[PATCH v3 00/12] KVM: selftests: Add more syscall wrappers, fix hardware_disable_test

[PATCH v3 00/12] KVM: selftests: Add more syscall wrappers, fix hardware_disable_test

[Sean Christopherson]

Third time's a charm.  Sorry for the spam.

Buried in the middle of this is a fix for the hardware_disable_test where
it can end up with a runtime of multiple minutes when running a preemptible
kernels on a large NUMA system.  Turns out my analysis/guess in v1 was wrong;
the problem isn't direct CPU contention, it's that letting the workers do any
amount of "real" work on any CPUs leads to pthread_create() slowdowns from
the perspective of the caller.

The vast, vast majority of the changes are adding kvm_xxx() syscall wrappers
so that tests don't need to manually assert that a syscall succeeds (the
hardware disable test had some ugly equivalents).

v3:
 - Key off _GNU_SOURCE, not __USE_GNU. [Sashiko]
 - Clear consumed CPUs from the allowed set. [Sashiko]
 - Assert that settings affinity attributes succeeds. [Sasiko]

v2:
 - https://lore.kernel.org/all/20260731170619.2620845-1-seanjc%40google.com
 - pause() the child after creating workers so as not to busy loop, and to
   maintain equivalent functionality. [Sashiko]
 - Don't bother pinning the parent of immediate child, and instead focus on
   affining workers to their CPUs asap.

v1: https://lore.kernel.org/all/20260730191908.2084803-1-seanjc@google.com


Sean Christopherson (12):
  KVM: selftests: Add a KVM syscall wrapper for sched_setaffinity()
  KVM: selftests: Set threads CPU affinity before doing work in hardware
    disable test
  KVM: selftests: Pre-set threads affinity in hardware disable test when
    possible
  KVM: selftests: Return the target CPU from pin_task_to_random_cpu()
  KVM: selftests: Extract picking of random CPU from cpu_set_t to
    separate API
  KVM: selftests: Affine threads to random CPUs in hardware disable test
  KVM: selftests: Drop unreachable, dead code from hardware disable test
  KVM: selftests: Add KVM syscall wrapper for pthread_create()
  KVM: selftests: Add KVM syscall wrappers for pthread_{cancel,join}()
  KVM: selftests: Add helper APIs to cancel+join pthreads
  KVM: selftests: Add KVM syscall wrappers for
    pthread_{g,s}etaffinity_np()
  KVM: selftests: Clean up global constants in hardware disable test

 tools/testing/selftests/kvm/arch_timer.c      |  18 +--
 tools/testing/selftests/kvm/arm64/vgic_irq.c  |  14 +--
 .../selftests/kvm/arm64/vgic_lpi_stress.c     |   8 +-
 tools/testing/selftests/kvm/dirty_log_test.c  |   4 +-
 .../selftests/kvm/hardware_disable_test.c     | 109 ++++++++----------
 .../selftests/kvm/include/kvm_syscalls.h      |  48 ++++++++
 .../testing/selftests/kvm/include/kvm_util.h  |  11 +-
 tools/testing/selftests/kvm/irq_test.c        |   4 +-
 tools/testing/selftests/kvm/irqfd_test.c      |   4 +-
 .../selftests/kvm/kvm_page_table_test.c       |   6 +-
 tools/testing/selftests/kvm/lib/kvm_util.c    |  10 +-
 tools/testing/selftests/kvm/lib/memstress.c   |   4 +-
 .../selftests/kvm/lib/userfaultfd_util.c      |   7 +-
 .../testing/selftests/kvm/memslot_perf_test.c |   4 +-
 tools/testing/selftests/kvm/mmu_stress_test.c |   4 +-
 .../selftests/kvm/pre_fault_memory_test.c     |   4 +-
 tools/testing/selftests/kvm/rseq_test.c       |  12 +-
 tools/testing/selftests/kvm/s390/memop.c      |   4 +-
 .../selftests/kvm/set_memory_region_test.c    |   6 +-
 tools/testing/selftests/kvm/steal_time.c      |   6 +-
 tools/testing/selftests/kvm/x86/hyperv_ipi.c  |  32 +----
 .../selftests/kvm/x86/hyperv_tlb_flush.c      |  30 +----
 .../kvm/x86/private_mem_conversions_test.c    |   4 +-
 .../kvm/x86/private_mem_kvm_exits_test.c      |   7 +-
 .../selftests/kvm/x86/recalc_apic_map_test.c  |   5 +-
 .../selftests/kvm/x86/sev_migrate_tests.c     |   4 +-
 .../selftests/kvm/x86/sync_regs_test.c        |   5 +-
 .../selftests/kvm/x86/tsc_scaling_sync.c      |   4 +-
 .../selftests/kvm/x86/xapic_ipi_test.c        |  31 +----
 .../selftests/kvm/x86/xen_shinfo_test.c       |  10 +-
 30 files changed, 187 insertions(+), 232 deletions(-)


base-commit: 3c7d7f908d574277a845423ec32250a8d8df44c8
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 01/12] KVM: selftests: Add a KVM syscall wrapper for sched_setaffinity()

[Sean Christopherson]

Add and use a KVM wrapper for sched_setaffinity() so that selftests don't
need to manually assert that the syscall succeeded.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/include/kvm_syscalls.h | 1 +
 tools/testing/selftests/kvm/rseq_test.c            | 6 ++----
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
index 5dae6143ddb0..01dca99009c2 100644
--- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
+++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
@@ -95,6 +95,7 @@ __KVM_SYSCALL_DEFINE(fallocate, 4, int, fd, int, mode, loff_t, offset, loff_t, l
 __KVM_SYSCALL_DEFINE(ftruncate, 2, unsigned int, fd, off_t, length);
 __KVM_SYSCALL_DEFINE(madvise, 3, void *, addr, size_t, length, int, advice);
 __KVM_SYSCALL_DEFINE(sched_getaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
+__KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
 
 #define kvm_free_fd(fd)		\
 do {				\
diff --git a/tools/testing/selftests/kvm/rseq_test.c b/tools/testing/selftests/kvm/rseq_test.c
index 557e393c223b..edcaf0fbe20c 100644
--- a/tools/testing/selftests/kvm/rseq_test.c
+++ b/tools/testing/selftests/kvm/rseq_test.c
@@ -75,7 +75,7 @@ static void *migration_worker(void *__rseq_tid)
 {
 	pid_t rseq_tid = (pid_t)(unsigned long)__rseq_tid;
 	cpu_set_t allowed_mask;
-	int r, i, cpu;
+	int i, cpu;
 
 	CPU_ZERO(&allowed_mask);
 
@@ -96,9 +96,7 @@ static void *migration_worker(void *__rseq_tid)
 		 * stable, i.e. while changing affinity is in-progress.
 		 */
 		smp_wmb();
-		r = sched_setaffinity(rseq_tid, sizeof(allowed_mask), &allowed_mask);
-		TEST_ASSERT(!r, "sched_setaffinity failed, errno = %d (%s)",
-			    errno, strerror(errno));
+		kvm_sched_setaffinity(rseq_tid, sizeof(allowed_mask), &allowed_mask);
 		smp_wmb();
 		atomic_inc(&seq_cnt);
 
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 02/12] KVM: selftests: Set threads CPU affinity before doing work in hardware disable test

[Sean Christopherson]

Affine worker threads to their target pCPU set *before* doing any work in
the hardware disable test.  When running a preemptible kernel on a NUMA
system with a large number of CPUs (100+ of CPUs), waiting to set the
affinity of the grandchildren until after the grandchild is fully spun up
results in total test runtimes of over a minute, presumably due to
cross-node NUMA penalties and scheduler lock contention.

E.g. from the perspective of the caller, when waiting to affine the tasks,
pthread_create() takes ~100x longer.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/hardware_disable_test.c     | 23 +++++++------------
 1 file changed, 8 insertions(+), 15 deletions(-)

diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index 3147f5c97e94..f088b4af3ccc 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -21,6 +21,7 @@
 #define FORK_NUM (1ULL << 9)
 #define DELAY_US_MAX 2000
 
+static cpu_set_t threads_cpu_set;
 sem_t *sem;
 
 static void guest_code(void)
@@ -35,6 +36,8 @@ static void *run_vcpu(void *arg)
 	struct kvm_vcpu *vcpu = arg;
 	struct kvm_run *run = vcpu->run;
 
+	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
+
 	vcpu_run(vcpu);
 
 	TEST_ASSERT(false, "%s: exited with reason %d: %s",
@@ -47,6 +50,8 @@ static void *sleeping_thread(void *arg)
 {
 	int fd;
 
+	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
+
 	while (true) {
 		fd = open("/dev/null", O_RDWR);
 		close(fd);
@@ -64,14 +69,6 @@ static inline void check_create_thread(pthread_t *thread, pthread_attr_t *attr,
 	TEST_ASSERT(r == 0, "%s: failed to create thread", __func__);
 }
 
-static inline void check_set_affinity(pthread_t thread, cpu_set_t *cpu_set)
-{
-	int r;
-
-	r = pthread_setaffinity_np(thread, sizeof(cpu_set_t), cpu_set);
-	TEST_ASSERT(r == 0, "%s: failed set affinity", __func__);
-}
-
 static inline void check_join(pthread_t thread, void **retval)
 {
 	int r;
@@ -84,16 +81,11 @@ static void run_test(u32 run)
 {
 	struct kvm_vcpu *vcpu;
 	struct kvm_vm *vm;
-	cpu_set_t cpu_set;
 	pthread_t threads[VCPU_NUM];
 	pthread_t throw_away;
 	void *b;
 	u32 i, j;
 
-	CPU_ZERO(&cpu_set);
-	for (i = 0; i < VCPU_NUM; i++)
-		CPU_SET(i, &cpu_set);
-
 	vm = vm_create(VCPU_NUM);
 
 	pr_debug("%s: [%d] start vcpus\n", __func__, run);
@@ -101,12 +93,10 @@ static void run_test(u32 run)
 		vcpu = vm_vcpu_add(vm, i, guest_code);
 
 		check_create_thread(&threads[i], NULL, run_vcpu, vcpu);
-		check_set_affinity(threads[i], &cpu_set);
 
 		for (j = 0; j < SLEEPING_THREAD_NUM; ++j) {
 			check_create_thread(&throw_away, NULL, sleeping_thread,
 					    (void *)NULL);
-			check_set_affinity(throw_away, &cpu_set);
 		}
 	}
 	pr_debug("%s: [%d] all threads launched\n", __func__, run);
@@ -153,6 +143,9 @@ int main(int argc, char **argv)
 	int s, r;
 	pid_t pid;
 
+	for (i = 0; i < VCPU_NUM; i++)
+		CPU_SET(i, &threads_cpu_set);
+
 	sem = sem_open("vm_sem", O_CREAT | O_EXCL, 0644, 0);
 	sem_unlink("vm_sem");
 
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 03/12] KVM: selftests: Pre-set threads affinity in hardware disable test when possible

[Sean Christopherson]

When building against a libc that provides pthread_attr_setaffinity_np(),
pre-set the grandchildren threads CPU affinity in the hardware disable test
to further reduce the perceived latency of pthread_create().  On large NUMA
systems, this reduces the average runtime from ~10s to ~5s.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../testing/selftests/kvm/hardware_disable_test.c  | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index f088b4af3ccc..b4edeadfa3f4 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -36,7 +36,9 @@ static void *run_vcpu(void *arg)
 	struct kvm_vcpu *vcpu = arg;
 	struct kvm_run *run = vcpu->run;
 
+#ifndef _GNU_SOURCE
 	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
+#endif
 
 	vcpu_run(vcpu);
 
@@ -50,7 +52,9 @@ static void *sleeping_thread(void *arg)
 {
 	int fd;
 
+#ifndef _GNU_SOURCE
 	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
+#endif
 
 	while (true) {
 		fd = open("/dev/null", O_RDWR);
@@ -80,22 +84,28 @@ static inline void check_join(pthread_t thread, void **retval)
 static void run_test(u32 run)
 {
 	struct kvm_vcpu *vcpu;
+	pthread_attr_t attr;
 	struct kvm_vm *vm;
 	pthread_t threads[VCPU_NUM];
 	pthread_t throw_away;
 	void *b;
 	u32 i, j;
 
+	TEST_ASSERT_EQ(pthread_attr_init(&attr), 0);
+#ifdef _GNU_SOURCE
+	TEST_ASSERT_EQ(pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &threads_cpu_set), 0);
+#endif
+
 	vm = vm_create(VCPU_NUM);
 
 	pr_debug("%s: [%d] start vcpus\n", __func__, run);
 	for (i = 0; i < VCPU_NUM; ++i) {
 		vcpu = vm_vcpu_add(vm, i, guest_code);
 
-		check_create_thread(&threads[i], NULL, run_vcpu, vcpu);
+		check_create_thread(&threads[i], &attr, run_vcpu, vcpu);
 
 		for (j = 0; j < SLEEPING_THREAD_NUM; ++j) {
-			check_create_thread(&throw_away, NULL, sleeping_thread,
+			check_create_thread(&throw_away, &attr, sleeping_thread,
 					    (void *)NULL);
 		}
 	}
-- 
2.55.0.508.g3f0d502094-goog

Re: [PATCH v3 03/12] KVM: selftests: Pre-set threads affinity in hardware disable test when possible

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The fallback logic for setting thread CPU affinity is guarded by `#ifndef _GNU_SOURCE`, which is always false because KVM selftests unconditionally define `_GNU_SOURCE`. This renders the fallback dead code and breaks builds on non-glibc systems.
--

--- Patch [3]: [PATCH v3 03/12] KVM: selftests: Pre-set threads affinity in hardware disable test when possible ---
commit 242db3bb38cdcdb7592d9d34951b2c689c2c39b5
Author: Sean Christopherson <seanjc@google.com>

KVM: selftests: Pre-set threads affinity in hardware disable test when possible

This commit pre-sets thread affinity using pthread_attr_setaffinity_np to
reduce pthread creation latency on large NUMA systems. It attempts to add a
fallback using kvm_sched_setaffinity when _GNU_SOURCE is not defined.

> diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
> index f088b4af3ccc7..b4edeadfa3f45 100644
> --- a/tools/testing/selftests/kvm/hardware_disable_test.c
> +++ b/tools/testing/selftests/kvm/hardware_disable_test.c
> @@ -36,7 +36,9 @@ static void *run_vcpu(void *arg)
>  	struct kvm_vcpu *vcpu = arg;
>  	struct kvm_run *run = vcpu->run;
>  
> +#ifndef _GNU_SOURCE
>  	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
> +#endif

[Severity: Low]
Since the KVM selftests build system unconditionally defines _GNU_SOURCE,
will this fallback code ever be reached?

[ ... ]

> @@ -80,22 +84,28 @@ static inline void check_join(pthread_t thread, void **retval)
>  static void run_test(u32 run)
>  {
>  	struct kvm_vcpu *vcpu;
> +	pthread_attr_t attr;
>  	struct kvm_vm *vm;
>  	pthread_t threads[VCPU_NUM];
>  	pthread_t throw_away;
>  	void *b;
>  	u32 i, j;
>  
> +	TEST_ASSERT_EQ(pthread_attr_init(&attr), 0);
> +#ifdef _GNU_SOURCE
> +	TEST_ASSERT_EQ(pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &threads_cpu_set), 0);
> +#endif

[Severity: Low]
Could this break the build on systems using alternative libc implementations
like musl?

If _GNU_SOURCE is unconditionally defined by the Makefile but the libc
doesn't provide pthread_attr_setaffinity_np, this might result in a
compilation error rather than using the intended fallback.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260731195612.2697986-1-seanjc@google.com?part=3

Re: [PATCH v3 03/12] KVM: selftests: Pre-set threads affinity in hardware disable test when possible

[Sean Christopherson]

+Aqib

On Fri, Jul 31, 2026, sashiko-bot@kernel.org wrote:
> > diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
> > index f088b4af3ccc7..b4edeadfa3f45 100644
> > --- a/tools/testing/selftests/kvm/hardware_disable_test.c
> > +++ b/tools/testing/selftests/kvm/hardware_disable_test.c
> > @@ -36,7 +36,9 @@ static void *run_vcpu(void *arg)
> >  	struct kvm_vcpu *vcpu = arg;
> >  	struct kvm_run *run = vcpu->run;
> >  
> > +#ifndef _GNU_SOURCE
> >  	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
> > +#endif
> 
> [Severity: Low]
> Since the KVM selftests build system unconditionally defines _GNU_SOURCE,
> will this fallback code ever be reached?
> 
> [ ... ]
> 
> > @@ -80,22 +84,28 @@ static inline void check_join(pthread_t thread, void **retval)
> >  static void run_test(u32 run)
> >  {
> >  	struct kvm_vcpu *vcpu;
> > +	pthread_attr_t attr;
> >  	struct kvm_vm *vm;
> >  	pthread_t threads[VCPU_NUM];
> >  	pthread_t throw_away;
> >  	void *b;
> >  	u32 i, j;
> >  
> > +	TEST_ASSERT_EQ(pthread_attr_init(&attr), 0);
> > +#ifdef _GNU_SOURCE
> > +	TEST_ASSERT_EQ(pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &threads_cpu_set), 0);
> > +#endif

Aqib,

Do you happen to definitively know which is the "right" macro to check to detect
support for pthread_attr_setaffinity_np()?  Belatedly, I saw that you used __GLIBC__
in a patch that ultimately wasn't applied[*].

I initially used __USE_GNU here, but then Sashiko yelled at me, so I switched to
_GNU_SOURCE based on Sashiko's suggestion, and now it's yelling at me again :-)


[*] https://lore.kernel.org/all/20250829142556.72577-4-aqibaf@amazon.com

> [Severity: Low]
> Could this break the build on systems using alternative libc implementations
> like musl?
> 
> If _GNU_SOURCE is unconditionally defined by the Makefile but the libc
> doesn't provide pthread_attr_setaffinity_np, this might result in a
> compilation error rather than using the intended fallback.
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20260731195612.2697986-1-seanjc@google.com?part=3

[PATCH v3 04/12] KVM: selftests: Return the target CPU from pin_task_to_random_cpu()

[Sean Christopherson]

When pinning a task to a random CPU, return which CPU the task was pinned
to so that the caller can do things like avoid running other tasks on the
target CPU.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/include/kvm_util.h | 2 +-
 tools/testing/selftests/kvm/lib/kvm_util.c     | 5 +++--
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h
index b39e713c30a4..61d72e1e7601 100644
--- a/tools/testing/selftests/kvm/include/kvm_util.h
+++ b/tools/testing/selftests/kvm/include/kvm_util.h
@@ -1094,7 +1094,7 @@ static inline void pin_task_to_cpu(pthread_t task, int cpu)
 	TEST_ASSERT(!r, "Failed to set thread affinity to pCPU '%u'", cpu);
 }
 
-void pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus);
+int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus);
 
 static inline int pin_task_to_any_cpu(pthread_t task)
 {
diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c
index 3794575d2ca0..af2bad0553f9 100644
--- a/tools/testing/selftests/kvm/lib/kvm_util.c
+++ b/tools/testing/selftests/kvm/lib/kvm_util.c
@@ -668,7 +668,7 @@ void kvm_print_vcpu_pinning_help(void)
 	       "     (default: no pinning)\n", name, name);
 }
 
-void pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
+int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
 {
 	int target_idx;
 	int nr_cpus;
@@ -682,11 +682,12 @@ void pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
 	for (cpu = 0; cpu < CPU_SETSIZE; cpu++) {
 		if (CPU_ISSET(cpu, possible_cpus) && target_idx-- == 0) {
 			pin_task_to_cpu(task, cpu);
-			return;
+			return cpu;
 		}
 	}
 
 	TEST_FAIL("Failed to find random CPU in possible_cpus");
+	return -1;
 }
 
 void kvm_parse_vcpu_pinning(const char *pcpus_string, u32 vcpu_to_pcpu[],
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 05/12] KVM: selftests: Extract picking of random CPU from cpu_set_t to separate API

[Sean Christopherson]

Extract kvm_pick_random_cpu() out of pin_task_to_random_cpu() so that tests
can choose a random CPU without having to immediately pin a task to that
CPU.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/include/kvm_util.h | 11 ++++++++++-
 tools/testing/selftests/kvm/lib/kvm_util.c     |  7 ++-----
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h
index 61d72e1e7601..5a7a455b5387 100644
--- a/tools/testing/selftests/kvm/include/kvm_util.h
+++ b/tools/testing/selftests/kvm/include/kvm_util.h
@@ -1084,6 +1084,8 @@ struct kvm_vcpu *vm_recreate_with_one_vcpu(struct kvm_vm *vm);
 
 void kvm_set_files_rlimit(u32 nr_vcpus);
 
+int kvm_pick_random_cpu(cpu_set_t *possible_cpus);
+
 int __pin_task_to_cpu(pthread_t task, int cpu);
 
 static inline void pin_task_to_cpu(pthread_t task, int cpu)
@@ -1094,7 +1096,14 @@ static inline void pin_task_to_cpu(pthread_t task, int cpu)
 	TEST_ASSERT(!r, "Failed to set thread affinity to pCPU '%u'", cpu);
 }
 
-int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus);
+static inline int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
+{
+	int cpu;
+
+	cpu = kvm_pick_random_cpu(possible_cpus);
+	pin_task_to_cpu(task, cpu);
+	return cpu;
+}
 
 static inline int pin_task_to_any_cpu(pthread_t task)
 {
diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c
index af2bad0553f9..9ddc047d5c27 100644
--- a/tools/testing/selftests/kvm/lib/kvm_util.c
+++ b/tools/testing/selftests/kvm/lib/kvm_util.c
@@ -668,7 +668,7 @@ void kvm_print_vcpu_pinning_help(void)
 	       "     (default: no pinning)\n", name, name);
 }
 
-int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
+int kvm_pick_random_cpu(cpu_set_t *possible_cpus)
 {
 	int target_idx;
 	int nr_cpus;
@@ -680,12 +680,9 @@ int pin_task_to_random_cpu(pthread_t task, cpu_set_t *possible_cpus)
 	target_idx = kvm_random_u64(&kvm_rng) % nr_cpus;
 
 	for (cpu = 0; cpu < CPU_SETSIZE; cpu++) {
-		if (CPU_ISSET(cpu, possible_cpus) && target_idx-- == 0) {
-			pin_task_to_cpu(task, cpu);
+		if (CPU_ISSET(cpu, possible_cpus) && target_idx-- == 0)
 			return cpu;
-		}
 	}
-
 	TEST_FAIL("Failed to find random CPU in possible_cpus");
 	return -1;
 }
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 06/12] KVM: selftests: Affine threads to random CPUs in hardware disable test

[Sean Christopherson]

Affine the worker threads to random CPUs in the hardware disable test, and
honor the starting CPU set in the process.  Hardcoding to CPUs 0-3 is both
wrong and less interesting than running on random CPUs.

Opportunistically convert the local 'i' to an int, as the macros it tests
against are signed, and using 'i' as an iterator is conventionally declared
as an int.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/hardware_disable_test.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index b4edeadfa3f4..fcc06c602fca 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -149,12 +149,17 @@ void wait_for_child_setup(pid_t pid)
 
 int main(int argc, char **argv)
 {
-	u32 i;
-	int s, r;
+	cpu_set_t allowed_cpu_set;
+	int s, r, cpu, i;
 	pid_t pid;
 
-	for (i = 0; i < VCPU_NUM; i++)
-		CPU_SET(i, &threads_cpu_set);
+	kvm_sched_getaffinity(0, sizeof(cpu_set_t), &allowed_cpu_set);
+
+	for (i = 0; i < VCPU_NUM && CPU_COUNT(&allowed_cpu_set); i++) {
+		cpu = kvm_pick_random_cpu(&allowed_cpu_set);
+		CPU_CLR(cpu, &allowed_cpu_set);
+		CPU_SET(cpu, &threads_cpu_set);
+	}
 
 	sem = sem_open("vm_sem", O_CREAT | O_EXCL, 0644, 0);
 	sem_unlink("vm_sem");
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 07/12] KVM: selftests: Drop unreachable, dead code from hardware disable test

[Sean Christopherson]

Drop all of the code that is unreachable in the hardware disable test, as
it adds a lot of noise that makes the test seem far more complicated than
it actually is.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/hardware_disable_test.c     | 41 +++++++------------
 1 file changed, 14 insertions(+), 27 deletions(-)

diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index fcc06c602fca..de8e63da7b58 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -15,6 +15,7 @@
 #include <test_util.h>
 
 #include "kvm_util.h"
+#include "ucall_common.h"
 
 #define VCPU_NUM 4
 #define SLEEPING_THREAD_NUM (1 << 4)
@@ -28,7 +29,7 @@ static void guest_code(void)
 {
 	for (;;)
 		;  /* Some busy work */
-	printf("Should not be reached.\n");
+	GUEST_ASSERT(0);
 }
 
 static void *run_vcpu(void *arg)
@@ -42,10 +43,8 @@ static void *run_vcpu(void *arg)
 
 	vcpu_run(vcpu);
 
-	TEST_ASSERT(false, "%s: exited with reason %d: %s",
-		    __func__, run->exit_reason,
-		    exit_reason_str(run->exit_reason));
-	pthread_exit(NULL);
+	TEST_FAIL("vCPU%d exited with reason %d: %s",
+		  vcpu->id, run->exit_reason, exit_reason_str(run->exit_reason));
 }
 
 static void *sleeping_thread(void *arg)
@@ -56,12 +55,11 @@ static void *sleeping_thread(void *arg)
 	kvm_sched_setaffinity(0, sizeof(cpu_set_t), &threads_cpu_set);
 #endif
 
