On Tue, Oct 07, 2014 at 06:14:36PM +0100, Mark Brown wrote:
On Mon, Oct 06, 2014 at 02:37:53PM +0200, Takashi Iwai wrote:
Mark Brown wrote:
However, this resulted in a few new build warnings, and some of them look correctly reported.
- sound/soc/codecs/mc13783.c:787:13: warning: 'np' may be used uninitialized in this function [-Wuninitialized]
The call of_node_put(np) is done unconditionally even for non-NULL pdata where np isn't initialized. This may lead to a real crash.
- sound/soc/fsl/eukrea-tlv320.c:221:14: warning: 'ssi_np' may be used uninitialized in this function [-Wuninitialized]
A similar bug: of_node_put(ssi_np) is reached even before ssi_np is initialized.
- sound/soc/fsl/imx-es8328.c:196:13: warning: 'codec_np' may be used uninitialized in this function [-Wuninitialized]
- sound/soc/fsl/imx-es8328.c:195:13: warning: 'ssi_np' may be used uninitialized in this function [-Wuninitialized]
Ditto.
Relevant people put in Cc. I hope I'll get a new pull request soon later :)
You missed CCing the Freescale guys and Timur so you're missing most of the relevant people here; adding them now.
Generally we fetch ssi_np and codec_np before other operations. But somehow these two are doing the fetching a bit later.
However, I think the below change can be a quick reasonable fix: - struct device_node *ssi_np, *codec_np; + struct device_node *codec_np = NULL; + struct device_node *ssi_np = NULL;
I'm patching and taking build test. Will send them soon.
Thanks Nicolin