At Fri, 25 Apr 2008 08:29:08 +0200 (CEST),
Jaroslav Kysela wrote:
>
> changeset: 6120:e2ff47e8771b
> tag: tip
> user: perex
> date: Fri Apr 25 08:29:05 2008 +0200
> files: include/ac97_codec.h kernel/drivers/input/touchscreen/ucb1400_ts.c
> description:
> add a private field for ac97-device drivers and let ucb1400 be its first user
> From: Sebastian Siewior <bigeasy@xxxxxxxxxxxxx>
>
> Currently the UCB1400 driver discovers the interrupt via probing. This works
> probably only on x86. This patch adds a private field to the ac97 struct
> where the ac97 driver can deposit informations for the device driver that
> serves a device which is attached to the ac97 bus.
> This patch also converts the UCB1400 driver to use this information if
> available.
>
> Signed-off-by: Sebastian Siewior <bigeasy@xxxxxxxxxxxxx>
Jaroslav, as I posted yesterday, I don't think it's good to use a void
pointer blindly there. In this case, it's different from a normal
private_data, which is asigned and used by the same driver.
This is a data passed by ac97 controller driver to another device
driver. Thus, it should be a strong type.
Also, a cast from a pointer to an int causes a compile warning.
So, could you revert it until all issues are solved?
thanks,
Takashi
>
> diff -r 9c74d781f388 -r e2ff47e8771b include/ac97_codec.h
> --- a/include/ac97_codec.h Fri Apr 25 08:15:12 2008 +0200
> +++ b/include/ac97_codec.h Fri Apr 25 08:29:05 2008 +0200
> @@ -475,6 +475,10 @@ struct snd_ac97 {
> struct snd_ac97_build_ops * build_ops;
> void *private_data;
> void (*private_free) (struct snd_ac97 *ac97);
> + /* This field is used by device drivers which serve devices which are
> + * attached to the AC97 bus.
> + */
> + void *device_private_data;
> /* --- */
> struct snd_ac97_bus *bus;
> struct pci_dev *pci; /* assigned PCI device - used for quirks */
> diff -r 9c74d781f388 -r e2ff47e8771b kernel/drivers/input/touchscreen/ucb1400_ts.c
> --- a/kernel/drivers/input/touchscreen/ucb1400_ts.c Fri Apr 25 08:15:12 2008 +0200
> +++ b/kernel/drivers/input/touchscreen/ucb1400_ts.c Fri Apr 25 08:29:05 2008 +0200
> @@ -492,10 +492,14 @@ static int ucb1400_ts_probe(struct devic
> goto err_free_devs;
> }
>
> - error = ucb1400_detect_irq(ucb);
> - if (error) {
> - printk(KERN_ERR "UCB1400: IRQ probe failed\n");
> - goto err_free_devs;
> + if (!ucb->ac97->device_private_data) {
> + error = ucb1400_detect_irq(ucb);
> + if (error) {
> + printk(KERN_ERR "UCB1400: IRQ probe failed\n");
> + goto err_free_devs;
> + }
> + } else {
> + ucb->irq = (int) ucb->ac97->device_private_data;
> }
>
> error = request_irq(ucb->irq, ucb1400_hard_irq, IRQF_TRIGGER_RISING,
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Don't miss this year's exciting event. There's still time to save $100.
> Use priority code J8TL2D2.
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Alsa-cvslog mailing list
> Alsa-cvslog@xxxxxxxxxxxxxxxxxxxxx
> https://lists.sourceforge.net/lists/listinfo/alsa-cvslog
>
_______________________________________________
Alsa-devel mailing list
Alsa-devel@xxxxxxxxxxxxxxxx
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
