On 2021-01-28 15:16, Andre Przywara wrote:
On Tue, 26 Jan 2021 15:15:20 +0000
Marc Zyngier <maz@xxxxxxxxxx> wrote:
Hi Marc,
We currently find out about the presence of a HW PMU (or the handling
of that PMU by perf, which amounts to the same thing) in a fairly
roundabout way, by checking the number of counters available to perf.
That's good enough for now, but we will soon need to find about about
that on paths where perf is out of reach (in the world switch).
Instead, let's turn kvm_arm_support_pmu_v3() into a static key.
I am sure the pesky build bot has told you about it already, but this
fails when ARM_PMU is not defined, as perf_num_counters() is not
defined. It's bit nasty, since it's a generic function, so we
can't easily stub it in its original header.
No sign from the bot yet, but that's indeed a problem. Well spotted.
Shall we find a place somewhere in arch/arm64 and provide a stub
implementation there, #ifndef CONFIG_ARM_PMU? Sounds ugly, though.
Or something else entirely?
How about:
diff --git a/arch/arm64/kvm/perf.c b/arch/arm64/kvm/perf.c
index 198fa4266b2d..739164324afe 100644
--- a/arch/arm64/kvm/perf.c
+++ b/arch/arm64/kvm/perf.c
@@ -55,7 +55,7 @@ int kvm_perf_init(void)
* hardware performance counters. This could ensure the presence of
* a physical PMU and CONFIG_PERF_EVENT is selected.
*/
- if (perf_num_counters() > 0)
+ if (IS_ENABLED(CONFIG_ARM_PMU) && perf_num_counters() > 0)
static_branch_enable(&kvm_arm_pmu_available);
return perf_register_guest_info_callbacks(&kvm_guest_cbs);
It certainly compiles here.
M.
--
Jazz is not dead. It just smells funny...
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
