On 26/03/20 21:08, Thomas Voegtle wrote:
On Thu, 26 Mar 2020, Paolo Bonzini wrote:
On 26/03/20 18:43, Thomas Voegtle wrote:
Hello,
the following one line commit
commit 7df003c85218b5f5b10a7f6418208f31e813f38f Author: Zhuang Yanying ann.zhuangyanying@huawei.com Date: Sat Oct 12 11:37:31 2019 +0800
KVM: fix overflow of zero page refcount with ksm running
applies cleanly to v5.5.y, v5.4.y, 4.19.y, v4.14.y and v4.9.y.
I actually ran into that bug on 4.9.y
Thanks in advance,
Thomas
Yes, indeed. It's not a trivial backport though, so I prefer to do it manually. I can help with that, or with reviews if Yanying already has patches ready.
Are you sure we are talking about the same commit? Or do I really see something wrong here? It's an one liner. Applies cleanly. Compiles. Don't know if it fixes the bug, though.
It's a one liner, but usage of the function it touches has changed subtly over the years. So before committing it to stable I need to check that all uses of kvm_is_reserved_pfn match 5.6 (and also kvm_is_zone_device_pfn is probably not there in older kernels, though I might be missing more stable backports there).
Paolo