On Thu 31-05-18 19:04:46, Qing Huang wrote:
GFP_ATOMIC is a non-sleeping (aka no reclaim) context with an access to
On 5/31/2018 2:10 AM, Michal Hocko wrote:
On Thu 31-05-18 10:55:32, Michal Hocko wrote:Just curious, how about GFP_ATOMIC flag? Would it work in a similar fashion?
On Thu 31-05-18 04:35:31, Eric Dumazet wrote:[...]
OK, so this is an example of an incremental development ;).I merely copied/pasted from alloc_skb_with_frags() :/I will have a look at it. Thanks!
__GFP_NORETRY was added by ed98df3361f0 ("net: use __GFP_NORETRY for
high order allocations") to prevent from OOM killer. Yet this was
not enough because fb05e7a89f50 ("net: don't wait for order-3 page
allocation") didn't want an excessive reclaim for non-costly orders
so it made it completely NOWAIT while it preserved __GFP_NORETRY in
place which is now redundant. Should I send a patch?
We experimented
with it a bit in the past but it seemed to cause other issue in our tests.
:-)
memory reserves. So the risk is that you deplete those reserves and
cause issues to other subsystems which need them as well.
By the way, we didn't encounter any OOM killer events. It seemed that theThe compaction was able to make a reasonable forward progress for you.
mlx4_alloc_icm() triggered slowpath.
We still had about 2GB free memory while it was highly fragmented.
But considering mlx4_alloc_icm is called with GFP_KERNEL resp. GFP_HIGHUSER
then the OOM killer is clearly possible as long as the order is lower
than 4.
Â#0 [ffff8801f308b380] remove_migration_pte at ffffffff811f0e0b
Â#1 [ffff8801f308b3e0] rmap_walk_file at ffffffff811cb890
Â#2 [ffff8801f308b440] rmap_walk at ffffffff811cbaf2
Â#3 [ffff8801f308b450] remove_migration_ptes at ffffffff811f0db0
Â#4 [ffff8801f308b490] __unmap_and_move at ffffffff811f2ea6
Â#5 [ffff8801f308b4e0] unmap_and_move at ffffffff811f2fc5
Â#6 [ffff8801f308b540] migrate_pages at ffffffff811f3219
Â#7 [ffff8801f308b5c0] compact_zone at ffffffff811b707e
Â#8 [ffff8801f308b650] compact_zone_order at ffffffff811b735d
Â#9 [ffff8801f308b6e0] try_to_compact_pages at ffffffff811b7485
#10 [ffff8801f308b770] __alloc_pages_direct_compact at ffffffff81195f96
#11 [ffff8801f308b7b0] __alloc_pages_slowpath at ffffffff811978a1
#12 [ffff8801f308b890] __alloc_pages_nodemask at ffffffff81197ec1
#13 [ffff8801f308b970] alloc_pages_current at ffffffff811e261f
#14 [ffff8801f308b9e0] mlx4_alloc_icm at ffffffffa01f39b2 [mlx4_core]
Thanks!