Message ID | 1343202228-1507-1-git-send-email-zonque@gmail.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
On Wed, Jul 25, 2012 at 09:43:47AM +0200, Daniel Mack wrote: > @@ -263,12 +243,7 @@ exit_free_mem: > static int __devexit rotary_encoder_remove(struct platform_device *pdev) > { > struct rotary_encoder *encoder = platform_get_drvdata(pdev); > - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; > > - free_irq(encoder->irq_a, encoder); > - free_irq(encoder->irq_b, encoder); > - gpio_free(pdata->gpio_a); > - gpio_free(pdata->gpio_b); > input_unregister_device(encoder->input); Another botched devm_ conversion. *sigh* Input device gone, IRQ arrives, kernel goes oops, machine hangs hard. Please, do not use devm_ interfaces unless... Actually, just do not use nor suggest devm_interfaces until all resources are devm-ized. Mixing 2 styles of releasing resources leads to trouble. Thanks.
On 25.07.2012 18:05, Dmitry Torokhov wrote: > On Wed, Jul 25, 2012 at 09:43:47AM +0200, Daniel Mack wrote: >> @@ -263,12 +243,7 @@ exit_free_mem: >> static int __devexit rotary_encoder_remove(struct platform_device *pdev) >> { >> struct rotary_encoder *encoder = platform_get_drvdata(pdev); >> - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; >> >> - free_irq(encoder->irq_a, encoder); >> - free_irq(encoder->irq_b, encoder); >> - gpio_free(pdata->gpio_a); >> - gpio_free(pdata->gpio_b); >> input_unregister_device(encoder->input); > > Another botched devm_ conversion. *sigh* > > Input device gone, IRQ arrives, kernel goes oops, machine hangs hard. > > Please, do not use devm_ interfaces unless... Actually, just do not use > nor suggest devm_interfaces until all resources are devm-ized. Mixing 2 > styles of releasing resources leads to trouble. Ok, makes sense. Thanks for noticing. Are you happy with the first version I submitted then? Daniel -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday, July 25, 2012 9:05 AM, Dmitry Torokhov wrote: > On Wed, Jul 25, 2012 at 09:43:47AM +0200, Daniel Mack wrote: >> @@ -263,12 +243,7 @@ exit_free_mem: >> static int __devexit rotary_encoder_remove(struct platform_device *pdev) >> { >> struct rotary_encoder *encoder = platform_get_drvdata(pdev); >> - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; >> >> - free_irq(encoder->irq_a, encoder); >> - free_irq(encoder->irq_b, encoder); >> - gpio_free(pdata->gpio_a); >> - gpio_free(pdata->gpio_b); >> input_unregister_device(encoder->input); > > Another botched devm_ conversion. *sigh* > > Input device gone, IRQ arrives, kernel goes oops, machine hangs hard. > > Please, do not use devm_ interfaces unless... Actually, just do not use > nor suggest devm_interfaces until all resources are devm-ized. Mixing 2 > styles of releasing resources leads to trouble. Oops.. Sorry about that. I didn't think about the interaction with input_unregister_device. Regards, Hartley -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Jul 25, 2012 at 06:07:24PM +0200, Daniel Mack wrote: > On 25.07.2012 18:05, Dmitry Torokhov wrote: > > On Wed, Jul 25, 2012 at 09:43:47AM +0200, Daniel Mack wrote: > >> @@ -263,12 +243,7 @@ exit_free_mem: > >> static int __devexit rotary_encoder_remove(struct platform_device *pdev) > >> { > >> struct rotary_encoder *encoder = platform_get_drvdata(pdev); > >> - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; > >> > >> - free_irq(encoder->irq_a, encoder); > >> - free_irq(encoder->irq_b, encoder); > >> - gpio_free(pdata->gpio_a); > >> - gpio_free(pdata->gpio_b); > >> input_unregister_device(encoder->input); > > > > Another botched devm_ conversion. *sigh* > > > > Input device gone, IRQ arrives, kernel goes oops, machine hangs hard. > > > > Please, do not use devm_ interfaces unless... Actually, just do not use > > nor suggest devm_interfaces until all resources are devm-ized. Mixing 2 > > styles of releasing resources leads to trouble. > > Ok, makes sense. Thanks for noticing. Are you happy with the first > version I submitted then? I have a few issues with that version as well: 1. It writes over platform_data field in the device which does nto belong to the driver (this field belongs to the platform/board code); 2. I believe we should favor kernel supplied data over firmware allowing users to override DT bindings, if needed; 3. It still uses devm_* to allocate memory and as I said I do not like mixing 2 styles of managign resources in one driver. Thanks.
