Acer Chromebook CB5-311 mainline

[fedup]

We've now reached the EOL for the nyan chromebooks so I'm planning to flash the SPI with uboot later this month.
Just to go over some of the issues raised by @vman and others.
When using nouveau for me results in poor performance. Whilst I'm able to play h.264 video the playback with good timings the image suffers flickering. In xfce4 when switching between virtual workspaces I can seen a sort of diagonal tearing effect so something's not working properly with vsynch. I have to use "PageFlip" "off" in xorg.conf to stop problems with the compositor. Setting firefox to use gpu acceleration improves the firefox experience if you push up the pstate but there is no improvements in xfce4 as it does not support accelerated graphics. In the end I set up 2 entries is my extlinux.conf file; one to boot the kernel with nouveau.modeset=0, which I use most times; the other entry allows me to boot with noveau enabled. I find the fbset performace much better so boot up without nouveau. I don't think nouveau will improve to allow us to make any satisfactory use of the tegra gpu.

I recently tried fedora 30 on a usb flash drive and thought I'd let you all know of some issues I ran into. Fedora 30 on arm has shipped without Firefox on arm as they are now using it with wayland on x86. Apparently they couldn't get a build for arm ready for the release so dropped it. I tried Epiphany, now the only web browser and it worked well but this is not great really. I wanted to try gnome shell with nouveau but gnome seems to be unavailable on arm. Then I tried kde-plasma but this gave loads of nouveau errors and crashed. xfce4 works with nouveau but seems somewhat unstable as I got a black screen a couple of times. Fedora 30 is best avoided on arm based chrome books. I'll install arch next to see how the new mesa 1.9 is working.

[jerrimu]

Decided to redo everything so wiped and put chromeos on fresh. When I went to install to sd csrd, getting an ioctl error. "ioctl error on BLKRRPART: Device or resource busy"


A quick google show a reboot worked on some systems to clear the error, but my chromeook just remounts at boot, so not sure how to reset.


Any tips?

[fedup]

I've just Flashed u-boot to spi with the tegra flashing tools, described on page 13 of this thread, so can confirm this method works. I needed to use the bct file that reey created so it looks as though I have a different revision of the nyan-big as I'm unable to use the norrin bct.

I am having problems with kernels using initramfs. I tried the debian kernel from the buster backport repo. It fails to boot. I also tried my custom kernel with an initramfs and this also failed to load. this is with the exlinux.conf. Not sure what I'm missing here.

I tried using the u-boot _out directory from reey's github https://github.com/reey/nyan-big-u-boot/releases/tag/native-u-boot When I flashed this to the spi the debian kernel/initrd booted fine. Debian kernel is missing the mwifiex wifi modules and also the battery charger module but otherwise it works fine. have yet to try arch and fedora as yet.

[vman]

Hi everyone,

thanks for the update, @fedup! To my annoyance I haven't been able to play around with computers much recently... and the time I had I mostly used for trying to get my NanoPiNeo to better play video. By now I have too many computers struggling with video, like the CB5-311 I'm typing these lines with...

Jerrimu, sorry to hear about your troubles. I hope you'll find a solution!

[fedup]

hello @vman
Yes it's a real shame about how poor the tegra performs with nouveau. I just run with nouveau.modeset=0 now and I'm using i3 as my window manager now and I'm happy with. I recently got a raspberry pi4 and it runs all the video, shame our acer didn't uses the same gpu. I'll never buy anything with nvidea again.

I sorted out the problems with uboot and initramfs. I now wrap the zImage , dtb and initramfs up in a fit image with mkimage. It now launches fine and I was able to start the debian kernel. I had to build a couple of modules but certainly quicker than compiling an entire kernel.

One problem I've noticed though is that the framebuffer is not starting the console early in the boot. I only get the messages once systemd starts launching services. Is difficuly to know what is going on with out the kernel printing to tty1. I know the when I first tried the mainline kernel around 2017 I used the terga_defconfig and penguin boot logo was their, infact I was getting 3 or 4 as the config compiles all the logos available. The framebuffer works but somethings is going wrong.C hecking the journal and DMESG I see early on
$this->bbcode_second_pass_code('', '[ +0.000604] Console: colour dummy device 80x30')
Later in the boot I see
$this->bbcode_second_pass_code('', '[ +0.006219] Console: switching to colour frame buffer device 240x67')

I would like to get the frambuffer early so the I can try and use the LUKS encryption on the rootfs. Need the framebuffer on in order to type the passphrase.

