Commit 56ba745d authored by Christian Borntraeger's avatar Christian Borntraeger Committed by Jiri Slaby

KVM: async_pf: do not warn on page allocation failures

commit d7444794 upstream.

In async_pf we try to allocate with NOWAIT to get an element quickly
or fail. This code also handle failures gracefully. Lets silence
potential page allocation failures under load.

qemu-system-s39: page allocation failure: order:0,mode:0x2200000
[...]
Call Trace:
([<00000000001146b8>] show_trace+0xf8/0x148)
[<000000000011476a>] show_stack+0x62/0xe8
[<00000000004a36b8>] dump_stack+0x70/0x98
[<0000000000272c3a>] warn_alloc_failed+0xd2/0x148
[<000000000027709e>] __alloc_pages_nodemask+0x94e/0xb38
[<00000000002cd36a>] new_slab+0x382/0x400
[<00000000002cf7ac>] ___slab_alloc.constprop.30+0x2dc/0x378
[<00000000002d03d0>] kmem_cache_alloc+0x160/0x1d0
[<0000000000133db4>] kvm_setup_async_pf+0x6c/0x198
[<000000000013dee8>] kvm_arch_vcpu_ioctl_run+0xd48/0xd58
[<000000000012fcaa>] kvm_vcpu_ioctl+0x372/0x690
[<00000000002f66f6>] do_vfs_ioctl+0x3be/0x510
[<00000000002f68ec>] SyS_ioctl+0xa4/0xb8
[<0000000000781c5e>] system_call+0xd6/0x264
[<000003ffa24fa06a>] 0x3ffa24fa06a
Signed-off-by: default avatarChristian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: default avatarDominik Dingel <dingel@linux.vnet.ibm.com>
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
Signed-off-by: default avatarJiri Slaby <jslaby@suse.cz>
parent 3fa6f32a
...@@ -161,7 +161,7 @@ int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, gfn_t gfn, ...@@ -161,7 +161,7 @@ int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, gfn_t gfn,
* do alloc nowait since if we are going to sleep anyway we * do alloc nowait since if we are going to sleep anyway we
* may as well sleep faulting in page * may as well sleep faulting in page
*/ */
work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT); work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT | __GFP_NOWARN);
if (!work) if (!work)
return 0; return 0;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment