Skip to contentSkip to conversation

perf/core: fix the bug in the event multiplexing

PATCH In review Huang Shijie 6 participants 9 messages Aug 9, 2023

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf/core: fix the bug in the event multiplexing

1.) Background.
1.1) In arm64, run a virtual guest with Qemu, and bind the guest
to core 33 and run program "a" in guest.
The code of "a" shows below:
----------------------------------------------------------

		#include <stdio.h>

		int main()
		{
			unsigned long i = 0;

			for (;;) {
				i++;
			}

			printf("i:%ld\n", i);
			return 0;
		}

----------------------------------------------------------

1.2) Use the following perf command in host:

      #perf stat -e cycles:G,cycles:H -C 33 -I 1000 sleep 1
          #           time             counts unit events
               1.000817400      3,299,471,572      cycles:G
               1.000817400          3,240,586      cycles:H

       This result is correct, my cpu's frequency is 3.3G.

1.3) Use the following perf command in host:

      #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
            time             counts unit events
     1.000831480        153,634,097      cycles:G                                                                (70.03%)
     1.000831480      3,147,940,599      cycles:H                                                                (70.03%)
     1.000831480      1,143,598,527      L1-dcache-loads                                                         (70.03%)
     1.000831480              9,986      L1-dcache-load-misses            #    0.00% of all L1-dcache accesses   (70.03%)
     1.000831480    <not supported>      LLC-loads
     1.000831480    <not supported>      LLC-load-misses
     1.000831480        580,887,696      L1-icache-loads                                                         (70.03%)
     1.000831480             77,855      L1-icache-load-misses            #    0.01% of all L1-icache accesses   (70.03%)
     1.000831480      6,112,224,612      dTLB-loads                                                              (70.03%)
     1.000831480             16,222      dTLB-load-misses                 #    0.00% of all dTLB cache accesses  (69.94%)
     1.000831480        590,015,996      iTLB-loads                                                              (59.95%)
     1.000831480                505      iTLB-load-misses                 #    0.00% of all iTLB cache accesses  (59.95%)

       This result is wrong. The "cycle:G" should be nearly 3.3G.

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

So we got wrong result.

3.) About this patch.

        3.1) Add arch_perf_rotate_pmu_set()
        3.2) Add is_guest().
	     Check the context for hrtimer.
	3.3) In arm64's arch_perf_rotate_pmu_set(),
       	     set the PMU registers by the context.

4.) Test result of this patch:

      #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
            time             counts unit events
     1.000817360      3,297,898,244      cycles:G                                                                (70.03%)
     1.000817360          2,719,941      cycles:H                                                                (70.03%)
     1.000817360            883,764      L1-dcache-loads                                                         (70.03%)
     1.000817360             17,517      L1-dcache-load-misses            #    1.98% of all L1-dcache accesses   (70.03%)
     1.000817360    <not supported>      LLC-loads
     1.000817360    <not supported>      LLC-load-misses
     1.000817360          1,033,816      L1-icache-loads                                                         (70.03%)
     1.000817360            103,839      L1-icache-load-misses            #   10.04% of all L1-icache accesses   (70.03%)
     1.000817360            982,401      dTLB-loads                                                              (70.03%)
     1.000817360             28,272      dTLB-load-misses                 #    2.88% of all dTLB cache accesses  (69.94%)
     1.000817360            972,072      iTLB-loads                                                              (59.95%)
     1.000817360                772      iTLB-load-misses                 #    0.08% of all iTLB cache accesses  (59.95%)

    The result is correct. The "cycle:G" is nearly 3.3G now.
---
 arch/arm64/kvm/pmu.c     | 8 ++++++++
 include/linux/kvm_host.h | 1 +
 kernel/events/core.c     | 5 +++++
 virt/kvm/kvm_main.c      | 9 +++++++++
 4 files changed, 23 insertions(+)
arch/arm64/kvm/pmu.c +8 -0
index 121f1a14c829..a6815c3f0c4e 100644
@@ -210,6 +210,14 @@ void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu)
 	kvm_vcpu_pmu_disable_el0(events_guest);
 }
 
