net: use kvmalloc with __GFP_REPEAT rather than open coded variant
authorMichal Hocko <mhocko@suse.com>
Mon, 8 May 2017 22:57:31 +0000 (15:57 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 9 May 2017 00:15:13 +0000 (17:15 -0700)
commitda6bc57a8f02dd90d07071b4cd067f2de26c9192
tree356110f1f8791a2a4b56988f7a5fa473a58f56b3
parent752ade68cbd81d0321dfecc188f655a945551b25
net: use kvmalloc with __GFP_REPEAT rather than open coded variant

fq_alloc_node, alloc_netdev_mqs and netif_alloc* open code kmalloc with
vmalloc fallback.  Use the kvmalloc variant instead.  Keep the
__GFP_REPEAT flag based on explanation from Eric:

 "At the time, tests on the hardware I had in my labs showed that
  vmalloc() could deliver pages spread all over the memory and that was
  a small penalty (once memory is fragmented enough, not at boot time)"

The way how the code is constructed means, however, that we prefer to go
and hit the OOM killer before we fall back to the vmalloc for requests
<=32kB (with 4kB pages) in the current code.  This is rather disruptive
for something that can be achived with the fallback.  On the other hand
__GFP_REPEAT doesn't have any useful semantic for these requests.  So
the effect of this patch is that requests which fit into 32kB will fall
back to vmalloc easier now.

Link: http://lkml.kernel.org/r/20170306103327.2766-3-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Eric Dumazet <edumazet@google.com>
Cc: David Miller <davem@davemloft.net>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
net/core/dev.c
net/sched/sch_fq.c