Skip to contentSkip to conversation

9pfs: test suite fixes

PATCH Superseded Christian Schoenebeck 4 participants 10 messages Oct 30, 2020 – Nov 9, 2020

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover 9pfs: test suite fixes

Fixes two bugs with the 9pfs 'local' tests as discussed with latest 9P PR (2020-10-23). See the discussion of that PR for details.

Christian Schoenebeck (2):
tests/9pfs: fix test dir for parallel tests tests/9pfs: fix coverity error in create_local_test_dir()

 tests/qtest/libqos/virtio-9p.c | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/2 tests/9pfs: fix test dir for parallel tests

Use mkdtemp() to generate a unique directory for the 9p 'local' tests.

This fixes occasional 9p test failures when running 'make check -jN' if QEMU was compiled for multiple target architectures, because the individual architecture's test suites would run in parallel and interfere with each other's data as the test directory was previously hard coded and hence the same directory was used by all of them simultaniously.

This also requires a change how the test directory is created and deleted:
As the test path is now randomized and virtio_9p_register_nodes() being called in a somewhat undeterministic way, that's no longer an appropriate place to create and remove the test directory. Use a constructor and destructor function for creating and removing the test directory instead. Unfortunately libqos currently does not support setup/teardown callbacks to handle this more cleanly.

---
 tests/qtest/libqos/virtio-9p.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)
tests/qtest/libqos/virtio-9p.c +19 -6
index d43647b3b7..6b22fa0e9a 100644
@@ -35,7 +35,12 @@ static char *concat_path(const char* a, const char* b)
 static void init_local_test_path(void)
 {
     char *pwd = g_get_current_dir();
-    local_test_path = concat_path(pwd, "qtest-9p-local");
+    char *template = concat_path(pwd, "qtest-9p-local-XXXXXX");
+    local_test_path = mkdtemp(template);
+    if (!local_test_path) {
+        g_test_message("mkdtemp('%s') failed: %s", template, strerror(errno));
+    }
+    g_assert(local_test_path);
     g_free(pwd);
 }
 
@@ -246,11 +251,6 @@ static void virtio_9p_register_nodes(void)
     const char *str_simple = "fsdev=fsdev0,mount_tag=" MOUNT_TAG;
     const char *str_addr = "fsdev=fsdev0,addr=04.0,mount_tag=" MOUNT_TAG;
 
-    /* make sure test dir for the 'local' tests exists and is clean */
-    init_local_test_path();
-    remove_local_test_dir();
-    create_local_test_dir();
-
     QPCIAddress addr = {
         .devfn = QPCI_DEVFN(4, 0),
     };
@@ -278,3 +278,16 @@ static void virtio_9p_register_nodes(void)
 }
 
 libqos_init(virtio_9p_register_nodes);
+
+static void __attribute__((constructor)) construct_virtio_9p(void)
+{
+    /* make sure test dir for the 'local' tests exists */
+    init_local_test_path();
+    create_local_test_dir();
+}
+
+static void __attribute__((destructor)) destruct_virtio_9p(void)
+{
+    /* remove previously created test dir when test suite completed */
+    remove_local_test_dir();
+}
Signature · 1 line
2.20.1
#4 lore raw

PATCH 2/2 tests/9pfs: fix coverity error in create_local_test_dir()

In reply to Christian Schoenebeck

Coverity wants the return value of mkdir() to be checked, so let's pretend to do that. We're actually just making a dummy check and ignore the result, because we actually only care if the required directory exists and we have an existence check for that in place already.

---
 tests/qtest/libqos/virtio-9p.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
tests/qtest/libqos/virtio-9p.c +5 -1
index 6b22fa0e9a..0a7c0ee5d8 100644
@@ -48,9 +48,13 @@ static void init_local_test_path(void)
 static void create_local_test_dir(void)
 {
     struct stat st;
+    int res;
 
     g_assert(local_test_path != NULL);
-    mkdir(local_test_path, 0777);
+    res = mkdir(local_test_path, 0777);
+    if (res < 0) {
+        /* ignore error, dummy check to prevent error by coverity */
+    }
 
     /* ensure test directory exists now ... */
     g_assert(stat(local_test_path, &st) == 0);
Signature · 1 line
2.20.1
Tested-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] tests/9pfs: fix test dir for parallel tests

In reply to Christian Schoenebeck
Quoted 16 lines · On Fri, 30 Oct 2020 09:19:46 +0100 Christian Schoenebeck…

Use mkdtemp() to generate a unique directory for the 9p 'local' tests.