On 25.07.2012 18:29, Dmitry Torokhov wrote: > On Wed, Jul 25, 2012 at 06:07:24PM +0200, Daniel Mack wrote: >> On 25.07.2012 18:05, Dmitry Torokhov wrote: >>> On Wed, Jul 25, 2012 at 09:43:47AM +0200, Daniel Mack wrote: >>>> @@ -263,12 +243,7 @@ exit_free_mem: >>>> static int __devexit rotary_encoder_remove(struct platform_device *pdev) >>>> { >>>> struct rotary_encoder *encoder = platform_get_drvdata(pdev); >>>> - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; >>>> >>>> - free_irq(encoder->irq_a, encoder); >>>> - free_irq(encoder->irq_b, encoder); >>>> - gpio_free(pdata->gpio_a); >>>> - gpio_free(pdata->gpio_b); >>>> input_unregister_device(encoder->input); >>> >>> Another botched devm_ conversion. *sigh* >>> >>> Input device gone, IRQ arrives, kernel goes oops, machine hangs hard. >>> >>> Please, do not use devm_ interfaces unless... Actually, just do not use >>> nor suggest devm_interfaces until all resources are devm-ized. Mixing 2 >>> styles of releasing resources leads to trouble. >> >> Ok, makes sense. Thanks for noticing. Are you happy with the first >> version I submitted then? > > I have a few issues with that version as well: > > 1. It writes over platform_data field in the device which does nto > belong to the driver (this field belongs to the platform/board code); > > 2. I believe we should favor kernel supplied data over firmware > allowing users to override DT bindings, if needed; Ok, this is different from what I saw in other drivers, but I'm fine with that. > 3. It still uses devm_* to allocate memory and as I said I do not like > mixing 2 styles of managign resources in one driver. Ok, sorry again. I didn't closely follow that new development, hence I wasn't aware of these pitfalls. New patch series coming up. -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/input/misc/rotary_encoder.c b/drivers/input/misc/rotary_encoder.c index f07f784..d4e0abb 100644 --- a/drivers/input/misc/rotary_encoder.c +++ b/drivers/input/misc/rotary_encoder.c @@ -145,6 +145,7 @@ static int __devinit rotary_encoder_probe(struct platform_device *pdev) struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; struct rotary_encoder *encoder; struct input_dev *input; + struct device *dev = &pdev->dev; irq_handler_t handler; int err; @@ -163,13 +164,11 @@ static int __devinit rotary_encoder_probe(struct platform_device *pdev) encoder->input = input; encoder->pdata = pdata; - encoder->irq_a = gpio_to_irq(pdata->gpio_a); - encoder->irq_b = gpio_to_irq(pdata->gpio_b); /* create and register the input driver */ input->name = pdev->name; input->id.bustype = BUS_HOST; - input->dev.parent = &pdev->dev; + input->dev.parent = dev; if (pdata->relative_axis) { input->evbit[0] = BIT_MASK(EV_REL); @@ -182,38 +181,27 @@ static int __devinit rotary_encoder_probe(struct platform_device *pdev) err = input_register_device(input); if (err) { - dev_err(&pdev->dev, "failed to register input device\n"); + dev_err(dev, "failed to register input device\n"); goto exit_free_mem; } /* request the GPIOs */ - err = gpio_request(pdata->gpio_a, DRV_NAME); + err = devm_gpio_request_one(dev, pdata->gpio_a, + GPIOF_IN, dev_name(dev)); if (err) { - dev_err(&pdev->dev, "unable to request GPIO %d\n", - pdata->gpio_a); + dev_err(dev, "unable to request GPIO %d\n", pdata->gpio_a); goto exit_unregister_input; } - err = gpio_direction_input(pdata->gpio_a); + err = devm_gpio_request_one(dev, pdata->gpio_b, + GPIOF_IN, dev_name(dev)); if (err) { - dev_err(&pdev->dev, "unable to set GPIO %d for input\n", - pdata->gpio_a); + dev_err(dev, "unable to request GPIO %d\n", pdata->gpio_b); goto exit_unregister_input; } - err = gpio_request(pdata->gpio_b, DRV_NAME); - if (err) { - dev_err(&pdev->dev, "unable to request GPIO %d\n", - pdata->gpio_b); - goto exit_free_gpio_a; - } - - err = gpio_direction_input(pdata->gpio_b); - if (err) { - dev_err(&pdev->dev, "unable to set GPIO %d for input\n", - pdata->gpio_b); - goto exit_free_gpio_a; - } + encoder->irq_a = gpio_to_irq(pdata->gpio_a); + encoder->irq_b = gpio_to_irq(pdata->gpio_b); /* request the IRQs */ if (pdata->half_period) { @@ -223,34 +211,26 @@ static int __devinit rotary_encoder_probe(struct platform_device *pdev) handler = &rotary_encoder_irq; } - err = request_irq(encoder->irq_a, handler, - IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING, - DRV_NAME, encoder); + err = devm_request_irq(dev, encoder->irq_a, handler, + IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING, + dev_name(dev), encoder); if (err) { - dev_err(&pdev->dev, "unable to request IRQ %d\n", - encoder->irq_a); - goto exit_free_gpio_b; + dev_err(dev, "unable to request IRQ %d\n", encoder->irq_a); + goto exit_unregister_input; } - err = request_irq(encoder->irq_b, handler, - IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING, - DRV_NAME, encoder); + err = devm_request_irq(dev, encoder->irq_b, handler, + IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING, + dev_name(dev), encoder); if (err) { - dev_err(&pdev->dev, "unable to request IRQ %d\n", - encoder->irq_b); - goto exit_free_irq_a; + dev_err(dev, "unable to request IRQ %d\n", encoder->irq_b); + goto exit_unregister_input; } platform_set_drvdata(pdev, encoder); return 0; -exit_free_irq_a: - free_irq(encoder->irq_a, encoder); -exit_free_gpio_b: - gpio_free(pdata->gpio_b); -exit_free_gpio_a: - gpio_free(pdata->gpio_a); exit_unregister_input: input_unregister_device(input); input = NULL; /* so we don't try to free it */ @@ -263,12 +243,7 @@ exit_free_mem: static int __devexit rotary_encoder_remove(struct platform_device *pdev) { struct rotary_encoder *encoder = platform_get_drvdata(pdev); - struct rotary_encoder_platform_data *pdata = pdev->dev.platform_data; - free_irq(encoder->irq_a, encoder); - free_irq(encoder->irq_b, encoder); - gpio_free(pdata->gpio_a); - gpio_free(pdata->gpio_b); input_unregister_device(encoder->input); platform_set_drvdata(pdev, NULL); kfree(encoder);
This simplifies the error unwind pathes. Also, don't call gpio_to_irq() on GPIOs before dev_gpio_request_one() succeeded on them. This avoids Ooopses with incorrect DT bindings. Signed-off-by: Daniel Mack <zonque@gmail.com> --- drivers/input/misc/rotary_encoder.c | 67 +++++++++++------------------------ 1 file changed, 21 insertions(+), 46 deletions(-)