[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <609f2de6-b91f-0018-cf98-18524a34b5d2@oracle.com>
Date: Fri, 24 Jul 2020 10:41:14 -0700
From: Mike Kravetz <mike.kravetz@...cle.com>
To: Muchun Song <songmuchun@...edance.com>, akpm@...ux-foundation.org,
mhocko@...nel.org
Cc: rientjes@...gle.com, mgorman@...e.de, walken@...gle.com,
linux-mm@...ck.org, linux-kernel@...r.kernel.org,
Jianchao Guo <guojianchao@...edance.com>
Subject: Re: [PATCH v2] mm/hugetlb: add mempolicy check in the reservation
routine
On 7/24/20 3:03 AM, Muchun Song wrote:
> In the reservation routine, we only check whether the cpuset meets
> the memory allocation requirements. But we ignore the mempolicy of
> MPOL_BIND case. If someone mmap hugetlb succeeds, but the subsequent
> memory allocation may fail due to mempolicy restrictions and receives
> the SIGBUS signal. This can be reproduced by the follow steps.
>
> 1) Compile the test case.
> cd tools/testing/selftests/vm/
> gcc map_hugetlb.c -o map_hugetlb
>
> 2) Pre-allocate huge pages. Suppose there are 2 numa nodes in the
> system. Each node will pre-allocate one huge page.
> echo 2 > /proc/sys/vm/nr_hugepages
>
> 3) Run test case(mmap 4MB). We receive the SIGBUS signal.
> numactl --membind=0 ./map_hugetlb 4
>
> With this patch applied, the mmap will fail in the step 3) and throw
> "mmap: Cannot allocate memory".
>
> Reported-by: Jianchao Guo <guojianchao@...edance.com>
> Signed-off-by: Muchun Song <songmuchun@...edance.com>
Michal has already provides some good suggestions.
> @@ -3653,7 +3666,7 @@ static int hugetlb_acct_memory(struct hstate *h, long delta)
> if (gather_surplus_pages(h, delta) < 0)
> goto out;
>
> - if (delta > cpuset_mems_nr(h->free_huge_pages_node)) {
> + if (delta > allowed_mems_nr(h)) {
> return_unused_surplus_pages(h, delta);
> goto out;
> }
There is a big comment before this code in hugetlb_acct_memory. The comment
only talks about cpusets. We should probably update that to include mempolicy
as well. It could be as simple as s/cpuset/cpuset or mempolicy/.
--
Mike Kravetz
Powered by blists - more mailing lists