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}
+
 /*
  * With VHE, keep track of the PMUSERENR_EL0 value for the host EL0 on the pCPU
  * where PMUSERENR_EL0 for the guest is loaded, since PMUSERENR_EL0 is switched
include/linux/kvm_host.h +1 -0
index 9d3ac7720da9..e350cbc8190f 100644
@@ -931,6 +931,7 @@ void kvm_destroy_vcpus(struct kvm *kvm);
 
 void vcpu_load(struct kvm_vcpu *vcpu);
 void vcpu_put(struct kvm_vcpu *vcpu);
+bool is_guest(void);
 
 #ifdef __KVM_HAVE_IOAPIC
 void kvm_arch_post_irq_ack_notifier_list_update(struct kvm *kvm);
kernel/events/core.c +5 -0
index 6fd9272eec6e..fe78f9d17eba 100644
@@ -4229,6 +4229,10 @@ ctx_event_to_rotate(struct perf_event_pmu_context *pmu_ctx)
 	return event;
 }
 
+void __weak arch_perf_rotate_pmu_set(void)
+{
+}
+
 static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
 {
 	struct perf_cpu_context *cpuctx = this_cpu_ptr(&perf_cpu_context);
@@ -4282,6 +4286,7 @@ static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
 	if (task_event || (task_epc && cpu_event))
 		__pmu_ctx_sched_in(task_epc->ctx, pmu);
 
+	arch_perf_rotate_pmu_set();
 	perf_pmu_enable(pmu);
 	perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
 
virt/kvm/kvm_main.c +9 -0
index dfbaafbe3a00..a77d336552be 100644
@@ -218,6 +218,15 @@ void vcpu_load(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(vcpu_load);
 
+/* Do we in the guest? */
+bool is_guest(void)
+{
+	struct kvm_vcpu *vcpu;
+
+	vcpu = __this_cpu_read(kvm_running_vcpu);
+	return !!vcpu;
+}
+
 void vcpu_put(struct kvm_vcpu *vcpu)
 {
 	preempt_disable();
Signature · 1 line
2.39.2
#1 lore raw
kernel test robot noticed the following build warnings:

Hi Huang,

kernel test robot noticed the following build warnings:

[auto build test WARNING on kvm/queue]
[also build test WARNING on kvmarm/next tip/perf/core linus/master v6.5-rc5 next-20230808] [cannot apply to acme/perf/core kvm/linux-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Huang-Shijie/perf-core-fix-the-bug-in-the-event-multiplexing/20230809-094637
base:   https://git.kernel.org/pub/scm/virt/kvm/kvm.git queue
patch link:    https://lore.kernel.org/r/20230809013953.7692-1-shijie%40os.amperecomputing.com
patch subject: [PATCH] perf/core: fix the bug in the event multiplexing
config: loongarch-allyesconfig (https://download.01.org/0day-ci/archive/20230809/[email protected]/config)
compiler: loongarch64-linux-gcc (GCC) 12.3.0
reproduce: (https://download.01.org/0day-ci/archive/20230809/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All warnings (new ones prefixed by >>):

>> kernel/events/core.c:4232:13: warning: no previous prototype for 'arch_perf_rotate_pmu_set' [-Wmissing-prototypes]
    4232 | void __weak arch_perf_rotate_pmu_set(void)
         |             ^~~~~~~~~~~~~~~~~~~~~~~~

vim +/arch_perf_rotate_pmu_set +4232 kernel/events/core.c

4231

> 4232	void __weak arch_perf_rotate_pmu_set(void)

4233 {
4234 }
4235

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
#2 lore raw
In reply to Huang Shijie

Hi Huang,

Quoted 10 lines · Huang Shijie

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

Quoted 6 lines

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

So we got wrong result.

This is a rather vague description of the problem. AFAICT, the issue here is on VHE systems we wind up getting the EL0 count enable/disable bits backwards when entering the guest, which is corroborated by the data you have below.