EDIT
Was able to modify my kernel config to get early frambuffer working so now able to have an encrypted emmc rootfs. The debian kernel works, I can't see anything as framebuffer isn't starting early enough but if I blindly enter the LUKS passphrase it unlocks the roots and boot fine.

[vman]

This week I tried to update the somewhat rusty 5.1 kernel of my archlinux CB5-311 to the current 5.4. First I once again could not get U-Boot to work. Then I tried the old way: update reey's PKGBUILD, create and install the packages and "dd" copy the vmlinux.kpart to a boot partition. This way I got to the login screen, but had no keyboard! An external keyboard would not work either. Anyway, 5.3.13, created pretty much the same way, is now working. In the PKGBUILD I also changed the kernel command line parameters to include nouveau.modeset=0... but that does not do anything really, in my case, it seems: I still get sluggish video playback performance when scaling to full-screen. "xvinfo" still says "no adaptors present". Funnily enough there seems to be some 3D support. "gears" performed even better than on a PineBookPro, which I could compare this system to yesterday.

From journalctl:

$this->bbcode_second_pass_code('', 'Nov 21 01:29:49 seagull kernel: [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
Nov 21 01:29:49 seagull kernel: [drm] No driver support for vblank timestamp query.
Nov 21 01:29:49 seagull kernel: Console: switching to colour frame buffer device 240x67
Nov 21 01:29:49 seagull kernel: input: tegra-hda HDMI/DP,pcm=3 as /devices/soc0/70030000.hda/sound/card0/input3
Nov 21 01:29:49 seagull kernel: drm drm: fb0: tegradrmfb frame buffer device
Nov 21 01:29:49 seagull kernel: [drm] Initialized tegra 0.0.0 20120330 for drm on minor 0
')

And from /var/log/Xorg.0.log:
$this->bbcode_second_pass_code('', '
[ 104.959] (II) LoadModule: "modesetting"
[ 104.959] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[ 104.961] (II) Module modesetting: vendor="X.Org Foundation"
[ 104.961] compiled for 1.20.6, module version = 1.20.6
[ 104.961] Module class: X.Org Video Driver
[ 104.961] ABI class: X.Org Video Driver, version 24.0
[ 104.961] (II) LoadModule: "fbdev"
[ 104.961] (WW) Warning, couldn't open module fbdev
[ 104.961] (EE) Failed to load module "fbdev" (module does not exist, 0)
[ 104.961] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[ 104.971] (II) modeset(0): using drv /dev/dri/card0
[ 104.971] (II) modeset(0): Creating default Display subsection in Screen section
"Default Screen Section" for depth/fbbpp 24/32
[ 104.971] (==) modeset(0): Depth 24, (==) framebuffer bpp 32
[ 104.971] (==) modeset(0): RGB weight 888
[ 104.971] (==) modeset(0): Default visual is TrueColor
[ 104.972] (II) Loading sub module "glamoregl"
[ 104.972] (II) LoadModule: "glamoregl"
[ 104.972] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[ 104.991] (II) Module glamoregl: vendor="X.Org Foundation"
[ 104.991] compiled for 1.20.6, module version = 1.0.1
[ 104.991] ABI class: X.Org ANSI C Emulation, version 0.4
[ 105.160] (EE) modeset(0): eglGetDisplay() failed
[ 105.160] (EE) modeset(0): glamor initialization failed
[ 105.160] (II) modeset(0): ShadowFB: preferred NO, enabled NO
')

More PineBook comparison:

*PineBookPro looks a little better, with its slimmer, black case.
*The PineBook's screen is better, I would say, not as blurry as the Acer CB.
*Even though it's five years older, the Acer currently outperforms the PB in an aquarium WebGL demo (30 fps when using Chrome OS with the proprietary nvidia drivers, 1fps in Arch).
*The PineBook I saw and heard was making some audible noise, whereas the ChromeBook is completely quiet.

