| Message ID | 20201207175940.10631-1-andriy.shevchenko@linux.intel.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v1] pinctrl: intel: Actually disable Tx and Rx buffers on GPIO request | expand |
On Mon, Dec 7, 2020 at 8:02 PM Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote: > > Mistakenly the buffers (input and output) become together enabled for a little > period of time during GPIO request. This is problematic, because instead of > initial motive to disable them in the commit af7e3eeb84e2 > ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), > the driven value on the pin, which might be used as an IRQ line, brings > firmware of some touch pads in an awkward state that needs a full power off > to recover. Fix this by, as stated in a culprit commit, disabling the buffers. Linus, since we are at rc7 I'm not going to send a PR for this single fix (I truly do not expect anything to come). Please, proceed as usual (either to v5.10 or to v5.11 cycle, there is not much difference b/c culptin is sitting in v5.9).
> On Dec 8, 2020, at 01:59, Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote: > > Mistakenly the buffers (input and output) become together enabled for a little > period of time during GPIO request. This is problematic, because instead of > initial motive to disable them in the commit af7e3eeb84e2 > ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), > the driven value on the pin, which might be used as an IRQ line, brings > firmware of some touch pads in an awkward state that needs a full power off > to recover. Fix this by, as stated in a culprit commit, disabling the buffers. > > Fixes: af7e3eeb84e2 ("pinctrl: intel: Disable input and output buffer when switching to GPIO") > BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=210497 > Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com> > --- > drivers/pinctrl/intel/pinctrl-intel.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/pinctrl/intel/pinctrl-intel.c b/drivers/pinctrl/intel/pinctrl-intel.c > index e77145e3b31b..5cd720f5b0e1 100644 > --- a/drivers/pinctrl/intel/pinctrl-intel.c > +++ b/drivers/pinctrl/intel/pinctrl-intel.c > @@ -452,8 +452,8 @@ static void intel_gpio_set_gpio_mode(void __iomem *padcfg0) > value |= PADCFG0_PMODE_GPIO; > > /* Disable input and output buffers */ > - value &= ~PADCFG0_GPIORXDIS; > - value &= ~PADCFG0_GPIOTXDIS; > + value |= PADCFG0_GPIORXDIS; > + value |= PADCFG0_GPIOTXDIS; > > /* Disable SCI/SMI/NMI generation */ > value &= ~(PADCFG0_GPIROUTIOXAPIC | PADCFG0_GPIROUTSCI); > -- > 2.29.2 > >
On Mon, Dec 7, 2020 at 7:07 PM Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > On Mon, Dec 7, 2020 at 8:02 PM Andy Shevchenko > <andriy.shevchenko@linux.intel.com> wrote: > > > > Mistakenly the buffers (input and output) become together enabled for a little > > period of time during GPIO request. This is problematic, because instead of > > initial motive to disable them in the commit af7e3eeb84e2 > > ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), > > the driven value on the pin, which might be used as an IRQ line, brings > > firmware of some touch pads in an awkward state that needs a full power off > > to recover. Fix this by, as stated in a culprit commit, disabling the buffers. > > Linus, since we are at rc7 I'm not going to send a PR for this single > fix (I truly do not expect anything to come). Please, proceed as usual > (either to v5.10 or to v5.11 cycle, there is not much difference b/c > culptin is sitting in v5.9). OK no problem, I applied the patch directly for fixes so we get it in ASAP. Yours, Linus Walleij
On Tue, Dec 08, 2020 at 09:34:43AM +0100, Linus Walleij wrote: > On Mon, Dec 7, 2020 at 7:07 PM Andy Shevchenko > <andy.shevchenko@gmail.com> wrote: > > On Mon, Dec 7, 2020 at 8:02 PM Andy Shevchenko > > <andriy.shevchenko@linux.intel.com> wrote: > > > > > > Mistakenly the buffers (input and output) become together enabled for a little > > > period of time during GPIO request. This is problematic, because instead of > > > initial motive to disable them in the commit af7e3eeb84e2 > > > ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), > > > the driven value on the pin, which might be used as an IRQ line, brings > > > firmware of some touch pads in an awkward state that needs a full power off > > > to recover. Fix this by, as stated in a culprit commit, disabling the buffers. > > > > Linus, since we are at rc7 I'm not going to send a PR for this single > > fix (I truly do not expect anything to come). Please, proceed as usual > > (either to v5.10 or to v5.11 cycle, there is not much difference b/c > > culptin is sitting in v5.9). > > OK no problem, I applied the patch directly for fixes so we get it in ASAP. Linus, thanks, however something happened to it. i.e. BugLink tag disappeared. It's not good, because due to that link and Kai's report I quickly understood what was the issue. Besides that commit message has some grammar / style issues. Can you please fix above and use below (fixed) text as commit message? ---8<---8<--- Mistakenly the buffers (input and output) become together enabled for a short period of time during GPIO request. This is problematic, because instead of initial motive to disable them in the commit af7e3eeb84e2 ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), the driven value on the pin, which might be used as an IRQ line, brings firmwares of some touch pads to an awkward state that needs a full power off to recover. Fix this by, as stated in the culprit commit, disabling the buffers. P.S. I'm looking at 5f4592bc20c8 in your pin control tree, fixes branch. -- With Best Regards, Andy Shevchenko
On Tue, Dec 08, 2020 at 11:48:53AM +0200, Andy Shevchenko wrote: > On Tue, Dec 08, 2020 at 09:34:43AM +0100, Linus Walleij wrote: > > On Mon, Dec 7, 2020 at 7:07 PM Andy Shevchenko > > <andy.shevchenko@gmail.com> wrote: > > > On Mon, Dec 7, 2020 at 8:02 PM Andy Shevchenko > > > <andriy.shevchenko@linux.intel.com> wrote: ... > > > Linus, since we are at rc7 I'm not going to send a PR for this single > > > fix (I truly do not expect anything to come). Please, proceed as usual > > > (either to v5.10 or to v5.11 cycle, there is not much difference b/c > > > culptin is sitting in v5.9). > > > > OK no problem, I applied the patch directly for fixes so we get it in ASAP. > > Linus, thanks, however something happened to it. i.e. BugLink tag disappeared. > It's not good, because due to that link and Kai's report I quickly understood > what was the issue. > > Besides that commit message has some grammar / style issues. > Can you please fix above and use below (fixed) text as commit message? > > ---8<---8<--- > > Mistakenly the buffers (input and output) become together enabled for a short > period of time during GPIO request. This is problematic, because instead of > initial motive to disable them in the commit af7e3eeb84e2 > ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), > the driven value on the pin, which might be used as an IRQ line, brings > firmwares of some touch pads to an awkward state that needs a full power off > to recover. Fix this by, as stated in the culprit commit, disabling the buffers. > > P.S. I'm looking at 5f4592bc20c8 in your pin control tree, fixes branch. I can send a v2 for your convenience. -- With Best Regards, Andy Shevchenko
On Tue, Dec 8, 2020 at 10:47 AM Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > Linus, thanks, however something happened to it. i.e. BugLink tag disappeared. > It's not good, because due to that link and Kai's report I quickly understood > what was the issue. Oh I think that is an issue with the b4 tool that I use to extract patches. It was reported some time back and might not be fixed in the version I'm running. Yours, Linus Walleij
On Tue, Dec 08, 2020 at 11:50:13AM +0200, Andy Shevchenko wrote: > On Tue, Dec 08, 2020 at 11:48:53AM +0200, Andy Shevchenko wrote: ... > I can send a v2 for your convenience. v2 has been sent. -- With Best Regards, Andy Shevchenko
diff --git a/drivers/pinctrl/intel/pinctrl-intel.c b/drivers/pinctrl/intel/pinctrl-intel.c index e77145e3b31b..5cd720f5b0e1 100644 --- a/drivers/pinctrl/intel/pinctrl-intel.c +++ b/drivers/pinctrl/intel/pinctrl-intel.c @@ -452,8 +452,8 @@ static void intel_gpio_set_gpio_mode(void __iomem *padcfg0) value |= PADCFG0_PMODE_GPIO; /* Disable input and output buffers */ - value &= ~PADCFG0_GPIORXDIS; - value &= ~PADCFG0_GPIOTXDIS; + value |= PADCFG0_GPIORXDIS; + value |= PADCFG0_GPIOTXDIS; /* Disable SCI/SMI/NMI generation */ value &= ~(PADCFG0_GPIROUTIOXAPIC | PADCFG0_GPIROUTSCI);
Mistakenly the buffers (input and output) become together enabled for a little period of time during GPIO request. This is problematic, because instead of initial motive to disable them in the commit af7e3eeb84e2 ("pinctrl: intel: Disable input and output buffer when switching to GPIO"), the driven value on the pin, which might be used as an IRQ line, brings firmware of some touch pads in an awkward state that needs a full power off to recover. Fix this by, as stated in a culprit commit, disabling the buffers. Fixes: af7e3eeb84e2 ("pinctrl: intel: Disable input and output buffer when switching to GPIO") BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=210497 Reported-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- drivers/pinctrl/intel/pinctrl-intel.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)