On Thu, Sep 17, 2020 at 09:04:55AM +0100, Christoph Hellwig wrote: > On Wed, Sep 16, 2020 at 09:07:14AM -0400, Brian Foster wrote: > > Dave described the main purpose earlier in this thread [1]. The initial > > motivation is that we've had downstream reports of soft lockup problems > > in writeback bio completion down in the bio -> bvec loop of > > iomap_finish_ioend() that has to finish writeback on each individual > > page of insanely large bios and/or chains. We've also had an upstream > > reports of a similar problem on linux-xfs [2]. > > > > The magic number itself was just pulled out of a hat. I picked it > > because it seemed conservative enough to still allow large contiguous > > bios (1GB w/ 4k pages) while hopefully preventing I/O completion > > problems, but was hoping for some feedback on that bit if the general > > approach was acceptable. I was also waiting for some feedback on either > > of the two users who reported the problem but I don't think I've heard > > back on that yet... > > I think the saner answer is to always run large completions in the > workqueue, and add a bunch of cond_resched() calls, rather than > arbitrarily breaking up the I/O size. > That wouldn't address the latency concern Dave brought up. That said, I have no issue with this as a targeted solution for the softlockup issue. iomap_finish_ioend[s]() is common code for both the workqueue and ->bi_end_io() contexts so that would require either some kind of context detection (and my understanding is in_atomic() is unreliable/frowned upon) or a new "atomic" parameter through iomap_finish_ioend[s]() to indicate whether it's safe to reschedule. Preference? Brian
