Skip to contentSkip to conversation

tests/migration: Skip tests for ppc tcg

PATCH Applied Dr. David Alan Gilbert (git) 4 participants 5 messages Jul 6, 2018 – Jul 9, 2018

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 tests/migration: Skip tests for ppc tcg

From: "Dr. David Alan Gilbert" <[email protected]>

PPC tcg seems to be failing migration tests quite regularly; we believe this is TCG bugs in dirty bit updating; it's not clear why PPC fails more but lets skip for the moment.

$ ./tests/migration-test
/ppc64/migration/deprecated: OK
/ppc64/migration/bad_dest: Skipping test: kvm_hv not available OK
/ppc64/migration/postcopy/unix: Skipping test: kvm_hv not available OK
/ppc64/migration/precopy/unix: Skipping test: kvm_hv not available OK
---
 tests/migration-test.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

v2
use a print rather than g_test_skip since gtester is broken in
some version

tests/migration-test.c +17 -6
index 3a85446f95..331efb0fe5 100644
@@ -360,7 +360,7 @@ static void migrate_start_postcopy(QTestState *who)
     qobject_unref(rsp);
 }
 
-static void test_migrate_start(QTestState **from, QTestState **to,
+static int test_migrate_start(QTestState **from, QTestState **to,
                                const char *uri, bool hide_stderr)
 {
     gchar *cmd_src, *cmd_dst;
@@ -385,9 +385,13 @@ static void test_migrate_start(QTestState **from, QTestState **to,
                                   accel, tmpfs, bootpath, uri);
     } else if (strcmp(arch, "ppc64") == 0) {
 
-        /* On ppc64, the test only works with kvm-hv, but not with kvm-pr */
+        /* On ppc64, the test only works with kvm-hv, but not with kvm-pr
+         * and TCG is touchy due to race conditions on dirty bits
+         * (especially on PPC for some reason)
+         */
         if (access("/sys/module/kvm_hv", F_OK)) {
-            accel = "tcg";
+            g_print("Skipping test: kvm_hv not available ");
+            return -1;
         }
         cmd_src = g_strdup_printf("-machine accel=%s -m 256M"
                                   " -name source,debug-threads=on"
@@ -424,6 +428,7 @@ static void test_migrate_start(QTestState **from, QTestState **to,
 
     *to = qtest_init(cmd_dst);
     g_free(cmd_dst);
+    return 0;
 }
 
 static void test_migrate_end(QTestState *from, QTestState *to, bool test_dest)
@@ -510,7 +515,9 @@ static void test_postcopy(void)
     char *uri = g_strdup_printf("unix:%s/migsocket", tmpfs);
     QTestState *from, *to;
 
-    test_migrate_start(&from, &to, uri, false);
+    if (test_migrate_start(&from, &to, uri, false)) {
+        return;
+    }
 
     migrate_set_capability(from, "postcopy-ram", "true");
     migrate_set_capability(to, "postcopy-ram", "true");
@@ -556,7 +563,9 @@ static void test_baddest(void)
     const char *status;
     bool failed;
 
-    test_migrate_start(&from, &to, "tcp:0:0", true);
+    if (test_migrate_start(&from, &to, "tcp:0:0", true)) {
+        return;
+    }
     migrate(from, "tcp:0:0");
     do {
         rsp = wait_command(from, "{ 'execute': 'query-migrate' }");
@@ -585,7 +594,9 @@ static void test_precopy_unix(void)
     char *uri = g_strdup_printf("unix:%s/migsocket", tmpfs);
     QTestState *from, *to;
 
-    test_migrate_start(&from, &to, uri, false);
+    if (test_migrate_start(&from, &to, uri, false)) {
+        return;
+    }
 
     /* We want to pick a speed slow enough that the test completes
      * quickly, but that it doesn't complete precopy even on a slow
Signature · 1 line
2.17.1
Reviewed-by on v2 1/1
#1 lore raw
Quoted 12 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

PPC tcg seems to be failing migration tests quite regularly; we believe this is TCG bugs in dirty bit updating; it's not clear why PPC fails more but lets skip for the moment.

