Fedlet: a Fedora Remix for Bay Trail tablets

Fedlet development is currently DORMANT

PLEASE NOTE: At present (October 2016), I do not have any time to work on Fedlet. Through the magic of Google, this looks like a current effort to get Linux (Ubuntu) running on Baytrail / Cherrytrail devices; you might want to try that.

EDIT FEB 2017: I still don’t have time for Fedlet, sorry. However, Nemanja Milosevic is doing stuff, and he links to this page, of another person who is…doing stuff. Hope this helps you.

Fedlet

Can you run Linux on a tablet? Sure you can!

Here’s video proof.

This is Fedlet, a Fedora remix for Intel Bay Trail-based tablet devices with 32-bit firmwares. Particularly the Dell Venue 8 Pro, which is what I have. It has been reported to work on the Lenovo Miix 2 and Asus T100. It may work on the Toshiba Encore and any other 32-bit firmware Bay Trail-based tablet.

It’s based on Fedora 23, more or less – but it has a slightly patched kernel, and a few other tweaks. Some of the Bay Trail support is not yet complete, and testers on various devices have reported instability. So this is not yet stable release quality, but it’s appropriate for playing around with these devices. Seriously, I mean it’s pretty experimental and nothing is guaranteed. This is for playing around and helping to make things better, it’s not a production OS. Please don’t install this if for some reason your Intel tablet is your primary work device or something.

Releases

20150810

Eleventh release of Fedlet.

  • Fedlet 20150810 for all 32-bit Baytrail hardware – SHA256SUM: 7b2b6a45df4738865481b9af17536f2f011b6c24a3c29fb9d6644dd21da2e545
  1. Fedora 23 Alpha-ish userland
  2. Kernel 4.2rc6
  3. Seems to be a bug where no OSK appears for Firefox; try installing Epiphany

20141209

Tenth release of Fedlet.

  • Fedora 20141209 for all 32-bit Baytrail hardware – SHA256SUM: c1fcf78e883d28345074bb48c814732a2dced2c79f54ee007c769e31a4fd134b
  1. Fedora 21(ish) userland – with 0-day updates
  2. Kernel 3.18.0, with Fedlet patches

20141124

Ninth release of Fedlet.

  • Fedlet 20141124 for all 32-bit Baytrail hardware – SHA256SUM: c64b859d5ec08dd1c6e15eb8e0553d1b775475e22c852062166a9ae63bfeac6a
  1. ~Fedora 21 Final TC4 userland
  2. Kernel 3.18rc6
  3. NVRAM map for Broadcom brcmfmac43241b4 (thanks “Brainwreck” of the Ubuntu T100 project) which may make wifi work OOTB on Asus T100
  4. Accelerometer-based rotation support for Venue 8 Pro (i.e. display rotates when you rotate the tablet)
  5. Backlight level control on Venue 8 Pro when booted with i915.force_backlight_pmic=1

20141111

Eighth release of Fedlet.

  • Fedlet 20141111 for all 32-bit Baytrail hardware – SHA256SUM: 3a4078db12b1ed17d9c330ef9c16d2690299212ec6f58867d64f2589a6afc088
  1. ~Fedora 21 Final TC1 userland
  2. Kernel 3.18rc4
  3. Patch from Jan-Michael Brummer to make Venue 8 Pro wifi work
  4. Has generic-release-cloud not generic-release-workstation, sorry, I’ll fix it for the next release

20140929

Seventh release of Fedlet, aka the “God, I’m still doing this?” release

  • Fedlet 20140929 for all 32-bit Baytrail hardware – SHA256SUM: aa2f1150e40965471fc2888db6aad7da52d98f36ce1224b630ba5ed99b28fd5e
  1. Current Fedora 21 userland, ~F21 Alpha
  2. Kernel 3.17rc6
  3. efibootmgr is back, so install might work (not tested)
  4. Patch from Jan-Michael Brummer for ‘Home’ button on V8P to act as ‘Super’ (start) key
  5. Patch from Jan-Michael Brummer for mic input (not tested yet)

20140911

Sixth release of Fedlet

  • Fedlet 20140911 for all 32-bit Baytrail hardware – SHA256SUM: d76574e38d5afab1cb84ac95dde3945376c76518d2610c8afdaa84305ba3f43e
  1. Updated to current Fedora 21 userland, ~= Fedora 21 Alpha TC7
  2. Based on Workstation kickstart
  3. Updated to 3.16 kernel with small Baytrail patch set, native modesetting should now work, no more hard-coded resolution hacks needed
  4. Sound driver and firmware included (but you still need to load a correct ALSA config to hear sound)
  5. Partial support for Venue 8 Pro built-in wireless (firmware included)
  6. Hardware button support for Venue 8 Pro
  7. Battery status support
  8. Install broken (missing efibootmgr)

