Package Base Details: linux-galliumos-braswell

Git Clone URL: https://aur.archlinux.org/linux-galliumos-braswell.git (read-only)
Submitter: barnacs
Maintainer: barnacs (alaskanarcher)
Last Packager: barnacs
Votes: 3
Popularity: 0.194576
First Submitted: 2017-03-22 05:14
Last Updated: 2017-07-12 22:12

Pinned Comments

barnacs commented on 2017-07-12 22:35

ValNyz, thanks, updated the branch name.

As the GalliumOS kernel releases no longer treat the braswell platform specially from v4.12.0, this package has been superseded by linux-galliumos [0].

Some braswell machines may still require or benefit from the GalliumOS kernel patches/config but hopefully the official arch kernel releases will be good enough for most of us starting from 4.12.

[0] https://aur.archlinux.org/packages/linux-galliumos/

Latest Comments

techdude300 commented on 2017-07-14 23:27

Has anyone tried compiling this from another machine then copying the file to the Chromebook? I've tried compiling both this package and linux-galliumos on my desktop, but even though they build and install fine, my C300SA freezes while loading the initramfs. I'm not sure what could be causing it.

EDIT: I was able to compile the regular Arch kernel from testing (4.12) from my desktop, install it onto the Chromebook, and it boots just fine. So I think this means there's something broken about this package and linux-galliumos, because neither works on the Chromebook, and I don't think it's related to how I compiled. That being said, vanilla 4.12 actually makes my keyboard work, so I'm probably not going to be messing with this much further.

Let me know if there's any debug information I can provide that would be helpful. I'm not sure how to get the kernel verbose/debuggable at such an early stage of boot though.

barnacs commented on 2017-07-12 22:35

ValNyz, thanks, updated the branch name.

As the GalliumOS kernel releases no longer treat the braswell platform specially from v4.12.0, this package has been superseded by linux-galliumos [0].

Some braswell machines may still require or benefit from the GalliumOS kernel patches/config but hopefully the official arch kernel releases will be good enough for most of us starting from 4.12.

[0] https://aur.archlinux.org/packages/linux-galliumos/

ValNyz commented on 2017-07-12 13:50

I'm trying to run Arch linux on an asus C300SA with your kernel.
Using makepkg, i had to change the branch name from
https://github.com/GalliumOS/linux.git#branch=v4.10.5
to
https://github.com/GalliumOS/linux.git#branch=v4.10.5-galliumos

Thanks for your work, i'll tell you if it works great for this computer

barnacs commented on 2017-06-01 23:04

changelog | linux-galliumos-braswell-4.10.5-6 | 2017-06-01 23:02

- fixed build error when compiling with gcc7 ( https://gcc.gnu.org/bugzilla/show_bug.cgi?id=72785 )

Wulfsta commented on 2017-06-01 21:51

Build is now failing:

> ...
> LD init/built-in.o
> LD vmlinux.o
> MODPOST vmlinux.o
> kernel/built-in.o: In function `update_wall_time':
> (.text+0x7bfa9): undefined reference to `____ilog2_NaN'
> make: *** [Makefile:969: vmlinux] Error 1
> ==> ERROR: A failure occurred in build().
> Aborting...

adams commented on 2017-04-30 13:39

@barnacs, it's stock kernel, sorry for the stupid mistake, I forgot to update syslinux.

Both keyboard and touchpad work now, thanks a lot.

barnacs commented on 2017-04-30 13:12

adams, your latest dmesg pastebin says:

> Linux version 4.10.11-1-ARCH

That looks like the stock arch kernel. Are you sure you are booting the one built from this package?

adams commented on 2017-04-30 12:34

Keyboard and touchpad don't work.

Keyboard can type when I move touchpad simultaneously, and the pointer doesn't move.