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 220

Quoted 8 lines
arch/arm64/kvm/pmu.c +8 -0
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}
+

This sort of hook is rather nasty, and I'd strongly prefer a solution that's confined to KVM. I don't think the !is_guest() branch is necessary at all. Regardless of how the pmu context is changed, we need to go through vcpu_put() before getting back out to userspace.

We can check for a running vCPU (ick) from kvm_set_pmu_events() and either do the EL0 bit flip there or make a request on the vCPU to call kvm_vcpu_pmu_restore_guest() immediately before reentering the guest. I'm slightly leaning towards the latter, unless anyone has a better idea here.

Signature · 2 lines
Thanks,
Oliver
#3 lore raw

Hi Oliver,

Quoted 37 lines · Oliver Upton

Hi Huang,

Quoted 10 lines · Huang Shijie

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger
event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

Quoted 6 lines

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

So we got wrong result.

This is a rather vague description of the problem. AFAICT, the issue here is on VHE systems we wind up getting the EL0 count enable/disable bits backwards when entering the guest, which is corroborated by the data you have below.

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 220

Quoted 8 lines
arch/arm64/kvm/pmu.c +8 -0
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}
+

This sort of hook is rather nasty, and I'd strongly prefer a solution that's confined to KVM. I don't think the !is_guest() branch is necessary at all. Regardless of how the pmu context is changed, we need to go through vcpu_put() before getting back out to userspace.

Quoted 5 lines

We can check for a running vCPU (ick) from kvm_set_pmu_events() and either do the EL0 bit flip there or make a request on the vCPU to call kvm_vcpu_pmu_restore_guest() immediately before reentering the guest. I'm slightly leaning towards the latter, unless anyone has a better idea here.

Thanks a lot, I will check the code about the latter one, and try to fix it again.

Thanks

Huang Shijie

reviewer
#4 lore raw
In reply to Oliver Upton
Quoted 18 lines · Oliver Upton

Hi Huang,

Quoted 15 lines · Huang Shijie

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

Quoted 6 lines
Quoted 1 line

So we got wrong result.

This is a rather vague description of the problem. AFAICT, the issue here is on VHE systems we wind up getting the EL0 count enable/disable bits backwards when entering the guest, which is corroborated by the data you have below.

Yep; IIUC the issue here is that when we take an IRQ from a guest and reprogram the PMU in the IRQ handler, the IRQ handler will program the PMU with appropriate host/guest/user/etc filters for a *host* context, and then we'll return back into the guest without reconfigurign the event filtering for a *guest* context.

That can happen for perf_rotate_context(), or when we install an event into a running context, as that'll happen via an IPI.

Quoted 14 lines

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 220

Quoted 8 lines
arch/arm64/kvm/pmu.c +8 -0
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}
+

This sort of hook is rather nasty, and I'd strongly prefer a solution that's confined to KVM. I don't think the !is_guest() branch is necessary at all. Regardless of how the pmu context is changed, we need to go through vcpu_put() before getting back out to userspace.

Quoted 5 lines

We can check for a running vCPU (ick) from kvm_set_pmu_events() and either do the EL0 bit flip there or make a request on the vCPU to call kvm_vcpu_pmu_restore_guest() immediately before reentering the guest. I'm slightly leaning towards the latter, unless anyone has a better idea here.

The latter sounds reasonable to me.

I suspect we need to take special care here to make sure we leave *all* events in a good state when re-entering the guest or if we get to kvm_sched_out() after *removing* an event via an IPI -- it'd be easy to mess either case up and leave some events in a bad state.

Thanks,
Mark.

#5 lore raw

Hi Mark,

Quoted 24 lines · Mark Rutland
Quoted 18 lines · Oliver Upton

Hi Huang,

Quoted 15 lines · Huang Shijie

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger
event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

Quoted 5 lines
Quoted 1 line

So we got wrong result.

This is a rather vague description of the problem. AFAICT, the issue here is on VHE systems we wind up getting the EL0 count enable/disable bits backwards when entering the guest, which is corroborated by the data you have below.

