4.5 regression: ACPI hangs when backlight hotkeys are pressed

4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Andy Lutomirski]

I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
the backlight hotkeys a couple times, things start to break.  Poking
at hung thread stacks, it appears that calls to
acpi_video_handles_brightness_key_presses never return.

It appears that the culprit is:

commit aecbd9b1bff6afbe349921b406b99d285b412820
Author: Adrien Schildknecht <adrien+dev@schischi.me>
Date:   Mon Jan 4 23:22:28 2016 +0100

    ACPI / video: driver must be registered before checking for keypresses

Reverting that patch makes my laptop work again.

I don't know what that patch is trying to do, but it looks highly
questionable to me.  Why should an accessor wait for a potential
different driver to register something?  I see no guarantee that it
will eve happen.

This is a Dell XPS 13 9350.

--Andy

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Rafael J. Wysocki]

On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
> the backlight hotkeys a couple times, things start to break.  Poking
> at hung thread stacks, it appears that calls to
> acpi_video_handles_brightness_key_presses never return.
> 
> It appears that the culprit is:
> 
> commit aecbd9b1bff6afbe349921b406b99d285b412820
> Author: Adrien Schildknecht <adrien+dev@schischi.me>
> Date:   Mon Jan 4 23:22:28 2016 +0100
> 
>     ACPI / video: driver must be registered before checking for keypresses
> 
> Reverting that patch makes my laptop work again.
> 
> I don't know what that patch is trying to do, but it looks highly
> questionable to me.  Why should an accessor wait for a potential
> different driver to register something?  I see no guarantee that it
> will eve happen.
> 
> This is a Dell XPS 13 9350.

Should be fixed in my linux-next branch, can you please try that one?

Rafael

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Andy Lutomirski]

On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
>> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
>> the backlight hotkeys a couple times, things start to break.  Poking
>> at hung thread stacks, it appears that calls to
>> acpi_video_handles_brightness_key_presses never return.
>>
>> It appears that the culprit is:
>>
>> commit aecbd9b1bff6afbe349921b406b99d285b412820
>> Author: Adrien Schildknecht <adrien+dev@schischi.me>
>> Date:   Mon Jan 4 23:22:28 2016 +0100
>>
>>     ACPI / video: driver must be registered before checking for keypresses
>>
>> Reverting that patch makes my laptop work again.
>>
>> I don't know what that patch is trying to do, but it looks highly
>> questionable to me.  Why should an accessor wait for a potential
>> different driver to register something?  I see no guarantee that it
>> will eve happen.
>>
>> This is a Dell XPS 13 9350.
>
> Should be fixed in my linux-next branch, can you please try that one?
>

To avoid churning my laptop tree too much, I cherry-picked:

ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
Satellite R830
ACPI / video: Revert "thinkpad_acpi: Use
acpi_video_handles_brightness_key_presses()"
ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
ACPI / video: Fix using an uninitialized mutex / list_head in
acpi_video_handles_brightness_key_presses()
ACPI / video: Revert "ACPI / video: driver must be registered before
checking for keypresses"

The result seems to work.  Are you planning on sending these for 4.5?

--Andy

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Rafael J. Wysocki]

On Monday, January 18, 2016 03:04:28 PM Andy Lutomirski wrote:
> On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
> >> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
> >> the backlight hotkeys a couple times, things start to break.  Poking
> >> at hung thread stacks, it appears that calls to
> >> acpi_video_handles_brightness_key_presses never return.
> >>
> >> It appears that the culprit is:
> >>
> >> commit aecbd9b1bff6afbe349921b406b99d285b412820
> >> Author: Adrien Schildknecht <adrien+dev@schischi.me>
> >> Date:   Mon Jan 4 23:22:28 2016 +0100
> >>
> >>     ACPI / video: driver must be registered before checking for keypresses
> >>
> >> Reverting that patch makes my laptop work again.
> >>
> >> I don't know what that patch is trying to do, but it looks highly
> >> questionable to me.  Why should an accessor wait for a potential
> >> different driver to register something?  I see no guarantee that it
> >> will eve happen.
> >>
> >> This is a Dell XPS 13 9350.
> >
> > Should be fixed in my linux-next branch, can you please try that one?
> >
> 
> To avoid churning my laptop tree too much, I cherry-picked:
> 
> ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
> Satellite R830
> ACPI / video: Revert "thinkpad_acpi: Use
> acpi_video_handles_brightness_key_presses()"
> ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
> ACPI / video: Fix using an uninitialized mutex / list_head in
> acpi_video_handles_brightness_key_presses()
> ACPI / video: Revert "ACPI / video: driver must be registered before
> checking for keypresses"
> 
> The result seems to work.  Are you planning on sending these for 4.5?

Yes, I am.

Most likely on Wednesday (or Thursday if anything urgent shows up in the meantime).