This fixes occasional 9p test failures when running 'make check -jN' if QEMU was compiled for multiple target architectures, because the individual architecture's test suites would run in parallel and interfere with each other's data as the test directory was previously hard coded and hence the same directory was used by all of them simultaniously.

This also requires a change how the test directory is created and deleted:
As the test path is now randomized and virtio_9p_register_nodes() being called in a somewhat undeterministic way, that's no longer an appropriate place to create and remove the test directory. Use a constructor and destructor function for creating and removing the test directory instead. Unfortunately libqos currently does not support setup/teardown callbacks to handle this more cleanly.

Quoted 2 lines

---

LGTM

I've been running 'make check-qtest -j' with 4 archs for 2 hours without hitting the issue.

Quoted 8 lines
 tests/qtest/libqos/virtio-9p.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)
tests/qtest/libqos/virtio-9p.c +0 -0
index d43647b3b7..6b22fa0e9a 100644

tests/qtest/libqos/virtio-9p.c static char *concat_path(const char* a, const char* b) · line 41

Quoted 8 lines
tests/qtest/libqos/virtio-9p.c +4 -1
@@ -35,7 +35,12 @@ static char *concat_path(const char* a, const char* b)
 static void init_local_test_path(void)
 {
     char *pwd = g_get_current_dir();
-    local_test_path = concat_path(pwd, "qtest-9p-local");
+    char *template = concat_path(pwd, "qtest-9p-local-XXXXXX");
+    local_test_path = mkdtemp(template);
+    if (!local_test_path) {
+        g_test_message("mkdtemp('%s') failed: %s", template, strerror(errno));

Just per curiosity, is there a preferred way to output error messages ?

Cc'ing Thomas and Laurent.

[...]

#5 lore raw

Re: [PATCH 2/2] tests/9pfs: fix coverity error in create_local_test_dir()

In reply to Christian Schoenebeck
Quoted 6 lines · On Fri, 30 Oct 2020 09:19:46 +0100 Christian Schoenebeck…

Coverity wants the return value of mkdir() to be checked, so let's pretend to do that. We're actually just making a dummy check and ignore the result, because we actually only care if the required directory exists and we have an existence check for that in place already.

I see that sometimes changelog shows a copy of the original coverity report (e.g. commit df1a312fea58).

Quoted 1 line

Please give credits to coverity, not me :-)

And most importantly, we want to mention the CID in the changelog.

e.g.

Reported-by Coverity (CID 1435963)
Quoted 10 lines
---
 tests/qtest/libqos/virtio-9p.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
tests/qtest/libqos/virtio-9p.c +0 -0
index 6b22fa0e9a..0a7c0ee5d8 100644

tests/qtest/libqos/virtio-9p.c static void init_local_test_path(void) · line 56

Quoted 10 lines
tests/qtest/libqos/virtio-9p.c +4 -1
@@ -48,9 +48,13 @@ static void init_local_test_path(void)
 static void create_local_test_dir(void)
 {
     struct stat st;
+    int res;
 
     g_assert(local_test_path != NULL);
-    mkdir(local_test_path, 0777);
+    res = mkdir(local_test_path, 0777);
+    if (res < 0) {
+        /* ignore error, dummy check to prevent error by coverity */

Why not printing an error message with errno there like you did in the previous patch ?

Quoted 4 lines
tests/qtest/libqos/virtio-9p.c +1 -0
+    }
 
     /* ensure test directory exists now ... */
     g_assert(stat(local_test_path, &st) == 0);
#6 lore raw

Re: [PATCH 2/2] tests/9pfs: fix coverity error in create_local_test_dir()

Quoted 11 lines · On Freitag, 30. Oktober 2020 12:44:18 CET Greg Kurz wrote:

On Fri, 30 Oct 2020 09:19:46 +0100

Quoted 5 lines · Christian Schoenebeck

Coverity wants the return value of mkdir() to be checked, so let's pretend to do that. We're actually just making a dummy check and ignore the result, because we actually only care if the required directory exists and we have an existence check for that in place already.

