On Mon, Jun 14, 2021 at 10:24:23AM +0530, Naresh Kamboju wrote: > On Sun, 13 Jun 2021 at 20:35, Greg Kroah-Hartman > <gregkh@xxxxxxxxxxxxxxxxxxx> wrote: > > > > On Sun, Jun 13, 2021 at 08:25:19PM +0530, Naresh Kamboju wrote: > > > The following error was noticed on stable-rc 5.12, 5.10, 5.4, 4.19, > > > 4.14, 4.9 and 4.4 > > > for i386 and arm. > > > > > > make --silent --keep-going --jobs=8 > > > O=/home/tuxbuild/.cache/tuxmake/builds/current ARCH=arm > > > CROSS_COMPILE=arm-linux-gnueabihf- 'CC=sccache > > > arm-linux-gnueabihf-gcc' 'HOSTCC=sccache gcc' > > > In file included from /builds/linux/include/linux/kernel.h:11, > > > from /builds/linux/include/linux/list.h:9, > > > from /builds/linux/include/linux/preempt.h:11, > > > from /builds/linux/include/linux/hardirq.h:5, > > > from /builds/linux/include/linux/kvm_host.h:7, > > > from > > > /builds/linux/arch/arm/kvm/../../../virt/kvm/kvm_main.c:18: > > > In function '__gfn_to_hva_memslot', > > > inlined from '__gfn_to_hva_many.part.6' at > > > /builds/linux/arch/arm/kvm/../../../virt/kvm/kvm_main.c:1446:9, > > > inlined from '__gfn_to_hva_many' at > > > /builds/linux/arch/arm/kvm/../../../virt/kvm/kvm_main.c:1434:22: > > > /builds/linux/include/linux/compiler.h:417:38: error: call to > > > '__compiletime_assert_59' declared with attribute error: BUILD_BUG_ON > > > failed: sizeof(_i) > sizeof(long) > > > _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__) > > > ^ > > > Reported-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx> > > > > > > ref: > > > https://gitlab.com/Linaro/lkft/mirrors/stable/linux-stable-rc/-/jobs/1342604370#L389 > > > > Odd. Does Linus's tree have this problem? > > > > The only arm changes was in arch/arm/include/asm/cpuidle.h in the tree > > right now. There are some kvm changes, but they are tiny... > > > > Can you bisect this? > > The bisect script pointing to, > > commit 1aa1b47db53e0a66899d63103b3ac1d7f54816bc > Author: Paolo Bonzini <pbonzini@xxxxxxxxxx> > Date: Tue Jun 8 15:31:42 2021 -0400 > kvm: avoid speculation-based attacks from out-of-range memslot accesses > > commit da27a83fd6cc7780fea190e1f5c19e87019da65c upstream. Ah, so is Linus's tree also broken the same way? thanks, greg k-h