20140310

Fifth release of Fedlet

  • Fedlet 20140310 for 8″, 800×1280 tablets
  1. Updated base packages and kernel
  2. Xorg hack to allow windows with integrated title bars to be dragged in GNOME (from Jan-Michael Brummer)

20140226

Fourth release of Fedlet

  1. Latest Fedora Rawhide base
  2. Kernel update: based on latest Rawhide, sound (and LPSS) support built in (but not working until you provide fw_sst_0f28.bin* in /usr/lib/firmware/intel and apply this mixer config), shutdown/reboot should work on Venue 8 Pro, T100 and Miix 2
  3. GNOME Terminal added to the Dash for convenience
  4. Updated the patched anaconda to latest Rawhide
  5. LibreOffice dropped to save space (I doubt anyone wants to use it on a tablet much…)

20140221

Third release of Fedlet

  1. Fix kernel performance regression
  2. Touch input rotation seems to work automagically now, so drop it from v8p-rotate (it’s now just a simple xrandr wrapper)
  3. Add a 10in (T100) build (untested)

20140220

Second release of Fedlet

  1. Repository configuration added (package: fedlet-repo)
  2. Useless custom build of xorg-x11-drv-intel dropped
  3. Kernel up to 3.14rc3 with some patches upstreamed, display hotplug reversion patch dropped and video= parameter adjusted to allow display to work with the hotplug reversion patch dropped
  4. anaconda bumped to latest version (with fedlet patch applied)

20140207

First release of Fedlet

Working

  1. Boot
  2. X
  3. GNOME
  4. 2D, 3D and video playback acceleration
  5. Touchscreen
  6. CPU frequency scaling (pstates)
  7. USB (you can use a USB wifi adapter)
  8. Power monitoring (battery status)
  9. Sound (with ALSA config file)
  10. Wifi (on Venue 8 Pro at least, possibly also Asus T100)
  11. Hardware buttons (on Venue 8 Pro)
  12. Backlight control (on Venue 8 Pro, when booted with i915.force_backlight_pmic=1)
  13. Installation and boot of installed system (if you’re very brave)

Partly working

  1. KMS (hence accelerated video) on Venue 8 Pro seems to have quirks related to boot process, see notes below
  2. Suspend (kinda works since kernel 3.16 or so, but screen backlight may stay on, and various things may not survive the resume, e.g. touchscreen or rotation)

Not working

  1. Venue 8 Pro onboard Bluetooth
  2. Icon for rotation app is invisible with recent GNOME
  3. Most likely lots of other things

Unknown (please let me know!)

  1. Hardware support (wireless, bluetooth etc) on devices other than Venue 8 Pro

Usage

Not for 64-bit firmwares

64-bit firmware Bay Trail devices are showing up now: I wouldn’t recommend using Fedlet on those, probably, as most of the point of Fedlet is to be a 32-bit UEFI image for the 32-bit firmware Bay Trail devices. If you have a 64-bit firmware Bay Trail device, I’d probably suggest installing Fedora 21 Beta (or a Final TC/RC) then updating to a 3.18 kernel from the rawhide-kernel-nodebug repository. I could do a 64-bit build of the Fedlet kernel and the few other divergent packages, I guess.

Writing the image to USB

You can follow the standard Fedora USB writing instructions – both livecd-iso-to-disk --format --reset-mbr --efi and dd like methods should work. Do not use Rufus, unetbootin or any other ‘smart’ third party USB stick writer. They rarely work correctly, especially for UEFI booting. Tools that work like dd (several are mentioned on the page linked above) are fine.

Booting from USB on Venue 8 Pro

To boot from USB on the Venue 8 Pro, turn it off, connect the USB stick, then hold down the volume up button immediately after pressing the power button, until you see the Dell logo. This should take you into a boot menu from which you can pick your USB stick. You can also hold volume down to get into the firmware UI, where you can go to the Boot tab and move the USB stick up to the top position in the boot order (see note above about how different boot paths impact graphics).

Notes and tips

Native graphics on Venue 8 Pro

It seems to vary between devices, but I have found that graphics don’t work properly on the Venue 8 Pro (screen goes black when KMS kicks in) if you boot normally or through the firmware UI (hold volume down on boot). KMS always works if you boot through the boot device menu (hold volume up on boot). If you have a V8P and you’re getting the black-screen-on-boot problem, try different boot paths.

