| Message ID | 20210906132508.15180-1-raphael.norwitz@nutanix.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] Prevent vhost-user-blk-test hang | expand |
On Mon, Sep 06, 2021 at 01:25:20PM +0000, Raphael Norwitz wrote: > In the vhost-user-blk-test, as of now there is nothing stoping > vhost-user-blk in QEMU writing to the socket right after forking off the > storage daemon before it has a chance to come up properly, leaving the > test hanging forever. This intermittently hanging test has caused QEMU > automation failures reported multiple times on the mailing list [1]. > > This change makes the storage-daemon notify the vhost-user-blk-test > that it is fully initialized and ready to handle client connections by > creating a pidfile on initialiation. This ensures that the storage-daemon > backend won't miss vhost-user messages and thereby resolves the hang. > > [1] https://lore.kernel.org/qemu-devel/CAFEAcA8kYpz9LiPNxnWJAPSjc=nv532bEdyfynaBeMeohqBp3A@mail.gmail.com/ > > Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com> > --- > tests/qtest/vhost-user-blk-test.c | 33 ++++++++++++++++++++++++++++++- > 1 file changed, 32 insertions(+), 1 deletion(-) > > diff --git a/tests/qtest/vhost-user-blk-test.c b/tests/qtest/vhost-user-blk-test.c > index 6f108a1b62..78140e6f28 100644 > --- a/tests/qtest/vhost-user-blk-test.c > +++ b/tests/qtest/vhost-user-blk-test.c > @@ -24,6 +24,9 @@ > #define TEST_IMAGE_SIZE (64 * 1024 * 1024) > #define QVIRTIO_BLK_TIMEOUT_US (30 * 1000 * 1000) > #define PCI_SLOT_HP 0x06 > +#define PIDFILE_RETRIES 5 > + > +const char *pidfile_format = "/tmp/daemon-%d"; Why is this not static? In fact... > > typedef struct { > pid_t pid; > @@ -885,7 +888,8 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, > int num_queues) > { > const char *vhost_user_blk_bin = qtest_qemu_storage_daemon_binary(); > - int i; > + int i, err, retries; > + char *daemon_pidfile_path; > gchar *img_path; > GString *storage_daemon_command = g_string_new(NULL); > QemuStorageDaemonState *qsd; > @@ -898,6 +902,12 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, > " -object memory-backend-memfd,id=mem,size=256M,share=on " > " -M memory-backend=mem -m 256M "); > > + err = asprintf(&daemon_pidfile_path, pidfile_format, getpid()); ...action at a distance makes it harder for gcc to warn about bad formats. I'd just inline "/tmp/daemon-%d" here in the asprintf call, and drop pidfile_format altogether. Why are we using bare asprintf instead of glib's g_strdup_printf? > + if (err == -1) { > + fprintf(stderr, "Failed to format storage-daemon pidfile name %m"); %m in printf is a glibc-ism; not portable to non-Linux. Do we care? > + abort(); Rather than directly abort, since this is a glib test runner, is there a glib function we should be using? For example, using g_assert_cmpint((err, !=, -1)? > + } > + > for (i = 0; i < vus_instances; i++) { > int fd; > char *sock_path = create_listen_socket(&fd); > @@ -914,6 +924,9 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, > i + 1, sock_path); > } > > + g_string_append_printf(storage_daemon_command, "--pidfile %s", Missing a space on the tail end if there are more arguments to append to the command line. > + daemon_pidfile_path); > + > g_test_message("starting vhost-user backend: %s", > storage_daemon_command->str); > pid_t pid = fork(); > @@ -930,7 +943,25 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, > execlp("/bin/sh", "sh", "-c", storage_daemon_command->str, NULL); > exit(1); > } > + > + /* > + * Ensure the storage-daemon has come up properly before allowing the > + * test to proceed. > + */ > + retries = 0; > + while (access(daemon_pidfile_path, F_OK) != 0) { > + if (retries > PIDFILE_RETRIES) { > + fprintf(stderr, "The storage-daemon failed to come up after %d " > + "seconds - killing the test", PIDFILE_RETRIES); > + abort(); Again, would some form of g_assert*() be better than bare abort? > + } > + > + retries++; > + usleep(1000); We're inconsistent on whether qtest files use bare usleep or g_usleep. > + } > + > g_string_free(storage_daemon_command, true); > + free(daemon_pidfile_path); Do you want to unlink() the file (if it exists)? > > qsd = g_new(QemuStorageDaemonState, 1); > qsd->pid = pid; > -- > 2.20.1 >
diff --git a/tests/qtest/vhost-user-blk-test.c b/tests/qtest/vhost-user-blk-test.c index 6f108a1b62..78140e6f28 100644 --- a/tests/qtest/vhost-user-blk-test.c +++ b/tests/qtest/vhost-user-blk-test.c @@ -24,6 +24,9 @@ #define TEST_IMAGE_SIZE (64 * 1024 * 1024) #define QVIRTIO_BLK_TIMEOUT_US (30 * 1000 * 1000) #define PCI_SLOT_HP 0x06 +#define PIDFILE_RETRIES 5 + +const char *pidfile_format = "/tmp/daemon-%d"; typedef struct { pid_t pid; @@ -885,7 +888,8 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, int num_queues) { const char *vhost_user_blk_bin = qtest_qemu_storage_daemon_binary(); - int i; + int i, err, retries; + char *daemon_pidfile_path; gchar *img_path; GString *storage_daemon_command = g_string_new(NULL); QemuStorageDaemonState *qsd; @@ -898,6 +902,12 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, " -object memory-backend-memfd,id=mem,size=256M,share=on " " -M memory-backend=mem -m 256M "); + err = asprintf(&daemon_pidfile_path, pidfile_format, getpid()); + if (err == -1) { + fprintf(stderr, "Failed to format storage-daemon pidfile name %m"); + abort(); + } + for (i = 0; i < vus_instances; i++) { int fd; char *sock_path = create_listen_socket(&fd); @@ -914,6 +924,9 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, i + 1, sock_path); } + g_string_append_printf(storage_daemon_command, "--pidfile %s", + daemon_pidfile_path); + g_test_message("starting vhost-user backend: %s", storage_daemon_command->str); pid_t pid = fork(); @@ -930,7 +943,25 @@ static void start_vhost_user_blk(GString *cmd_line, int vus_instances, execlp("/bin/sh", "sh", "-c", storage_daemon_command->str, NULL); exit(1); } + + /* + * Ensure the storage-daemon has come up properly before allowing the + * test to proceed. + */ + retries = 0; + while (access(daemon_pidfile_path, F_OK) != 0) { + if (retries > PIDFILE_RETRIES) { + fprintf(stderr, "The storage-daemon failed to come up after %d " + "seconds - killing the test", PIDFILE_RETRIES); + abort(); + } + + retries++; + usleep(1000); + } + g_string_free(storage_daemon_command, true); + free(daemon_pidfile_path); qsd = g_new(QemuStorageDaemonState, 1); qsd->pid = pid;
In the vhost-user-blk-test, as of now there is nothing stoping vhost-user-blk in QEMU writing to the socket right after forking off the storage daemon before it has a chance to come up properly, leaving the test hanging forever. This intermittently hanging test has caused QEMU automation failures reported multiple times on the mailing list [1]. This change makes the storage-daemon notify the vhost-user-blk-test that it is fully initialized and ready to handle client connections by creating a pidfile on initialiation. This ensures that the storage-daemon backend won't miss vhost-user messages and thereby resolves the hang. [1] https://lore.kernel.org/qemu-devel/CAFEAcA8kYpz9LiPNxnWJAPSjc=nv532bEdyfynaBeMeohqBp3A@mail.gmail.com/ Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com> --- tests/qtest/vhost-user-blk-test.c | 33 ++++++++++++++++++++++++++++++- 1 file changed, 32 insertions(+), 1 deletion(-)