I see that sometimes changelog shows a copy of the original coverity report (e.g. commit df1a312fea58).

Ok, I'll add that.

Quoted 9 lines
Quoted 1 line

Please give credits to coverity, not me :-)

And most importantly, we want to mention the CID in the changelog.

e.g.

Reported-by Coverity (CID 1435963)

Ok.

It's not clear to me where this coverity report is accessible online. A quick search only brought me to statistics about its latest check, but not the details of the report you quoted.

And more importantly: is there coverity CI support that one could enable on github, so that pending patches were checked before upstream merge?

Quoted 22 lines
Quoted 22 lines

---

 tests/qtest/libqos/virtio-9p.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/tests/qtest/libqos/virtio-9p.c index 6b22fa0e9a..0a7c0ee5d8 100644

tests/qtest/libqos/virtio-9p.c +1 -0
@@ -48,9 +48,13 @@ static void init_local_test_path(void)
 
 static void create_local_test_dir(void)
 {
 
     struct stat st;
 
+    int res;
 
     g_assert(local_test_path != NULL);
 
Quoted 7 lines

tests/qtest/libqos/virtio-9p.c static void init_local_test_path(void) · line 56

Quoted 4 lines
tests/qtest/libqos/virtio-9p.c +3 -1
-    mkdir(local_test_path, 0777);
+    res = mkdir(local_test_path, 0777);
+    if (res < 0) {
+        /* ignore error, dummy check to prevent error by coverity */

Why not printing an error message with errno there like you did in the previous patch ?

Yeah, originally I didn't want to trigger false positives on automated CIs if mkdir() failed just because the directory already exists. But OTOH g_test_message() is just an info-level message, so it is Ok to bark silently and I will add it.

Quoted 5 lines
Quoted 5 lines

+ }

     /* ensure test directory exists now ... */
     g_assert(stat(local_test_path, &st) == 0);

Thanks!

Best regards,
Christian Schoenebeck

#7 lore raw

Re: [PATCH 2/2] tests/9pfs: fix coverity error in create_local_test_dir()

Quoted 4 lines · Christian Schoenebeck

On Freitag, 30. Oktober 2020 12:44:18 CET Greg Kurz wrote:
It's not clear to me where this coverity report is accessible online. A quick search only brought me to statistics about its latest check, but not the details of the report you quoted.

https://scan.coverity.com/projects/qemu . To see the actual defect reports you need to create an account and request access to the QEMU project (we happily give access to developers, but it is a manual-approval process).

Quoted 2 lines

And more importantly: is there coverity CI support that one could enable on github, so that pending patches were checked before upstream merge?

No, unfortunately not. The Coverity free-for-open-source-projects system has a very limited number of scans it allows (for a project the size of ours just one a day) so we can't open it up to submaintainer branches or even use it on pull requests pre merge; the best we can do is running it on master daily.

thanks
-- PMM

#9 lore raw

Re: [PATCH 2/2] tests/9pfs: fix coverity error in create_local_test_dir()

In reply to Christian Schoenebeck
Quoted 23 lines · On Fri, 30 Oct 2020 12:59:48 +0100 Christian Schoenebeck…
Quoted 11 lines · On Freitag, 30. Oktober 2020 12:44:18 CET Greg Kurz wrote:

On Fri, 30 Oct 2020 09:19:46 +0100

Quoted 5 lines · Christian Schoenebeck

Coverity wants the return value of mkdir() to be checked, so let's pretend to do that. We're actually just making a dummy check and ignore the result, because we actually only care if the required directory exists and we have an existence check for that in place already.

I see that sometimes changelog shows a copy of the original coverity report (e.g. commit df1a312fea58).

Ok, I'll add that.

Quoted 8 lines
Quoted 1 line

Please give credits to coverity, not me :-)

And most importantly, we want to mention the CID in the changelog.

e.g.

Quoted 8 lines
Quoted 1 line
Reported-by Coverity (CID 1435963)

Ok.

It's not clear to me where this coverity report is accessible online. A quick search only brought me to statistics about its latest check, but not the details of the report you quoted.

I've been notified by mail because I have an account there.

https://scan.coverity.com/users/sign_up

Quoted 3 lines

And more importantly: is there coverity CI support that one could enable on github, so that pending patches were checked before upstream merge?