Sound

On most hardware, you should be able to make sound work with this ALSA state file. Download it and run alsactl -f /path/to/t100_B.state restore.

Connecting USB devices

If you don’t know this already you probably shouldn’t be playing with Fedlet, but in order to connect any USB devices, you need something called a “USB OTG cable”, which basically turns the micro-USB port on the tablet into a ‘regular’ USB port you can plug keyboards and USB sticks and things into. Available at any decent parts retailer for about $5, or any big box electronics store for about $25. Your choice.

If wifi isn’t working on your device, you can plug in a wireless USB adapter if you have a USB OTG adapter. I’m using an Asus USB-N10, it should work out of the box.

For ease of testing it’s probably a good idea to have a USB hub you can plug a wireless adapter (if needed), USB stick(s) and keyboard into.

Firefox extensions

The grab-and-drag and Go-Mobile extensions for Firefox are probably useful things to have.

Video playback acceleration

If you are legally allowed to – I can’t tell you whether you are or not, I am not a lawyer – you can install the libva-intel-driver package from RPM Fusion’s free repository. This will enable hardware-accelerated video playback in any app which speaks libva (for me, it fails on quite a few videos, have to dig into that).

Installation

If you’re very, very bold, you should be able to install Fedlet. On the Venue 8 Pro, the internal storage has a fairly big NTFS partition with Windows on it, and a bunch of smaller partitions. I’d recommend just destroying the big Windows partition and installing into that space: the other partitions are system and recovery partitions, if you leave them intact, it should be possible to recover the Windows installation later if you want to (I have not tested this).

If you do install this, get kernel updates from [my repository][19], and don’t install official kernel updates from the Fedora repos. We’re trying to get all the patches merged ASAP. I’ll try and remember to put updated kernel builds in my repo regularly. Stock kernels will now boot, at least, but (as of 3.16) shutdown/reboot may not work, battery status won’t work, and Venue 8 Pro wifi won’t work.

On the Venue 8 Pro at least, the firmware has an irritating habit of putting the Windows boot loader back at the top of the UEFI boot manager list if you attach or remove USB sticks (or sometimes, just for giggles). If you boot it with this setup it’ll go into Windows auto-recovery. I haven’t been brave enough to see what this does yet, I just force power off and go back into the firmware and put Fedlet (“Generic”) back at the top of the list.

What’s in it that’s different from Fedora?

The ‘sources’ for the outside-of-Fedora stuff that’s included in the image can be found in this github repository. There are:

  • Some kernel patches in the kernel/ directory which are applied to the kernel package in the image
  • Some Xorg config snippets and a trivial utility for rotating the screen on the Venue 8 Pro, in the xorg/ directory
  • The kickstart used to build the image, and a patch to python-imgcreate for building UEFI bootable 32-bit live images, in the ks/ directory
  • The patch that (hopefully) makes installation work smoothly in the anaconda/ directory
  • The necessary firmware for the Venue 8 Pro’s wifi adapter in the baytrail-firmware/ directory
  • The repository definition for the fedlet repo in the fedlet-repo/ directory
  • A (hopefully) working ALSA configuration file in the alsa/ directory

The packages that differ from pure Fedora Rawhide are all available from [this repository][19]. There is:

  • A patched anaconda which should allow installation to work cleanly
  • The kernel package patched with the patches from the git repository
  • The fedlet-repo package containing the repository definition
  • A package of the v8p-rotate utility
  • The baytrail-firmware package, currently containing only Venue 8 Pro wireless firmware (license proprietary-but-freely-redistributable)
  • A backport of Rawhide’s linux-firmware package, which contains the firmware needed for the sound adapter

All the variant packages have the dist tag ‘awb’ to make them easily distinguishable from official Fedora packages (except the linux-firmware package, which is just a backport).

The image should be roughly reproducible by just building a live image, using the kickstart, from a running Fedora 21 system, after applying the patch to python-imgcreate’s /usr/lib/python2.7/site-packages/imgcreate/live.py.

I cannot take any of the credit for the hard work on this: all I’ve done is write silly little scripts and stick the bits together. Multiple folks at Intel, Red Hat and elsewhere have done the tough work. An especial big thanks to Alan Coxm Aubrey Li, and Mika Westerberg at Intel, Jan-Michael Brummer at IAV (formerly of Intel), and Kalle Valo at Qualcomm (for the V8P wifi) who are really pushing the thing along.