-	while (true) {
+	while (1) {
 		fd = open("/dev/null", O_RDWR);
 		close(fd);
 	}
-	TEST_ASSERT(false, "%s: exited", __func__);
-	pthread_exit(NULL);
+	TEST_FAIL("%s: exited", __func__);
 }
 
 static inline void check_create_thread(pthread_t *thread, pthread_attr_t *attr,
@@ -73,22 +71,12 @@ static inline void check_create_thread(pthread_t *thread, pthread_attr_t *attr,
 	TEST_ASSERT(r == 0, "%s: failed to create thread", __func__);
 }
 
-static inline void check_join(pthread_t thread, void **retval)
-{
-	int r;
-
-	r = pthread_join(thread, retval);
-	TEST_ASSERT(r == 0, "%s: failed to join thread", __func__);
-}
-
 static void run_test(u32 run)
 {
 	struct kvm_vcpu *vcpu;
 	pthread_attr_t attr;
 	struct kvm_vm *vm;
-	pthread_t threads[VCPU_NUM];
-	pthread_t throw_away;
-	void *b;
+	pthread_t thread;
 	u32 i, j;
 
 	TEST_ASSERT_EQ(pthread_attr_init(&attr), 0);
@@ -102,19 +90,18 @@ static void run_test(u32 run)
 	for (i = 0; i < VCPU_NUM; ++i) {
 		vcpu = vm_vcpu_add(vm, i, guest_code);
 
-		check_create_thread(&threads[i], &attr, run_vcpu, vcpu);
+		check_create_thread(&thread, &attr, run_vcpu, vcpu);
 
-		for (j = 0; j < SLEEPING_THREAD_NUM; ++j) {
-			check_create_thread(&throw_away, &attr, sleeping_thread,
+		for (j = 0; j < SLEEPING_THREAD_NUM; ++j)
+			check_create_thread(&thread, &attr, sleeping_thread,
 					    (void *)NULL);
-		}
 	}
 	pr_debug("%s: [%d] all threads launched\n", __func__, run);
 	sem_post(sem);
-	for (i = 0; i < VCPU_NUM; ++i)
-		check_join(threads[i], &b);
-	/* Should not be reached */
-	TEST_ASSERT(false, "%s: [%d] child escaped the ninja", __func__, run);
+
+	/* Wait for the parent to SIGKILL this child. */
+	while (1)
+		pause();
 }
 
 void wait_for_child_setup(pid_t pid)
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 08/12] KVM: selftests: Add KVM syscall wrapper for pthread_create()

[Sean Christopherson]

Add and use a KVM wrapper for pthread_create() syscall so that selftests
don't need to manually assert that the syscall succeeded.

Note, most tests don't actually assert success, but they all obviously
rely on the syscall to succeed.

Other than explicitly failing if pthread_create() fails, no functional
change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/arch_timer.c         | 14 ++++----------
 tools/testing/selftests/kvm/arm64/vgic_irq.c     | 10 +++-------
 .../selftests/kvm/arm64/vgic_lpi_stress.c        |  4 ++--
 tools/testing/selftests/kvm/dirty_log_test.c     |  2 +-
 .../selftests/kvm/hardware_disable_test.c        | 16 +++-------------
 .../testing/selftests/kvm/include/kvm_syscalls.h |  5 +++++
 tools/testing/selftests/kvm/irq_test.c           |  2 +-
 tools/testing/selftests/kvm/irqfd_test.c         |  2 +-
 .../testing/selftests/kvm/kvm_page_table_test.c  |  4 ++--
 tools/testing/selftests/kvm/lib/memstress.c      |  2 +-
 .../testing/selftests/kvm/lib/userfaultfd_util.c |  4 ++--
 tools/testing/selftests/kvm/memslot_perf_test.c  |  2 +-
 tools/testing/selftests/kvm/mmu_stress_test.c    |  2 +-
 .../selftests/kvm/pre_fault_memory_test.c        |  2 +-
 tools/testing/selftests/kvm/rseq_test.c          |  4 ++--
 tools/testing/selftests/kvm/s390/memop.c         |  2 +-
 .../selftests/kvm/set_memory_region_test.c       |  2 +-
 tools/testing/selftests/kvm/steal_time.c         |  2 +-
 tools/testing/selftests/kvm/x86/hyperv_ipi.c     | 11 ++++-------
 .../testing/selftests/kvm/x86/hyperv_tlb_flush.c |  9 +++------
 .../kvm/x86/private_mem_conversions_test.c       |  2 +-
 .../kvm/x86/private_mem_kvm_exits_test.c         |  5 ++---
 .../selftests/kvm/x86/recalc_apic_map_test.c     |  2 +-
 .../selftests/kvm/x86/sev_migrate_tests.c        |  2 +-
 tools/testing/selftests/kvm/x86/sync_regs_test.c |  2 +-
 .../testing/selftests/kvm/x86/tsc_scaling_sync.c |  2 +-
 tools/testing/selftests/kvm/x86/xapic_ipi_test.c |  8 ++------
 .../testing/selftests/kvm/x86/xen_shinfo_test.c  |  3 +--
 28 files changed, 50 insertions(+), 77 deletions(-)

diff --git a/tools/testing/selftests/kvm/arch_timer.c b/tools/testing/selftests/kvm/arch_timer.c
index f8b02597897b..a6a66b5b7ea9 100644
--- a/tools/testing/selftests/kvm/arch_timer.c
+++ b/tools/testing/selftests/kvm/arch_timer.c
@@ -141,28 +141,22 @@ static void test_run(struct kvm_vm *vm)
 {
 	pthread_t pt_vcpu_migration;
 	unsigned int i;
-	int ret;
 
 	pthread_mutex_init(&vcpu_done_map_lock, NULL);
 	vcpu_done_map = bitmap_zalloc(test_args.nr_vcpus);
 	TEST_ASSERT(vcpu_done_map, "Failed to allocate vcpu done bitmap");
 
-	for (i = 0; i < (unsigned long)test_args.nr_vcpus; i++) {
-		ret = pthread_create(&pt_vcpu_run[i], NULL, test_vcpu_run,
-				     (void *)(unsigned long)i);
-		TEST_ASSERT(!ret, "Failed to create vCPU-%d pthread", i);
-	}
+	for (i = 0; i < (unsigned long)test_args.nr_vcpus; i++)
+		kvm_pthread_create(&pt_vcpu_run[i], NULL, test_vcpu_run,
+				   (void *)(unsigned long)i);
 
 	/* Spawn a thread to control the vCPU migrations */
 	if (test_args.migration_freq_ms) {
 		srand(time(NULL));
 
-		ret = pthread_create(&pt_vcpu_migration, NULL,
-					test_vcpu_migration, NULL);
-		TEST_ASSERT(!ret, "Failed to create the migration pthread");
+		kvm_pthread_create(&pt_vcpu_migration, NULL, test_vcpu_migration, NULL);
 	}
 
-
 	for (i = 0; i < test_args.nr_vcpus; i++)
 		pthread_join(pt_vcpu_run[i], NULL);
 
diff --git a/tools/testing/selftests/kvm/arm64/vgic_irq.c b/tools/testing/selftests/kvm/arm64/vgic_irq.c
index 5e231998617e..2d6f20d5c117 100644
--- a/tools/testing/selftests/kvm/arm64/vgic_irq.c
+++ b/tools/testing/selftests/kvm/arm64/vgic_irq.c
@@ -988,7 +988,7 @@ static void test_vgic_two_cpus(void *gcode)
 	struct test_args args = {};
 	struct kvm_vm *vm;
 	gva_t args_gva;
-	int gic_fd, ret;
+	int gic_fd;
 
 	vm = vm_create_with_vcpus(2, gcode, vcpus);
 
@@ -1004,12 +1004,8 @@ static void test_vgic_two_cpus(void *gcode)
 
 	gic_fd = vgic_v3_setup(vm, 2, 64);
 
-	ret = pthread_create(&thr[0], NULL, test_vcpu_run, vcpus[0]);
-	if (ret)
-		TEST_FAIL("Can't create thread for vcpu 0 (%d)\n", ret);
-	ret = pthread_create(&thr[1], NULL, test_vcpu_run, vcpus[1]);
-	if (ret)
-		TEST_FAIL("Can't create thread for vcpu 1 (%d)\n", ret);
+	kvm_pthread_create(&thr[0], NULL, test_vcpu_run, vcpus[0]);
+	kvm_pthread_create(&thr[1], NULL, test_vcpu_run, vcpus[1]);
 
 	pthread_join(thr[0], NULL);
 	pthread_join(thr[1], NULL);
diff --git a/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c b/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
index d64d434d3f06..549e0547d905 100644
--- a/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
+++ b/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
@@ -311,10 +311,10 @@ static void run_test(void)
 	pthread_barrier_init(&test_setup_barrier, NULL, nr_vcpus + nr_devices + 1);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_create(&vcpu_threads[i], NULL, vcpu_worker_thread, vcpus[i]);
+		kvm_pthread_create(&vcpu_threads[i], NULL, vcpu_worker_thread, vcpus[i]);
 
 	for (i = 0; i < nr_devices; i++)
-		pthread_create(&lpi_threads[i], NULL, lpi_worker_thread, (void *)i);
+		kvm_pthread_create(&lpi_threads[i], NULL, lpi_worker_thread, (void *)i);
 
 	pthread_barrier_wait(&test_setup_barrier);
 
diff --git a/tools/testing/selftests/kvm/dirty_log_test.c b/tools/testing/selftests/kvm/dirty_log_test.c
index e8419d7da1ea..ca22e5f09ac2 100644
--- a/tools/testing/selftests/kvm/dirty_log_test.c
+++ b/tools/testing/selftests/kvm/dirty_log_test.c
@@ -691,7 +691,7 @@ static void run_test(enum vm_guest_mode mode, void *arg)
 
 	TEST_ASSERT_EQ(vcpu_stop, false);
 
-	pthread_create(&vcpu_thread, NULL, vcpu_worker, vcpu);
+	kvm_pthread_create(&vcpu_thread, NULL, vcpu_worker, vcpu);
 
 	for (iteration = 1; iteration <= p->iterations; iteration++) {
 		unsigned long i, reap_i;
diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index de8e63da7b58..72bac346bdd9 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -5,7 +5,6 @@
  * return notifiers.
  */
 #include <fcntl.h>
-#include <pthread.h>
 #include <semaphore.h>
 #include <stdint.h>
 #include <stdlib.h>
@@ -14,6 +13,7 @@
 
 #include <test_util.h>
 
+#include "kvm_syscalls.h"
 #include "kvm_util.h"
 #include "ucall_common.h"
 
@@ -62,15 +62,6 @@ static void *sleeping_thread(void *arg)
 	TEST_FAIL("%s: exited", __func__);
 }
 
-static inline void check_create_thread(pthread_t *thread, pthread_attr_t *attr,
-				       void *(*f)(void *), void *arg)
-{
-	int r;
-
-	r = pthread_create(thread, attr, f, arg);
-	TEST_ASSERT(r == 0, "%s: failed to create thread", __func__);
-}
-
 static void run_test(u32 run)
 {
 	struct kvm_vcpu *vcpu;
@@ -90,11 +81,10 @@ static void run_test(u32 run)
 	for (i = 0; i < VCPU_NUM; ++i) {
 		vcpu = vm_vcpu_add(vm, i, guest_code);
 
-		check_create_thread(&thread, &attr, run_vcpu, vcpu);
+		kvm_pthread_create(&thread, &attr, run_vcpu, vcpu);
 
 		for (j = 0; j < SLEEPING_THREAD_NUM; ++j)
-			check_create_thread(&thread, &attr, sleeping_thread,
-					    (void *)NULL);
+			kvm_pthread_create(&thread, &attr, sleeping_thread, (void *)NULL);
 	}
 	pr_debug("%s: [%d] all threads launched\n", __func__, run);
 	sem_post(sem);
diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
index 01dca99009c2..e75be8378779 100644
--- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
+++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
@@ -12,6 +12,7 @@
 #include <sys/mman.h>
 #include <sys/syscall.h>
 
+#include <pthread.h>
 #include <sched.h>
 #include <test_util.h>
 
@@ -97,6 +98,10 @@ __KVM_SYSCALL_DEFINE(madvise, 3, void *, addr, size_t, length, int, advice);
 __KVM_SYSCALL_DEFINE(sched_getaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
 __KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
 