Same error in dmesg[1]
> genirq: Flags mismatch irq 182. 00002002 (ELAN0000:00) vs. 00000080 (i8042)
> elan_i2c i2c-ELAN0000:00: cannot register irq=182

[1]
https://pastebin.com/raw/z1Sype1f

barnacs commented on 2017-04-30 09:35

changelog | linux-galliumos-braswell-4.10.5-5 | 2017-04-30 09:24

- keyboard should now work on all systems that report dmi product family as "Intel_Strago" plus Samsung Chromebook 3 (CELES), which doesn't report product family, irrespective of bios date

barnacs commented on 2017-04-30 08:35

@adams, @alaskanarcher

The patch doesn't work on that system due to the different BIOS date. From the dmesg:

> genirq: Flags mismatch irq 182. 00002002 (ELAN0000:00) vs. 00000080 (i8042)
...
> cherryview-pinctrl INT33FF:01: Failed to translate GPIO to IRQ

There is a new patch that will hopefully fix it for all chromebooks. I will update the package soon.

alaskanarcher commented on 2017-04-30 08:06

@adams,
This is strange because I have the same computer and this patch works for me. Our dmidecode output is identical except for the BIOS Release Date. Yours is 11/06/206, mine is 05/21/2016.

The IRQ for the keyboard GPIO appears to be correct, which is what the cherryview patch has to do with...
> serio: i8042 KBD port at 0x60,0x64 irq 182

But this looks problematic...
> i8042: PNP: PS/2 Controller [PNP0303:PS2K] at 0x60,0x64 irq 182
> i8042: PNP: PS/2 appears to have AUX port disabled, if this is incorrect please boot with i8042.nopnp
> i8042: Warning: Keylock active

adams commented on 2017-04-30 04:28

I am using EDGAR(Acer Chromebook 14), this patch cannot fix keyboard and touchpad, however the related "galliumos-braswell-config" fixed my sound.

dmesg
https://pastebin.com/raw/0XYZ4Tv5

dmidecode
https://pastebin.com/raw/Ljg788FF

alaskanarcher commented on 2017-04-19 17:08

@aguitel, @barnacs, Can we please move this discussion over to either a) the galliumos-braswell-config AUR package, or b) the Newbie Corner in the Arch Linux Forums.

@aguitel, with Arch Linux you are expected to become familiar with the basics of Arch, like pacman's options and the meaning of its output, and also with the fundamentals of PKGBUILDs and makepkg BEFORE you use a lot of packages from the AUR. You also need to help yourself by reading the wiki and man pages instead of relying on AUR package maintainers to teach you the basics. There are folks watching the Newbie Corner of the Arch Linux Forums who help folks with the basics, but you still have to do your own research and homework before asking a question.

Pacman's output is describing the problem exactly:
> error: failed to commit transaction (conflicting files)

So no, you don't have galliumos-braswell-config installed. There is a good reason why pacman isn't letting you install that. Understand why it stops you so you can determine for yourself whether it is appropriate to override it and how.

If you want a distro that just works without all this in depth learning required, try the actual GalliumOS distro.

barnacs commented on 2017-04-19 17:01

@aguitel use "pacman --force" or the equivalent option of your aur wrapper. Let's move any further discussion about that package to its comments.

aguitel commented on 2017-04-19 16:48

i installed pulseaudio first but when install galliumos-braswell-config error appears

Packages (1) galliumos-braswell-config-1.0.0-1

Total Installed Size: 0.02 MiB

:: Proceed with installation? [Y/n] y
(1/1) checking keys in keyring [######################] 100%
(1/1) checking package integrity [######################] 100%
(1/1) loading package files [######################] 100%
(1/1) checking for file conflicts [######################] 100%
error: failed to commit transaction (conflicting files)
galliumos-braswell-config: /etc/pulse/default.pa exists in filesystem
galliumos-braswell-config: /usr/share/pulseaudio/alsa-mixer/paths/analog-output-headphones.conf exists in filesystem
galliumos-braswell-config: /usr/share/pulseaudio/alsa-mixer/paths/analog-output-speaker.conf exists in filesystem
Errors occurred, no packages were upgraded.

barnacs commented on 2017-04-19 16:36

@aguitel No, that shouldn't matter as long as you are using any version of this kernel package. The important thing is to install pulseaudio first and then install galliumos-braswell-config to let it overwrite some configs. If you don't get audio this way, you could try resetting your alsa mixer state.

aguitel commented on 2017-04-19 15:45

i tried with kernel 4.10.5-3
need to install again ?

barnacs commented on 2017-04-19 15:37

@aguitel Great! For the audio, have you tried https://aur.archlinux.org/packages/galliumos-braswell-config/ ? You should let it replace some pulseaudio files.

aguitel commented on 2017-04-19 14:34

with last kernel 4.10.5-4 keyboard and touchpad working fine

only sound remain to fix

barnacs commented on 2017-04-18 21:35

4.10.5-4

aguitel commented on 2017-04-18 21:31

sure , what is the last version ?

barnacs commented on 2017-04-18 21:23

@aguitel Thanks!

