Hi all,
this series ensures I/O is quiesced before a cpu and thus the managed
interrupt handler is shut down.
This patchset tries to address the issue by the following approach:
- before the last cpu in hctx->cpumask is going to offline, mark this
hctx as inactive
- disable preempt during allocating tag for request, and after tag is
allocated, check if this hctx is inactive. If yes, give up the
allocation and try remote allocation from online CPUs
- before hctx becomes inactive, drain all allocated requests on this
hctx
The guts of the changes are from Ming Lei, I just did a bunch of prep
cleanups so that they can fit in more nicely. The series also depends
on my "avoid a few q_usage_counter roundtrips v3" series.
Thanks John Garry for running lots of tests on arm64 with this previous
version patches and co-working on investigating all kinds of issues.
A git tree is available here:
git://git.infradead.org/users/hch/block.git blk-mq-hotplug.3
Gitweb:
http://git.infradead.org/users/hch/block.git/shortlog/refs/heads/blk-mq-hotplug.3
Changes since v4:
- move the blk_mq_all_tag_iter kerneldoc comment to make it clear what is
documented
- add a WARN_ON_ONCE
- use BLK_MQ_NO_TAG in one more place instead of -1
- remove the PF_KTHREAD special casing
Changes since v3:
- add two new patches to clean up the magic -1 tag values
- improve a few commit messages and comments
- cleanup the blk_mq_all_tag_iter implementation
- add a msleep to the cpu hot unplug case in __blk_mq_alloc_request
Changes since v2:
- don't disable preemption and use smp calls