+typedef void *(*pthread_fn_t)(void *);
+__KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
+		     const pthread_attr_t *, attr, pthread_fn_t, fn, void *, arg);
+
 #define kvm_free_fd(fd)		\
 do {				\
 	kvm_close(fd);		\
diff --git a/tools/testing/selftests/kvm/irq_test.c b/tools/testing/selftests/kvm/irq_test.c
index 240f6f0fdbe4..7da5f8d75e85 100644
--- a/tools/testing/selftests/kvm/irq_test.c
+++ b/tools/testing/selftests/kvm/irq_test.c
@@ -296,7 +296,7 @@ int main(int argc, char **argv)
 		kvm_sched_getaffinity(0, sizeof(available_cpus), &available_cpus);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_create(&vcpu_threads[i], NULL, vcpu_thread_main, vcpus[i]);
+		kvm_pthread_create(&vcpu_threads[i], NULL, vcpu_thread_main, vcpus[i]);
 
 	for (i = 0; i < nr_vcpus; i++) {
 		struct kvm_vcpu *vcpu = vcpus[i];
diff --git a/tools/testing/selftests/kvm/irqfd_test.c b/tools/testing/selftests/kvm/irqfd_test.c
index 5d7590d01868..8b39144a2834 100644
--- a/tools/testing/selftests/kvm/irqfd_test.c
+++ b/tools/testing/selftests/kvm/irqfd_test.c
@@ -128,7 +128,7 @@ int main(int argc, char *argv[])
 
 	close(__eventfd);
 
-	pthread_create(&racing_thread, NULL, secondary_irqfd_juggler, vm2);
+	kvm_pthread_create(&racing_thread, NULL, secondary_irqfd_juggler, vm2);
 
 	for (i = 0; i < 10000; i++) {
 		WRITE_ONCE(__eventfd, kvm_new_eventfd());
diff --git a/tools/testing/selftests/kvm/kvm_page_table_test.c b/tools/testing/selftests/kvm/kvm_page_table_test.c
index a910e3abb8c7..8222bdb63e49 100644
--- a/tools/testing/selftests/kvm/kvm_page_table_test.c
+++ b/tools/testing/selftests/kvm/kvm_page_table_test.c
@@ -364,8 +364,8 @@ static void run_test(enum vm_guest_mode mode, void *arg)
 	*current_stage = KVM_BEFORE_MAPPINGS;
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_create(&vcpu_threads[i], NULL, vcpu_worker,
-			       test_args.vcpus[i]);
+		kvm_pthread_create(&vcpu_threads[i], NULL, vcpu_worker,
+				   test_args.vcpus[i]);
 
 	vcpus_complete_new_stage(*current_stage);
 	pr_info("Started all vCPUs successfully\n");
diff --git a/tools/testing/selftests/kvm/lib/memstress.c b/tools/testing/selftests/kvm/lib/memstress.c
index 3599b75d97c9..0bfa623ce9b5 100644
--- a/tools/testing/selftests/kvm/lib/memstress.c
+++ b/tools/testing/selftests/kvm/lib/memstress.c
@@ -294,7 +294,7 @@ void memstress_start_vcpu_threads(int nr_vcpus,
 		vcpu->vcpu_idx = i;
 		WRITE_ONCE(vcpu->running, false);
 
-		pthread_create(&vcpu->thread, NULL, vcpu_thread_main, vcpu);
+		kvm_pthread_create(&vcpu->thread, NULL, vcpu_thread_main, vcpu);
 	}
 
 	for (i = 0; i < nr_vcpus; i++) {
diff --git a/tools/testing/selftests/kvm/lib/userfaultfd_util.c b/tools/testing/selftests/kvm/lib/userfaultfd_util.c
index ef8d76f71f83..4b3e3158ed87 100644
--- a/tools/testing/selftests/kvm/lib/userfaultfd_util.c
+++ b/tools/testing/selftests/kvm/lib/userfaultfd_util.c
@@ -167,8 +167,8 @@ struct uffd_desc *uffd_setup_demand_paging(int uffd_mode, useconds_t delay,
 		uffd_desc->reader_args[i].handler = handler;
 		uffd_desc->reader_args[i].pipe = pipes[0];
 
-		pthread_create(&uffd_desc->readers[i], NULL, uffd_handler_thread_fn,
-			       &uffd_desc->reader_args[i]);
+		kvm_pthread_create(&uffd_desc->readers[i], NULL, uffd_handler_thread_fn,
+				   &uffd_desc->reader_args[i]);
 
 		PER_VCPU_DEBUG("Created uffd thread %i for HVA range [%p, %p)\n",
 			       i, hva, hva + len);
diff --git a/tools/testing/selftests/kvm/memslot_perf_test.c b/tools/testing/selftests/kvm/memslot_perf_test.c
index 4d9ad6104a6e..eb49305db404 100644
--- a/tools/testing/selftests/kvm/memslot_perf_test.c
+++ b/tools/testing/selftests/kvm/memslot_perf_test.c
@@ -366,7 +366,7 @@ static void launch_vm(struct vm_data *data)
 {
 	pr_info_v("Launching the test VM\n");
 
-	pthread_create(&data->vcpu_thread, NULL, vcpu_worker, data);
+	kvm_pthread_create(&data->vcpu_thread, NULL, vcpu_worker, data);
 
 	/* Ensure the guest thread is spun up. */
 	wait_for_vcpu();
diff --git a/tools/testing/selftests/kvm/mmu_stress_test.c b/tools/testing/selftests/kvm/mmu_stress_test.c
index 3d5f33a63b2b..9448498849ff 100644
--- a/tools/testing/selftests/kvm/mmu_stress_test.c
+++ b/tools/testing/selftests/kvm/mmu_stress_test.c
@@ -222,7 +222,7 @@ static pthread_t *spawn_workers(struct kvm_vm *vm, struct kvm_vcpu **vcpus,
 		info[i].vcpu = vcpus[i];
 		info[i].start_gpa = gpa;
 		info[i].end_gpa = gpa + nr_bytes;
-		pthread_create(&threads[i], NULL, vcpu_worker, &info[i]);
+		kvm_pthread_create(&threads[i], NULL, vcpu_worker, &info[i]);
 	}
 	return threads;
 }
diff --git a/tools/testing/selftests/kvm/pre_fault_memory_test.c b/tools/testing/selftests/kvm/pre_fault_memory_test.c
index a0fcae3cb7a8..88287ccb9093 100644
--- a/tools/testing/selftests/kvm/pre_fault_memory_test.c
+++ b/tools/testing/selftests/kvm/pre_fault_memory_test.c
@@ -84,7 +84,7 @@ static void pre_fault_memory(struct kvm_vcpu *vcpu, u64 base_gpa, u64 offset,
 	 * Concurrently delete (and recreate) the slot to test KVM's handling
 	 * of a racing memslot deletion with prefaulting.
 	 */
-	pthread_create(&slot_worker, NULL, delete_slot_worker, &data);
+	kvm_pthread_create(&slot_worker, NULL, delete_slot_worker, &data);
 
 	while (!READ_ONCE(data.worker_ready))
 		cpu_relax();
diff --git a/tools/testing/selftests/kvm/rseq_test.c b/tools/testing/selftests/kvm/rseq_test.c
index edcaf0fbe20c..0921b8d90d4a 100644
--- a/tools/testing/selftests/kvm/rseq_test.c
+++ b/tools/testing/selftests/kvm/rseq_test.c
@@ -239,8 +239,8 @@ int main(int argc, char *argv[])
 	 */
 	vm = vm_create_with_one_vcpu(&vcpu, guest_code);
 
-	pthread_create(&migration_thread, NULL, migration_worker,
-		       (void *)(unsigned long)kvm_gettid());
+	kvm_pthread_create(&migration_thread, NULL, migration_worker,
+			   (void *)(unsigned long)kvm_gettid());
 
 	if (latency >= 0) {
 		/*
diff --git a/tools/testing/selftests/kvm/s390/memop.c b/tools/testing/selftests/kvm/s390/memop.c
index 0244848621b3..96b32bd74009 100644
--- a/tools/testing/selftests/kvm/s390/memop.c
+++ b/tools/testing/selftests/kvm/s390/memop.c
@@ -678,7 +678,7 @@ static void test_cmpxchg_key_concurrent(void)
 	HOST_SYNC(t.vcpu, STAGE_SKEYS_SET);
 	prepare_mem12();
 	MOP(t.vcpu, LOGICAL, WRITE, mem1, max_block, GADDR_V(mem2));
-	pthread_create(&thread, NULL, run_guest, &t.vcpu);
+	kvm_pthread_create(&thread, NULL, run_guest, &t.vcpu);
 
 	for (int i = 0; i < cmpxchg_iter_outer; i++) {
 		do {
diff --git a/tools/testing/selftests/kvm/set_memory_region_test.c b/tools/testing/selftests/kvm/set_memory_region_test.c
index a152ab65c657..12f5022bb164 100644
--- a/tools/testing/selftests/kvm/set_memory_region_test.c
+++ b/tools/testing/selftests/kvm/set_memory_region_test.c
@@ -133,7 +133,7 @@ static struct kvm_vm *spawn_vm(struct kvm_vcpu **vcpu, pthread_t *vcpu_thread,
 	hva = addr_gpa2hva(vm, MEM_REGION_GPA);
 	memset(hva, 0, 2 * 4096);
 
-	pthread_create(vcpu_thread, NULL, vcpu_worker, *vcpu);
+	kvm_pthread_create(vcpu_thread, NULL, vcpu_worker, *vcpu);
 
 	/* Ensure the guest thread is spun up. */
 	wait_for_vcpu();
diff --git a/tools/testing/selftests/kvm/steal_time.c b/tools/testing/selftests/kvm/steal_time.c
index 0b88c743ee7d..f5f708be5aa0 100644
--- a/tools/testing/selftests/kvm/steal_time.c
+++ b/tools/testing/selftests/kvm/steal_time.c
@@ -548,7 +548,7 @@ 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, NULL, do_steal_time, NULL);
+		kvm_pthread_create(&thread, NULL, do_steal_time, NULL);
 		pthread_getaffinity_np(thread, sizeof(cpuset), &cpuset);
 		TEST_ASSERT(CPU_COUNT(&cpuset) == 1 && CPU_ISSET(cpu, &cpuset),
 			    "Worker failed to inherit parent's CPU affinity");
diff --git a/tools/testing/selftests/kvm/x86/hyperv_ipi.c b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
index 771535f9aad3..3cf451f3153e 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_ipi.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
@@ -245,7 +245,7 @@ int main(int argc, char *argv[])
 	struct kvm_vcpu *vcpu[3];
 	gva_t hcall_page;
 	pthread_t threads[2];
-	int stage = 1, r;
+	int stage = 1;
 	struct ucall uc;
 
 	TEST_REQUIRE(kvm_has_cap(KVM_CAP_HYPERV_SEND_IPI));
@@ -272,11 +272,8 @@ int main(int argc, char *argv[])
 	vcpu_args_set(vcpu[0], 2, hcall_page, addr_gva2gpa(vm, hcall_page));
 	vcpu_set_hv_cpuid(vcpu[0]);
 
-	r = pthread_create(&threads[0], NULL, vcpu_thread, vcpu[1]);
-	TEST_ASSERT(!r, "pthread_create failed errno=%d", r);
-
-	r = pthread_create(&threads[1], NULL, vcpu_thread, vcpu[2]);
-	TEST_ASSERT(!r, "pthread_create failed errno=%d", errno);
+	kvm_pthread_create(&threads[0], NULL, vcpu_thread, vcpu[1]);
+	kvm_pthread_create(&threads[1], NULL, vcpu_thread, vcpu[2]);
 
 	while (true) {
 		vcpu_run(vcpu[0]);
@@ -306,5 +303,5 @@ int main(int argc, char *argv[])
 	cancel_join_vcpu_thread(threads[1], vcpu[2]);
 	kvm_vm_free(vm);
 
-	return r;
+	return 0;
 }
diff --git a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
index b4be9a175379..87fac1968a47 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
@@ -575,7 +575,7 @@ int main(int argc, char *argv[])
 	u64 *pte;
 	struct test_data *data;
 	struct ucall uc;
-	int stage = 1, r, i;
+	int stage = 1, i;
 
 	TEST_REQUIRE(kvm_has_cap(KVM_CAP_HYPERV_TLBFLUSH));
 
@@ -632,11 +632,8 @@ int main(int argc, char *argv[])
 	vcpu_set_msr(vcpu[2], HV_X64_MSR_VP_INDEX, WORKER_VCPU_ID_2);
 	vcpu_set_hv_cpuid(vcpu[2]);
 