$ ./tests/migration-test
/ppc64/migration/deprecated: OK
/ppc64/migration/bad_dest: Skipping test: kvm_hv not available OK
/ppc64/migration/postcopy/unix: Skipping test: kvm_hv not available OK
/ppc64/migration/precopy/unix: Skipping test: kvm_hv not available OK
Quoted 5 lines
---
 tests/migration-test.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)
#2 lore raw
In reply to Dr. David Alan Gilbert (git)
Quoted 12 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

PPC tcg seems to be failing migration tests quite regularly; we believe this is TCG bugs in dirty bit updating; it's not clear why PPC fails more but lets skip for the moment.

$ ./tests/migration-test
/ppc64/migration/deprecated: OK
/ppc64/migration/bad_dest: Skipping test: kvm_hv not available OK
/ppc64/migration/postcopy/unix: Skipping test: kvm_hv not available OK
/ppc64/migration/precopy/unix: Skipping test: kvm_hv not available OK
Quoted 5 lines
---
 tests/migration-test.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

Applied to master, thanks -- hopefully this will reduce the number of spurious test failures I encounter.

-- PMM

#3 lore raw
Quoted 14 lines · * Peter Maydell ([email protected]) wrote:
Quoted 12 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

PPC tcg seems to be failing migration tests quite regularly; we believe this is TCG bugs in dirty bit updating; it's not clear why PPC fails more but lets skip for the moment.

$ ./tests/migration-test
/ppc64/migration/deprecated: OK
/ppc64/migration/bad_dest: Skipping test: kvm_hv not available OK
/ppc64/migration/postcopy/unix: Skipping test: kvm_hv not available OK
/ppc64/migration/precopy/unix: Skipping test: kvm_hv not available OK
Quoted 8 lines
Quoted 5 lines
---
 tests/migration-test.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

Applied to master, thanks -- hopefully this will reduce the number of spurious test failures I encounter.

Yes; although note I believe the failures are actually real TCG errors that do need fixing, so not actually spurious as such.

Dave

Quoted 1 line

-- PMM

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK

#4 lore raw
In reply to Dr. David Alan Gilbert (git)
Quoted 37 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

PPC tcg seems to be failing migration tests quite regularly; we believe this is TCG bugs in dirty bit updating; it's not clear why PPC fails more but lets skip for the moment.

$ ./tests/migration-test
/ppc64/migration/deprecated: OK
/ppc64/migration/bad_dest: Skipping test: kvm_hv not available OK
/ppc64/migration/postcopy/unix: Skipping test: kvm_hv not available OK
/ppc64/migration/precopy/unix: Skipping test: kvm_hv not available OK
---
 tests/migration-test.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

v2
use a print rather than g_test_skip since gtester is broken in
some version

tests/migration-test.c +1 -1
index 3a85446f95..331efb0fe5 100644
@@ -360,7 +360,7 @@ static void migrate_start_postcopy(QTestState *who)
     qobject_unref(rsp);
 }
 
-static void test_migrate_start(QTestState **from, QTestState **to,
+static int test_migrate_start(QTestState **from, QTestState **to,
                                const char *uri, bool hide_stderr)
 {
     gchar *cmd_src, *cmd_dst;
@@ -385,9 +385,13 @@ static void test_migrate_start(QTestState **from, QTestState **to,
                                   accel, tmpfs, bootpath, uri);
     } else if (strcmp(arch, "ppc64") == 0) {

tests/migration-test.c static void test_migrate_start(QTestState **from, QTestState **to, · line 394

Quoted 10 lines
tests/migration-test.c +6 -2
 
-        /* On ppc64, the test only works with kvm-hv, but not with kvm-pr */
+        /* On ppc64, the test only works with kvm-hv, but not with kvm-pr
+         * and TCG is touchy due to race conditions on dirty bits
+         * (especially on PPC for some reason)
+         */
         if (access("/sys/module/kvm_hv", F_OK)) {
-            accel = "tcg";
+            g_print("Skipping test: kvm_hv not available ");
+            return -1;

That "Skipping test: ..." message is now popping up a couple of times during "make check-qtest" here on my x86 laptop. That's kind of ugly. Could you maybe turn that into a g_debug() instead?

Thomas