linux/fs/cifs
Michal Hocko 063d99b4fa mm, fs: obey gfp_mapping for add_to_page_cache()
Commit 6afdb859b7 ("mm: do not ignore mapping_gfp_mask in page cache
allocation paths") has caught some users of hardcoded GFP_KERNEL used in
the page cache allocation paths.  This, however, wasn't complete and
there were others which went unnoticed.

Dave Chinner has reported the following deadlock for xfs on loop device:
: With the recent merge of the loop device changes, I'm now seeing
: XFS deadlock on my single CPU, 1GB RAM VM running xfs/073.
:
: The deadlocked is as follows:
:
: kloopd1: loop_queue_read_work
:       xfs_file_iter_read
:       lock XFS inode XFS_IOLOCK_SHARED (on image file)
:       page cache read (GFP_KERNEL)
:       radix tree alloc
:       memory reclaim
:       reclaim XFS inodes
:       log force to unpin inodes
:       <wait for log IO completion>
:
: xfs-cil/loop1: <does log force IO work>
:       xlog_cil_push
:       xlog_write
:       <loop issuing log writes>
:               xlog_state_get_iclog_space()
:               <blocks due to all log buffers under write io>
:               <waits for IO completion>
:
: kloopd1: loop_queue_write_work
:       xfs_file_write_iter
:       lock XFS inode XFS_IOLOCK_EXCL (on image file)
:       <wait for inode to be unlocked>
:
: i.e. the kloopd, with it's split read and write work queues, has
: introduced a dependency through memory reclaim. i.e. that writes
: need to be able to progress for reads make progress.
:
: The problem, fundamentally, is that mpage_readpages() does a
: GFP_KERNEL allocation, rather than paying attention to the inode's
: mapping gfp mask, which is set to GFP_NOFS.
:
: The didn't used to happen, because the loop device used to issue
: reads through the splice path and that does:
:
:       error = add_to_page_cache_lru(page, mapping, index,
:                       GFP_KERNEL & mapping_gfp_mask(mapping));

This has changed by commit aa4d86163e ("block: loop: switch to VFS
ITER_BVEC").

This patch changes mpage_readpage{s} to follow gfp mask set for the
mapping.  There are, however, other places which are doing basically the
same.

lustre:ll_dir_filler is doing GFP_KERNEL from the function which
apparently uses GFP_NOFS for other allocations so let's make this
consistent.

cifs:readpages_get_pages is called from cifs_readpages and
__cifs_readpages_from_fscache called from the same path obeys mapping
gfp.

ramfs_nommu_expand_for_mapping is hardcoding GFP_KERNEL as well
regardless it uses mapping_gfp_mask for the page allocation.

ext4_mpage_readpages is the called from the page cache allocation path
same as read_pages and read_cache_pages