Quoted 5 lines

Yep; IIUC the issue here is that when we take an IRQ from a guest and reprogram the PMU in the IRQ handler, the IRQ handler will program the PMU with appropriate host/guest/user/etc filters for a *host* context, and then we'll return back into the guest without reconfigurign the event filtering for a *guest* context.

Yes.

Quoted 17 lines

That can happen for perf_rotate_context(), or when we install an event into a running context, as that'll happen via an IPI.

Quoted 13 lines

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 220

Quoted 8 lines
arch/arm64/kvm/pmu.c +8 -0
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}
+

This sort of hook is rather nasty, and I'd strongly prefer a solution that's confined to KVM. I don't think the !is_guest() branch is necessary at all. Regardless of how the pmu context is changed, we need to go through vcpu_put() before getting back out to userspace.

Quoted 6 lines
Quoted 5 lines

We can check for a running vCPU (ick) from kvm_set_pmu_events() and either do the EL0 bit flip there or make a request on the vCPU to call kvm_vcpu_pmu_restore_guest() immediately before reentering the guest. I'm slightly leaning towards the latter, unless anyone has a better idea here.

The latter sounds reasonable to me.

okay. I prefer the latter one now. :)

Thanks

Huang Shijie

Quoted 8 lines

I suspect we need to take special care here to make sure we leave *all* events in a good state when re-entering the guest or if we get to kvm_sched_out() after *removing* an event via an IPI -- it'd be easy to mess either case up and leave some events in a bad state.

Thanks,
Mark.

maintainer
#6 lore raw
In reply to Huang Shijie

On Wed, 09 Aug 2023 02:39:53 +0100,
Huang Shijie <[email protected]> wrote:

For a start, please provide a sensible subject line for your patch. "fix the bug" is not exactly descriptive, and I'd argue that if there was only one bug left, I'd have taken an early retirement by now.

Quoted 3 lines

1.) Background.
1.1) In arm64, run a virtual guest with Qemu, and bind the guest

Is that with QEMU in system emulation mode? Or QEMU as a VMM for KVM?

Quoted 1 line

to core 33 and run program "a" in guest.

Is core 33 significant? Is the program itself significant?

Quoted 98 lines

The code of "a" shows below:

   	----------------------------------------------------------
		#include <stdio.h>

		int main()
		{
			unsigned long i = 0;

			for (;;) {
				i++;
			}

			printf("i:%ld\n", i);
			return 0;
		}
   	----------------------------------------------------------

   1.2) Use the following perf command in host:
      #perf stat -e cycles:G,cycles:H -C 33 -I 1000 sleep 1
          #           time             counts unit events
               1.000817400      3,299,471,572      cycles:G
               1.000817400          3,240,586      cycles:H

       This result is correct, my cpu's frequency is 3.3G.

   1.3) Use the following perf command in host:
      #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
            time             counts unit events
     1.000831480        153,634,097      cycles:G                                                                (70.03%)
     1.000831480      3,147,940,599      cycles:H                                                                (70.03%)
     1.000831480      1,143,598,527      L1-dcache-loads                                                         (70.03%)
     1.000831480              9,986      L1-dcache-load-misses            #    0.00% of all L1-dcache accesses   (70.03%)
     1.000831480    <not supported>      LLC-loads
     1.000831480    <not supported>      LLC-load-misses
     1.000831480        580,887,696      L1-icache-loads                                                         (70.03%)
     1.000831480             77,855      L1-icache-load-misses            #    0.01% of all L1-icache accesses   (70.03%)
     1.000831480      6,112,224,612      dTLB-loads                                                              (70.03%)
     1.000831480             16,222      dTLB-load-misses                 #    0.00% of all dTLB cache accesses  (69.94%)
     1.000831480        590,015,996      iTLB-loads                                                              (59.95%)
     1.000831480                505      iTLB-load-misses                 #    0.00% of all iTLB cache accesses  (59.95%)

       This result is wrong. The "cycle:G" should be nearly 3.3G.

