[alsa-devel] [PATCH] ASoC: imx-sgtl5000: do not use devres on a foreign device

Shawn Guo shawn.guo at linaro.org
Wed Sep 25 14:14:30 CEST 2013


On Wed, Sep 25, 2013 at 12:16:08PM +0200, Philipp Zabel wrote:
> Calling devm_clk_get with any device pointer other than our own confuses
> devres. Use clk_get instead. This avoids hitting the following warning in
> the imx-sgtl5000 error path:
> 
> imx-sgtl5000 sound.12: snd_soc_register_card failed (-517)
> platform sound.12: Driver imx-sgtl5000 requests probe deferral
> ------------[ cut here ]------------
> WARNING: CPU: 0 PID: 75 at drivers/base/dd.c:272 driver_probe_device+0x194/0x218()
> Modules linked in: snd_soc_sgtl5000(+) snd_soc_imx_sgtl5000 coda snd_soc_imx_audmux imx_sdma snd_soc_fsl_spdif snd_soc_fsl_ssi
> CPU: 0 PID: 75 Comm: udevd Not tainted 3.11.0-rc6+ #4682
> Backtrace:
> [<80010bc4>] (dump_backtrace+0x0/0x10c) from [<80010d60>] (show_stack+0x18/0x1c)
>  r6:00000110 r5:00000009 r4:00000000 r3:00000000
> [<80010d48>] (show_stack+0x0/0x1c) from [<804f0764>] (dump_stack+0x20/0x28)
> [<804f0744>] (dump_stack+0x0/0x28) from [<8001a4a4>] (warn_slowpath_common+0x6c/0x8c)
> [<8001a438>] (warn_slowpath_common+0x0/0x8c) from [<8001a4e8>] (warn_slowpath_null+0x24/0x2c)
>  r8:7f032000 r7:7f02f93c r6:cf8eaa54 r5:cf8eaa20 r4:80728a0c
> [<8001a4c4>] (warn_slowpath_null+0x0/0x2c) from [<80286bdc>] (driver_probe_device+0x194/0x218)
> [<80286a48>] (driver_probe_device+0x0/0x218) from [<80286cf4>] (__driver_attach+0x94/0x98)
>  r7:00000000 r6:cf8eaa54 r5:7f02f93c r4:cf8eaa20
> [<80286c60>] (__driver_attach+0x0/0x98) from [<802851c8>] (bus_for_each_dev+0x5c/0x90)
>  r6:80286c60 r5:7f02f93c r4:00000000 r3:cf8ef03c
> [<8028516c>] (bus_for_each_dev+0x0/0x90) from [<80286654>] (driver_attach+0x24/0x28)
>  r6:806d0424 r5:cf16a580 r4:7f02f93c
> [<80286630>] (driver_attach+0x0/0x28) from [<802861e4>] (bus_add_driver+0xdc/0x234)
> [<80286108>] (bus_add_driver+0x0/0x234) from [<802871d4>] (driver_register+0x80/0x154)
>  r8:7f032000 r7:00000001 r6:7f02fa68 r5:7f02fa74 r4:7f02f93c
> [<80287154>] (driver_register+0x0/0x154) from [<8033c278>] (i2c_register_driver+0x34/0xbc)
> [<8033c244>] (i2c_register_driver+0x0/0xbc) from [<7f032018>] (sgtl5000_i2c_driver_init+0x18/0x24 [snd_soc_sgtl5000])
>  r5:7f02fa74 r4:cfb7ff48
> [<7f032000>] (sgtl5000_i2c_driver_init+0x0/0x24 [snd_soc_sgtl5000]) from [<80008738>] (do_one_initcall+0xf4/0x150)
> [<80008644>] (do_one_initcall+0x0/0x150) from [<80053f64>] (load_module+0x174c/0x1db4)
> [<80052818>] (load_module+0x0/0x1db4) from [<800546ac>] (SyS_init_module+0xe0/0xf4)
> [<800545cc>] (SyS_init_module+0x0/0xf4) from [<8000e540>] (ret_fast_syscall+0x0/0x30)
>  r6:00005b22 r5:00afed68 r4:00000000
> ---[ end trace b24c5c3bb145dbdd ]---

This is a known issue [1].  Thanks for fixing it.

> 
> Signed-off-by: Philipp Zabel <p.zabel at pengutronix.de>

Acked-by: Shawn Guo <shawn.guo at linaro.org>

Shawn

[1] http://thread.gmane.org/gmane.linux.alsa.devel/109093/focus=109101

> ---
>  sound/soc/fsl/imx-sgtl5000.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/sound/soc/fsl/imx-sgtl5000.c b/sound/soc/fsl/imx-sgtl5000.c
> index 46c5b4f..52df7d5 100644
> --- a/sound/soc/fsl/imx-sgtl5000.c
> +++ b/sound/soc/fsl/imx-sgtl5000.c
> @@ -128,7 +128,7 @@ static int imx_sgtl5000_probe(struct platform_device *pdev)
>  		goto fail;
>  	}
>  
> -	data->codec_clk = devm_clk_get(&codec_dev->dev, NULL);
> +	data->codec_clk = clk_get(&codec_dev->dev, NULL);
>  	if (IS_ERR(data->codec_clk)) {
>  		ret = PTR_ERR(data->codec_clk);
>  		goto fail;
> @@ -172,6 +172,8 @@ static int imx_sgtl5000_probe(struct platform_device *pdev)
>  	return 0;
>  
>  fail:
> +	if (!IS_ERR(data->codec_clk))
> +		clk_put(data->codec_clk);
>  	if (ssi_np)
>  		of_node_put(ssi_np);
>  	if (codec_np)
> @@ -185,6 +187,7 @@ static int imx_sgtl5000_remove(struct platform_device *pdev)
>  	struct imx_sgtl5000_data *data = platform_get_drvdata(pdev);
>  
>  	snd_soc_unregister_card(&data->card);
> +	clk_put(data->codec_clk);
>  
>  	return 0;
>  }
> -- 
> 1.8.4.rc3
> 



More information about the Alsa-devel mailing list