Re: [Qemu-devel] [PATCH] tests: Exit boot-serial-test loop if child dies

[Wainer dos Santos Moschetta]

On 12/17/2018 01:58 AM, Richard Henderson wrote:
> There's no point in waiting 5 full minutes when there will be
> no more output.  Compute timeout based on elapsed wall clock
> time instead of N * delays, as the delay is a minimum sleep time.
>
> Cc: Thomas Huth <address@hidden>
> Cc: Laurent Vivier <address@hidden>
> Cc: Paolo Bonzini <address@hidden>
> Signed-off-by: Richard Henderson <address@hidden>
> ---
>   tests/libqtest.h         |  8 +++++
>   tests/boot-serial-test.c | 16 +++++++--
>   tests/libqtest.c         | 70 +++++++++++++++++++++++++---------------
>   3 files changed, 65 insertions(+), 29 deletions(-)

Reviewed-by: Wainer dos Santos Moschetta <address@hidden>

> diff --git a/tests/libqtest.h b/tests/libqtest.h
> index 96a6c01352..9758c51be6 100644
> --- a/tests/libqtest.h
> +++ b/tests/libqtest.h
> @@ -1011,4 +1011,12 @@ bool qmp_rsp_is_err(QDict *rsp);
>    */
>   void qmp_assert_error_class(QDict *rsp, const char *class);
>   
> +/**
> + * qtest_probe_child:
> + * @s: QTestState instance to operate on.
> + *
> + * Returns: true if the child is still alive.
> + */
> +bool qtest_probe_child(QTestState *s);
> +
>   #endif
> diff --git a/tests/boot-serial-test.c b/tests/boot-serial-test.c
> index 8ec6aed35d..54b9347325 100644
> --- a/tests/boot-serial-test.c
> +++ b/tests/boot-serial-test.c
> @@ -130,11 +130,12 @@ static testdef_t tests[] = {
>   
>   static bool check_guest_output(const testdef_t *test, int fd)
>   {
> -    int i, nbr = 0, pos = 0, ccnt;
> +    int nbr = 0, pos = 0, ccnt;
> +    time_t now, start = time(NULL);
>       char ch;
>   
> -    /* Poll serial output... Wait at most 360 seconds */
> -    for (i = 0; i < 36000; ++i) {
> +    /* Poll serial output... */
> +    while (1) {
>           ccnt = 0;
>           while (ccnt++ < 512 && (nbr = read(fd, &ch, 1)) == 1) {
>               if (ch == test->expect[pos]) {
> @@ -148,6 +149,15 @@ static bool check_guest_output(const testdef_t *test, int 
> fd)
>               }
>           }
>           g_assert(nbr >= 0);
> +        /* Wait only if the child is still alive.  */
> +        if (!qtest_probe_child(global_qtest)) {
> +            break;
> +        }
> +        /* Wait at most 360 seconds.  */
> +        now = time(NULL);
> +        if (now - start >= 360) {
> +            break;
> +        }
>           g_usleep(10000);
>       }
>   
> diff --git a/tests/libqtest.c b/tests/libqtest.c
> index 43be078518..1d75d3c936 100644
> --- a/tests/libqtest.c
> +++ b/tests/libqtest.c
> @@ -39,10 +39,11 @@ struct QTestState
>   {
>       int fd;
>       int qmp_fd;
> +    pid_t qemu_pid;  /* our child QEMU process */
> +    int wstatus;
> +    bool big_endian;
>       bool irq_level[MAX_IRQ];
>       GString *rx;
> -    pid_t qemu_pid;  /* our child QEMU process */
> -    bool big_endian;
>   };
>   
>   static GHookList abrt_hooks;
> @@ -96,36 +97,52 @@ static int socket_accept(int sock)
>       return ret;
>   }
>   
> +bool qtest_probe_child(QTestState *s)
> +{
> +    pid_t pid = s->qemu_pid;
> +
> +    if (pid != -1) {
> +        pid = waitpid(pid, &s->wstatus, WNOHANG);
> +        if (pid == 0) {
> +            return true;
> +        }
> +        s->qemu_pid = -1;
> +    }
> +    return false;
> +}
> +
>   static void kill_qemu(QTestState *s)
>   {
> -    if (s->qemu_pid != -1) {
> -        int wstatus = 0;
> -        pid_t pid;
> -
> -        kill(s->qemu_pid, SIGTERM);
> -        TFR(pid = waitpid(s->qemu_pid, &wstatus, 0));
> +    pid_t pid = s->qemu_pid;
> +    int wstatus;
>   
> +    /* Skip wait if qtest_probe_child already reaped.  */
> +    if (pid != -1) {
> +        kill(pid, SIGTERM);
> +        TFR(pid = waitpid(s->qemu_pid, &s->wstatus, 0));
>           assert(pid == s->qemu_pid);
> -        /*
> -         * We expect qemu to exit with status 0; anything else is
> -         * fishy and should be logged with as much detail as possible.
> -         */
> -        if (wstatus) {
> -            if (WIFEXITED(wstatus)) {
> -                fprintf(stderr, "%s:%d: kill_qemu() tried to terminate QEMU "
> -                        "process but encountered exit status %d\n",
> -                        __FILE__, __LINE__, WEXITSTATUS(wstatus));
> -            } else if (WIFSIGNALED(wstatus)) {
> -                int sig = WTERMSIG(wstatus);
> -                const char *signame = strsignal(sig) ?: "unknown ???";
> -                const char *dump = WCOREDUMP(wstatus) ? " (core dumped)" : "";
> +    }
>   
> -                fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
> -                        "from signal %d (%s)%s\n",
> -                        __FILE__, __LINE__, sig, signame, dump);
> -            }
> -            abort();
> +    /*
> +     * We expect qemu to exit with status 0; anything else is
> +     * fishy and should be logged with as much detail as possible.
> +     */
> +    wstatus = s->wstatus;
> +    if (wstatus) {
> +        if (WIFEXITED(wstatus)) {
> +            fprintf(stderr, "%s:%d: kill_qemu() tried to terminate QEMU "
> +                    "process but encountered exit status %d\n",
> +                    __FILE__, __LINE__, WEXITSTATUS(wstatus));
> +        } else if (WIFSIGNALED(wstatus)) {
> +            int sig = WTERMSIG(wstatus);
> +            const char *signame = strsignal(sig) ?: "unknown ???";
> +            const char *dump = WCOREDUMP(wstatus) ? " (core dumped)" : "";
> +
> +            fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
> +                    "from signal %d (%s)%s\n",
> +                    __FILE__, __LINE__, sig, signame, dump);
>           }
> +        abort();
>       }
>   }
>   
> @@ -228,6 +245,7 @@ QTestState *qtest_init_without_qmp_handshake(const char *extra_args)
>   
>       g_test_message("starting QEMU: %s", command);
>   
> +    s->wstatus = 0;
>       s->qemu_pid = fork();
>       if (s->qemu_pid == 0) {
>           setenv("QEMU_AUDIO_DRV", "none", true);