On Fri, 2011-03-18 at 15:05 +0100, ext Clemens Ladisch wrote:
Sudarshan Bisht wrote:
On Thu, 2011-03-17 at 13:08 +0100, ext Clemens Ladisch wrote:
sudarshan.bisht@nokia.com wrote:
--- a/src/conf.c +++ b/src/conf.c @@ -3321,9 +3321,11 @@ static int snd_config_hooks_call(snd_config_t *root, snd_config_t *config, snd_c snd_config_delete(func_conf); if (err >= 0) { snd_config_t *nroot;
err = func(root, config, &nroot, private_data);if (err < 0)SNDERR("function %s returned error: %s", func_name, snd_strerror(err));
if (func) {err = func(root, config, &nroot, private_data);if (err < 0)SNDERR("function %s returned error: %s", func_name, snd_strerror(err));}The preceding "!func" and "err >= 0" checks already guarantee that func is valid.
But in case of "goto _err" , "!func" and "err >= 0" are not going to be checked.
"err >= 0" is checked, and in all error cases, err must be negative; if not, this is bug at that place and should be fixed before that goto. I now see that the TYPE_COMPOUND check forgets to set "err = -EINVAL;".
Yes, now I got your point, and I will take care of this in all cases.
Regards, Clemens