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

So we got wrong result.

3.) About this patch.

        3.1) Add arch_perf_rotate_pmu_set()
        3.2) Add is_guest().
	     Check the context for hrtimer.
	3.3) In arm64's arch_perf_rotate_pmu_set(),
       	     set the PMU registers by the context.

4.) Test result of this patch:

      #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
            time             counts unit events
     1.000817360      3,297,898,244      cycles:G                                                                (70.03%)
     1.000817360          2,719,941      cycles:H                                                                (70.03%)
     1.000817360            883,764      L1-dcache-loads                                                         (70.03%)
     1.000817360             17,517      L1-dcache-load-misses            #    1.98% of all L1-dcache accesses   (70.03%)
     1.000817360    <not supported>      LLC-loads
     1.000817360    <not supported>      LLC-load-misses
     1.000817360          1,033,816      L1-icache-loads                                                         (70.03%)
     1.000817360            103,839      L1-icache-load-misses            #   10.04% of all L1-icache accesses   (70.03%)
     1.000817360            982,401      dTLB-loads                                                              (70.03%)
     1.000817360             28,272      dTLB-load-misses                 #    2.88% of all dTLB cache accesses  (69.94%)
     1.000817360            972,072      iTLB-loads                                                              (59.95%)
     1.000817360                772      iTLB-load-misses                 #    0.08% of all iTLB cache accesses  (59.95%)

    The result is correct. The "cycle:G" is nearly 3.3G now.
---
 arch/arm64/kvm/pmu.c     | 8 ++++++++
 include/linux/kvm_host.h | 1 +
 kernel/events/core.c     | 5 +++++
 virt/kvm/kvm_main.c      | 9 +++++++++
 4 files changed, 23 insertions(+)
arch/arm64/kvm/pmu.c +0 -0
index 121f1a14c829..a6815c3f0c4e 100644

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 219

Quoted 10 lines
arch/arm64/kvm/pmu.c +7 -0
@@ -210,6 +210,14 @@ void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu)
 	kvm_vcpu_pmu_disable_el0(events_guest);
 }
 
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}

So we're now randomly poking at the counters even when no guest is running, based on whatever is stashed in internal KVM data structures? I'm sure this is going to work really well.

Hint: even if these functions don't directly look at the vcpu pointer, passing NULL is a really bad idea. It is a sure sign that you don't have the context on which to perform the action you're trying to do.

This really shouldn't do *anything* when the rotation process is not preempting a guest.

include/linux/kvm_host.h void kvm_destroy_vcpus(struct kvm *kvm); · line 934

Quoted 13 lines
arch/arm64/kvm/pmu.c +1 -0
+
 /*
  * With VHE, keep track of the PMUSERENR_EL0 value for the host EL0 on the pCPU
  * where PMUSERENR_EL0 for the guest is loaded, since PMUSERENR_EL0 is switched
include/linux/kvm_host.h +1 -0
index 9d3ac7720da9..e350cbc8190f 100644
@@ -931,6 +931,7 @@ void kvm_destroy_vcpus(struct kvm *kvm);
 
 void vcpu_load(struct kvm_vcpu *vcpu);
 void vcpu_put(struct kvm_vcpu *vcpu);
+bool is_guest(void);

Why do we need this (not to mention the poor choice of name)? We already have kvm_get_running_vcpu(), which does everything you need (and gives you the actual context).

