Messages in this thread | | | Subject | Re: [PATCH] mm: Fix kswapd livelock on single core, no preempt kernel | From | James Bottomley <> | Date | Wed, 14 Dec 2011 14:51:16 +0400 |
| |
On Tue, 2011-12-13 at 09:44 -0800, Mike Waychison wrote: > On a single core system with kernel preemption disabled, it is possible > for the memory system to be so taxed that kswapd cannot make any forward > progress. This can happen when most of system memory is tied up as > anonymous memory without swap enabled, causing kswapd to consistently > fail to achieve its watermark goals. In turn, sleeping_prematurely() > will consistently return true and kswapd_try_to_sleep() to never invoke > schedule(). This causes the kswapd thread to stay on the CPU in > perpetuity and keeps other threads from processing oom-kills to reclaim > memory. > > The cond_resched() instance in balance_pgdat() is never called as the > loop that iterates from DEF_PRIORITY down to 0 will always set > all_zones_ok to true, and not set it to false once we've passed > DEF_PRIORITY as zones that are marked ->all_unreclaimable are not > considered in the "all_zones_ok" evaluation. > > This change modifies kswapd_try_to_sleep to ensure that we enter > scheduler at least once per invocation if needed. This allows kswapd to > get off the CPU and allows other threads to die off from the OOM killer > (freeing memory that is otherwise unavailable in the process).
This keeps cropping up. I think it's not the same as the last time I saw it (which was on a multi-core system) but it was definitely caused by an issue with sleeping_prematurely(). For reference, this is the thread:
http://marc.info/?t=130436700400001
And this was the eventual fix that worked for me:
http://marc.info/?t=130892304300003
James
| |