Re: [PATCH v2] tests/9pfs: Use g_autofree and g_autoptr where possible

[Christian Schoenebeck <qemu_oss@crudebyte.com>]

On Montag, 31. Januar 2022 17:39:30 CET Greg Kurz wrote:
> It is recommended to use g_autofree or g_autoptr as it reduces
> the odds of introducing memory leaks in future changes. Let's do
> that everywhere a glib allocation is performed.
> 
> The virtio_9p_create_local_test_dir() function needs some extra
> care though : the template pointer is eventually cached into the
> local_test_path global variable for the duration of the tests and
> should not be freed. Add the g_autofree annotation but negate it
> with g_steal_pointer() to make it clear that the pointer ownership
> is dropped on purpose.
> 
> Based-on:
> <f6602123c6f7d0d593466231b04fba087817abbd.1642879848.git.qemu_oss@crudebyte
> .com> Signed-off-by: Greg Kurz <groug@kaod.org>
> ---
> v2: - fix crash with local_test_path
> ---
>  tests/qtest/libqos/virtio-9p.c | 30 +++++++++++++++---------------
>  1 file changed, 15 insertions(+), 15 deletions(-)
> 
> diff --git a/tests/qtest/libqos/virtio-9p.c b/tests/qtest/libqos/virtio-9p.c
> index ef96ef006adc..f0ffbc583492 100644
> --- a/tests/qtest/libqos/virtio-9p.c
> +++ b/tests/qtest/libqos/virtio-9p.c
> @@ -37,18 +37,22 @@ static char *concat_path(const char* a, const char* b)
>      return g_build_filename(a, b, NULL);
>  }
> 
> -void virtio_9p_create_local_test_dir(void)
> +static char *make_temp_dir(char *template)
>  {
> -    struct stat st;
> -    char *pwd = g_get_current_dir();
> -    char *template = concat_path(pwd, "qtest-9p-local-XXXXXX");
> -
> -    local_test_path = mkdtemp(template);
> -    if (!local_test_path) {
> +    char *path = mkdtemp(template);
> +    if (!path) {
>          g_test_message("mkdtemp('%s') failed: %s", template,
> strerror(errno)); }
> -    g_free(pwd);
> +    return path;
> +}
> +
> +void virtio_9p_create_local_test_dir(void)
> +{
> +    g_autofree char *pwd = g_get_current_dir();
> +    g_autofree char *template = concat_path(pwd, "qtest-9p-local-XXXXXX");
> +    struct stat st;
> 
> +    local_test_path = make_temp_dir(g_steal_pointer(&template));

Quite noisy diff and under the bottom line, it does not fix what it originally 
supposed to: addressing the leak of the global variable 'local_test_path'. 
g_steal_pointer() simply makes the previous g_autofree prefix pointless.

Why not just moving g_autofree from 'template' to the global variable 
'local_test_path' and avoid all that noise? 

>      g_assert(local_test_path != NULL);
> 
>      /* ensure test directory exists now ... */
> @@ -60,12 +64,11 @@ void virtio_9p_create_local_test_dir(void)
>  void virtio_9p_remove_local_test_dir(void)
>  {
>      g_assert(local_test_path != NULL);
> -    char *cmd = g_strdup_printf("rm -fr '%s'\n", local_test_path);
> +    g_autofree char *cmd = g_strdup_printf("rm -fr '%s'\n",
> local_test_path); int res = system(cmd);
>      if (res < 0) {
>          /* ignore error, dummy check to prevent compiler error */
>      }
> -    g_free(cmd);
>  }
> 
>  char *virtio_9p_test_path(const char *path)
> @@ -209,8 +212,8 @@ static void *virtio_9p_pci_create(void *pci_bus,
> QGuestAllocator *t_alloc, static void regex_replace(GString *haystack,
> const char *pattern, const char *replace_fmt, ...)
>  {
> -    GRegex *regex;
> -    char *replace, *s;
> +    g_autoptr(GRegex) regex = NULL;
> +    g_autofree char *replace = NULL, *s = NULL;
>      va_list argp;
> 
>      va_start(argp, replace_fmt);
> @@ -220,9 +223,6 @@ static void regex_replace(GString *haystack, const char
> *pattern, regex = g_regex_new(pattern, 0, 0, NULL);
>      s = g_regex_replace(regex, haystack->str, -1, 0, replace, 0, NULL);
>      g_string_assign(haystack, s);
> -    g_free(s);
> -    g_regex_unref(regex);
> -    g_free(replace);
>  }
> 
>  void virtio_9p_assign_local_driver(GString *cmd_line, const char *args)