Hi Linus,
could you please pull the following two commits to fix the gfs2
deadlocks resulting from the conversion to ->readahead in commit
d4388340ae0b ("fs: convert mpage_readpages to mpage_readahead")?
The first commit adds a new IOCB_NOIO flag to generic_file_read_iter.
In the previous version [1] which you've acked [2] and Matthew Willcox
has reviewed [3], ->readpage could still be called even when IOCB_NOIO
was set, so I've added an additional check above that call and I've
dropped the ack and reviewed-by tags. In addition, bit 8 is now left
unused for the new IOCB_WAITQ flag in the block tree per Jens Axboe's
request.
[1]
https://lore.kernel.org/linux-fsdevel/cahc6fu6lmr7m_8uhmb_77jupyno-kgcz-1ytlqya-ppqvvb...@mail.gmail.com/
[2]
https://lore.kernel.org/linux-fsdevel/CAHk-=whbk-jym6_hbxbu6+gs7gtw3hwg4islncq0qtwshm6...@mail.gmail.com/
[3]
https://lore.kernel.org/linux-fsdevel/20200703114108.ge25...@casper.infradead.org/
Thanks a lot,
Andreas
The following changes since commit dcb7fd82c75ee2d6e6f9d8cc71c52519ed52e258:
Linux 5.8-rc4 (2020-07-05 16:20:22 -0700)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git
tags/gfs2-v5.8-rc4.fixes
for you to fetch changes up to 20f82c38b18e3d17f9e40dea3a28f721fac4:
gfs2: Rework read and page fault locking (2020-07-07 23:40:12 +0200)
Fix gfs2 readahead deadlocks
Andreas Gruenbacher (2):
fs: Add IOCB_NOIO flag for generic_file_read_iter
gfs2: Rework read and page fault locking
fs/gfs2/aops.c | 45 +
fs/gfs2/file.c | 52 ++--
include/linux/fs.h | 1 +
mm/filemap.c | 23 +--
4 files changed, 73 insertions(+), 48 deletions(-)