-	r = pthread_create(&threads[0], NULL, vcpu_thread, vcpu[1]);
-	TEST_ASSERT(!r, "pthread_create() failed");
-
-	r = pthread_create(&threads[1], NULL, vcpu_thread, vcpu[2]);
-	TEST_ASSERT(!r, "pthread_create() failed");
+	kvm_pthread_create(&threads[0], NULL, vcpu_thread, vcpu[1]);
+	kvm_pthread_create(&threads[1], NULL, vcpu_thread, vcpu[2]);
 
 	while (true) {
 		vcpu_run(vcpu[0]);
diff --git a/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c b/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
index 1d2f5d4fd45d..e419b08f1fa6 100644
--- a/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
+++ b/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
@@ -412,7 +412,7 @@ static void test_mem_conversions(enum vm_mem_backing_src_type src_type, u32 nr_v
 		 */
 		virt_map(vm, gpa, gpa, PER_CPU_DATA_SIZE / vm->page_size);
 
-		pthread_create(&threads[i], NULL, __test_mem_conversions, vcpus[i]);
+		kvm_pthread_create(&threads[i], NULL, __test_mem_conversions, vcpus[i]);
 	}
 
 	WRITE_ONCE(run_vcpus, true);
diff --git a/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c b/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
index 10db9fe6d906..e26524fb4b50 100644
--- a/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
+++ b/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
@@ -65,9 +65,8 @@ static void test_private_access_memslot_deleted(void)
 	/* Request to access page privately */
 	vm_mem_set_private(vm, EXITS_TEST_GPA, EXITS_TEST_SIZE);
 
-	pthread_create(&vm_thread, NULL,
-		       (void *(*)(void *))run_vcpu_get_exit_reason,
-		       (void *)vcpu);
+	kvm_pthread_create(&vm_thread, NULL,
+			   (pthread_fn_t)run_vcpu_get_exit_reason, (void *)vcpu);
 
 	vm_mem_region_delete(vm, EXITS_TEST_SLOT);
 
diff --git a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
index cbc92a862ea9..13abb65620d5 100644
--- a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
+++ b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
@@ -57,7 +57,7 @@ int main(void)
 	for (i = 0; i < KVM_MAX_VCPUS; i++)
 		vcpu_set_msr(vcpus[i], MSR_IA32_APICBASE, LAPIC_X2APIC);
 
-	TEST_ASSERT_EQ(pthread_create(&thread, NULL, race, vcpus[0]), 0);
+	kvm_pthread_create(&thread, NULL, race, vcpus[0]);
 
 	vcpuN = vcpus[KVM_MAX_VCPUS - 1];
 	for (t = time(NULL) + TIMEOUT; time(NULL) < t;) {
diff --git a/tools/testing/selftests/kvm/x86/sev_migrate_tests.c b/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
index d59abb198d86..66c7ce49620a 100644
--- a/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
+++ b/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
@@ -128,7 +128,7 @@ static void test_sev_migrate_locking(void)
 		       sizeof(input[i].source_vms));
 
 	for (i = 0; i < NR_LOCK_TESTING_THREADS; ++i)
-		pthread_create(&pt[i], NULL, locking_test_thread, &input[i]);
+		kvm_pthread_create(&pt[i], NULL, locking_test_thread, &input[i]);
 
 	for (i = 0; i < NR_LOCK_TESTING_THREADS; ++i)
 		pthread_join(pt[i], NULL);
diff --git a/tools/testing/selftests/kvm/x86/sync_regs_test.c b/tools/testing/selftests/kvm/x86/sync_regs_test.c
index 5b0c2359bbb4..f7228af7d594 100644
--- a/tools/testing/selftests/kvm/x86/sync_regs_test.c
+++ b/tools/testing/selftests/kvm/x86/sync_regs_test.c
@@ -181,7 +181,7 @@ static void race_sync_regs(struct kvm_vcpu *vcpu, void *racer)
 		    !!(run->s.regs.sregs.cr4 & X86_CR4_PAE),
 		    !!(run->s.regs.sregs.efer & EFER_LME));
 
-	TEST_ASSERT_EQ(pthread_create(&thread, NULL, racer, (void *)run), 0);
+	kvm_pthread_create(&thread, NULL, racer, (void *)run);
 
 	for (t = time(NULL) + TIMEOUT; time(NULL) < t;) {
 		/*
diff --git a/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c b/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
index 59da8d4da607..54b4b8d06d3b 100644
--- a/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
+++ b/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
@@ -94,7 +94,7 @@ int main(int argc, char *argv[])
 	pthread_t cpu_threads[NR_TEST_VCPUS];
 	unsigned long cpu;
 	for (cpu = 0; cpu < NR_TEST_VCPUS; cpu++)
-		pthread_create(&cpu_threads[cpu], NULL, run_vcpu, (void *)cpu);
+		kvm_pthread_create(&cpu_threads[cpu], NULL, run_vcpu, (void *)cpu);
 
 	unsigned long failures = 0;
 	for (cpu = 0; cpu < NR_TEST_VCPUS; cpu++) {
diff --git a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
index 3a326c5e74ca..271ec09c629e 100644
--- a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
+++ b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
@@ -387,7 +387,6 @@ void get_cmdline_args(int argc, char *argv[], int *run_secs,
 
 int main(int argc, char *argv[])
 {
-	int r;
 	int wait_secs;
 	const int max_halter_wait = 10;
 	int run_secs = 0;
@@ -428,9 +427,7 @@ int main(int argc, char *argv[])
 	params[1].pipis_rcvd = pipis_rcvd;
 
 	/* Start halter vCPU thread and wait for it to execute first HLT. */
-	r = pthread_create(&threads[0], NULL, vcpu_thread, &params[0]);
-	TEST_ASSERT(r == 0,
-		    "pthread_create halter failed errno=%d", errno);
+	kvm_pthread_create(&threads[0], NULL, vcpu_thread, &params[0]);
 	fprintf(stderr, "Halter vCPU thread started\n");
 
 	wait_secs = 0;
@@ -447,8 +444,7 @@ int main(int argc, char *argv[])
 		"Halter vCPU thread reported its APIC ID: %u after %d seconds.\n",
 		data->halter_apic_id, wait_secs);
 
-	r = pthread_create(&threads[1], NULL, vcpu_thread, &params[1]);
-	TEST_ASSERT(r == 0, "pthread_create sender failed errno=%d", errno);
+	kvm_pthread_create(&threads[1], NULL, vcpu_thread, &params[1]);
 
 	fprintf(stderr,
 		"IPI sender vCPU thread started. Letting vCPUs run for %d seconds.\n",
diff --git a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
index 5076f6a75455..20f518fe2450 100644
--- a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
@@ -894,8 +894,7 @@ int main(int argc, char *argv[])
 				if (verbose)
 					printf("Testing shinfo lock corruption (KVM_XEN_HVM_EVTCHN_SEND)\n");
 
-				ret = pthread_create(&thread, NULL, &juggle_shinfo_state, (void *)vm);
-				TEST_ASSERT(ret == 0, "pthread_create() failed: %s", strerror(ret));
+				kvm_pthread_create(&thread, NULL, &juggle_shinfo_state, (void *)vm);
 
 				struct kvm_irq_routing_xen_evtchn uxe = {
 					.port = 1,
-- 
2.55.0.508.g3f0d502094-goog

Re: [PATCH v3 08/12] KVM: selftests: Add KVM syscall wrapper for pthread_create()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The `kvm_pthread_create` wrapper incorrectly uses `errno` for error reporting, which `pthread_create` does not set.
--

--- Patch [8]: [PATCH v3 08/12] KVM: selftests: Add KVM syscall wrapper for pthread_create() ---
commit d35383198b73b658a7a90f8c3560d48c6d71d62b
Author: Sean Christopherson <seanjc@google.com>

KVM: selftests: Add KVM syscall wrapper for pthread_create()

This commit adds a wrapper for pthread_create() in KVM selftests to
automatically assert success. It converts various manual checks across
multiple tests to use the new kvm_pthread_create() wrapper.

> diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
> --- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
> +++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
[ ... ]
> @@ -97,6 +98,10 @@ __KVM_SYSCALL_DEFINE(madvise, 3, void *, addr, size_t, length, int, advice);
>  __KVM_SYSCALL_DEFINE(sched_getaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
>  __KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
>  
> +typedef void *(*pthread_fn_t)(void *);
> +__KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
> +		     const pthread_attr_t *, attr, pthread_fn_t, fn, void *, arg);
> +

[Severity: Medium]
Since pthread_create() returns the error code directly rather than setting
errno, will this __KVM_SYSCALL_DEFINE wrapper print stale or zero errno values
on failure?

The __KVM_SYSCALL_DEFINE macro relies on __KVM_SYSCALL_ERROR, which formats
the error using errno:

#define __KVM_SYSCALL_ERROR(_name, _ret) \
    "%s failed, rc: %i errno: %i (%s)", (_name), (_ret), errno, strerror(errno)

If pthread_create() fails, should it be handled differently to ensure the
actual returned error code is passed to strerror() instead of the unmodified
errno?

>  #define kvm_free_fd(fd)		\
>  do {				\
>  	kvm_close(fd);		\

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260731195612.2697986-1-seanjc@google.com?part=8

[PATCH v3 09/12] KVM: selftests: Add KVM syscall wrappers for pthread_{cancel,join}()

[Sean Christopherson]

Add and use KVM wrappers for pthread_{cancel,join}() so that selftests
don't need to manually assert that the syscalls succeeded.

Note, the vast majority tests don't actually assert success, but they all
obviously rely on the syscall to succeed.

Other than explicitly failing if a syscall fails, no functional change
intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/kvm/arch_timer.c             |  4 ++--
 tools/testing/selftests/kvm/arm64/vgic_irq.c         |  4 ++--
 tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c  |  4 ++--
 tools/testing/selftests/kvm/dirty_log_test.c         |  2 +-
 tools/testing/selftests/kvm/include/kvm_syscalls.h   |  2 ++
 tools/testing/selftests/kvm/irq_test.c               |  2 +-
 tools/testing/selftests/kvm/irqfd_test.c             |  2 +-
 tools/testing/selftests/kvm/kvm_page_table_test.c    |  2 +-
 tools/testing/selftests/kvm/lib/memstress.c          |  2 +-
 tools/testing/selftests/kvm/lib/userfaultfd_util.c   |  3 +--
 tools/testing/selftests/kvm/memslot_perf_test.c      |  2 +-
 tools/testing/selftests/kvm/mmu_stress_test.c        |  2 +-
 tools/testing/selftests/kvm/pre_fault_memory_test.c  |  2 +-
 tools/testing/selftests/kvm/rseq_test.c              |  2 +-
 tools/testing/selftests/kvm/s390/memop.c             |  2 +-
 tools/testing/selftests/kvm/set_memory_region_test.c |  4 ++--
 tools/testing/selftests/kvm/steal_time.c             |  2 +-
 tools/testing/selftests/kvm/x86/hyperv_ipi.c         | 10 ++--------
 tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c   | 10 ++--------
 .../selftests/kvm/x86/private_mem_conversions_test.c |  2 +-
 .../selftests/kvm/x86/private_mem_kvm_exits_test.c   |  2 +-
 .../testing/selftests/kvm/x86/recalc_apic_map_test.c |  4 ++--
 tools/testing/selftests/kvm/x86/sev_migrate_tests.c  |  2 +-
 tools/testing/selftests/kvm/x86/sync_regs_test.c     |  4 ++--
 tools/testing/selftests/kvm/x86/tsc_scaling_sync.c   |  2 +-
 tools/testing/selftests/kvm/x86/xapic_ipi_test.c     | 12 ++----------
 tools/testing/selftests/kvm/x86/xen_shinfo_test.c    |  8 ++------
 27 files changed, 38 insertions(+), 61 deletions(-)

diff --git a/tools/testing/selftests/kvm/arch_timer.c b/tools/testing/selftests/kvm/arch_timer.c
index a6a66b5b7ea9..03160ad666f5 100644
--- a/tools/testing/selftests/kvm/arch_timer.c
+++ b/tools/testing/selftests/kvm/arch_timer.c
@@ -158,10 +158,10 @@ static void test_run(struct kvm_vm *vm)
 	}
 
 	for (i = 0; i < test_args.nr_vcpus; i++)
-		pthread_join(pt_vcpu_run[i], NULL);
+		kvm_pthread_join(pt_vcpu_run[i], NULL);
 
 	if (test_args.migration_freq_ms)
-		pthread_join(pt_vcpu_migration, NULL);
+		kvm_pthread_join(pt_vcpu_migration, NULL);
 
 	bitmap_free(vcpu_done_map);
 }
diff --git a/tools/testing/selftests/kvm/arm64/vgic_irq.c b/tools/testing/selftests/kvm/arm64/vgic_irq.c
index 2d6f20d5c117..3cb85932ee85 100644
--- a/tools/testing/selftests/kvm/arm64/vgic_irq.c
+++ b/tools/testing/selftests/kvm/arm64/vgic_irq.c
@@ -1007,8 +1007,8 @@ static void test_vgic_two_cpus(void *gcode)
 	kvm_pthread_create(&thr[0], NULL, test_vcpu_run, vcpus[0]);
 	kvm_pthread_create(&thr[1], NULL, test_vcpu_run, vcpus[1]);
 
