diff options
| author | Chunwei Chen <[email protected]> | 2015-12-17 18:31:58 -0800 |
|---|---|---|
| committer | Brian Behlendorf <[email protected]> | 2015-12-18 13:24:52 -0800 |
| commit | b4ad50ac5f16de9220452f0da493e67c060d701b (patch) | |
| tree | 2242714bb50ea85f87792f859d20bbfefcbc9f7c /module/spl/spl-kmem-cache.c | |
| parent | 200366f23f1a16874d78a07936c5a33f2d488022 (diff) | |
Use spl_fstrans_mark instead of memalloc_noio_save
For earlier versions of the kernel with memalloc_noio_save, it only turns
off __GFP_IO but leaves __GFP_FS untouched during direct reclaim. This
would cause threads to direct reclaim into ZFS and cause deadlock.
Instead, we should stick to using spl_fstrans_mark. Since we would
explicitly turn off both __GFP_IO and __GFP_FS before allocation, it
will work on every version of the kernel.
This impacts kernel versions 3.9-3.17, see upstream kernel commit
torvalds/linux@934f307 for reference.
Signed-off-by: Chunwei Chen <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Signed-off-by: Tim Chase <[email protected]>
Closes #515
Issue zfsonlinux/zfs#4111
Diffstat (limited to 'module/spl/spl-kmem-cache.c')
| -rw-r--r-- | module/spl/spl-kmem-cache.c | 9 |
1 files changed, 2 insertions, 7 deletions
diff --git a/module/spl/spl-kmem-cache.c b/module/spl/spl-kmem-cache.c index 846d24300..e3edca5a0 100644 --- a/module/spl/spl-kmem-cache.c +++ b/module/spl/spl-kmem-cache.c @@ -201,7 +201,7 @@ kv_alloc(spl_kmem_cache_t *skc, int size, int flags) ASSERT(ISP2(size)); ptr = (void *)__get_free_pages(lflags, get_order(size)); } else { - ptr = spl_vmalloc(size, lflags | __GFP_HIGHMEM, PAGE_KERNEL); + ptr = __vmalloc(size, lflags | __GFP_HIGHMEM, PAGE_KERNEL); } /* Resulting allocated memory will be page aligned */ @@ -1156,15 +1156,10 @@ spl_cache_grow_work(void *data) spl_kmem_cache_t *skc = ska->ska_cache; spl_kmem_slab_t *sks; -#if defined(PF_MEMALLOC_NOIO) - unsigned noio_flag = memalloc_noio_save(); - sks = spl_slab_alloc(skc, ska->ska_flags); - memalloc_noio_restore(noio_flag); -#else fstrans_cookie_t cookie = spl_fstrans_mark(); sks = spl_slab_alloc(skc, ska->ska_flags); spl_fstrans_unmark(cookie); -#endif + spin_lock(&skc->skc_lock); if (sks) { skc->skc_slab_total++; |