Re: [PATCH v7 9/9] block: Avoid deadlocks with bio allocation by stacking drivers

On Wed, Aug 29, 2012 at 09:50:06AM -0700, Kent Overstreet wrote:

[..]

quoted

The problem is that majority of device mapper code assumes that if we 
submit a bio, that bio will be finished in a finite time. The commit 
d89d87965dcbe6fe4f96a2a7e8421b3a75f634d1 in 2.6.22 broke this assumption.

I suggest - instead of writing workarounds for this current->bio_list 
misbehavior, why not remove current->bio_list at all? We could revert 
d89d87965dcbe6fe4f96a2a7e8421b3a75f634d1, allocate a per-device workqueue, 
test stack usage in generic_make_request, and if it is too high (more than 
half of the stack used, or so), put the bio to the target device's 
blockqueue.

That could be simpler than allocating per-bioset workqueue and it also 
solves more problems (possible deadlocks in dm).

It certainly would be simpler, but honestly the potential for
performance regressions scares me (and bcache at least is used on fast
enough devices where it's going to matter). Also it's not so much the
performance overhead - we can just measure that - it's that if we're
just using the workqueue code the scheduler's getting involved and we
can't just measure what the effects of that are going to be in
production.

Are workqueues not getting involved already in your solution of punting
to rescuer thread. In the proposal above also, workers get involved
only if stack depth is too deep. So for normal stack usage performance
should not be impacted.

Performance aside, punting submission to per device worker in case of deep
stack usage sounds cleaner solution to me.

Thanks
Vivek