-	pthread_join(thr[0], NULL);
-	pthread_join(thr[1], NULL);
+	kvm_pthread_join(thr[0], NULL);
+	kvm_pthread_join(thr[1], NULL);
 
 	close(gic_fd);
 	kvm_vm_free(vm);
diff --git a/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c b/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
index 549e0547d905..9a69c439a0f8 100644
--- a/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
+++ b/tools/testing/selftests/kvm/arm64/vgic_lpi_stress.c
@@ -321,13 +321,13 @@ static void run_test(void)
 	clock_gettime(CLOCK_MONOTONIC, &start);
 
 	for (i = 0; i < nr_devices; i++)
-		pthread_join(lpi_threads[i], NULL);
+		kvm_pthread_join(lpi_threads[i], NULL);
 
 	delta = timespec_elapsed(start);
 	write_guest_global(vm, test_data.request_vcpus_stop, true);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(vcpu_threads[i], NULL);
+		kvm_pthread_join(vcpu_threads[i], NULL);
 
 	report_stats(delta);
 }
diff --git a/tools/testing/selftests/kvm/dirty_log_test.c b/tools/testing/selftests/kvm/dirty_log_test.c
index ca22e5f09ac2..af5eb0334a74 100644
--- a/tools/testing/selftests/kvm/dirty_log_test.c
+++ b/tools/testing/selftests/kvm/dirty_log_test.c
@@ -805,7 +805,7 @@ static void run_test(enum vm_guest_mode mode, void *arg)
 	WRITE_ONCE(host_quit, true);
 	sem_post(&sem_vcpu_cont);
 
-	pthread_join(vcpu_thread, NULL);
+	kvm_pthread_join(vcpu_thread, NULL);
 
 	pr_info("Total bits checked: dirty (%lu), clear (%lu)\n",
 		host_dirty_count, host_clear_count);
diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
index e75be8378779..d6c18e13b9c8 100644
--- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
+++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
@@ -101,6 +101,8 @@ __KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_s
 typedef void *(*pthread_fn_t)(void *);
 __KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
 		     const pthread_attr_t *, attr, pthread_fn_t, fn, void *, arg);
+__KVM_SYSCALL_DEFINE(pthread_join, 2, pthread_t, thread, void **, thread_return);
+__KVM_SYSCALL_DEFINE(pthread_cancel, 1, pthread_t, thread);
 
 #define kvm_free_fd(fd)		\
 do {				\
diff --git a/tools/testing/selftests/kvm/irq_test.c b/tools/testing/selftests/kvm/irq_test.c
index 7da5f8d75e85..168c52734fef 100644
--- a/tools/testing/selftests/kvm/irq_test.c
+++ b/tools/testing/selftests/kvm/irq_test.c
@@ -356,7 +356,7 @@ int main(int argc, char **argv)
 	WRITE_AND_SYNC_TO_GUEST(vm, done, true);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(vcpu_threads[i], NULL);
+		kvm_pthread_join(vcpu_threads[i], NULL);
 
 	return 0;
 }
diff --git a/tools/testing/selftests/kvm/irqfd_test.c b/tools/testing/selftests/kvm/irqfd_test.c
index 8b39144a2834..ebb2478878de 100644
--- a/tools/testing/selftests/kvm/irqfd_test.c
+++ b/tools/testing/selftests/kvm/irqfd_test.c
@@ -139,5 +139,5 @@ int main(int argc, char *argv[])
 	}
 
 	WRITE_ONCE(done, true);
-	pthread_join(racing_thread, NULL);
+	kvm_pthread_join(racing_thread, NULL);
 }
diff --git a/tools/testing/selftests/kvm/kvm_page_table_test.c b/tools/testing/selftests/kvm/kvm_page_table_test.c
index 8222bdb63e49..974813ea8953 100644
--- a/tools/testing/selftests/kvm/kvm_page_table_test.c
+++ b/tools/testing/selftests/kvm/kvm_page_table_test.c
@@ -413,7 +413,7 @@ static void run_test(enum vm_guest_mode mode, void *arg)
 	}
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(vcpu_threads[i], NULL);
+		kvm_pthread_join(vcpu_threads[i], NULL);
 
 	ret = sem_destroy(&test_stage_updated);
 	TEST_ASSERT(ret == 0, "Error in sem_destroy");
diff --git a/tools/testing/selftests/kvm/lib/memstress.c b/tools/testing/selftests/kvm/lib/memstress.c
index 0bfa623ce9b5..2864562f57b4 100644
--- a/tools/testing/selftests/kvm/lib/memstress.c
+++ b/tools/testing/selftests/kvm/lib/memstress.c
@@ -312,7 +312,7 @@ void memstress_join_vcpu_threads(int nr_vcpus)
 	WRITE_ONCE(memstress_args.stop_vcpus, true);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(vcpu_threads[i].thread, NULL);
+		kvm_pthread_join(vcpu_threads[i].thread, NULL);
 }
 
 static void toggle_dirty_logging(struct kvm_vm *vm, int slots, bool enable)
diff --git a/tools/testing/selftests/kvm/lib/userfaultfd_util.c b/tools/testing/selftests/kvm/lib/userfaultfd_util.c
index 4b3e3158ed87..f7ce5a6ddcc2 100644
--- a/tools/testing/selftests/kvm/lib/userfaultfd_util.c
+++ b/tools/testing/selftests/kvm/lib/userfaultfd_util.c
@@ -187,8 +187,7 @@ void uffd_stop_demand_paging(struct uffd_desc *uffd)
 			    "Unable to write to pipefd %i for uffd_desc %p", i, uffd);
 
 	for (i = 0; i < uffd->num_readers; ++i)
-		TEST_ASSERT(!pthread_join(uffd->readers[i], NULL),
-			    "Pthread_join failed on reader %i for uffd_desc %p", i, uffd);
+		kvm_pthread_join(uffd->readers[i], NULL);
 
 	close(uffd->uffd);
 
diff --git a/tools/testing/selftests/kvm/memslot_perf_test.c b/tools/testing/selftests/kvm/memslot_perf_test.c
index eb49305db404..e1299611be32 100644
--- a/tools/testing/selftests/kvm/memslot_perf_test.c
+++ b/tools/testing/selftests/kvm/memslot_perf_test.c
@@ -381,7 +381,7 @@ static void free_vm(struct vm_data *data)
 
 static void wait_guest_exit(struct vm_data *data)
 {
-	pthread_join(data->vcpu_thread, NULL);
+	kvm_pthread_join(data->vcpu_thread, NULL);
 }
 
 static void let_guest_run(struct sync_area *sync)
diff --git a/tools/testing/selftests/kvm/mmu_stress_test.c b/tools/testing/selftests/kvm/mmu_stress_test.c
index 9448498849ff..741e343962ed 100644
--- a/tools/testing/selftests/kvm/mmu_stress_test.c
+++ b/tools/testing/selftests/kvm/mmu_stress_test.c
@@ -409,7 +409,7 @@ int main(int argc, char *argv[])
 
 	/* Sanity check that the vCPUs actually ran. */
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(threads[i], NULL);
+		kvm_pthread_join(threads[i], NULL);
 
 	/*
 	 * Deliberately exit without deleting the remaining memslots or closing
diff --git a/tools/testing/selftests/kvm/pre_fault_memory_test.c b/tools/testing/selftests/kvm/pre_fault_memory_test.c
index 88287ccb9093..c57631aab3d3 100644
--- a/tools/testing/selftests/kvm/pre_fault_memory_test.c
+++ b/tools/testing/selftests/kvm/pre_fault_memory_test.c
@@ -116,7 +116,7 @@ static void pre_fault_memory(struct kvm_vcpu *vcpu, u64 base_gpa, u64 offset,
 		 */
 		if (!slot_recreated) {
 			WRITE_ONCE(data.recreate_slot, true);
-			pthread_join(slot_worker, NULL);
+			kvm_pthread_join(slot_worker, NULL);
 			slot_recreated = true;
 
 			/*
diff --git a/tools/testing/selftests/kvm/rseq_test.c b/tools/testing/selftests/kvm/rseq_test.c
index 0921b8d90d4a..b386aece68fd 100644
--- a/tools/testing/selftests/kvm/rseq_test.c
+++ b/tools/testing/selftests/kvm/rseq_test.c
@@ -312,7 +312,7 @@ int main(int argc, char *argv[])
 		    "  e.g. via cpuidle.off=1 or via -l <latency>, or run with -u to\n"
 		    "  disable this sanity check.", i);
 
-	pthread_join(migration_thread, NULL);
+	kvm_pthread_join(migration_thread, NULL);
 
 	kvm_vm_free(vm);
 
diff --git a/tools/testing/selftests/kvm/s390/memop.c b/tools/testing/selftests/kvm/s390/memop.c
index 96b32bd74009..ce8db96ccf7e 100644
--- a/tools/testing/selftests/kvm/s390/memop.c
+++ b/tools/testing/selftests/kvm/s390/memop.c
@@ -701,7 +701,7 @@ static void test_cmpxchg_key_concurrent(void)
 		}
 	}
 
-	pthread_join(thread, NULL);
+	kvm_pthread_join(thread, NULL);
 
 	MOP(t.vcpu, LOGICAL, READ, mem2, max_block, GADDR_V(mem2));
 	TEST_ASSERT(popcount_eq(*(__uint128_t *)mem1, *(__uint128_t *)mem2),
diff --git a/tools/testing/selftests/kvm/set_memory_region_test.c b/tools/testing/selftests/kvm/set_memory_region_test.c
index 12f5022bb164..dffe9654f45b 100644
--- a/tools/testing/selftests/kvm/set_memory_region_test.c
+++ b/tools/testing/selftests/kvm/set_memory_region_test.c
@@ -216,7 +216,7 @@ static void test_move_memory_region(bool disable_slot_zap_quirk)
 	/* Defered sync from when the memslot was misaligned (above). */
 	wait_for_vcpu();
 
-	pthread_join(vcpu_thread, NULL);
+	kvm_pthread_join(vcpu_thread, NULL);
 
 	kvm_vm_free(vm);
 }
@@ -302,7 +302,7 @@ static void test_delete_memory_region(bool disable_slot_zap_quirk)
 	 */
 	vm_mem_region_delete(vm, 0);
 
-	pthread_join(vcpu_thread, NULL);
+	kvm_pthread_join(vcpu_thread, NULL);
 
 	run = vcpu->run;
 
diff --git a/tools/testing/selftests/kvm/steal_time.c b/tools/testing/selftests/kvm/steal_time.c
index f5f708be5aa0..6cb1eb4c40c9 100644
--- a/tools/testing/selftests/kvm/steal_time.c
+++ b/tools/testing/selftests/kvm/steal_time.c
@@ -556,7 +556,7 @@ int main(int ac, char **av)
 		do
 			sched_yield();
 		while (get_run_delay() - run_delay < MIN_RUN_DELAY_NS);
-		pthread_join(thread, NULL);
+		kvm_pthread_join(thread, NULL);
 		run_delay = get_run_delay() - run_delay;
 		TEST_ASSERT(run_delay >= MIN_RUN_DELAY_NS,
 			    "Expected run_delay >= %ld, got %ld",
diff --git a/tools/testing/selftests/kvm/x86/hyperv_ipi.c b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
index 3cf451f3153e..d2efb1043534 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_ipi.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
@@ -225,15 +225,9 @@ static void *vcpu_thread(void *arg)
 static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
 {
 	void *retval;
-	int r;
 
-	r = pthread_cancel(thread);
-	TEST_ASSERT(!r, "pthread_cancel on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
-
-	r = pthread_join(thread, &retval);
-	TEST_ASSERT(!r, "pthread_join on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, &retval);
 	TEST_ASSERT(retval == PTHREAD_CANCELED,
 		    "expected retval=%p, got %p", PTHREAD_CANCELED,
 		    retval);
diff --git a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
index 87fac1968a47..e4fc9bb6f329 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
@@ -551,15 +551,9 @@ static void *vcpu_thread(void *arg)
 static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
 {
 	void *retval;
-	int r;
 