As I've noticed in my previous post I cannot say I would be happy about
sprinkling mapping_gfp_mask all over the place and it sounds like we
should drop gfp_mask argument altogether and use it internally in
__add_to_page_cache_locked that would require all the filesystems to use
mapping gfp consistently which I am not sure is the case here.  From a
quick glance it seems that some file system use it all the time while
others are selective.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Dave Chinner <david@fromorbit.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Ming Lei <ming.lei@canonical.com>
Cc: Andreas Dilger <andreas.dilger@intel.com>
Cc: Oleg Drokin <oleg.drokin@intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-10-16 11:42:28 -07:00
..
Kconfig Allow parsing vers=3.11 on cifs mount 2015-06-27 20:23:32 -07:00
Makefile
asn1.c
cache.c
cifs_debug.c cifs: fix MUST SecurityFlags filtering 2015-01-26 19:38:26 -06:00
cifs_debug.h cifs: convert printk(LEVEL...) to pr_<level> 2014-12-07 22:48:07 -06:00
cifs_dfs_ref.c Fix that several functions handle incorrect value of mapchars 2015-05-10 19:56:35 -05:00
cifs_fs_sb.h Allow conversion of characters in Mac remap range. Part 1 2014-10-16 15:20:20 -05:00
cifs_ioctl.h Add way to query server fs info for smb3 2015-08-20 10:19:25 -05:00
cifs_spnego.c KEYS: Remove key_type::match in favour of overriding default by match_preparse 2014-09-16 17:36:06 +01:00
cifs_spnego.h
cifs_unicode.c Fix to convert SURROGATE PAIR 2015-05-20 13:12:51 -05:00
cifs_unicode.h Remap reserved posix characters by default (part 3/3) 2014-10-16 15:20:20 -05:00
cifs_uniupr.h
cifsacl.c fs/cifs: remove obsolete __constant 2014-12-10 17:41:02 -08:00
cifsacl.h
cifsencrypt.c cifs: use server timestamp for ntlmv2 authentication 2015-09-22 15:24:02 -05:00
cifsfs.c [CIFS] mount option sec=none not displayed properly in /proc/mounts 2015-09-11 19:37:06 -05:00
cifsfs.h [CIFS] Update cifs version number 2015-10-03 16:54:17 -05:00
cifsglob.h Add ioctl to set integrity 2015-06-28 21:15:45 -05:00
cifspdu.h Add way to query server fs info for smb3 2015-08-20 10:19:25 -05:00
cifsproto.h Fix that several functions handle incorrect value of mapchars 2015-05-10 19:56:35 -05:00
cifssmb.c cifs: Fix use-after-free on mid_q_entry 2015-08-20 10:19:25 -05:00
connect.c cifs: Unset CIFS_MOUNT_POSIX_PATHS flag when following dfs mounts 2015-06-29 14:50:22 -05:00
dir.c Fix that several functions handle incorrect value of mapchars 2015-05-10 19:56:35 -05:00
dns_resolve.c
dns_resolve.h
export.c
file.c mm, fs: obey gfp_mapping for add_to_page_cache() 2015-10-16 11:42:28 -07:00
fscache.c
fscache.h
inode.c [SMB3] Do not fall back to SMBWriteX in set_file_size error cases 2015-10-01 22:48:37 -05:00
ioctl.c CIFS: fix type confusion in copy offload ioctl 2015-09-11 09:54:03 -05:00
link.c Merge branch 'for-linus-1' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2015-06-22 12:51:21 -07:00
misc.c VFS: normal filesystems (and lustre): d_inode() annotations 2015-04-15 15:06:57 -04:00
netmisc.c Fix signed/unsigned pointer warning 2014-12-14 14:55:57 -06:00
nterr.c
nterr.h
ntlmssp.h
readdir.c cifs: Don't replace dentries for dfs mounts 2015-05-10 19:56:05 -05:00
rfc1002pdu.h
sess.c Merge branch 'akpm' (patchbomb from Andrew) 2014-12-10 18:34:42 -08:00
smb1ops.c Fix that several functions handle incorrect value of mapchars 2015-05-10 19:56:35 -05:00
smb2file.c VFS: normal filesystems (and lustre): d_inode() annotations 2015-04-15 15:06:57 -04:00
smb2glob.h
smb2inode.c CIFS: Fix wrong filename length for SMB2 2014-08-25 16:45:17 -05:00
smb2maperror.c Fix problem recognizing symlinks 2014-10-02 14:10:04 -05:00
smb2misc.c VFS: normal filesystems (and lustre): d_inode() annotations 2015-04-15 15:06:57 -04:00
smb2ops.c disabling oplocks/leases via module parm enable_oplocks broken for SMB3 2015-09-22 15:23:57 -05:00
smb2pdu.c [SMB3] Missing null tcon check 2015-09-26 09:48:58 -05:00
smb2pdu.h Update negotiate protocol for SMB3.11 dialect 2015-06-28 21:15:48 -05:00
smb2proto.h mfsymlinks support for SMB2.1/SMB3. Part 2 query symlink 2014-10-16 15:20:20 -05:00
smb2status.h
smb2transport.c Convert MessageID in smb2_hdr to LE 2014-12-14 14:55:45 -06:00
smbencrypt.c cifs: use memzero_explicit to clear stack buffer 2015-01-19 15:32:13 -06:00
smberr.h
smbfsctl.h Add Get/Set Integrity Information structure definitions 2015-06-28 21:15:41 -05:00
transport.c cifs: Fix use-after-free on mid_q_entry 2015-08-20 10:19:25 -05:00
winucase.c
xattr.c VFS: normal filesystems (and lustre): d_inode() annotations 2015-04-15 15:06:57 -04:00