Seeing that I'd say the CB could still be good for years to come. I actually still use it almost daily, and only have to charge it after a few days usage.

I'd really appreciate it if someone could give a concise account on how to best install U-Boot. Partitioning? Where to take U-Boot from, and which version? How to compile, and which resulting file to copy to the boot partition? When I test installed it some time ago, I had to patch source files... that should not be necessary anymore?

Then, if I could get this machine running better, I'd be inclined to somewhat fill in the gaps in the DebianOn manual (aimed at advanced users, not Linux novices), so other CB owners can also make better use of their hardware.

[fedup]

Hello @vman
Sorry to hear you are having problems with u-boot.
I wiped coreboot from my chromebook, installing uboot with nvidia tegra tools. This means I'm now not chainloading uboot. From memory I had problems using extlinux as the boot loader, so I now create a fit image similar to the way google does. This seems to be the prefered method now. You wrap the dtb, zImage and an initramfs into one image. I think I tried this with u-boot chainloded by coreboot but without success. I'm now unable to test this as coreboot and cromeos have been wiped from my system.
I have also thought about updating the Debian insallation guide. I've made some notes but it's pulling it all together. Debian stable ie buster seems a good choice as there are not to many updates. Fedora and arch are probably to cutting edge. I use a usb flash drive to experiment with them but now keep debian buster on the internal emmc.

[vman]

Hi CB5-311 users,

change of year is a time to look back, and time to tinker with computers... so I now see this is my 52nd post on this forum - hard to believe for me!

What I'd really still like to see on this device is better video playback support in Linux.

But I don't want to flash the device, wiping ChromeOS from it, if this can be achieved by just booting from an SD card. Has anyone here succeeded in doing so?

I experimented with u-boot some more, and with Debian, but without much success. Looking back at my older posts I saw I had been booting with u-boot as recently as August 1st (can't believe I did not remember that anymore... too much stress this year), and getting back to booting it, even in a 2019-10 version, was not hard, after all. I pretty much just used the official source, and followed the instructions in doc/README.chromium-chainload. I managed to boot my custom 5.3 kernel from Arch, also starting a debootstrap'd Debian sid with it. I only got to the shell login, though, no X. Here are the errors from nouveau I saw in syslog:

$this->bbcode_second_pass_code('', '
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 5.3.13-1 (x@x) (gcc version 8.3.0 (GCC)) #1 SMP PREEMPT Wed Nov 27 08:54:55 CET 2019
[ 0.000000] CPU: ARMv7 Processor [413fc0f3] revision 3 (ARMv7), cr=30c5387d
[ 0.000000] CPU: div instructions available: patching division code
[ 0.000000] CPU: PIPT / VIPT nonaliasing data cache, PIPT instruction cache
[ 0.000000] OF: fdt: Machine model: Acer Chromebook 13 CB5-311
[ 0.000000] Memory policy: Data cache writealloc
[ 0.000000] cma: Reserved 64 MiB at 0x00000000fbc00000
[ 0.000000] On node 0 totalpages: 524288
[ 0.000000] DMA zone: 1920 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 196608 pages, LIFO batch:63
[ 0.000000] HighMem zone: 327680 pages, LIFO batch:63
[ 0.000000] percpu: Embedded 20 pages/cpu s51916 r8192 d21812 u81920
[ 0.000000] pcpu-alloc: s51916 r8192 d21812 u81920 alloc=20*4096
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3
[ 0.000000] Built 1 zonelists, mobility grouping on. Total pages: 522368
[ 0.000000] Kernel command line: rw noinitrd root=/dev/mmcblk2p2
[ 0.000000] Dentry cache hash table entries: 131072 (order: 7, 524288 bytes, linear)
[ 0.000000] Inode-cache hash table entries: 65536 (order: 6, 262144 bytes, linear)
[ 0.000000] mem auto-init: stack:off, heap alloc:off, heap free:off
[ 0.000000] software IO TLB: mapped [mem 0xaaabd000-0xaeabd000] (64MB)
[ 0.000000] Memory: 1924372K/2097152K available (12288K kernel code, 744K rwdata, 2684K rodata, 2048K init, 294K bss, 107244K reserved, 65536K cma-reserved, 1245184K highmem)
[...]
[ 24.414732] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.427153] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.439542] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 24.451938] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.464337] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.764728] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.777163] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.789561] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 24.801938] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.814311] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.826692] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.839110] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.851481] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.863930] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 24.876331] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.888793] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.901201] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.913616] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.926032] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.938427] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 24.950819] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.963222] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.975635] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 24.988025] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.000401] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.012872] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 25.025301] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.037704] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.050083] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.062482] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.074903] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.087305] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[ 25.099686] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.112089] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.124485] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.211218] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.223673] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 0d78 data 00000004
[ 25.236082] nouveau 57000000.gpu: gr: DATA_ERROR 0000009c [] ch 2 [0400165000 gnome-shell[620]] subc 0 class a297 mthd 17e0 data 0000006c
[...]
')

