]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
arm/arm64: KVM: Keep elrsr/aisr in sync with software model
authorChristoffer Dall <christoffer.dall@linaro.org>
Mon, 4 May 2015 01:25:26 +0000 (09:25 +0800)
committerSasha Levin <sasha.levin@oracle.com>
Mon, 11 May 2015 11:07:37 +0000 (07:07 -0400)
commit5c0ac4b58b3a99ff1c624b799f416c2afc31e845
treef799c9d28bf63037a5a3690893e30c8338156a78
parent74e0fedea4b7dd63f6955835f3e7455fdfc64ea7
arm/arm64: KVM: Keep elrsr/aisr in sync with software model

commit ae705930fca6322600690df9dc1c7d0516145a93 upstream.

There is an interesting bug in the vgic code, which manifests itself
when the KVM run loop has a signal pending or needs a vmid generation
rollover after having disabled interrupts but before actually switching
to the guest.

In this case, we flush the vgic as usual, but we sync back the vgic
state and exit to userspace before entering the guest.  The consequence
is that we will be syncing the list registers back to the software model
using the GICH_ELRSR and GICH_EISR from the last execution of the guest,
potentially overwriting a list register containing an interrupt.

This showed up during migration testing where we would capture a state
where the VM has masked the arch timer but there were no interrupts,
resulting in a hung test.

Cc: Marc Zyngier <marc.zyngier@arm.com>
Reported-by: Alex Bennee <alex.bennee@linaro.org>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
include/kvm/arm_vgic.h
virt/kvm/arm/vgic-v2.c
virt/kvm/arm/vgic-v3.c
virt/kvm/arm/vgic.c