[  228.636487] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  228.636489] CR2: 00007fe530dffe38 CR3: 0000000103fc8001 CR4: 00000000003706f0
[  228.636492] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  228.636494] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400


On Mon, Feb 5, 2024 at 4:57 PM David Howells <dhowells@redhat.com> wrote:
>
> Hi Steve,
>
> Here are patches to convert cifs to use the netfslib library.  With this I
> can run a certain amount of xfstests on CIFS, though not all the tests work
> correctly because of fallocate issues.
>
> The patches remove around 2000 lines from CIFS
>
> To do:
>
>  (*) Implement write-retry.  Currently, netfslib errors out on a failed DIO
>      write and relies on the VM to drive writepages again on a failed
>      buffered write.  This needs some retry logic adding into
>      fs/netfs/output.c.
>
>      I'm not sure what the best way to handle this is.  One way is to
>      resend each failing subreq as it fails, offloading this to a kernel
>      thread that re-splits the subreq, calling out to a rreq->op to do the
>      splitting, thereby allowing cifs to renegotiate credits.  If a subreq
>      is split, the two parts need to be adjacent in the rreq->subrequests
>      list.
>
>      An alternative way might be to try and combine failing tests and then
>      split them.
>
>      Yet a third way might be to try each failing subreq a smaller bit at a
>      time and keep track of what has been sent in
>      wdata->subreq.transferred.
>
>      Whichever way is chosen, NETFS_SREQ_RETRYING should be set in
>      wdata->subreq.flags instead of setting wdata->replay.
>
> Notes:
>
>  (1) CIFS is made to use unbuffered I/O for unbuffered caching modes and
>      write-through caching for cache=strict.
>
>  (2) Various cifs fallocate() function implementations have issues that
>      aren't easily fixed without enhanced protocol support.
>
>  (3) It should be possible to turn on multipage folio support in CIFS now.
>
>  (4) The then-unused CIFS code is removed in three patches, not one, to
>      avoid the git patch generator from producing confusing patches in
>      which it thinks code is being moved around rather than just being
>      removed.
>
> The patches can be found here also:
>
>         https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=cifs-netfs
>
> Changes
> =======
> ver #5)
>  - Rebased to -rc3 plus SteveF's for-next branch as netfslib is now
>    upstream, as are a couple of patches from this series.
>  - Replace the ->replay bool Shyam added with a flag on the netfs
>    subrequest.  This is tested by the code, but not currently set (see
>    above).
>
> ver #4)
>  - Slimmed down the branch:
>    - Split the cifs-related patches off to a separate branch (cifs-netfs)
>    - Deferred the content-encryption to the in-progress ceph changes.
>    - Deferred the use-PG_writeback rather than PG_fscache patch
>  - Rebased on a later linux-next with afs-rotation patches.
>
> ver #3)
>  - Moved the fscache module into netfslib to avoid export cycles.
>  - Fixed a bunch of bugs.
>  - Got CIFS to pass as much of xfstests as possible.
>  - Added a patch to make 9P use all the helpers.
>  - Added a patch to stop using PG_fscache, but rather dirty pages on
>    reading and have writepages write to the cache.
>
> ver #2)
>  - Folded the addition of NETFS_RREQ_NONBLOCK/BLOCKED into first patch that
>    uses them.
>  - Folded addition of rsize member into first user.
>  - Don't set rsize in ceph (yet) and set it in kafs to 256KiB.  cifs sets
>    it dynamically.
>  - Moved direct_bv next to direct_bv_count in struct netfs_io_request and
>    labelled it with a __counted_by().
>  - Passed flags into netfs_xa_store_and_mark() rather than two bools.
>  - Removed netfs_set_up_buffer() as it wasn't used.
>
> David
>
> Link: https://lore.kernel.org/r/20231213152350.431591-1-dhowells@redhat.com/ [1]
> Link: https://lore.kernel.org/r/20231013160423.2218093-1-dhowells@redhat.com/ # v1
> Link: https://lore.kernel.org/r/20231117211544.1740466-1-dhowells@redhat.com/ # v2
> Link: https://lore.kernel.org/r/20231207212206.1379128-1-dhowells@redhat.com/ # v3
> Link: https://lore.kernel.org/r/20231213154139.432922-1-dhowells@redhat.com/ # v4
>
> David Howells (12):
>   cifs: Replace cifs_readdata with a wrapper around netfs_io_subrequest
>   cifs: Set zero_point in the copy_file_range() and remap_file_range()
>   cifs: Replace cifs_writedata with a wrapper around netfs_io_subrequest
>   cifs: Use more fields from netfs_io_subrequest
>   cifs: Make wait_mtu_credits take size_t args
>   cifs: Implement netfslib hooks
>   cifs: Replace the writedata replay bool with a netfs sreq flag
>   cifs: Move cifs_loose_read_iter() and cifs_file_write_iter() to file.c
>   cifs: Cut over to using netfslib
>   cifs: Remove some code that's no longer used, part 1
>   cifs: Remove some code that's no longer used, part 2
>   cifs: Remove some code that's no longer used, part 3
>
>  fs/netfs/buffered_write.c    |    3 +
>  fs/netfs/io.c                |    7 +-
>  fs/smb/client/Kconfig        |    1 +
>  fs/smb/client/cifsfs.c       |   69 +-
>  fs/smb/client/cifsfs.h       |   10 +-
>  fs/smb/client/cifsglob.h     |   59 +-
>  fs/smb/client/cifsproto.h    |   14 +-
>  fs/smb/client/cifssmb.c      |  111 +-
>  fs/smb/client/file.c         | 2911 ++++++----------------------------
>  fs/smb/client/fscache.c      |  109 --
>  fs/smb/client/fscache.h      |   54 -
>  fs/smb/client/inode.c        |   19 +-
>  fs/smb/client/smb2ops.c      |   10 +-
>  fs/smb/client/smb2pdu.c      |  169 +-
>  fs/smb/client/smb2proto.h    |    5 +-
>  fs/smb/client/trace.h        |  144 +-
>  fs/smb/client/transport.c    |   17 +-
>  include/linux/netfs.h        |    2 +
>  include/trace/events/netfs.h |    1 +
>  19 files changed, 852 insertions(+), 2863 deletions(-)
>
>