-	r = pthread_cancel(thread);
-	TEST_ASSERT(!r, "pthread_cancel on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
-
-	r = pthread_join(thread, &retval);
-	TEST_ASSERT(!r, "pthread_join on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, &retval);
 	TEST_ASSERT(retval == PTHREAD_CANCELED,
 		    "expected retval=%p, got %p", PTHREAD_CANCELED,
 		    retval);
diff --git a/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c b/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
index e419b08f1fa6..3aa0673f79f3 100644
--- a/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
+++ b/tools/testing/selftests/kvm/x86/private_mem_conversions_test.c
@@ -418,7 +418,7 @@ static void test_mem_conversions(enum vm_mem_backing_src_type src_type, u32 nr_v
 	WRITE_ONCE(run_vcpus, true);
 
 	for (i = 0; i < nr_vcpus; i++)
-		pthread_join(threads[i], NULL);
+		kvm_pthread_join(threads[i], NULL);
 
 	kvm_vm_free(vm);
 
diff --git a/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c b/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
index e26524fb4b50..e6d16e9d6120 100644
--- a/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
+++ b/tools/testing/selftests/kvm/x86/private_mem_kvm_exits_test.c
@@ -70,7 +70,7 @@ static void test_private_access_memslot_deleted(void)
 
 	vm_mem_region_delete(vm, EXITS_TEST_SLOT);
 
-	pthread_join(vm_thread, &thread_return);
+	kvm_pthread_join(vm_thread, &thread_return);
 	exit_reason = (u32)(u64)thread_return;
 
 	TEST_ASSERT_EQ(exit_reason, KVM_EXIT_MEMORY_FAULT);
diff --git a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
index 13abb65620d5..e3e397f32bff 100644
--- a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
+++ b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
@@ -65,8 +65,8 @@ int main(void)
 		vcpu_set_msr(vcpuN, MSR_IA32_APICBASE, LAPIC_DISABLED);
 	}
 
-	TEST_ASSERT_EQ(pthread_cancel(thread), 0);
-	TEST_ASSERT_EQ(pthread_join(thread, NULL), 0);
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, NULL);
 
 	kvm_vm_free(vm);
 
diff --git a/tools/testing/selftests/kvm/x86/sev_migrate_tests.c b/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
index 66c7ce49620a..de006e1c861c 100644
--- a/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
+++ b/tools/testing/selftests/kvm/x86/sev_migrate_tests.c
@@ -131,7 +131,7 @@ static void test_sev_migrate_locking(void)
 		kvm_pthread_create(&pt[i], NULL, locking_test_thread, &input[i]);
 
 	for (i = 0; i < NR_LOCK_TESTING_THREADS; ++i)
-		pthread_join(pt[i], NULL);
+		kvm_pthread_join(pt[i], NULL);
 	for (i = 0; i < NR_LOCK_TESTING_THREADS; ++i)
 		kvm_vm_free(input[i].vm);
 }
diff --git a/tools/testing/selftests/kvm/x86/sync_regs_test.c b/tools/testing/selftests/kvm/x86/sync_regs_test.c
index f7228af7d594..99a5bbeffc5d 100644
--- a/tools/testing/selftests/kvm/x86/sync_regs_test.c
+++ b/tools/testing/selftests/kvm/x86/sync_regs_test.c
@@ -199,8 +199,8 @@ static void race_sync_regs(struct kvm_vcpu *vcpu, void *racer)
 		}
 	}
 
-	TEST_ASSERT_EQ(pthread_cancel(thread), 0);
-	TEST_ASSERT_EQ(pthread_join(thread, NULL), 0);
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, NULL);
 
 	kvm_x86_state_cleanup(state);
 }
diff --git a/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c b/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
index 54b4b8d06d3b..2a3a9bd8b338 100644
--- a/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
+++ b/tools/testing/selftests/kvm/x86/tsc_scaling_sync.c
@@ -99,7 +99,7 @@ int main(int argc, char *argv[])
 	unsigned long failures = 0;
 	for (cpu = 0; cpu < NR_TEST_VCPUS; cpu++) {
 		void *this_cpu_failures;
-		pthread_join(cpu_threads[cpu], &this_cpu_failures);
+		kvm_pthread_join(cpu_threads[cpu], &this_cpu_failures);
 		failures += (unsigned long)this_cpu_failures;
 	}
 
diff --git a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
index 271ec09c629e..33e78b284c67 100644
--- a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
+++ b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
@@ -231,17 +231,9 @@ static void *vcpu_thread(void *arg)
 static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
 {
 	void *retval;
-	int r;
 
-	r = pthread_cancel(thread);
-	TEST_ASSERT(r == 0,
-		    "pthread_cancel on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
-
-	r = pthread_join(thread, &retval);
-	TEST_ASSERT(r == 0,
-		    "pthread_join on vcpu_id=%d failed with errno=%d",
-		    vcpu->id, r);
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, &retval);
 	TEST_ASSERT(retval == PTHREAD_CANCELED,
 		    "expected retval=%p, got %p", PTHREAD_CANCELED,
 		    retval);
diff --git a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
index 20f518fe2450..3ec5e67b21ef 100644
--- a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
@@ -437,7 +437,6 @@ int main(int argc, char *argv[])
 	struct kvm_vm *vm;
 	pthread_t thread;
 	bool verbose;
-	int ret;
 
 	verbose = argc > 1 && (!strncmp(argv[1], "-v", 3) ||
 			       !strncmp(argv[1], "--verbose", 10));
@@ -948,11 +947,8 @@ int main(int argc, char *argv[])
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 
-				ret = pthread_cancel(thread);
-				TEST_ASSERT(ret == 0, "pthread_cancel() failed: %s", strerror(ret));
-
-				ret = pthread_join(thread, 0);
-				TEST_ASSERT(ret == 0, "pthread_join() failed: %s", strerror(ret));
+				kvm_pthread_cancel(thread);
+				kvm_pthread_join(thread, 0);
 				goto done;
 
 			case TEST_GUEST_SAW_IRQ:
-- 
2.55.0.508.g3f0d502094-goog

Re: [PATCH v3 09/12] KVM: selftests: Add KVM syscall wrappers for pthread_{cancel,join}()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The `kvm_pthread_join` and `kvm_pthread_cancel` wrappers generate misleading error messages on failure because they use `__KVM_SYSCALL_ERROR`, which prints `errno` and `strerror(errno)`. However, `pthread_join` and `pthread_cancel` return error numbers directly and do not set `errno`.
--

--- Patch [9]: [PATCH v3 09/12] KVM: selftests: Add KVM syscall wrappers for pthread_{cancel,join}() ---
commit 26558313a1f5853d420a8a8544d24882f9b4835b
Author: Sean Christopherson <seanjc@google.com>

KVM: selftests: Add KVM syscall wrappers for pthread_{cancel,join}()

This commit adds and uses wrappers for pthread_cancel() and pthread_join()
across KVM selftests. The wrappers aim to automatically assert on failure,
removing the need for manual assertions in the test code.

> diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
> index e75be83787798..d6c18e13b9c8b 100644
> --- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
> +++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
> @@ -101,6 +101,8 @@ __KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_s
>  typedef void *(*pthread_fn_t)(void *);
>  __KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
>  		     const pthread_attr_t *, attr, pthread_fn_t, fn, void *, arg);
> +__KVM_SYSCALL_DEFINE(pthread_join, 2, pthread_t, thread, void **, thread_return);
> +__KVM_SYSCALL_DEFINE(pthread_cancel, 1, pthread_t, thread);

[Severity: Medium]
Will this generate misleading error messages if the thread functions fail?

The pthread_join() and pthread_cancel() functions return their error codes
directly rather than setting errno. If the __KVM_SYSCALL_DEFINE macro
uses errno for its error reporting, it could log an unrelated error
string instead of the actual failure reason.

>  
>  #define kvm_free_fd(fd)		\
>  do {				\

[ ... ]

> diff --git a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
> index 20f518fe24506..3ec5e67b21efa 100644
> --- a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
> +++ b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
> @@ -948,11 +947,8 @@ int main(int argc, char *argv[])
>  				TEST_ASSERT(!evtchn_irq_expected,
>  					    "Expected event channel IRQ but it didn't happen");
>  
> -				ret = pthread_cancel(thread);
> -				TEST_ASSERT(ret == 0, "pthread_cancel() failed: %s", strerror(ret));
> -
> -				ret = pthread_join(thread, 0);
> -				TEST_ASSERT(ret == 0, "pthread_join() failed: %s", strerror(ret));
> +				kvm_pthread_cancel(thread);
> +				kvm_pthread_join(thread, 0);

[Severity: Medium]
Does this change replace an accurate error message with a potentially
misleading one?

The original code correctly used strerror(ret) because these functions
return the error code directly. If the new wrappers rely on errno, they
might print an unrelated error string when these functions fail.

>  				goto done;
>  
>  			case TEST_GUEST_SAW_IRQ:

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260731195612.2697986-1-seanjc@google.com?part=9

[PATCH v3 10/12] KVM: selftests: Add helper APIs to cancel+join pthreads

[Sean Christopherson]

Deduplicate test code that does cancel()+join() on pthreads by adding a
collection of syscall wrappers.  The tests that asynchronously cancel
pthreads, i.e. use PTHREAD_CANCEL_ASYNCHRONOUS, are particularly ugly,
especially since it's not immediately obvious why those tests assert on
the thread being CANCELED, whereas others do not.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/include/kvm_syscalls.h      | 24 +++++++++++++++++++
 tools/testing/selftests/kvm/x86/hyperv_ipi.c  | 15 ++----------
 .../selftests/kvm/x86/hyperv_tlb_flush.c      | 15 ++----------
 .../selftests/kvm/x86/recalc_apic_map_test.c  |  3 +--
 .../selftests/kvm/x86/sync_regs_test.c        |  3 +--
 .../selftests/kvm/x86/xapic_ipi_test.c        | 15 ++----------
 .../selftests/kvm/x86/xen_shinfo_test.c       |  3 +--
 7 files changed, 33 insertions(+), 45 deletions(-)

diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
index d6c18e13b9c8..c0657e54aee5 100644
--- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
+++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
@@ -104,6 +104,30 @@ __KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
 __KVM_SYSCALL_DEFINE(pthread_join, 2, pthread_t, thread, void **, thread_return);
 __KVM_SYSCALL_DEFINE(pthread_cancel, 1, pthread_t, thread);
 
+static inline void __kvm_pthread_cancel_join(pthread_t thread, void **r)
+{
+	kvm_pthread_cancel(thread);
+	kvm_pthread_join(thread, r);
+}
+
+static inline void kvm_pthread_cancel_join(pthread_t thread)
+{
+	__kvm_pthread_cancel_join(thread, NULL);
+}
+
+/*
+ * Cancel+Join a pthread that was configured with PTHREAD_CANCEL_ASYNCHRONOUS
+ * and is expected to exit only in response to cancellation.
+ */
+static inline void kvm_pthread_cancel_join_async(pthread_t thread)
+{
+	void *r;
+
+	__kvm_pthread_cancel_join(thread, &r);
+	TEST_ASSERT(r == PTHREAD_CANCELED,
+		    "expected retval=%p, got %p", PTHREAD_CANCELED, r);
+}
+
 #define kvm_free_fd(fd)		\
 do {				\
 	kvm_close(fd);		\
diff --git a/tools/testing/selftests/kvm/x86/hyperv_ipi.c b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
index d2efb1043534..a2fc0bca8011 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_ipi.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_ipi.c
@@ -222,17 +222,6 @@ static void *vcpu_thread(void *arg)
 	return NULL;
 }
 
-static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
-{
-	void *retval;
-
-	kvm_pthread_cancel(thread);
-	kvm_pthread_join(thread, &retval);
-	TEST_ASSERT(retval == PTHREAD_CANCELED,
-		    "expected retval=%p, got %p", PTHREAD_CANCELED,
-		    retval);
-}
-
 int main(int argc, char *argv[])
 {
 	struct kvm_vm *vm;
@@ -293,8 +282,8 @@ int main(int argc, char *argv[])
 	}
 
 done:
-	cancel_join_vcpu_thread(threads[0], vcpu[1]);
-	cancel_join_vcpu_thread(threads[1], vcpu[2]);
+	kvm_pthread_cancel_join_async(threads[0]);
+	kvm_pthread_cancel_join_async(threads[1]);
 	kvm_vm_free(vm);
 
 	return 0;
diff --git a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
index e4fc9bb6f329..dc9d40830a7f 100644
--- a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
+++ b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c
@@ -548,17 +548,6 @@ static void *vcpu_thread(void *arg)
 	return NULL;
 }
 
-static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
-{
-	void *retval;
-
-	kvm_pthread_cancel(thread);
-	kvm_pthread_join(thread, &retval);
-	TEST_ASSERT(retval == PTHREAD_CANCELED,
-		    "expected retval=%p, got %p", PTHREAD_CANCELED,
-		    retval);
-}
-
 int main(int argc, char *argv[])
 {
 	struct kvm_vm *vm;
@@ -652,8 +641,8 @@ int main(int argc, char *argv[])
 	}
 
 done:
-	cancel_join_vcpu_thread(threads[0], vcpu[1]);
-	cancel_join_vcpu_thread(threads[1], vcpu[2]);
+	kvm_pthread_cancel_join_async(threads[0]);
+	kvm_pthread_cancel_join_async(threads[1]);
 	kvm_vm_free(vm);
 
 	return 0;
diff --git a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
index e3e397f32bff..6252a96320ee 100644
--- a/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
+++ b/tools/testing/selftests/kvm/x86/recalc_apic_map_test.c
@@ -65,8 +65,7 @@ int main(void)
 		vcpu_set_msr(vcpuN, MSR_IA32_APICBASE, LAPIC_DISABLED);
 	}
 