Quoted 19 lines
include/linux/kvm_host.h +0 -0
 #ifdef __KVM_HAVE_IOAPIC
 void kvm_arch_post_irq_ack_notifier_list_update(struct kvm *kvm);
kernel/events/core.c +4 -0
index 6fd9272eec6e..fe78f9d17eba 100644
@@ -4229,6 +4229,10 @@ ctx_event_to_rotate(struct perf_event_pmu_context *pmu_ctx)
 	return event;
 }
 
+void __weak arch_perf_rotate_pmu_set(void)
+{
+}
+
 static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
 {
 	struct perf_cpu_context *cpuctx = this_cpu_ptr(&perf_cpu_context);

kernel/events/core.c static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc) · line 4289

Quoted 4 lines
kernel/events/core.c +1 -0
@@ -4282,6 +4286,7 @@ static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
 	if (task_event || (task_epc && cpu_event))
 		__pmu_ctx_sched_in(task_epc->ctx, pmu);
 
+	arch_perf_rotate_pmu_set();

KVM already supports hooking into the perf core using the perf_guest_info_callbacks structure. Why should we need a separate mechanism?

Quoted 13 lines
kernel/events/core.c +0 -0
 	perf_pmu_enable(pmu);
 	perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
 
virt/kvm/kvm_main.c +2 -0
index dfbaafbe3a00..a77d336552be 100644
@@ -218,6 +218,15 @@ void vcpu_load(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(vcpu_load);
 
+/* Do we in the guest? */
+bool is_guest(void)

virt/kvm/kvm_main.c void vcpu_load(struct kvm_vcpu *vcpu) · line 232

Quoted 10 lines
virt/kvm/kvm_main.c +7 -0
+{
+	struct kvm_vcpu *vcpu;
+
+	vcpu = __this_cpu_read(kvm_running_vcpu);
+	return !!vcpu;
+}
+
 void vcpu_put(struct kvm_vcpu *vcpu)
 {
 	preempt_disable();

It looks like you've identified an actual issue. However, I'm highly sceptical of the implementation. This really needs some more work.

Another question is how the same thing is handled on x86? Maybe they don't suffer from this problem thanks to specific architectural features, but it'd be good to find out, as this may guide the implementation in a different way.

Thanks,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#7 lore raw

On Wed, Aug 09, 2023 at 09:48:29AM +0100, Marc Zyngier wrote:

[...]

Quoted 4 lines

Another question is how the same thing is handled on x86? Maybe they don't suffer from this problem thanks to specific architectural features, but it'd be good to find out, as this may guide the implementation in a different way.

I'm pretty sure the bug here is arm64 specific. x86 (at least on intel) fetches the guest PMU context from the perf driver w/ irqs disabled immediately before entering the guest (see atomic_switch_perf_msrs()).

Signature · 2 lines
Thanks,
Oliver
#8 lore raw
In reply to Marc Zyngier

Hi Marc,

Quoted 6 lines · Marc Zyngier

On Wed, 09 Aug 2023 02:39:53 +0100,
Huang Shijie <[email protected]> wrote:

For a start, please provide a sensible subject line for your patch. "fix the bug" is not exactly descriptive, and I'd argue that if there was only one bug left, I'd have taken an early retirement by now.

okay, I will change the subject in the future.

Quoted 3 lines
Quoted 2 lines

1.) Background.
1.1) In arm64, run a virtual guest with Qemu, and bind the guest

Is that with QEMU in system emulation mode? Or QEMU as a VMM for KVM?

Run the Qemu as a VMM for KVM.

Quoted 3 lines
Quoted 2 lines

to core 33 and run program "a" in guest.

Is core 33 significant? Is the program itself significant?

It is not a significant one, just chosed by random.

The program is just used for testing the perf, not a significant one.

Quoted 112 lines
Quoted 98 lines
         The code of "a" shows below:
    	----------------------------------------------------------
		#include <stdio.h>

		int main()
		{
			unsigned long i = 0;

			for (;;) {
				i++;
			}

			printf("i:%ld\n", i);
			return 0;
		}
    	----------------------------------------------------------

    1.2) Use the following perf command in host:
       #perf stat -e cycles:G,cycles:H -C 33 -I 1000 sleep 1
           #           time             counts unit events
                1.000817400      3,299,471,572      cycles:G
                1.000817400          3,240,586      cycles:H

        This result is correct, my cpu's frequency is 3.3G.

