[PATCH rcu 0/4] Torture-test updates for smp_call_function()

[PATCH rcu 0/4] Torture-test updates for smp_call_function()

[Paul E. McKenney]

Hello!

This series contains updates for the smp_call_function() torture tests:

1.	Scale scftorture memory based on number of CPUs.

2.	Forgive memory-allocation failure if KASAN.

3.	Pause testing after memory-allocation failure.

4.	Add CONFIG_PREEMPT_DYNAMIC=n to NOPREEMPT scenario.

						Thanx, Paul

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

 b/kernel/scftorture.c                                      |    6 ++++--
 b/tools/testing/selftests/rcutorture/bin/torture.sh        |    4 +++-
 b/tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT |    2 ++
 kernel/scftorture.c                                        |    6 +++++-
 4 files changed, 14 insertions(+), 4 deletions(-)

[PATCH rcu 1/4] torture: Scale scftorture memory based on number of CPUs

[Paul E. McKenney]

As the number of CPUs increases, the number of outstanding no-wait
smp_call_function() handlers also increases, so that the default of
2G of memory is not always sufficient on 80-CPU systems.  This commit
therefore scales the amount of memory specified to qemu based on the
number of CPUs specified to the scftorture test instance.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 tools/testing/selftests/rcutorture/bin/torture.sh | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/rcutorture/bin/torture.sh b/tools/testing/selftests/rcutorture/bin/torture.sh
index 5a2ae2264403..9fa6526067f0 100755
--- a/tools/testing/selftests/rcutorture/bin/torture.sh
+++ b/tools/testing/selftests/rcutorture/bin/torture.sh
@@ -376,8 +376,10 @@ fi
 
 if test "$do_scftorture" = "yes"
 then
+	# Scale memory based on the number of CPUs.
+	scfmem=$((2+HALF_ALLOTED_CPUS/16))
 	torture_bootargs="scftorture.nthreads=$HALF_ALLOTED_CPUS torture.disable_onoff_at_boot csdlock_debug=1"
-	torture_set "scftorture" tools/testing/selftests/rcutorture/bin/kvm.sh --torture scf --allcpus --duration "$duration_scftorture" --configs "$configs_scftorture" --kconfig "CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --memory 2G --trust-make
+	torture_set "scftorture" tools/testing/selftests/rcutorture/bin/kvm.sh --torture scf --allcpus --duration "$duration_scftorture" --configs "$configs_scftorture" --kconfig "CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --memory ${scfmem}G --trust-make
 fi
 
 if test "$do_rt" = "yes"
-- 
2.40.1

[PATCH rcu 2/4] scftorture: Forgive memory-allocation failure if KASAN

[Paul E. McKenney]

Kernels built with CONFIG_KASAN=y quarantine newly freed memory in order
to better detect use-after-free errors.  However, this can exhaust memory
more quickly in allocator-heavy tests, which can result in spurious
scftorture failure.  This commit therefore forgives memory-allocation
failure in kernels built with CONFIG_KASAN=y, but continues counting
the errors for use in detailed test-result analyses.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/scftorture.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/scftorture.c b/kernel/scftorture.c
index 5d113aa59e77..83c33ba0ca7e 100644
--- a/kernel/scftorture.c
+++ b/kernel/scftorture.c
@@ -171,7 +171,8 @@ static void scf_torture_stats_print(void)
 		scfs.n_all_wait += scf_stats_p[i].n_all_wait;
 	}
 	if (atomic_read(&n_errs) || atomic_read(&n_mb_in_errs) ||
-	    atomic_read(&n_mb_out_errs) || atomic_read(&n_alloc_errs))
+	    atomic_read(&n_mb_out_errs) ||
+	    (!IS_ENABLED(CONFIG_KASAN) && atomic_read(&n_alloc_errs)))
 		bangstr = "!!! ";
 	pr_alert("%s %sscf_invoked_count %s: %lld resched: %lld single: %lld/%lld single_ofl: %lld/%lld single_rpc: %lld single_rpc_ofl: %lld many: %lld/%lld all: %lld/%lld ",
 		 SCFTORT_FLAG, bangstr, isdone ? "VER" : "ver", invoked_count, scfs.n_resched,
@@ -323,7 +324,8 @@ static void scftorture_invoke_one(struct scf_statistics *scfp, struct torture_ra
 		preempt_disable();
 	if (scfsp->scfs_prim == SCF_PRIM_SINGLE || scfsp->scfs_wait) {
 		scfcp = kmalloc(sizeof(*scfcp), GFP_ATOMIC);
-		if (WARN_ON_ONCE(!scfcp)) {
+		if (!scfcp) {
+			WARN_ON_ONCE(!IS_ENABLED(CONFIG_KASAN));
 			atomic_inc(&n_alloc_errs);
 		} else {
 			scfcp->scfc_cpu = -1;
-- 
2.40.1