--
Thanks,

Steve
David Howells Feb. 19, 2024, 3:30 p.m. UTC | #2
Steve French <smfrench@gmail.com> wrote:

> [  228.573783]  ? show_regs+0x6d/0x80
> [  228.573787]  ? die+0x37/0xa0
> [  228.573789]  ? do_trap+0xd4/0xf0
> [  228.573793]  ? do_error_trap+0x71/0xb0
> [  228.573795]  ? iov_iter_revert+0x114/0x120
> [  228.573798]  ? exc_invalid_op+0x52/0x80
> [  228.573801]  ? iov_iter_revert+0x114/0x120
> [  228.573803]  ? asm_exc_invalid_op+0x1b/0x20
> [  228.573808]  ? iov_iter_revert+0x114/0x120
> [  228.573813]  ? smb2_readv_callback+0x50f/0x5b0 [cifs]
> [  228.573874]  cifs_demultiplex_thread+0x46e/0xe40 [cifs]

I don't suppose you can tell me what line smb2_readv_callback+0x50f/0x5b0 is?

David
David Howells Feb. 19, 2024, 3:42 p.m. UTC | #3
David Howells <dhowells@redhat.com> wrote:

> I don't suppose you can tell me what line smb2_readv_callback+0x50f/0x5b0 is?

It's almost certainly the iov_iter_revert() here:

	switch (mid->mid_state) {
	case MID_RESPONSE_RECEIVED:
		credits.value = le16_to_cpu(shdr->CreditRequest);
		credits.instance = server->reconnect_instance;
		/* result already set, check signature */
		if (server->sign && !mid->decrypted) {
			int rc;

			iov_iter_revert(&rqst.rq_iter, rdata->got_bytes);
			iov_iter_truncate(&rqst.rq_iter, rdata->got_bytes);

The reason that the:

	[  228.573737] kernel BUG at lib/iov_iter.c:582!

happens is that we're trying to wind the iterator back before its start point.

Now, the iterator is reinitialised at the beginning of the function:

	if (rdata->got_bytes) {
		rqst.rq_iter	  = rdata->subreq.io_iter;
		rqst.rq_iter_size = iov_iter_count(&rdata->subreq.io_iter);
	}

so the reversion is probably unnecessary.

Note that this can only happen if we're using signed messages:

		if (server->sign && !mid->decrypted) {

as we wind back the iterator so that we can use it to feed the buffer to the
hashing algorithm.

David
David Howells Feb. 19, 2024, 4:12 p.m. UTC | #4
David Howells <dhowells@redhat.com> wrote:

> so the reversion is probably unnecessary.

Removing the iov_iter_revert() fixes the problem.

David