Thanks,
Rafael

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Andy Lutomirski]

On Mon, Jan 18, 2016 at 3:54 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> On Monday, January 18, 2016 03:04:28 PM Andy Lutomirski wrote:
>> On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
>> > On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
>> >> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
>> >> the backlight hotkeys a couple times, things start to break.  Poking
>> >> at hung thread stacks, it appears that calls to
>> >> acpi_video_handles_brightness_key_presses never return.
>> >>
>> >> It appears that the culprit is:
>> >>
>> >> commit aecbd9b1bff6afbe349921b406b99d285b412820
>> >> Author: Adrien Schildknecht <adrien+dev@schischi.me>
>> >> Date:   Mon Jan 4 23:22:28 2016 +0100
>> >>
>> >>     ACPI / video: driver must be registered before checking for keypresses
>> >>
>> >> Reverting that patch makes my laptop work again.
>> >>
>> >> I don't know what that patch is trying to do, but it looks highly
>> >> questionable to me.  Why should an accessor wait for a potential
>> >> different driver to register something?  I see no guarantee that it
>> >> will eve happen.
>> >>
>> >> This is a Dell XPS 13 9350.
>> >
>> > Should be fixed in my linux-next branch, can you please try that one?
>> >
>>
>> To avoid churning my laptop tree too much, I cherry-picked:
>>
>> ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
>> Satellite R830
>> ACPI / video: Revert "thinkpad_acpi: Use
>> acpi_video_handles_brightness_key_presses()"
>> ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
>> ACPI / video: Fix using an uninitialized mutex / list_head in
>> acpi_video_handles_brightness_key_presses()
>> ACPI / video: Revert "ACPI / video: driver must be registered before
>> checking for keypresses"
>>
>> The result seems to work.  Are you planning on sending these for 4.5?
>
> Yes, I am.
>
> Most likely on Wednesday (or Thursday if anything urgent shows up in the meantime).
>

Sounds good.  I'll keep testing.

Some day my laptop will Just Work on a distro kernel :)

--Andy

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Rafael J. Wysocki]

On Monday, January 18, 2016 03:57:49 PM Andy Lutomirski wrote:
> On Mon, Jan 18, 2016 at 3:54 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > On Monday, January 18, 2016 03:04:28 PM Andy Lutomirski wrote:
> >> On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> >> > On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
> >> >> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
> >> >> the backlight hotkeys a couple times, things start to break.  Poking
> >> >> at hung thread stacks, it appears that calls to
> >> >> acpi_video_handles_brightness_key_presses never return.
> >> >>
> >> >> It appears that the culprit is:
> >> >>
> >> >> commit aecbd9b1bff6afbe349921b406b99d285b412820
> >> >> Author: Adrien Schildknecht <adrien+dev@schischi.me>
> >> >> Date:   Mon Jan 4 23:22:28 2016 +0100
> >> >>
> >> >>     ACPI / video: driver must be registered before checking for keypresses
> >> >>
> >> >> Reverting that patch makes my laptop work again.
> >> >>
> >> >> I don't know what that patch is trying to do, but it looks highly
> >> >> questionable to me.  Why should an accessor wait for a potential
> >> >> different driver to register something?  I see no guarantee that it
> >> >> will eve happen.
> >> >>
> >> >> This is a Dell XPS 13 9350.
> >> >
> >> > Should be fixed in my linux-next branch, can you please try that one?
> >> >
> >>
> >> To avoid churning my laptop tree too much, I cherry-picked:
> >>
> >> ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
> >> Satellite R830
> >> ACPI / video: Revert "thinkpad_acpi: Use
> >> acpi_video_handles_brightness_key_presses()"
> >> ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
> >> ACPI / video: Fix using an uninitialized mutex / list_head in
> >> acpi_video_handles_brightness_key_presses()
> >> ACPI / video: Revert "ACPI / video: driver must be registered before
> >> checking for keypresses"
> >>
> >> The result seems to work.  Are you planning on sending these for 4.5?
> >
> > Yes, I am.
> >
> > Most likely on Wednesday (or Thursday if anything urgent shows up in the meantime).
> >
> 
> Sounds good.  I'll keep testing.
> 
> Some day my laptop will Just Work on a distro kernel :)

Well, even then it'll help a lot if you run the mainline on it, so we know that
it doesn't regress.

Thanks,
Rafael

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Andy Lutomirski]