[PATCH rcu 3/4] scftorture: Pause testing after memory-allocation failure

[Paul E. McKenney]

The scftorture test can quickly execute a large number of calls to no-wait
smp_call_function(), each of which holds a block of memory until the
corresponding handler is invoked.  Especially when the longwait module
parameter is specified, this can chew up an arbitrarily large amount
of memory.  This commit therefore blocks after each memory-allocation
failure, with the duration a function of longwait.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 kernel/scftorture.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/kernel/scftorture.c b/kernel/scftorture.c
index 83c33ba0ca7e..59032aaccd18 100644
--- a/kernel/scftorture.c
+++ b/kernel/scftorture.c
@@ -313,6 +313,7 @@ static void scf_handler_1(void *scfc_in)
 // Randomly do an smp_call_function*() invocation.
 static void scftorture_invoke_one(struct scf_statistics *scfp, struct torture_random_state *trsp)
 {
+	bool allocfail = false;
 	uintptr_t cpu;
 	int ret = 0;
 	struct scf_check *scfcp = NULL;
@@ -327,6 +328,7 @@ static void scftorture_invoke_one(struct scf_statistics *scfp, struct torture_ra
 		if (!scfcp) {
 			WARN_ON_ONCE(!IS_ENABLED(CONFIG_KASAN));
 			atomic_inc(&n_alloc_errs);
+			allocfail = true;
 		} else {
 			scfcp->scfc_cpu = -1;
 			scfcp->scfc_wait = scfsp->scfs_wait;
@@ -433,7 +435,9 @@ static void scftorture_invoke_one(struct scf_statistics *scfp, struct torture_ra
 		cpus_read_unlock();
 	else
 		preempt_enable();
-	if (!(torture_random(trsp) & 0xfff))
+	if (allocfail)
+		schedule_timeout_idle((1 + longwait) * HZ);  // Let no-wait handlers complete.
+	else if (!(torture_random(trsp) & 0xfff))
 		schedule_timeout_uninterruptible(1);
 }
 
-- 
2.40.1

[PATCH rcu 4/4] scftorture: Add CONFIG_PREEMPT_DYNAMIC=n to NOPREEMPT scenario

[Paul E. McKenney]

It is no longer possible to build a kernel with a preemption-disabled
RCU without use of CONFIG_PREEMPT_DYNAMIC=n.  This commit therefore
adds CONFIG_PREEMPT_DYNAMIC=n to the scf torture type's NOPREEMPT
scenario file.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT b/tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT
index 3a59346b3de7..6133f54ce2a7 100644
--- a/tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT
+++ b/tools/testing/selftests/rcutorture/configs/scf/NOPREEMPT
@@ -2,6 +2,8 @@ CONFIG_SMP=y
 CONFIG_PREEMPT_NONE=y
 CONFIG_PREEMPT_VOLUNTARY=n
 CONFIG_PREEMPT=n
+CONFIG_PREEMPT_DYNAMIC=n
+#CHECK#CONFIG_PREEMPT_RCU=n
 CONFIG_HZ_PERIODIC=n
 CONFIG_NO_HZ_IDLE=n
 CONFIG_NO_HZ_FULL=y
-- 
2.40.1