The error handling in neo1973_init is incorrect:
* If platform_device_add fails, we go on with the rest of the
initialization instead of bailing out. Things will break when the
module is removed (platform_device_unregister called on a device
that wasn't registered.)
* If i2c_add_driver fails, we return an error so the module will not
load, but we don't unregister neo1973_snd_device, so we are leaking
resources.
Add the missing error handling.
Signed-off-by: Jean Delvare <khali@xxxxxxxxxxxx>
Cc: Tim Niemeyer <reddog@xxxxxxxxxxxxxx>
Cc: Graeme Gregory <graeme@xxxxxxxxxxxx>
Cc: Mark Brown <broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
---
I couldn't even test-build that one. Tim, Graeme, can any of you
please review and test this patch? Thanks.
sound/soc/s3c24xx/neo1973_wm8753.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
--- linux-2.6.27-rc5.orig/sound/soc/s3c24xx/neo1973_wm8753.c 2008-09-01 15:13:17.000000000 +0200
+++ linux-2.6.27-rc5/sound/soc/s3c24xx/neo1973_wm8753.c 2008-09-01 15:22:42.000000000 +0200
@@ -717,12 +717,16 @@ static int __init neo1973_init(void)
neo1973_snd_devdata.dev = &neo1973_snd_device->dev;
ret = platform_device_add(neo1973_snd_device);
- if (ret)
+ if (ret) {
platform_device_put(neo1973_snd_device);
+ return ret;
+ }
ret = i2c_add_driver(&lm4857_i2c_driver);
- if (ret != 0)
+ if (ret != 0) {
printk(KERN_ERR "can't add i2c driver");
+ platform_device_unregister(neo1973_snd_device);
+ }
return ret;
}
--
Jean Delvare
_______________________________________________
Alsa-devel mailing list
Alsa-devel@xxxxxxxxxxxxxxxx
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