Also, the system again only had 2 GByte of RAM, then.

Booting the Debian kernel would not work, resulting only in a black screen early on.

Basically I was hoping for three things u-boot could do now:

1. making it possible to choose kernels at boot-up.
2. passing arguments such as "nomodeset" to the kernel.
3. not having to create kpart images.

Results / failures:
1. the menu does not work, apparently? I googled a little and found something about menu.c32 being required for it. Did anyone here have a usable menu?
2. this seems to work (as confirmed by running "journalctl -b", but contrary to @fedup's experience I still get "no adaptors present" from "xvinfo", even when I add "nomodeset" and "nouveau.modeset=0".
3. this is somewhat annoying, but I wrote down the "mkimage" und "vbutil_kernel" commands in a file, so I can repeat them any time without much effort.

So really I think I don't need u-boot. Realistically I don't think I'm going to use this device much longer anyway. I won an "old" 15" i7 ThinkPad at the Christmas giveaway drawing at work - which of course plays everything effortlessly. I guess I should just use that one for video, and the lighter and smaller CB5-311 with its superior battery runtime for browsing and typing (and wait till maybe someone posts easy-to-follow instructions that don't involve flashing the device).

Or maybe a newer kernel, or a specific kernel config could improve the situation at some point?

[Manizuca]

Hi there!

After some time i decided to take my CB5 and update the system, and everything is forking as expected. I'm using a full ALARM updated system, with custom kernel and xorg-server.

For the kernel, my 5.4.13 build is working fine and at this stage should be stable (i try to avoid new kernel versions up to X.Y.4).

xorg-server-git works fine with acceleration, probably because the git version has modifiers enabled by default (though the modifiers support has some minor bugs at the moment). For the stable version of xorg-server to work (without modifiers), you should revert commit c56fe4118a2dd991ff1b2a532c0f234eddd435e9 from mesa (that commit was backported to 18.3.3, thats why 18.2.x works but 18.3.x fails).

In the future, ill try to use uboot, as im currently using a libreboot build from late-2016

[vman]

Hi Manizuca,

thanks for posting your experience here! After my last, frustrated post about probably changing to a different device anyway I changed my mind again for three reasons:

$this->bbcode_list('1')

currently I don't have any spare money. I'll actually sell the ThinkPad I have been writing about, to get back some liquidity.

I can't even make out any really nice system on the web: silent, compact, long battery life, no (sw)in[td]el, not too expensive... there is nothing (other than the PineBook, which is no better than the cb, except for software support).

the cb5-311 theoretically is fast enough for my mobile and 'sofa' needs, probably for many years to come.

But the state of hardware support on my cb is quite bad, really. Maybe you can elaborate a little, Manizuca?

$this->bbcode_list('a')

how exactly did you install libreboot? I see there is "nyan big" config in the source from git. And what is the effect of it? Do you get a boot menu? Can you boot distribution kernels?

can you share your kernel config (/proc/config.gz)? In my last attempt with 5.4 I had not working keyboard, and currently I'm still running 5.3.

acceleration means "xvinfo" is showing more than "no adaptors present", and you can play video in full-screen?

what do you mean by "modifiers" in the context of xorg-server?

I'm also running a fully updated ALARM system, and that comes with mesa 19.3 already... should I worry about mesa 18.3 (and backports)?