I see that Peter already provided the details.

Quoted 30 lines
Quoted 22 lines
Quoted 22 lines

---

 tests/qtest/libqos/virtio-9p.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/tests/qtest/libqos/virtio-9p.c index 6b22fa0e9a..0a7c0ee5d8 100644

tests/qtest/libqos/virtio-9p.c +1 -0
@@ -48,9 +48,13 @@ static void init_local_test_path(void)
 
 static void create_local_test_dir(void)
 {
 
     struct stat st;
 
+    int res;
 
     g_assert(local_test_path != NULL);
 
Quoted 7 lines

tests/qtest/libqos/virtio-9p.c static void init_local_test_path(void) · line 56

Quoted 4 lines
tests/qtest/libqos/virtio-9p.c +3 -1
-    mkdir(local_test_path, 0777);
+    res = mkdir(local_test_path, 0777);
+    if (res < 0) {
+        /* ignore error, dummy check to prevent error by coverity */

Why not printing an error message with errno there like you did in the previous patch ?

Quoted 2 lines

Yeah, originally I didn't want to trigger false positives on automated CIs if mkdir() failed just because the directory already exists. But OTOH

mkdtemp() should buy you the directory doesn't exist.

Quoted 15 lines

g_test_message() is just an info-level message, so it is Ok to bark silently and I will add it.

Quoted 5 lines
Quoted 5 lines

+ }

     /* ensure test directory exists now ... */
     g_assert(stat(local_test_path, &st) == 0);

Thanks!

Best regards,
Christian Schoenebeck

#8 lore raw

Re: [PATCH 2/2] tests/9pfs: fix coverity error in create_local_test_dir()

In reply to Peter Maydell
Quoted 17 lines · On Freitag, 30. Oktober 2020 13:09:26 CET Peter Maydell wrote:

On Fri, 30 Oct 2020 at 12:02, Christian Schoenebeck

Quoted 4 lines · <[email protected]> wrote:

On Freitag, 30. Oktober 2020 12:44:18 CET Greg Kurz wrote:
It's not clear to me where this coverity report is accessible online. A quick search only brought me to statistics about its latest check, but not the details of the report you quoted.

https://scan.coverity.com/projects/qemu . To see the actual defect reports you need to create an account and request access to the QEMU project (we happily give access to developers, but it is a manual-approval process).

Quoted 3 lines

And more importantly: is there coverity CI support that one could enable on
github, so that pending patches were checked before upstream merge?

Quoted 8 lines

No, unfortunately not. The Coverity free-for-open-source-projects system has a very limited number of scans it allows (for a project the size of ours just one a day) so we can't open it up to submaintainer branches or even use it on pull requests pre merge; the best we can do is running it on master daily.

thanks
-- PMM

Thanks for the clarification Peter!

I try to sign up for Coverity next week.

Best regards,
Christian Schoenebeck

#3 lore raw

Re: [PATCH 1/2] tests/9pfs: fix test dir for parallel tests

In reply to Greg Kurz
Quoted 2 lines · Greg Kurz

On Fri, 30 Oct 2020 09:19:46 +0100
Christian Schoenebeck <[email protected]> wrote:

[...]

Quoted 7 lines
Quoted 7 lines
tests/qtest/libqos/virtio-9p.c +0 -0
index d43647b3b7..6b22fa0e9a 100644
@@ -35,7 +35,12 @@ static char *concat_path(const char* a, const char* b)
 static void init_local_test_path(void)
 {
Quoted 8 lines

tests/qtest/libqos/virtio-9p.c static char *concat_path(const char* a, const char* b) · line 41

Quoted 6 lines
tests/qtest/libqos/virtio-9p.c +4 -1
     char *pwd = g_get_current_dir();
-    local_test_path = concat_path(pwd, "qtest-9p-local");
+    char *template = concat_path(pwd, "qtest-9p-local-XXXXXX");
+    local_test_path = mkdtemp(template);
+    if (!local_test_path) {
+        g_test_message("mkdtemp('%s') failed: %s", template, strerror(errno));

Just per curiosity, is there a preferred way to output error messages ?

I don't think that we've ever agreed on a common way here... but as far as I can see, most tests rather use fprintf(stderr, ...) for error messages, while g_test_message() is rather for normal log messages?

Thomas