At Sun, 1 Nov 2009 16:50:38 +0100, Krzysztof Helt wrote:
From: Krzysztof Helt krzysztof.h1@wp.pl
The cs4236 was two step detection with call to the snd_wss_free() between two steps. The snd_wss_free() did not free a sound device created in the snd_wss_create(). This caused an OOPS during module removal as the same sound device was released twice. The same OOPS happened if the cs4236 module loading failed.
Fix this by adapting the snd_cs4236_create() to correctly work with chips less capable then cs4236. The snd_cs4236_create() behaves the same as the snd_wss_create() if the chip is less capable than the cs4236.
Signed-off-by: Krzysztof Helt krzysztof.h1@wp.pl
This is the second version of the patch with added snd_device_free() call in error path.
Thanks. The change looks good, but...
@@ -281,83 +285,89 @@ int snd_cs4236_create(struct snd_card *card,
...
err = snd_wss_create(card, port, cport, irq, dma1, dma2, hardware, hwshare, &chip); if (err < 0) return err;
- if (!(chip->hardware & WSS_HW_CS4236B_MASK)) {
snd_printk(KERN_ERR "CS4236+: MODE3 and extended registers ""not available, hardware=0x%x\n", chip->hardware);snd_device_free(card, chip);return -ENODEV;
I'd just return 0 here instead of error (with *rchip = chip, of course). Then the code flow gets simpler and the patch becomes smaller.
Could you fix and repost?
thanks,
Takashi