On Jul 13 2016 19:18, Takashi Sakamoto wrote:
Hi,
On Jul 13 2016 18:59, Dan Carpenter wrote:
My static checker complains that "resp" could be unitialized on error when we print its value.
Signed-off-by: Dan Carpenter dan.carpenter@oracle.com
diff --git a/sound/pci/mixart/mixart_mixer.c b/sound/pci/mixart/mixart_mixer.c index 58fd79eb..51e5349 100644 --- a/sound/pci/mixart/mixart_mixer.c +++ b/sound/pci/mixart/mixart_mixer.c @@ -965,7 +965,7 @@ static int mixart_update_monitoring(struct snd_mixart* chip, int channel) int err; struct mixart_msg request; struct mixart_set_out_audio_level audio_level;
- u32 resp;
u32 resp = 0;
if(chip->pipe_out_ana.status == PIPE_UNDEFINED) return -EINVAL; /* no pipe defined */
The 'resp' variable is firstly given to snd_mixart_send_msg().
static int mixart_update_monitoring(struct snd_mixart* chip, int channel) { ... err = snd_mixart_send_msg(chip->mgr, &request, sizeof(resp), &resp); if((err<0) || resp) { dev_dbg(chip->card->dev, "error MSG_CONNECTOR_SET_OUT_AUDIO_LEVEL card(%d) resp(%x)\n", chip->chip_idx, resp); return -EINVAL; } return 0; }
When the function, snd_mixart_send_msg(), assigns nothing to the variable in error cases, dev_dbg() prints uninitialized value.
Hm. I realized that this interpretation is not enough. When value of the error is not negative, snd_mixart_send_msg() can assign something to the resp variable, then dev_dbg() is called. In short, when the function is successful, there's a case that the message is processed. This is a bit strange...
I think a condition statement for the resp variable is needless here, as long as the variable is initialized ahead...
Reviewed-by: Takashi Sakamoto o-takashi@sakamocchi.jp
Anyway, it's better to initialize it.
Regards
Takashi Sakamoto