]> git.proxmox.com Git - mirror_ubuntu-kernels.git/commit
fs, seq_file: fallback to vmalloc instead of oom kill processes
authorDavid Rientjes <rientjes@google.com>
Sat, 13 Dec 2014 00:56:16 +0000 (16:56 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 13 Dec 2014 20:42:49 +0000 (12:42 -0800)
commit5cec38ac866bfb8775638e71a86e4d8cac30caae
treedbc2549883c7561640c64a72d9acdfe17732e543
parent6b4f7799c6a5703ac6b8c0649f4c22f00fa07513
fs, seq_file: fallback to vmalloc instead of oom kill processes

Since commit 058504edd026 ("fs/seq_file: fallback to vmalloc allocation"),
seq_buf_alloc() falls back to vmalloc() when the kmalloc() for contiguous
memory fails.  This was done to address order-4 slab allocations for
reading /proc/stat on large machines and noticed because
PAGE_ALLOC_COSTLY_ORDER < 4, so there is no infinite loop in the page
allocator when allocating new slab for such high-order allocations.

Contiguous memory isn't necessary for caller of seq_buf_alloc(), however.
Other GFP_KERNEL high-order allocations that are <=
PAGE_ALLOC_COSTLY_ORDER will simply loop forever in the page allocator and
oom kill processes as a result.

We don't want to kill processes so that we can allocate contiguous memory
in situations when contiguous memory isn't necessary.

This patch does the kmalloc() allocation with __GFP_NORETRY for high-order
allocations.  This still utilizes memory compaction and direct reclaim in
the allocation path, the only difference is that it will fail immediately
instead of oom kill processes when out of memory.

[akpm@linux-foundation.org: add comment]
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/seq_file.c