Hi Takashi.
Just see the snd_assert() removal. Much better as far as I'm concerned but just as a remark -- there's many combined ones of the form:
diff --git a/sound/isa/cs423x/cs4236_lib.c b/sound/isa/cs423x/cs4236_lib.c index 33e9cf1..6a85fdc 100644 --- a/sound/isa/cs423x/cs4236_lib.c +++ b/sound/isa/cs423x/cs4236_lib.c @@ -967,7 +967,8 @@ int snd_cs4236_mixer(struct snd_wss *chip) int err; struct snd_kcontrol_new *kcontrol;
- snd_assert(chip != NULL && chip->card != NULL, return -EINVAL); + if (snd_BUG_ON(!chip || !chip->card)) + return -EINVAL; card = chip->card; strcpy(card->mixername, snd_wss_chip_id(chip));
in there. Just thought I'd point that Andrew Morton has been pushing back on multiple conditions per regular BUG_ON()/WARN_ON() lately as they end up giving less useful information then split ones ("so what triggered?").
Just a remark as said but if you agree with him, maybe now is the best time to split them directly as well.
Rene.