diff options
author | Brian Behlendorf <[email protected]> | 2011-03-20 15:03:18 -0700 |
---|---|---|
committer | Brian Behlendorf <[email protected]> | 2011-03-20 15:12:08 -0700 |
commit | 2092cf68d89a51eb0d6193aeadabb579dfc4b4a0 (patch) | |
tree | 832474eb5e585d4853b6a2b6015e2826917efc27 /module/spl/spl-kmem.c | |
parent | cb255ae57278be3206693b5116eff95e612e809a (diff) |
Disable vmalloc() direct reclaim
As part of vmalloc() a __pte_alloc_kernel() allocation may occur. This
internal allocation does not honor the gfp flags passed to vmalloc().
This means even when vmalloc(GFP_NOFS) is called it is possible that a
synchronous reclaim will occur. This reclaim can trigger file IO which
can result in a deadlock. This issue can be avoided by explicitly
setting PF_MEMALLOC on the process to subvert synchronous reclaim when
vmalloc() is called with !__GFP_FS.
An example stack of the deadlock can be found here (1), along with the
upstream kernel bug (2), and the original bug discussion on the
linux-mm mailing list (3). This code can be properly autoconf'ed
when the upstream bug is fixed.
1) http://github.com/behlendorf/zfs/issues/labels/Vmalloc#issue/133
2) http://bugzilla.kernel.org/show_bug.cgi?id=30702
3) http://marc.info/?l=linux-mm&m=128942194520631&w=4
Diffstat (limited to 'module/spl/spl-kmem.c')
-rw-r--r-- | module/spl/spl-kmem.c | 24 |
1 files changed, 22 insertions, 2 deletions
diff --git a/module/spl/spl-kmem.c b/module/spl/spl-kmem.c index 100d674e5..2f533bdf6 100644 --- a/module/spl/spl-kmem.c +++ b/module/spl/spl-kmem.c @@ -842,11 +842,31 @@ kv_alloc(spl_kmem_cache_t *skc, int size, int flags) ASSERT(ISP2(size)); - if (skc->skc_flags & KMC_KMEM) + if (skc->skc_flags & KMC_KMEM) { ptr = (void *)__get_free_pages(flags, get_order(size)); - else + } else { + /* + * As part of vmalloc() an __pte_alloc_kernel() allocation + * may occur. This internal allocation does not honor the + * gfp flags passed to vmalloc(). This means even when + * vmalloc(GFP_NOFS) is called it is possible synchronous + * reclaim will occur. This reclaim can trigger file IO + * which can result in a deadlock. This issue can be avoided + * by explicitly setting PF_MEMALLOC on the process to + * subvert synchronous reclaim. The following bug has + * been filed at kernel.org to track the issue. + * + * https://bugzilla.kernel.org/show_bug.cgi?id=30702 + */ + if (!(flags & __GFP_FS)) + current->flags |= PF_MEMALLOC; + ptr = __vmalloc(size, flags | __GFP_HIGHMEM, PAGE_KERNEL); + if (!(flags & __GFP_FS)) + current->flags &= ~PF_MEMALLOC; + } + /* Resulting allocated memory will be page aligned */ ASSERT(IS_P2ALIGNED(ptr, PAGE_SIZE)); |