> cherryview-pinctrl INT33FF:01: Failed to translate GPIO to IRQ

Looks like the same issue. Updated the patch to include Samsung Chromebook 3 (CELES). Could you please try the latest version and let us know if it fixes your keyboard?

aguitel commented on 2017-04-18 20:58

dmesg
https://bpaste.net/show/a8f1f22200f3

dmidecode
https://bpaste.net/show/64090f06e7bd

aguitel commented on 2017-04-18 20:43

i think need to build in chromebook, i will install one more time and pastebin dmesg and dmidecode

alaskanarcher commented on 2017-04-18 20:35

@aguitel, yes, you can build it on any Arch Linux x86_64 system and copy over the .pkg.tar.xz file. I build it for myself on an AWS EC2 server. I recommend reading up on the Linux Kernel build process and Arch Linux PKGBUILDs and makepkg (and its build options, like --noextract for only doing the build process).

Why do you need to rebuild it though, did you delete the .pkg.tar.xz file? If it's because you only have 4.5.10-2 built, that one is just fine to use for the dmesg output.

aguitel commented on 2017-04-18 20:06

can i build linux-galliumos-braswell package in other pc than chromebook and then copy to pendrive and install in chromebook ? i ask this why chromebook are slow and with low resources and take long time to do that

alaskanarcher commented on 2017-04-18 19:55

@aguitel, please supply dmesg and dmidecode output from the linux-galliumos-braswell-4.10.5-3 kernel. In other words, the most up to date clone of this package. An external USB keyboard should work. If you don't have access to one, try continuously moving your finger across the trackpad while typing. That seems to trigger processing of the input buffer.

@barnacs, ah yes that was silly of me to not look at the FIRST line of dmesg, ha. You are right, 4.8.13 is definitely before this regression. And that alone makes it seem very likely that this is the same bug. Once we more positively confirm that it's the same bug I'll comment on that kernel bug report and add another chromebook to the list of the afflicted. Maybe someone there can see some common and unique DMI info that can be used to apply this quirk to all of these systems. Maybe all Intel_strago and this one Celes 3 exception.

aguitel commented on 2017-04-18 19:51

this is chromebook samsung 3 celes (braswell)

barnacs commented on 2017-04-18 19:37

@alaskanarcher

from linked the dmesg output

> Linux version 4.8.13-1-ARCH (builduser@tobias) (gcc version 6.2.1 20160830 (GCC) ) #1 SMP PREEMPT Fri Dec 9 07:24:34 CET 2016

As far as I know the bug first appeared in 4.9 so I'm assuming this output is from the last available working kernel version. Also, it seems to be a braswell system, which would further indicate that this may indeed be the same bug. Making a lot of assumptions here though...

from the dmidecode output:

> Family: Not Specified

Now, that is unfortunate. Also, the cpu id is slightly different from the other two.

aguitel commented on 2017-04-18 19:36

dmesg a d dmidecode was generated with arch kernel 4.8.13-1

alaskanarcher commented on 2017-04-18 19:11

@aguitel, your dmesg doesn't seem to indicate that the source of your problem is due to the cherryview-pinctrl bug that was affecting Wulfsta below. I'm not sure why the keyboard isn't working for you as your dmesg output seems to indicate the correct irq number and that the keyboard is being set up without error.

[ 0.792359] serio: i8042 KBD port at 0x60,0x64 irq 182
...
[ 0.833200] input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input0

Just to confirm, this output was generated after booting into the kernel built from the most up to date and unmodified clone of this AUR repo, and the builtin keyboard and trackpad was not working when you generated the output, correct?

aguitel commented on 2017-04-18 18:19

dmesg
https://bpaste.net/show/42a1f6e2dc9b

dmidecode
https://bpaste.net/show/2a3c284b7ebc

alaskanarcher commented on 2017-04-18 18:02

@aguitel, can you provide pastebins with `# dmidecode` and `$ dmesg` outputs from booting this kernel on your samsung 3 celes?