    1.3) Use the following perf command in host:
       #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
             time             counts unit events
      1.000831480        153,634,097      cycles:G                                                                (70.03%)
      1.000831480      3,147,940,599      cycles:H                                                                (70.03%)
      1.000831480      1,143,598,527      L1-dcache-loads                                                         (70.03%)
      1.000831480              9,986      L1-dcache-load-misses            #    0.00% of all L1-dcache accesses   (70.03%)
      1.000831480    <not supported>      LLC-loads
      1.000831480    <not supported>      LLC-load-misses
      1.000831480        580,887,696      L1-icache-loads                                                         (70.03%)
      1.000831480             77,855      L1-icache-load-misses            #    0.01% of all L1-icache accesses   (70.03%)
      1.000831480      6,112,224,612      dTLB-loads                                                              (70.03%)
      1.000831480             16,222      dTLB-load-misses                 #    0.00% of all dTLB cache accesses  (69.94%)
      1.000831480        590,015,996      iTLB-loads                                                              (59.95%)
      1.000831480                505      iTLB-load-misses                 #    0.00% of all iTLB cache accesses  (59.95%)

        This result is wrong. The "cycle:G" should be nearly 3.3G.

2.) Root cause.
There is only 7 counters in my arm64 platform:
(one cycle counter) + (6 normal counters)

In 1.3 above, we will use 10 event counters. Since we only have 7 counters, the perf core will trigger
event multiplexing in hrtimer:
merge_sched_in() -->perf_mux_hrtimer_restart() --> perf_rotate_context().

In the perf_rotate_context(), it does not restore some PMU registers as context_switch() does. In context_switch():
kvm_sched_in() --> kvm_vcpu_pmu_restore_guest() kvm_sched_out() --> kvm_vcpu_pmu_restore_host()

So we got wrong result.

3.) About this patch.

         3.1) Add arch_perf_rotate_pmu_set()
         3.2) Add is_guest().
	     Check the context for hrtimer.
	3.3) In arm64's arch_perf_rotate_pmu_set(),
        	     set the PMU registers by the context.

4.) Test result of this patch:

       #perf stat -e cycles:G,cycles:H -C 33 -d -d  -I 1000 sleep 1
             time             counts unit events
      1.000817360      3,297,898,244      cycles:G                                                                (70.03%)
      1.000817360          2,719,941      cycles:H                                                                (70.03%)
      1.000817360            883,764      L1-dcache-loads                                                         (70.03%)
      1.000817360             17,517      L1-dcache-load-misses            #    1.98% of all L1-dcache accesses   (70.03%)
      1.000817360    <not supported>      LLC-loads
      1.000817360    <not supported>      LLC-load-misses
      1.000817360          1,033,816      L1-icache-loads                                                         (70.03%)
      1.000817360            103,839      L1-icache-load-misses            #   10.04% of all L1-icache accesses   (70.03%)
      1.000817360            982,401      dTLB-loads                                                              (70.03%)
      1.000817360             28,272      dTLB-load-misses                 #    2.88% of all dTLB cache accesses  (69.94%)
      1.000817360            972,072      iTLB-loads                                                              (59.95%)
      1.000817360                772      iTLB-load-misses                 #    0.08% of all iTLB cache accesses  (59.95%)

     The result is correct. The "cycle:G" is nearly 3.3G now.
---
  arch/arm64/kvm/pmu.c     | 8 ++++++++
  include/linux/kvm_host.h | 1 +
  kernel/events/core.c     | 5 +++++
  virt/kvm/kvm_main.c      | 9 +++++++++
  4 files changed, 23 insertions(+)
arch/arm64/kvm/pmu.c +0 -0
index 121f1a14c829..a6815c3f0c4e 100644

arch/arm64/kvm/pmu.c void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu) · line 219

Quoted 10 lines
arch/arm64/kvm/pmu.c +7 -0
@@ -210,6 +210,14 @@ void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu)
  	kvm_vcpu_pmu_disable_el0(events_guest);
  }
 
+void arch_perf_rotate_pmu_set(void)
+{
+	if (is_guest())
+		kvm_vcpu_pmu_restore_guest(NULL);
+	else
+		kvm_vcpu_pmu_restore_host(NULL);
+}

So we're now randomly poking at the counters even when no guest is running, based on whatever is stashed in internal KVM data structures? I'm sure this is going to work really well.

