Re: [RFC 1/6] mm, page_alloc: fix more premature OOM due to race with cpuset update

From: Vlastimil Babka
Date: Thu Apr 13 2017 - 02:06:43 EST


On 04/13/2017 07:42 AM, Anshuman Khandual wrote:
> On 04/11/2017 07:36 PM, Vlastimil Babka wrote:
>> Commit e47483bca2cc ("mm, page_alloc: fix premature OOM when racing with cpuset
>> mems update") has fixed known recent regressions found by LTP's cpuset01
>> testcase. I have however found that by modifying the testcase to use per-vma
>> mempolicies via bind(2) instead of per-task mempolicies via set_mempolicy(2),
>> the premature OOM still happens and the issue is much older.
>
> Meanwhile while we are discussing this RFC, will it be better to WARN
> out these situations where we dont have node in the intersection,
> hence no usable zone during allocation. That might actually give
> a hint to the user before a premature OOM/allocation failure comes.

Well, the bug is very old and nobody reported it so far, AFAIK. So it's
not that urgent.

>>
>> The root of the problem is that the cpuset's mems_allowed and mempolicy's
>> nodemask can temporarily have no intersection, thus get_page_from_freelist()
>> cannot find any usable zone. The current semantic for empty intersection is to
>> ignore mempolicy's nodemask and honour cpuset restrictions. This is checked in
>> node_zonelist(), but the racy update can happen after we already passed the
>> check. Such races should be protected by the seqlock task->mems_allowed_seq,
>> but it doesn't work here, because 1) mpol_rebind_mm() does not happen under
>> seqlock for write, and doing so would lead to deadlock, as it takes mmap_sem
>> for write, while the allocation can have mmap_sem for read when it's taking the
>> seqlock for read. And 2) the seqlock cookie of callers of node_zonelist()
>> (alloc_pages_vma() and alloc_pages_current()) is different than the one of
>> __alloc_pages_slowpath(), so there's still a potential race window.
>>
>> This patch fixes the issue by having __alloc_pages_slowpath() check for empty
>> intersection of cpuset and ac->nodemask before OOM or allocation failure. If
>> it's indeed empty, the nodemask is ignored and allocation retried, which mimics
>> node_zonelist(). This works fine, because almost all callers of
>> __alloc_pages_nodemask are obtaining the nodemask via node_zonelist(). The only
>> exception is new_node_page() from hotplug, where the potential violation of
>> nodemask isn't an issue, as there's already a fallback allocation attempt
>> without any nodemask. If there's a future caller that needs to have its specific
>> nodemask honoured over task's cpuset restrictions, we'll have to e.g. add a gfp
>> flag for that.
>
> Did you really mean node_zonelist() in both the instances above. Because
> that function just picks up either FALLBACK_ZONELIST or NOFALLBACK_ZONELIST
> depending upon the passed GFP flags in the allocation request and does not
> deal with ignoring the passed nodemask.

Oops, I meant policy_zonelist(), thanks for noticing.