| Message ID | 20191213201340.7ekvehdt3olwpk4x@gallifrey |
|---|---|
| State | New |
| Headers | show |
| Series | [bionic:linux,PULL] LP:#1848739 - [linux] Patch to prevent possible data corruption | expand |
On 13.12.19 21:13, Marcelo Henrique Cerri wrote: > BugLink: https://bugs.launchpad.net/bugs/1848739 > > This patch set was already included into the 4.15 xenial:linux-azure > kernel, but it affects all our 4.15 kernels and the same issue was > already reported for the 4.15 linux-gcp kernel. > > The patch set was already tested against linux-gcp, as described on > the LP bug, with positive results. Since we have positive feedbacks > with both linux-azure and linux-gcp, the risk of regression for the > inclusion on the 4.15 master kernel should be small. > > --- > > The following changes since commit bdefbb8e170ff68940ffcaec6456ba3986cb0ce7: > > selftests/net: correct the return value for run_afpackettests (2019-12-10 15:30:55 +0100) > > are available in the Git repository at: > > git://git.launchpad.net/~mhcerri/ubuntu/+source/linux/+git/bionic lp1848739 > > for you to fetch changes up to f249c87a984b9f638fac9af5ed48d800ed107971: > > blk-mq: punt failed direct issue to dispatch list (2019-12-13 15:18:16 -0300) > > ---------------------------------------------------------------- > Bart Van Assche (1): > blk-mq: Rename blk_mq_request_direct_issue() into blk_mq_request_issue_directly() > > Jens Axboe (4): > blk-mq: move hctx lock/unlock into a helper > blk-mq: don't queue more if we get a busy return > blk-mq: fix corruption with direct issue > blk-mq: punt failed direct issue to dispatch list > > Mike Snitzer (2): > blk-mq: factor out a few helpers from __blk_mq_try_issue_directly > blk-mq-sched: remove unused 'can_block' arg from blk_mq_sched_insert_request > > Ming Lei (7): > blk-mq: quiesce queue during switching io sched and updating nr_requests > blk-mq: improve DM's blk-mq IO merging via blk_insert_cloned_request feedback > blk-mq: don't dispatch request in blk_mq_request_direct_issue if queue is busy > blk-mq: introduce BLK_STS_DEV_RESOURCE > blk-mq: dequeue request one by one from sw queue if hctx is busy > blk-mq: issue directly if hw queue isn't busy in case of 'none' > blk-mq: fail the request in case issue failure > > Yufen Yu (1): > dm mpath: fix missing call of path selector type->end_io > > block/blk-core.c | 4 +- > block/blk-exec.c | 2 +- > block/blk-mq-debugfs.c | 9 ++ > block/blk-mq-sched.c | 26 +++-- > block/blk-mq-sched.h | 2 +- > block/blk-mq.c | 258 +++++++++++++++++++++++++++++++++---------- > block/blk-mq.h | 5 + > block/elevator.c | 2 + > drivers/block/null_blk.c | 2 +- > drivers/block/virtio_blk.c | 2 +- > drivers/block/xen-blkfront.c | 2 +- > drivers/md/dm-rq.c | 20 +++- > drivers/nvme/host/fc.c | 12 +- > drivers/scsi/scsi_lib.c | 6 +- > include/linux/blk-mq.h | 3 +- > include/linux/blk_types.h | 18 +++ > 16 files changed, 281 insertions(+), 92 deletions(-) > > Isolated to specific usage and tested. Acked-by: Stefan Bader <stefan.bader@canonical.com>
On 2019-12-13 21:13, Marcelo Henrique Cerri wrote: > BugLink: https://bugs.launchpad.net/bugs/1848739 > > This patch set was already included into the 4.15 xenial:linux-azure > kernel, but it affects all our 4.15 kernels and the same issue was > already reported for the 4.15 linux-gcp kernel. > > The patch set was already tested against linux-gcp, as described on > the LP bug, with positive results. Since we have positive feedbacks > with both linux-azure and linux-gcp, the risk of regression for the > inclusion on the 4.15 master kernel should be small. > > --- > > The following changes since commit bdefbb8e170ff68940ffcaec6456ba3986cb0ce7: > > selftests/net: correct the return value for run_afpackettests (2019-12-10 15:30:55 +0100) > > are available in the Git repository at: > > git://git.launchpad.net/~mhcerri/ubuntu/+source/linux/+git/bionic lp1848739 > > for you to fetch changes up to f249c87a984b9f638fac9af5ed48d800ed107971: > > blk-mq: punt failed direct issue to dispatch list (2019-12-13 15:18:16 -0300) Changes already running on some cloud derivatives for a while and with good test results. So it seems relatively safe to apply the changes to the main bionic kernel. Acked-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com> > > ---------------------------------------------------------------- > Bart Van Assche (1): > blk-mq: Rename blk_mq_request_direct_issue() into blk_mq_request_issue_directly() > > Jens Axboe (4): > blk-mq: move hctx lock/unlock into a helper > blk-mq: don't queue more if we get a busy return > blk-mq: fix corruption with direct issue > blk-mq: punt failed direct issue to dispatch list > > Mike Snitzer (2): > blk-mq: factor out a few helpers from __blk_mq_try_issue_directly > blk-mq-sched: remove unused 'can_block' arg from blk_mq_sched_insert_request > > Ming Lei (7): > blk-mq: quiesce queue during switching io sched and updating nr_requests > blk-mq: improve DM's blk-mq IO merging via blk_insert_cloned_request feedback > blk-mq: don't dispatch request in blk_mq_request_direct_issue if queue is busy > blk-mq: introduce BLK_STS_DEV_RESOURCE > blk-mq: dequeue request one by one from sw queue if hctx is busy > blk-mq: issue directly if hw queue isn't busy in case of 'none' > blk-mq: fail the request in case issue failure > > Yufen Yu (1): > dm mpath: fix missing call of path selector type->end_io > > block/blk-core.c | 4 +- > block/blk-exec.c | 2 +- > block/blk-mq-debugfs.c | 9 ++ > block/blk-mq-sched.c | 26 +++-- > block/blk-mq-sched.h | 2 +- > block/blk-mq.c | 258 +++++++++++++++++++++++++++++++++---------- > block/blk-mq.h | 5 + > block/elevator.c | 2 + > drivers/block/null_blk.c | 2 +- > drivers/block/virtio_blk.c | 2 +- > drivers/block/xen-blkfront.c | 2 +- > drivers/md/dm-rq.c | 20 +++- > drivers/nvme/host/fc.c | 12 +- > drivers/scsi/scsi_lib.c | 6 +- > include/linux/blk-mq.h | 3 +- > include/linux/blk_types.h | 18 +++ > 16 files changed, 281 insertions(+), 92 deletions(-) > >
On 2019-12-13 21:13, Marcelo Henrique Cerri wrote: > BugLink: https://bugs.launchpad.net/bugs/1848739 > > This patch set was already included into the 4.15 xenial:linux-azure > kernel, but it affects all our 4.15 kernels and the same issue was > already reported for the 4.15 linux-gcp kernel. > > The patch set was already tested against linux-gcp, as described on > the LP bug, with positive results. Since we have positive feedbacks > with both linux-azure and linux-gcp, the risk of regression for the > inclusion on the 4.15 master kernel should be small. > > --- > > The following changes since commit bdefbb8e170ff68940ffcaec6456ba3986cb0ce7: > > selftests/net: correct the return value for run_afpackettests (2019-12-10 15:30:55 +0100) > > are available in the Git repository at: > > git://git.launchpad.net/~mhcerri/ubuntu/+source/linux/+git/bionic lp1848739 > > for you to fetch changes up to f249c87a984b9f638fac9af5ed48d800ed107971: > > blk-mq: punt failed direct issue to dispatch list (2019-12-13 15:18:16 -0300) Applied to bionic/linux. Thanks, Kleber > > ---------------------------------------------------------------- > Bart Van Assche (1): > blk-mq: Rename blk_mq_request_direct_issue() into blk_mq_request_issue_directly() > > Jens Axboe (4): > blk-mq: move hctx lock/unlock into a helper > blk-mq: don't queue more if we get a busy return > blk-mq: fix corruption with direct issue > blk-mq: punt failed direct issue to dispatch list > > Mike Snitzer (2): > blk-mq: factor out a few helpers from __blk_mq_try_issue_directly > blk-mq-sched: remove unused 'can_block' arg from blk_mq_sched_insert_request > > Ming Lei (7): > blk-mq: quiesce queue during switching io sched and updating nr_requests > blk-mq: improve DM's blk-mq IO merging via blk_insert_cloned_request feedback > blk-mq: don't dispatch request in blk_mq_request_direct_issue if queue is busy > blk-mq: introduce BLK_STS_DEV_RESOURCE > blk-mq: dequeue request one by one from sw queue if hctx is busy > blk-mq: issue directly if hw queue isn't busy in case of 'none' > blk-mq: fail the request in case issue failure > > Yufen Yu (1): > dm mpath: fix missing call of path selector type->end_io > > block/blk-core.c | 4 +- > block/blk-exec.c | 2 +- > block/blk-mq-debugfs.c | 9 ++ > block/blk-mq-sched.c | 26 +++-- > block/blk-mq-sched.h | 2 +- > block/blk-mq.c | 258 +++++++++++++++++++++++++++++++++---------- > block/blk-mq.h | 5 + > block/elevator.c | 2 + > drivers/block/null_blk.c | 2 +- > drivers/block/virtio_blk.c | 2 +- > drivers/block/xen-blkfront.c | 2 +- > drivers/md/dm-rq.c | 20 +++- > drivers/nvme/host/fc.c | 12 +- > drivers/scsi/scsi_lib.c | 6 +- > include/linux/blk-mq.h | 3 +- > include/linux/blk_types.h | 18 +++ > 16 files changed, 281 insertions(+), 92 deletions(-) > >
BugLink: https://bugs.launchpad.net/bugs/1848739 This patch set was already included into the 4.15 xenial:linux-azure kernel, but it affects all our 4.15 kernels and the same issue was already reported for the 4.15 linux-gcp kernel. The patch set was already tested against linux-gcp, as described on the LP bug, with positive results. Since we have positive feedbacks with both linux-azure and linux-gcp, the risk of regression for the inclusion on the 4.15 master kernel should be small. --- The following changes since commit bdefbb8e170ff68940ffcaec6456ba3986cb0ce7: selftests/net: correct the return value for run_afpackettests (2019-12-10 15:30:55 +0100) are available in the Git repository at: git://git.launchpad.net/~mhcerri/ubuntu/+source/linux/+git/bionic lp1848739 for you to fetch changes up to f249c87a984b9f638fac9af5ed48d800ed107971: blk-mq: punt failed direct issue to dispatch list (2019-12-13 15:18:16 -0300) ---------------------------------------------------------------- Bart Van Assche (1): blk-mq: Rename blk_mq_request_direct_issue() into blk_mq_request_issue_directly() Jens Axboe (4): blk-mq: move hctx lock/unlock into a helper blk-mq: don't queue more if we get a busy return blk-mq: fix corruption with direct issue blk-mq: punt failed direct issue to dispatch list Mike Snitzer (2): blk-mq: factor out a few helpers from __blk_mq_try_issue_directly blk-mq-sched: remove unused 'can_block' arg from blk_mq_sched_insert_request Ming Lei (7): blk-mq: quiesce queue during switching io sched and updating nr_requests blk-mq: improve DM's blk-mq IO merging via blk_insert_cloned_request feedback blk-mq: don't dispatch request in blk_mq_request_direct_issue if queue is busy blk-mq: introduce BLK_STS_DEV_RESOURCE blk-mq: dequeue request one by one from sw queue if hctx is busy blk-mq: issue directly if hw queue isn't busy in case of 'none' blk-mq: fail the request in case issue failure Yufen Yu (1): dm mpath: fix missing call of path selector type->end_io block/blk-core.c | 4 +- block/blk-exec.c | 2 +- block/blk-mq-debugfs.c | 9 ++ block/blk-mq-sched.c | 26 +++-- block/blk-mq-sched.h | 2 +- block/blk-mq.c | 258 +++++++++++++++++++++++++++++++++---------- block/blk-mq.h | 5 + block/elevator.c | 2 + drivers/block/null_blk.c | 2 +- drivers/block/virtio_blk.c | 2 +- drivers/block/xen-blkfront.c | 2 +- drivers/md/dm-rq.c | 20 +++- drivers/nvme/host/fc.c | 12 +- drivers/scsi/scsi_lib.c | 6 +- include/linux/blk-mq.h | 3 +- include/linux/blk_types.h | 18 +++ 16 files changed, 281 insertions(+), 92 deletions(-)