On Mon, Jan 18, 2016 at 4:01 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> On Monday, January 18, 2016 03:57:49 PM Andy Lutomirski wrote:
>> On Mon, Jan 18, 2016 at 3:54 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
>> > On Monday, January 18, 2016 03:04:28 PM Andy Lutomirski wrote:
>> >> On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
>> >> > On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
>> >> >> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
>> >> >> the backlight hotkeys a couple times, things start to break.  Poking
>> >> >> at hung thread stacks, it appears that calls to
>> >> >> acpi_video_handles_brightness_key_presses never return.
>> >> >>
>> >> >> It appears that the culprit is:
>> >> >>
>> >> >> commit aecbd9b1bff6afbe349921b406b99d285b412820
>> >> >> Author: Adrien Schildknecht <adrien+dev@schischi.me>
>> >> >> Date:   Mon Jan 4 23:22:28 2016 +0100
>> >> >>
>> >> >>     ACPI / video: driver must be registered before checking for keypresses
>> >> >>
>> >> >> Reverting that patch makes my laptop work again.
>> >> >>
>> >> >> I don't know what that patch is trying to do, but it looks highly
>> >> >> questionable to me.  Why should an accessor wait for a potential
>> >> >> different driver to register something?  I see no guarantee that it
>> >> >> will eve happen.
>> >> >>
>> >> >> This is a Dell XPS 13 9350.
>> >> >
>> >> > Should be fixed in my linux-next branch, can you please try that one?
>> >> >
>> >>
>> >> To avoid churning my laptop tree too much, I cherry-picked:
>> >>
>> >> ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
>> >> Satellite R830
>> >> ACPI / video: Revert "thinkpad_acpi: Use
>> >> acpi_video_handles_brightness_key_presses()"
>> >> ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
>> >> ACPI / video: Fix using an uninitialized mutex / list_head in
>> >> acpi_video_handles_brightness_key_presses()
>> >> ACPI / video: Revert "ACPI / video: driver must be registered before
>> >> checking for keypresses"
>> >>
>> >> The result seems to work.  Are you planning on sending these for 4.5?
>> >
>> > Yes, I am.
>> >
>> > Most likely on Wednesday (or Thursday if anything urgent shows up in the meantime).
>> >
>>
>> Sounds good.  I'll keep testing.
>>
>> Some day my laptop will Just Work on a distro kernel :)
>
> Well, even then it'll help a lot if you run the mainline on it, so we know that
> it doesn't regress.
>

So far the only other change I noticed is that I stopped seeing
warnings about S1 and S2 being missing.

--Andy

Re: 4.5 regression: ACPI hangs when backlight hotkeys are pressed

[Rafael J. Wysocki]

On Monday, January 18, 2016 04:09:27 PM Andy Lutomirski wrote:
> On Mon, Jan 18, 2016 at 4:01 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > On Monday, January 18, 2016 03:57:49 PM Andy Lutomirski wrote:
> >> On Mon, Jan 18, 2016 at 3:54 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> >> > On Monday, January 18, 2016 03:04:28 PM Andy Lutomirski wrote:
> >> >> On Mon, Jan 18, 2016 at 2:25 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> >> >> > On Monday, January 18, 2016 12:52:53 PM Andy Lutomirski wrote:
> >> >> >> I'm running 4.4 with the pm+acpi-4.5-rc1-1 merged in.  After pressing
> >> >> >> the backlight hotkeys a couple times, things start to break.  Poking
> >> >> >> at hung thread stacks, it appears that calls to
> >> >> >> acpi_video_handles_brightness_key_presses never return.
> >> >> >>
> >> >> >> It appears that the culprit is:
> >> >> >>
> >> >> >> commit aecbd9b1bff6afbe349921b406b99d285b412820
> >> >> >> Author: Adrien Schildknecht <adrien+dev@schischi.me>
> >> >> >> Date:   Mon Jan 4 23:22:28 2016 +0100
> >> >> >>
> >> >> >>     ACPI / video: driver must be registered before checking for keypresses
> >> >> >>
> >> >> >> Reverting that patch makes my laptop work again.
> >> >> >>
> >> >> >> I don't know what that patch is trying to do, but it looks highly
> >> >> >> questionable to me.  Why should an accessor wait for a potential
> >> >> >> different driver to register something?  I see no guarantee that it
> >> >> >> will eve happen.
> >> >> >>
> >> >> >> This is a Dell XPS 13 9350.
> >> >> >
> >> >> > Should be fixed in my linux-next branch, can you please try that one?
> >> >> >
> >> >>
> >> >> To avoid churning my laptop tree too much, I cherry-picked:
> >> >>
> >> >> ACPI / video: Add disable_backlight_sysfs_if quirk for the Toshiba
> >> >> Satellite R830
> >> >> ACPI / video: Revert "thinkpad_acpi: Use
> >> >> acpi_video_handles_brightness_key_presses()"
> >> >> ACPI / video: Document acpi_video_handles_brightness_key_presses() a bit
> >> >> ACPI / video: Fix using an uninitialized mutex / list_head in
> >> >> acpi_video_handles_brightness_key_presses()
> >> >> ACPI / video: Revert "ACPI / video: driver must be registered before
> >> >> checking for keypresses"
> >> >>
> >> >> The result seems to work.  Are you planning on sending these for 4.5?
> >> >
> >> > Yes, I am.
> >> >
> >> > Most likely on Wednesday (or Thursday if anything urgent shows up in the meantime).
> >> >
> >>
> >> Sounds good.  I'll keep testing.
> >>
> >> Some day my laptop will Just Work on a distro kernel :)
> >
> > Well, even then it'll help a lot if you run the mainline on it, so we know that
> > it doesn't regress.
> >
> 
> So far the only other change I noticed is that I stopped seeing
> warnings about S1 and S2 being missing.

That's intentional. :-)

Thanks,
Rafael