| Message ID | 1351439163-9575-1-git-send-email-anarsoul@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Vasily Khoruzhick <anarsoul@gmail.com> writes:
> Bootloader can leave interrupt bit pending, and it confuses driver.
OK, looks good to me.
One question though, just to make sure everything is covered :
Suppose that the resume is the consequence of keypad press. If anybody relied
previously on "reading the keypad" to know which key was pressed to wakeup the
SoC, is it still working after your patch ?
And I suppose you can't "convince" your bootloader to behave correctly, can you ?
Cheers.
--
Robert
On Sun, Oct 28, 2012 at 8:55 PM, Robert Jarzmik <robert.jarzmik@free.fr> wrote: > Vasily Khoruzhick <anarsoul@gmail.com> writes: > >> Bootloader can leave interrupt bit pending, and it confuses driver. > OK, looks good to me. > One question though, just to make sure everything is covered : > Suppose that the resume is the consequence of keypad press. If anybody relied > previously on "reading the keypad" to know which key was pressed to wakeup the > SoC, is it still working after your patch ? Hi Robert, Code from git master will not call pxa27x_keypad_config on resume if device can cause wakeup, so everything is OK (however it's another issue for pxa27x, because device is not resuming correctly) > And I suppose you can't "convince" your bootloader to behave correctly, can you ? It behaves correctly, it just can't handle last keypress (because it's booting Linux). Even if it would be possible to fix this issue in bootloader it's always nice to keep driver failure-proof, so it does not fail even if bootloader left something in non-consistent state. Regards Vasily
Vasily Khoruzhick <anarsoul@gmail.com> writes: > It behaves correctly, it just can't handle last keypress (because it's > booting Linux). > Even if it would be possible to fix this issue in bootloader it's > always nice to keep driver > failure-proof, so it does not fail even if bootloader left something > in non-consistent state. True. Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
On Mon, Oct 29, 2012 at 2:34 AM, Robert Jarzmik <robert.jarzmik@free.fr> wrote: > Vasily Khoruzhick <anarsoul@gmail.com> writes: > >> It behaves correctly, it just can't handle last keypress (because it's >> booting Linux). >> Even if it would be possible to fix this issue in bootloader it's >> always nice to keep driver >> failure-proof, so it does not fail even if bootloader left something >> in non-consistent state. > True. > Acked-by: Robert Jarzmik <robert.jarzmik@free.fr> > > -- > Robert Applied Thanks Haojian
diff --git a/drivers/input/keyboard/pxa27x_keypad.c b/drivers/input/keyboard/pxa27x_keypad.c index 803ff6f..cad9d5d 100644 --- a/drivers/input/keyboard/pxa27x_keypad.c +++ b/drivers/input/keyboard/pxa27x_keypad.c @@ -368,6 +368,9 @@ static void pxa27x_keypad_config(struct pxa27x_keypad *keypad) unsigned int mask = 0, direct_key_num = 0; unsigned long kpc = 0; + /* clear pending interrupt bit */ + keypad_readl(KPC); + /* enable matrix keys with automatic scan */ if (pdata->matrix_key_rows && pdata->matrix_key_cols) { kpc |= KPC_ASACT | KPC_MIE | KPC_ME | KPC_MS_ALL;
Bootloader can leave interrupt bit pending, and it confuses driver. Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com> --- Rebased against v3.7-rc2 drivers/input/keyboard/pxa27x_keypad.c | 3 +++ 1 file changed, 3 insertions(+)