On Mon, Jun 2, 2014 at 1:58 PM, Hans de Goede <hdegoede(a)redhat.com> wrote:
On 06/02/2014 06:53 PM, Josh Boyer wrote:
> On Mon, Jun 02, 2014 at 05:40:57PM +0200, Hans de Goede wrote:
>> Hi All,
>> I've a bunch of backlight handling fixes which for various reasons did not
>> make it upstream for 3.15, which I would like to carry in the Fedora kernel
>> package for now.
>> [PATCH 01/14] thinkpad_acpi: Add mappings for F9 - F12 hotkeys on
>> -Not a backlight fix, but a simple input fix, fixing some keys not working on
>> the latest thinkpads. Reviewed and acked upstream, waiting on the upstream
>> drivers/platform/x86 maintainer.
>> [PATCH 02/14] samsung-laptop: Add broken-acpi-video quirk for
>> -Not send upstream yet, waiting for the reporter to get back to me that it
>> actually fixes things, already in the Fedora 3.14 kernels
>> [PATCH 03/14] ideapad-laptop: Blacklist rfkill control on the Lenovo
>> -Created and tested myself on a laptop of a friend
>> -Waiting on the upstream drivers/platform/x86 maintainer
>> [PATCH 04/14] asus-wmi: Add a no backlight quirk
>> [PATCH 05/14] eeepc-wmi: Add no backlight quirk for Asus H87I-PLUS
>> -Waiting on the upstream drivers/platform/x86 maintainer
Yes, Matthew. I've already send him a ping on these (2 pings for the thinkpad keys
>> [PATCH 06/14] acpi-video: Don't register acpi_video_resume notifier
>> [PATCH 07/14] acpi-video: Add an acpi_video_unregister_backlight
>> [PATCH 08/14] acer-wmi: Switch to acpi_video_unregister_backlight
>> [PATCH 09/14] acer-wmi: Add Aspire 5741 to video_vendor_dmi_table
>> -Queued upstream for addition to 3.16, likely fixes a bunch of other Acer
>> models too
>> [PATCH 10/14] nouveau: Don't check acpi_video_backlight_support()
>> [PATCH 11/14] backlight: Add backlight device (un)registration
>> [PATCH 12/14] acpi-video: Unregister the backlight device if a raw device shows
>> [PATCH 13/14] acpi-video: Add use native backlight quirk for the ThinkPad W530
>> -Slightly adventurous, but only comes into play when using
>> video.use_native_backlight=1 on the commandline, or for models which do so
>> by a dmi quirk
>> -Likely this set + video.use_native_backlight=1 will be needed on a whole
>> bunch of recent laptops with nvidia or ati gfx. Already got confirmation on
>> the upstream list that this is also needed for some ati equipped laptops too.
>> -Queued upstream for addition to 3.16
>> [PATCH 14/14] acpi-video: Add use_native_backlight quirk for HP ProBook 4540s
>> -Want to add to Fedora kernels and ask the user to test before sending upstream
>> -Nothing special really, just another laptop needing
>> Please let me know if there are any objections against me adding this set
>> the Fedora 3.15 kernel pkg. If you don't object an ack would be nice too :)
> OK, I have a question on the backlight ones.
> 3.16 has some backlight rework in the ACPI layer to deal with the
> Windows2012 issue, which I believe fixes several models of laptops.
Correct, basically 3.16 is going to flip the video.use_native_backlight default
from 0 to 1.
> You got a number of quirk patches into 3.15 already that won't be in 3.16
> because of that rework (if I understood the situation correctly).
No, I've discussed this with upstream and it is unsure of the flipping of
that default will stick, or if it will cause too much regressions (it is almost
inevitable that it will break some models, and we will need reverse quirks).
So it was decided to keep the quirk list and keep updating it for now, even
though the quirks will become no-ops for 3.16. And then we will remove the
quirk list in the future if the new default sticks.
> So are these backlight patches additional quirks that were simply
> developed too late to get into 3.15, or are they backports of stuff in
> queued for 3.16?
Patches 6-9 are for an unrelated / orthogonal backlight problem seen on
Acer laptops with nvidia graphics. This is mostly a load / initialization
ordering problem, which does not show up with intel graphics because these
are actually handled in a special way in acpi/video.c . As such they do
introduce a new acpi_video_unregister_backlight function, which is also
used in patches 10-13, since these deal with a very similar ordering problem
when using use_native_backlight in combination with either ati or nouveau.
To answer your quirks or backports question, these 4 are more quirk patches
then a backport.
Patches 10-13 deal with and initialization ordering issues, fixing
use_native_backlight sofar being a no-op when used with ati or nouveau.
These 4 are more in the backport territory, but they are quirk-ish too,
in the sence that the new code does not come into play unless a laptop
model has the use_native_backlight quirk. In 3.16 as such they will
become much more important since their use_native_backlight=1 will be the
Patch 14 adds one more use_native_backlight quirk.
> If the latter, wouldn't they depend on that
> aforementioned rework?
There is no real rework, just the flipping of the default, although
patches 10-13 are important for making the flipping of the default actually
work on ati / nouveau using laptops.
> If not, are you going to attempt to get them into 3.15 stable?
The patches which only add a use_native_backlight quirk all have are
The two 4-patch patch-sets do not have a Cc: stable. I consider the chances of
these two sets causing regressions small, but not 0, hence no Cc: stable. This
is also why I'm suggesting adding them to 3.15 and not to 3.14, so that they
can sit through the Fedora 3.15 stabilization phase.
The benefit of adding these 2 patch-sets now, is that they do fix real world
problems reported by Fedora users, in the case of the Acer patch-set I already
have confirmation from 2 different Fedora users, using 2 different models, that
this helps them. In the case of the second set I've one Fedora user confirming
it helps him + one person on the linux-acpi list.
OK. Thanks for the further background info. I don't see any reason
to keep these out of rawhide. If they introduce regressions, they'll
probably be in a similar state upstream.
I can add these later today, or you can add them if you'd like. If
you do, please take a second and add the Bugzilla: and
Upstream-status: fields we've been keeping at the tops of the patches.