Skip to contentSkip to conversation

KVM: selftests: Use KVM's task pinning APIs in steal_time

PATCH In review Hisam Mehboob 3 participants 3 messages Jul 6 – Jul 7

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 KVM: selftests: Use KVM's task pinning APIs in steal_time

From: Sean Christopherson <[email protected]>

Use pin_self_to_cpu() and pin_task_to_cpu() to pin the vCPU thread and the stealer thread to pCPU0 in the steal_time test. Eliminating the usage of pthread_attr_setaffinity_np() in particular allows building the test against non-glibc C libraries.

---
 tools/testing/selftests/kvm/steal_time.c | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)
tools/testing/selftests/kvm/steal_time.c +4 -8
index 76fcdd1fd3cb..19a97d6a3607 100644
@@ -508,9 +508,7 @@ int main(int ac, char **av)
 {
 	struct kvm_vcpu *vcpus[NR_VCPUS];
 	struct kvm_vm *vm;
-	pthread_attr_t attr;
 	pthread_t thread;
-	cpu_set_t cpuset;
 	unsigned int gpages;
 	long stolen_time;
 	long run_delay;
@@ -520,11 +518,7 @@ int main(int ac, char **av)
 	verbose = ac > 1 && (!strncmp(av[1], "-v", 3) || !strncmp(av[1], "--verbose", 10));
 
 	/* Set CPU affinity so we can force preemption of the VCPU */
-	CPU_ZERO(&cpuset);
-	CPU_SET(0, &cpuset);
-	pthread_attr_init(&attr);
-	pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &cpuset);
-	pthread_setaffinity_np(pthread_self(), sizeof(cpu_set_t), &cpuset);
+	pin_self_to_cpu(0);
 
 	/* Create a VM and an identity mapped memslot for the steal time structure */
 	vm = vm_create_with_vcpus(NR_VCPUS, guest_code, vcpus);
@@ -558,7 +552,9 @@ int main(int ac, char **av)
 
 		/* Steal time from the VCPU. The steal time thread has the same CPU affinity as the VCPUs. */
 		run_delay = get_run_delay();
-		pthread_create(&thread, &attr, do_steal_time, NULL);
+		pthread_create(&thread, NULL, do_steal_time, NULL);
+		pin_task_to_cpu(thread, 0);
+
 		do
 			sched_yield();
 		while (get_run_delay() - run_delay < MIN_RUN_DELAY_NS);
Signature · 1 line
2.51.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] Race condition causing intermittent test failure (ESRCH) due to late thread pinning --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Race condition causing intermittent test failure (ESRCH) due to late thread pinning --

commit 27b6bbfae99d5564cda41cbc7b8c36c7b9a4ed64
Author: Sean Christopherson <[email protected]>

KVM: selftests: Use KVM's task pinning APIs in steal_time

This commit updates the steal_time test to use KVM's task pinning APIs (pin_self_to_cpu and pin_task_to_cpu) to pin the vCPU and stealer threads to pCPU0. This removes the use of pthread_attr_setaffinity_np, allowing the test to build against non-glibc C libraries.

Quoted 4 lines
tools/testing/selftests/kvm/steal_time.c +0 -0
index 76fcdd1fd3cb4..19a97d6a36076 100644

[ ... ]

tools/testing/selftests/kvm/steal_time.c int main(int ac, char **av) · line 556

Quoted 7 lines
tools/testing/selftests/kvm/steal_time.c +2 -1
@@ -558,7 +552,9 @@ int main(int ac, char **av)
 
 		/* Steal time from the VCPU. The steal time thread has the same CPU affinity as the VCPUs. */
 		run_delay = get_run_delay();
-		pthread_create(&thread, &attr, do_steal_time, NULL);
+		pthread_create(&thread, NULL, do_steal_time, NULL);
+		pin_task_to_cpu(thread, 0);

[Severity: Medium]
Could this introduce a race condition that causes intermittent test failures?

Since pthread_create is called before pinning, there is a small window where the child thread (do_steal_time) could execute and complete its very brief 200us run before the parent thread executes pin_task_to_cpu.

If the child thread exits before pin_task_to_cpu attempts to set its affinity, it will return ESRCH and trigger the TEST_ASSERT inside pin_task_to_cpu, abruptly aborting the test.

Is the explicit pin_task_to_cpu call necessary here? Since the main thread already called pin_self_to_cpu(0) earlier, the new thread created by pthread_create should automatically inherit the parent's CPU affinity mask.

Quoted 4 lines
tools/testing/selftests/kvm/steal_time.c +1 -0
+
 		do
 			sched_yield();
 		while (get_run_delay() - run_delay < MIN_RUN_DELAY_NS);
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#2 lore raw
Quoted 36 lines · On Mon, Jul 06, 2026, [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Race condition causing intermittent test failure (ESRCH) due to late thread pinning --

commit 27b6bbfae99d5564cda41cbc7b8c36c7b9a4ed64
Author: Sean Christopherson <[email protected]>

KVM: selftests: Use KVM's task pinning APIs in steal_time

This commit updates the steal_time test to use KVM's task pinning APIs (pin_self_to_cpu and pin_task_to_cpu) to pin the vCPU and stealer threads to pCPU0. This removes the use of pthread_attr_setaffinity_np, allowing the test to build against non-glibc C libraries.

Quoted 4 lines
tools/testing/selftests/kvm/steal_time.c +0 -0
index 76fcdd1fd3cb4..19a97d6a36076 100644

[ ... ]

tools/testing/selftests/kvm/steal_time.c int main(int ac, char **av) · line 556

Quoted 7 lines
tools/testing/selftests/kvm/steal_time.c +2 -1
@@ -558,7 +552,9 @@ int main(int ac, char **av)
 
 		/* Steal time from the VCPU. The steal time thread has the same CPU affinity as the VCPUs. */
 		run_delay = get_run_delay();
-		pthread_create(&thread, &attr, do_steal_time, NULL);
+		pthread_create(&thread, NULL, do_steal_time, NULL);
+		pin_task_to_cpu(thread, 0);

[Severity: Medium]
Could this introduce a race condition that causes intermittent test failures?

Since pthread_create is called before pinning, there is a small window where the child thread (do_steal_time) could execute and complete its very brief 200us run before the parent thread executes pin_task_to_cpu.

Quoted 5 lines

If the child thread exits before pin_task_to_cpu attempts to set its affinity, it will return ESRCH and trigger the TEST_ASSERT inside pin_task_to_cpu, abruptly aborting the test.

Is the explicit pin_task_to_cpu call necessary here?

Nope, which means we can simply drop the problematic pthread_attr_setaffinity_np(), though I think it makes sense to add a sanity check that the worker did indeed inherit the affinity, if only to document what's happening.

I'll send a v2.

Thanks for posting this Hisam!

Quoted 3 lines

Since the main thread already called pin_self_to_cpu(0) earlier, the new thread created by pthread_create should automatically inherit the parent's CPU affinity mask.