Re: [PATCH] KVM: arm: fix missing free_percpu_irq in kvm_timer_hyp_init()

From: Marc Zyngier
Date: Fri Dec 06 2019 - 05:56:38 EST


On 2019-11-23 02:30, linmiaohe wrote:
From: Miaohe Lin <linmiaohe@xxxxxxxxxx>

When host_ptimer_irq request irq resource failed, we forget
to release the host_vtimer_irq resource already requested.
Fix this missing irq release and other similar scenario.

That's really not a big deal, as nothing but KVM can use the
timers anyway, but I guess it doesn't hurt to be correct.


Fixes: 9e01dc76be6a ("KVM: arm/arm64: arch_timer: Assign the phys
timer on VHE systems")
Signed-off-by: Miaohe Lin <linmiaohe@xxxxxxxxxx>
---
virt/kvm/arm/arch_timer.c | 17 ++++++++++++-----
1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c
index f182b2380345..73867f97040c 100644
--- a/virt/kvm/arm/arch_timer.c
+++ b/virt/kvm/arm/arch_timer.c
@@ -935,7 +935,7 @@ int kvm_timer_hyp_init(bool has_gic)
kvm_get_running_vcpus());
if (err) {
kvm_err("kvm_arch_timer: error setting vcpu affinity\n");
- goto out_free_irq;
+ goto out_free_vtimer_irq;
}

static_branch_enable(&has_gic_active_state);
@@ -960,7 +960,7 @@ int kvm_timer_hyp_init(bool has_gic)
if (err) {
kvm_err("kvm_arch_timer: can't request ptimer interrupt %d (%d)\n",
host_ptimer_irq, err);
- return err;
+ goto out_disable_gic_state;
}

if (has_gic) {
@@ -968,7 +968,7 @@ int kvm_timer_hyp_init(bool has_gic)
kvm_get_running_vcpus());
if (err) {
kvm_err("kvm_arch_timer: error setting vcpu affinity\n");
- goto out_free_irq;
+ goto out_free_ptimer_irq;
}
}

@@ -977,15 +977,22 @@ int kvm_timer_hyp_init(bool has_gic)
kvm_err("kvm_arch_timer: invalid physical timer IRQ: %d\n",
info->physical_irq);
err = -ENODEV;
- goto out_free_irq;
+ goto out_disable_gic_state;
}

cpuhp_setup_state(CPUHP_AP_KVM_ARM_TIMER_STARTING,
"kvm/arm/timer:starting", kvm_timer_starting_cpu,
kvm_timer_dying_cpu);
return 0;
-out_free_irq:
+
+out_free_ptimer_irq:
+ free_percpu_irq(host_ptimer_irq, kvm_get_running_vcpus());
+out_disable_gic_state:
+ if (has_gic)
+ static_branch_disable(&has_gic_active_state);

Given that we're failing the init of KVM, this is totally
superfluous. Also, this state is still valid, no matter
what happens (the GIC is not going away from under our feet).

+out_free_vtimer_irq:
free_percpu_irq(host_vtimer_irq, kvm_get_running_vcpus());
+
return err;
}

Thanks,

M.
--
Jazz is not dead. It just smells funny...