@barnacs, it has been suggested on the bug report that this problem is likely to present on all systems in the 'Intel_strago' family. Next patch may just check for that.

aguitel commented on 2017-04-18 17:41

i test this package in samsung 3 celes and keyboard and touchpad is not working

Wulfsta commented on 2017-04-18 04:04

@barnacs The patch worked, thanks for the update.

alaskanarcher commented on 2017-04-17 18:53

@barnacs,

I wasn't having issues with the old dumber irq bump patch but I appreciate that you changed the new patch to include the afflicted Asus C301SA instead of reverting to the old patch. This way we'll be more likely to catch more systems where this change in IRQ numbering has caused an issue. That's the only way these things will get fixed upstream.

According to the pinctrl kernel devs this is really a result of the firmware on these systems using hardcoded IRQ numbers, which they consider to be a bug. This prompted the devs to make the patch apply specifically to the firmware version, not just the specific Acer CB 14 laptop. So if we get a firmware update that doesn't change how IRQ numbering is handled, don't be surprised if this breaks again.

@Wulfsta That atkb error looks like what I was observing before this patch was available. https://bbs.archlinux.org/viewtopic.php?pid=1689233#p1689233 It is interestingly not in the dmesg output I posted on the kernel bug report (but that was generated from a different kernel version when I had debugging turned on).

Please let me know if barnacs' patch fixes this for you. If so then we should leave another comment on the bug report updating them that their fix also works for your computer.

barnacs commented on 2017-04-17 01:17

@Wulfsta I'm not quite sure what I'm doing either but I updated the package anyway. I don't have a device to test it but hopefully it now works for Asus C301SA as well.

Wulfsta commented on 2017-04-16 23:45

@barnacs I'll report it in that thread as suggested - in the meantime I'll just modify the patch by replacing the function call with `false`. I tried to get the patch to recognize my device, but I'm unfamiliar with the dmi_system_id structure syntax.

barnacs commented on 2017-04-14 06:52

@Wulfsta Here's the related upstream ticket: https://bugzilla.kernel.org/show_bug.cgi?id=194945
You should probably raise your issue there along with your dmidecode output as the current patch only fixes the keyboard for a specific model (acer cb431).
In the mean time, you could just keep using the previous version I guess.

@alaskanarcher Were you having issues with the dumber irq bump patch? Or is it ok with you to revert to that for wider compatibility? Once we know the dmi strings for this asus I could alter the patch to work for both models but there may be more models affected out there...

Wulfsta commented on 2017-04-14 05:54

The 2017-04-12 17:05 update breaks the Asus C301SA.

It is a fairly interesting error - the standard kernel complains prior to recognizing a keyboard, but the update causes
`atkb serial0: Failed to enable keyboard on isa0060/serial0`

If you need more details please ask.

alaskanarcher commented on 2017-04-12 08:24

Changelog for 2017-04-12:
- Removed CONFIG_INPUT_EVBUG from config
- Changed CONFIG_INPUT_EVDEV to m
- Replaced patch for keyboard interrupt in the cherryview pinctrl driver on EDGAR Cyan Chromebooks. Now using a patch from kernel developer Mika Westerberg. This is expected to be merged upstream shortly

barnacs commented on 2017-03-28 06:06

Yeah, that evbug output is excessive, though you can just blacklist the evbug module without recompiling the kernel.

@alaskanarcher Added you as a maintainer. Thanks a lot for helping out and also for your efforts towards getting support for this platform into mainline!

alaskanarcher commented on 2017-03-28 00:49

It compiles with CONFIG_INPUT_EVDEV=y and CONFIG_INPUT_EVBUG=y or m which results in an overwhelming number (as in 100/s) of evbug debug messages in dmesg.
i.e.
evbug: Event. Dev: input13, Type: 0, Code: 0, Value: 0

For normal use I believe this parameter should be changed to CONFIG_INPUT_EVDEV=m, and CONFIG_INPUT_EVBUG left undefined, as it is in the mainline Arch Linux kernel.

I also replaced GalliumOS' increase-cherryview-num-irqs.diff patch for fixing the keyboard with the patch provided by the kernel devs which will be submitted upstream soon.
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=194945
Patch: https://bugzilla.kernel.org/attachment.cgi?id=255427

@barnacs I would be happy to share my changes. I use this platform and like to help maintain it. Open to collaborating on this package if you are interested.

alaskanarcher commented on 2017-03-26 17:36

Tested this with the package galliumos-braswell-config and I now have working keyboard and audio.

Thank you Barnacs