KVM: arm/arm64: Introduce kvm_pmu_vcpu_init() to setup PMU counter index
authorZenghui Yu <yuzenghui@huawei.com>
Thu, 18 Jul 2019 08:15:10 +0000 (08:15 +0000)
committerMarc Zyngier <marc.zyngier@arm.com>
Tue, 23 Jul 2019 08:23:44 +0000 (09:23 +0100)
commitbca031e2c8aa22a978a2452bf959e27e9fa73dc7
treec7e80e707e3abc3d924254432f4159b36b51631f
parent5f9e832c137075045d15cd6899ab0505cfb2ca4b
KVM: arm/arm64: Introduce kvm_pmu_vcpu_init() to setup PMU counter index

We use "pmc->idx" and the "chained" bitmap to determine if the pmc is
chained, in kvm_pmu_pmc_is_chained().  But idx might be uninitialized
(and random) when we doing this decision, through a KVM_ARM_VCPU_INIT
ioctl -> kvm_pmu_vcpu_reset(). And the test_bit() against this random
idx will potentially hit a KASAN BUG [1].

In general, idx is the static property of a PMU counter that is not
expected to be modified across resets, as suggested by Julien.  It
looks more reasonable if we can setup the PMU counter idx for a vcpu
in its creation time. Introduce a new function - kvm_pmu_vcpu_init()
for this basic setup. Oh, and the KASAN BUG will get fixed this way.

[1] https://www.spinics.net/lists/kvm-arm/msg36700.html

Fixes: 80f393a23be6 ("KVM: arm/arm64: Support chained PMU counters")
Suggested-by: Andrew Murray <andrew.murray@arm.com>
Suggested-by: Julien Thierry <julien.thierry@arm.com>
Acked-by: Julien Thierry <julien.thierry@arm.com>
Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
include/kvm/arm_pmu.h
virt/kvm/arm/arm.c
virt/kvm/arm/pmu.c