[PATCH 1/1] selftests/powerpc: Improve tm-resched-dscr

The tm-resched-dscr self test can, in some situations, run for
several minutes before being successfully interrupted by the context
switch it needs in order to perform the test. This often seems to
occur when the test is being run in a virtual machine.

Improve the test by running it under eat_cpu() to guarantee
contention for the CPU and increase the chance of a context switch.

In practice this seems to reduce the test time, in some cases, from
more than two minutes to under a second.

Also remove the "progress dots" so that if the test does run for a
long time, it doesn't produce large amounts of unnecessary output.

Signed-off-by: Sam Bobroff <redacted>
---
 tools/testing/selftests/powerpc/tm/Makefile          |  1 +
 tools/testing/selftests/powerpc/tm/tm-resched-dscr.c | 12 ++++++++----
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/powerpc/tm/Makefile b/tools/testing/selftests/powerpc/tm/Makefile
index 958c11c14acd..7bfcd454fb2a 100644
--- a/tools/testing/selftests/powerpc/tm/Makefile
+++ b/tools/testing/selftests/powerpc/tm/Makefile

@@ -15,6 +15,7 @@ $(OUTPUT)/tm-syscall: tm-syscall-asm.S
 $(OUTPUT)/tm-syscall: CFLAGS += -I../../../../../usr/include
 $(OUTPUT)/tm-tmspr: CFLAGS += -pthread
 $(OUTPUT)/tm-vmx-unavail: CFLAGS += -pthread -m64
+$(OUTPUT)/tm-resched-dscr: ../pmu/lib.o
 
 SIGNAL_CONTEXT_CHK_TESTS := $(patsubst %,$(OUTPUT)/%,$(SIGNAL_CONTEXT_CHK_TESTS))
 $(SIGNAL_CONTEXT_CHK_TESTS): tm-signal.S

diff --git a/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c b/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c
index e79ccd6aada1..a7ac2e4c60d9 100644
--- a/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c
+++ b/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c

@@ -30,6 +30,7 @@
 
 #include "utils.h"
 #include "tm.h"
+#include "../pmu/lib.h"
 
 #define SPRN_DSCR       0x03
 

@@ -75,8 +76,6 @@ int test_body(void)
 		);
 		assert(rv); /* make sure the transaction aborted */
 		if ((texasr >> 56) != TM_CAUSE_RESCHED) {
-			putchar('.');
-			fflush(stdout);
 			continue;
 		}
 		if (dscr2 != dscr1) {

@@ -89,7 +88,12 @@ int test_body(void)
 	}
 }
 
-int main(void)
+static int tm_resched_dscr(void)
 {
-	return test_harness(test_body, "tm_resched_dscr");
+	return eat_cpu(test_body);
+}
+
+int main(int argc, const char *argv[])
+{
+	return test_harness(tm_resched_dscr, "tm_resched_dscr");
 }

-- 
2.14.1.2.g4274c698f

On Thu, 2017-08-17 at 01:06:47 UTC, Sam bobroff wrote:

The tm-resched-dscr self test can, in some situations, run for
several minutes before being successfully interrupted by the context
switch it needs in order to perform the test. This often seems to
occur when the test is being run in a virtual machine.

Improve the test by running it under eat_cpu() to guarantee
contention for the CPU and increase the chance of a context switch.

In practice this seems to reduce the test time, in some cases, from
more than two minutes to under a second.

Also remove the "progress dots" so that if the test does run for a
long time, it doesn't produce large amounts of unnecessary output.

Signed-off-by: Sam Bobroff <redacted>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/99597ceda00b57396a176e08bd5e5c

cheers