[PATCH v6 00/19] gfs2: Fix mmap + page fault deadlocks

From: Andreas Gruenbacher
Date: Thu Aug 19 2021 - 15:41:19 EST


Hi all,

here's another update on top of v5.14-rc6. Changes:

* Per request from Linus, change fault_in_{readable,writeable} to
return the number of bytes *not* faulted in, like copy_to_user() and
copy_from_user() does. Convert fault_in_iov_iter_readable and
fault_in_iov_iter_writeable to those same semantics.

* Per suggestion from Linus, introduce a new FOLL_NOFAULT flag to
prevent get_user_pages from faulting in pages. This is similar to
FOLL_FAST_ONLY, but less fragile and available on all architectures.
Use that for turning off page faults during iov_iter_get_pages() and
iov_iter_get_pages_alloc().

* Introduce a new HIF_MAY_DEMOTE flag that allows a glock to be taken
away from a holder when a conflicting locking request comes in. This
allows glock holders to hang on to glocks as long as no conflicting
locking requests occur. This avoids returning short reads and writes
when pages need to be faulted in.

* Limit the number of pages that are faulted in at once to a more
sensible size instead of faulting in all pages at once. When
faulting in pages doesn't lead to success, fault in a single page
in the next attempt. When that still doesn't succeed, give up.
This should prevent endless loops when fault_in_iov_iter_*() and
bio_iov_iter_get_pages() disagree.

* It turns out that taking the inode glock in gfs2_write_lock and
releasing it in gfs2_write_unlock was entirely pointless, so move
the locking into gfs2_file_buffered_write instead. This then also
allows to eliminate ip->i_gh.


This iteration fixes the issues with fstest generic/208.


For immediate consideration by Al Viro:

iov_iter: Fix iov_iter_get_pages{,_alloc} page fault return value


For immediate consideration by Paul Mackerras:

powerpc/kvm: Fix kvm_use_magic_page


Thanks,
Andreas

Andreas Gruenbacher (16):
iov_iter: Fix iov_iter_get_pages{,_alloc} page fault return value
powerpc/kvm: Fix kvm_use_magic_page
Turn fault_in_pages_{readable,writeable} into
fault_in_{readable,writeable}
Turn iov_iter_fault_in_readable into fault_in_iov_iter_readable
iov_iter: Introduce fault_in_iov_iter_writeable
gfs2: Add wrapper for iomap_file_buffered_write
gfs2: Clean up function may_grant
gfs2: Move the inode glock locking to gfs2_file_buffered_write
gfs2: Fix mmap + page fault deadlocks for buffered I/O
iomap: Fix iomap_dio_rw return value for user copies
iomap: Support partial direct I/O on user copy failures
iomap: Add done_before argument to iomap_dio_rw
gup: Introduce FOLL_NOFAULT flag to disable page faults
iov_iter: Introduce nofault flag to disable page faults
gfs2: Fix mmap + page fault deadlocks for direct I/O
gfs2: Eliminate ip->i_gh

Bob Peterson (3):
gfs2: Eliminate vestigial HIF_FIRST
gfs2: Remove redundant check from gfs2_glock_dq
gfs2: Introduce flag for glock holder auto-demotion

arch/powerpc/kernel/kvm.c | 3 +-
arch/powerpc/kernel/signal_32.c | 4 +-
arch/powerpc/kernel/signal_64.c | 2 +-
arch/x86/kernel/fpu/signal.c | 7 +-
drivers/gpu/drm/armada/armada_gem.c | 7 +-
fs/btrfs/file.c | 7 +-
fs/btrfs/ioctl.c | 5 +-
fs/ext4/file.c | 5 +-
fs/f2fs/file.c | 2 +-
fs/fuse/file.c | 2 +-
fs/gfs2/bmap.c | 60 +----
fs/gfs2/file.c | 244 ++++++++++++++++++--
fs/gfs2/glock.c | 340 +++++++++++++++++++++-------
fs/gfs2/glock.h | 20 ++
fs/gfs2/incore.h | 5 +-
fs/iomap/buffered-io.c | 2 +-
fs/iomap/direct-io.c | 21 +-
fs/ntfs/file.c | 2 +-
fs/xfs/xfs_file.c | 6 +-
fs/zonefs/super.c | 4 +-
include/linux/iomap.h | 11 +-
include/linux/mm.h | 3 +-
include/linux/pagemap.h | 58 +----
include/linux/uio.h | 4 +-
lib/iov_iter.c | 103 +++++++--
mm/filemap.c | 4 +-
mm/gup.c | 139 +++++++++++-
27 files changed, 784 insertions(+), 286 deletions(-)

--
2.26.3