Quoted 6 lines

Hint: even if these functions don't directly look at the vcpu pointer, passing NULL is a really bad idea. It is a sure sign that you don't have the context on which to perform the action you're trying to do.

This really shouldn't do *anything* when the rotation process is not preempting a guest.

Thanks for explanation, I will try Oliver's second method to resolve this issue.

Quoted 10 lines
Quoted 10 lines
arch/arm64/kvm/pmu.c +1 -0
+
  /*
   * With VHE, keep track of the PMUSERENR_EL0 value for the host EL0 on the pCPU
   * where PMUSERENR_EL0 for the guest is loaded, since PMUSERENR_EL0 is switched
include/linux/kvm_host.h +0 -0
index 9d3ac7720da9..e350cbc8190f 100644
@@ -931,6 +931,7 @@ void kvm_destroy_vcpus(struct kvm *kvm);
 
Quoted 6 lines

include/linux/kvm_host.h void kvm_destroy_vcpus(struct kvm *kvm); · line 934

Quoted 3 lines
include/linux/kvm_host.h +1 -0
  void vcpu_load(struct kvm_vcpu *vcpu);
  void vcpu_put(struct kvm_vcpu *vcpu);
+bool is_guest(void);

Why do we need this (not to mention the poor choice of name)? We already have kvm_get_running_vcpu(), which does everything you need (and gives you the actual context).

yes.

Quoted 22 lines
Quoted 22 lines
include/linux/kvm_host.h +0 -0
  #ifdef __KVM_HAVE_IOAPIC
  void kvm_arch_post_irq_ack_notifier_list_update(struct kvm *kvm);
kernel/events/core.c +4 -0
index 6fd9272eec6e..fe78f9d17eba 100644
@@ -4229,6 +4229,10 @@ ctx_event_to_rotate(struct perf_event_pmu_context *pmu_ctx)
  	return event;
  }
 
+void __weak arch_perf_rotate_pmu_set(void)
+{
+}
+
  static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
  {
  	struct perf_cpu_context *cpuctx = this_cpu_ptr(&perf_cpu_context);
@@ -4282,6 +4286,7 @@ static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc)
  	if (task_event || (task_epc && cpu_event))
  		__pmu_ctx_sched_in(task_epc->ctx, pmu);
 
Quoted 4 lines

kernel/events/core.c static bool perf_rotate_context(struct perf_cpu_pmu_context *cpc) · line 4289

Quoted 1 line
kernel/events/core.c +1 -0
+	arch_perf_rotate_pmu_set();

KVM already supports hooking into the perf core using the perf_guest_info_callbacks structure. Why should we need a separate mechanism?

okay, I will check it too.

Quoted 12 lines
Quoted 12 lines
kernel/events/core.c +0 -0
  	perf_pmu_enable(pmu);
  	perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
 
virt/kvm/kvm_main.c +0 -0
index dfbaafbe3a00..a77d336552be 100644
@@ -218,6 +218,15 @@ void vcpu_load(struct kvm_vcpu *vcpu)
  }
  EXPORT_SYMBOL_GPL(vcpu_load);
 
Quoted 14 lines

virt/kvm/kvm_main.c void vcpu_load(struct kvm_vcpu *vcpu) · line 232

Quoted 12 lines
virt/kvm/kvm_main.c +9 -0
+/* Do we in the guest? */
+bool is_guest(void)
+{
+	struct kvm_vcpu *vcpu;
+
+	vcpu = __this_cpu_read(kvm_running_vcpu);
+	return !!vcpu;
+}
+
  void vcpu_put(struct kvm_vcpu *vcpu)
  {
  	preempt_disable();

It looks like you've identified an actual issue. However, I'm highly sceptical of the implementation. This really needs some more work.

Thanks again.

Huang Shijie