-	kvm_pthread_cancel(thread);
-	kvm_pthread_join(thread, NULL);
+	kvm_pthread_cancel_join(thread);
 
 	kvm_vm_free(vm);
 
diff --git a/tools/testing/selftests/kvm/x86/sync_regs_test.c b/tools/testing/selftests/kvm/x86/sync_regs_test.c
index 99a5bbeffc5d..ed0c21b427c9 100644
--- a/tools/testing/selftests/kvm/x86/sync_regs_test.c
+++ b/tools/testing/selftests/kvm/x86/sync_regs_test.c
@@ -199,8 +199,7 @@ static void race_sync_regs(struct kvm_vcpu *vcpu, void *racer)
 		}
 	}
 
-	kvm_pthread_cancel(thread);
-	kvm_pthread_join(thread, NULL);
+	kvm_pthread_cancel_join(thread);
 
 	kvm_x86_state_cleanup(state);
 }
diff --git a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
index 33e78b284c67..469e3ab16460 100644
--- a/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
+++ b/tools/testing/selftests/kvm/x86/xapic_ipi_test.c
@@ -228,17 +228,6 @@ static void *vcpu_thread(void *arg)
 	return NULL;
 }
 
-static void cancel_join_vcpu_thread(pthread_t thread, struct kvm_vcpu *vcpu)
-{
-	void *retval;
-
-	kvm_pthread_cancel(thread);
-	kvm_pthread_join(thread, &retval);
-	TEST_ASSERT(retval == PTHREAD_CANCELED,
-		    "expected retval=%p, got %p", PTHREAD_CANCELED,
-		    retval);
-}
-
 void do_migrations(struct test_data_page *data, int run_secs, int delay_usecs,
 		   u64 *pipis_rcvd)
 {
@@ -450,8 +439,8 @@ int main(int argc, char *argv[])
 	/*
 	 * Cancel threads and wait for them to stop.
 	 */
-	cancel_join_vcpu_thread(threads[0], params[0].vcpu);
-	cancel_join_vcpu_thread(threads[1], params[1].vcpu);
+	kvm_pthread_cancel_join_async(threads[0]);
+	kvm_pthread_cancel_join_async(threads[1]);
 
 	/*
 	 * If the host support Idle HLT, i.e. KVM *might* be using Idle HLT,
diff --git a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
index 3ec5e67b21ef..aaa414c677ad 100644
--- a/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86/xen_shinfo_test.c
@@ -947,8 +947,7 @@ int main(int argc, char *argv[])
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 
-				kvm_pthread_cancel(thread);
-				kvm_pthread_join(thread, 0);
+				kvm_pthread_cancel_join(thread);
 				goto done;
 
 			case TEST_GUEST_SAW_IRQ:
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 11/12] KVM: selftests: Add KVM syscall wrappers for pthread_{g,s}etaffinity_np()

[Sean Christopherson]

Add and use KVM wrappers for pthread_{g,s}etaffinity_np() so that selftests
don't need to manually assert that the syscalls succeeded, and so that they
don't need to manually pass in sizeof(cpu_set_t) for the size.

Note, the steal time test didn't assert success, but the TEST_ASSERT() one
line later makes it quite clear the test relies on pthread_getaffinity_np()
succeeding.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../testing/selftests/kvm/include/kvm_syscalls.h | 16 ++++++++++++++++
 tools/testing/selftests/kvm/steal_time.c         |  2 +-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/kvm/include/kvm_syscalls.h b/tools/testing/selftests/kvm/include/kvm_syscalls.h
index c0657e54aee5..dc7461894aef 100644
--- a/tools/testing/selftests/kvm/include/kvm_syscalls.h
+++ b/tools/testing/selftests/kvm/include/kvm_syscalls.h
@@ -98,6 +98,22 @@ __KVM_SYSCALL_DEFINE(madvise, 3, void *, addr, size_t, length, int, advice);
 __KVM_SYSCALL_DEFINE(sched_getaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
 __KVM_SYSCALL_DEFINE(sched_setaffinity, 3, pid_t, pid, size_t, cpusetsize, cpu_set_t *, mask);
 
+__KVM_SYSCALL_DEFINE(pthread_getaffinity_np, 3, pthread_t, thread,
+		     size_t, cpusetsize, cpu_set_t *, cpuset);
+__KVM_SYSCALL_DEFINE(pthread_setaffinity_np, 3, pthread_t, thread,
+		     size_t, cpusetsize, const cpu_set_t *, cpuset);
+
+static inline void kvm_pthread_getaffinity(pthread_t thread, cpu_set_t *cpuset)
+{
+	kvm_pthread_getaffinity_np(thread, sizeof(cpu_set_t), cpuset);
+}
+
+static inline void kvm_pthread_setaffinity(pthread_t thread,
+					   const cpu_set_t *cpuset)
+{
+	kvm_pthread_setaffinity_np(thread, sizeof(cpu_set_t), cpuset);
+}
+
 typedef void *(*pthread_fn_t)(void *);
 __KVM_SYSCALL_DEFINE(pthread_create, 4, pthread_t *, thread,
 		     const pthread_attr_t *, attr, pthread_fn_t, fn, void *, arg);
diff --git a/tools/testing/selftests/kvm/steal_time.c b/tools/testing/selftests/kvm/steal_time.c
index 6cb1eb4c40c9..bc3c62b72c58 100644
--- a/tools/testing/selftests/kvm/steal_time.c
+++ b/tools/testing/selftests/kvm/steal_time.c
@@ -549,7 +549,7 @@ 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();
 		kvm_pthread_create(&thread, NULL, do_steal_time, NULL);
-		pthread_getaffinity_np(thread, sizeof(cpuset), &cpuset);
+		kvm_pthread_getaffinity(thread, &cpuset);
 		TEST_ASSERT(CPU_COUNT(&cpuset) == 1 && CPU_ISSET(cpu, &cpuset),
 			    "Worker failed to inherit parent's CPU affinity");
 
-- 
2.55.0.508.g3f0d502094-goog

[PATCH v3 12/12] KVM: selftests: Clean up global constants in hardware disable test

[Sean Christopherson]

Rename the global constants in the hardware disable test to better
capture what they control, and open code the literal number of sleeping
tasks and iterations instead of using power-of-2 math to express values
that are 100% arbitrary.

Opportunistically tag the global semaphore with "static".

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/hardware_disable_test.c     | 20 +++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/tools/testing/selftests/kvm/hardware_disable_test.c b/tools/testing/selftests/kvm/hardware_disable_test.c
index 72bac346bdd9..43a36ef3ead8 100644
--- a/tools/testing/selftests/kvm/hardware_disable_test.c
+++ b/tools/testing/selftests/kvm/hardware_disable_test.c
@@ -17,13 +17,13 @@
 #include "kvm_util.h"
 #include "ucall_common.h"
 
-#define VCPU_NUM 4
-#define SLEEPING_THREAD_NUM (1 << 4)
-#define FORK_NUM (1ULL << 9)
-#define DELAY_US_MAX 2000
+#define NR_VCPUS		4
+#define NR_SLEEPERS_PER_VCPU	16
+#define NR_ITERATIONS		512
+#define DELAY_US_MAX		2000
 
 static cpu_set_t threads_cpu_set;
-sem_t *sem;
+static sem_t *sem;
 
 static void guest_code(void)
 {
@@ -75,15 +75,15 @@ static void run_test(u32 run)
 	TEST_ASSERT_EQ(pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &threads_cpu_set), 0);
 #endif
 
-	vm = vm_create(VCPU_NUM);
+	vm = vm_create(NR_VCPUS);
 
 	pr_debug("%s: [%d] start vcpus\n", __func__, run);
-	for (i = 0; i < VCPU_NUM; ++i) {
+	for (i = 0; i < NR_VCPUS; ++i) {
 		vcpu = vm_vcpu_add(vm, i, guest_code);
 
 		kvm_pthread_create(&thread, &attr, run_vcpu, vcpu);
 
-		for (j = 0; j < SLEEPING_THREAD_NUM; ++j)
+		for (j = 0; j < NR_SLEEPERS_PER_VCPU; ++j)
 			kvm_pthread_create(&thread, &attr, sleeping_thread, (void *)NULL);
 	}
 	pr_debug("%s: [%d] all threads launched\n", __func__, run);
@@ -132,7 +132,7 @@ int main(int argc, char **argv)
 
 	kvm_sched_getaffinity(0, sizeof(cpu_set_t), &allowed_cpu_set);
 
-	for (i = 0; i < VCPU_NUM && CPU_COUNT(&allowed_cpu_set); i++) {
+	for (i = 0; i < NR_VCPUS && CPU_COUNT(&allowed_cpu_set); i++) {
 		cpu = kvm_pick_random_cpu(&allowed_cpu_set);
 		CPU_CLR(cpu, &allowed_cpu_set);
 		CPU_SET(cpu, &threads_cpu_set);
@@ -141,7 +141,7 @@ int main(int argc, char **argv)
 	sem = sem_open("vm_sem", O_CREAT | O_EXCL, 0644, 0);
 	sem_unlink("vm_sem");
 
-	for (i = 0; i < FORK_NUM; ++i) {
+	for (i = 0; i < NR_ITERATIONS; ++i) {
 		pid = fork();
 		TEST_ASSERT(pid >= 0, "%s: unable to fork", __func__);
 		if (pid == 0)
-- 
2.55.0.508.g3f0d502094-goog

Re: [PATCH v3 00/12] KVM: selftests: Add more syscall wrappers, fix hardware_disable_test

[Yosry Ahmed]

On Fri, Jul 31, 2026 at 12:56 PM Sean Christopherson <seanjc@google.com> wrote:
>
> Third time's a charm.  Sorry for the spam.
>
> Buried in the middle of this is a fix for the hardware_disable_test where
> it can end up with a runtime of multiple minutes when running a preemptible
> kernels on a large NUMA system.  Turns out my analysis/guess in v1 was wrong;
> the problem isn't direct CPU contention, it's that letting the workers do any
> amount of "real" work on any CPUs leads to pthread_create() slowdowns from
> the perspective of the caller.

How long does the test take to run with this series?

Re: [PATCH v3 00/12] KVM: selftests: Add more syscall wrappers, fix hardware_disable_test

[Sean Christopherson]

On Fri, Jul 31, 2026, Yosry Ahmed wrote:
> On Fri, Jul 31, 2026 at 12:56 PM Sean Christopherson <seanjc@google.com> wrote:
> >
> > Third time's a charm.  Sorry for the spam.
> >
> > Buried in the middle of this is a fix for the hardware_disable_test where
> > it can end up with a runtime of multiple minutes when running a preemptible
> > kernels on a large NUMA system.  Turns out my analysis/guess in v1 was wrong;
> > the problem isn't direct CPU contention, it's that letting the workers do any
> > amount of "real" work on any CPUs leads to pthread_create() slowdowns from
> > the perspective of the caller.
> 
> How long does the test take to run with this series?

Less than 5 seconds if nothing else is running.  ~12 seconds if I run it
concurrently with mmu_stress_test using default settings, and ~25 seconds if I
run it on an overloaded system (naively running all selftests in parallel).

Re: [PATCH v3 00/12] KVM: selftests: Add more syscall wrappers, fix hardware_disable_test

[Yosry Ahmed]

On Fri, Jul 31, 2026 at 2:41 PM Sean Christopherson <seanjc@google.com> wrote:
>
> On Fri, Jul 31, 2026, Yosry Ahmed wrote:
> > On Fri, Jul 31, 2026 at 12:56 PM Sean Christopherson <seanjc@google.com> wrote:
> > >
> > > Third time's a charm.  Sorry for the spam.
> > >
> > > Buried in the middle of this is a fix for the hardware_disable_test where
> > > it can end up with a runtime of multiple minutes when running a preemptible
> > > kernels on a large NUMA system.  Turns out my analysis/guess in v1 was wrong;
> > > the problem isn't direct CPU contention, it's that letting the workers do any
> > > amount of "real" work on any CPUs leads to pthread_create() slowdowns from
> > > the perspective of the caller.
> >
> > How long does the test take to run with this series?
>
> Less than 5 seconds if nothing else is running.  ~12 seconds if I run it
> concurrently with mmu_stress_test using default settings, and ~25 seconds if I
> run it on an overloaded system (naively running all selftests in parallel).

Cool, I will stop skipping it after this series lands then :)