548 Responses

  1. Aurélien
    Aurélien December 8, 2014 at 2:21 am | | Reply

    Hello,

    I was trying to install fedlet yesterday on my asus t100 (32gb with 500gb hard drive) and there is something that I don’t understand about the bootloader (I was putting the system in the hardrive by the way). The program is yelling at me if :

    * I don’t put a bootloader (ok, it seems normal)
    * I put a bootloader on the mmcblk disk, saying it should be on GPT
    * I use the bootloader already existent on the mmcblk disk

    But, If I create a new partition on the hard drive and assign the bootloader to this one it seems to be ok. I’m really not used to this uefi system but my goal would be to have a bootable system with both fedlet and windows and therefore I would not like to loose my windows bootloader. Moreover, already some people manage to install it on an asus without hard drive (right?), so this should not be a problem in any case.

    Any clues of what could be the good way to setup the bootloader ?

    Thanks a lot for your work guys 🙂

    Aurélien.

  2. Markus Larsson
    Markus Larsson December 10, 2014 at 11:48 am | | Reply

    I’m sad to say that there is something fishy going on when running on Lenovo Miix 2 10.
    The system hangs in what seems to be the lockscreen going bananas every like 1-3 minutes.
    I’m getting the usual can’t write to emmc and the wifi driver seems to be acting up at the same time.
    If in a text console before the hangs I usually get some kind of error message from brcmfmac, like “invalid key index”, right before it hangs.
    I’m running 20141124 with the kernel from 20141209 since I haven’t been able to keep the system stable for long enough to install all packages that could be updated.
    I’m not really sure where to start but hopefully I can help getting this solved.

  3. Paul Terhune
    Paul Terhune December 10, 2014 at 10:13 pm | | Reply

    Works on my HP Omni 10. No Bluetooth though. Rotate is pretty sketchy. Screen goes black for a long time before it comes back and then the desktop gets really slow and unresponsive. Wifi works! Big plus as it’s Broadcom. Just got this working and was pleasantly surprised I could get ANYTHING other than Win 8.1 to boot. Tried the “Install to hard drive” in the apps menu and anaconda crashed almost immediately. But I’m thrilled it’s working at all. I’ll keep messing with it.

    Thanks for the work!

  4. DDS Central
    DDS Central December 11, 2014 at 10:38 am | | Reply

    Ubuntu mostly works on my T100 with 3.18 plus Fedlet patches.

    Problems:
    * Random freezes, requiring a hard reset
    * Slow WiFi speed (can’t do a stable 700KB/s on LAN)
    * Sometimes, very slow UI response (usually fixed by rebooting)
    * Strange errors in dmesg, probably related to the backlight patch.

    Not working:
    * Bluetooth:
    no idea how to fix…

    * Orientation sensor
    It’s a Invensys MPU6500, but inv-mpu6050 driver (which is supposed to support it) returns error -121 on probe.

    * Camera
    The sensor is MT9M114, there’s a driver for this in Chromium repos, tried to patch it to work with 3.18. It detects the camera but does not create any device for it.

    * Volume buttons
    These will probably require device specific patches to work.

    * Ambient light sensor.
    It’s a CM3218. It works fine with driver from jfwells repo, but I have no idea how to make Ubuntu use it.

    Also, I’m testing Android x86 4.4.4 with Fedlet kernel (and a few additional patches from from https://github.com/pstglia/linux/tree/v3.18-rc6_android).
    Surprisingly, it mostly works (thanks to the mainlining project), but has the same issues as in Ubuntu.

    Will need to watch Linux git repo for additional baytrail patches…

  5. jekader
    jekader December 12, 2014 at 9:44 am | | Reply

    Wow, nice project! I just got a cheap chinese atom-based tablet and am playing around with it trying to install Linux. Great to see that you have already progressed with this. Good luck, I’ll be glad to test stuff.

    PS: will ping you on Monday as it seems we work at the same company 🙂

  6. Olivier
    Olivier December 14, 2014 at 6:49 am | | Reply

    Seems to works on an odys wintab (the wifi is a 8723bs and works more or less with this https://github.com/not-aml/openlinux-amlogic-wifi-drivers/tree/2014-08-12/rtl8723BS (tested on a standard 3.17.6-1 for now).

    A question: why not provide a 64 bit distribution? Actually all tablets that I am aware of support it (there are based on a 64 bit atom processor), it is just the UEFI that is 32 bit but you can load a 64 bit kernel from the 32 bit grub. Recent 64 bit kernel can even access 32 bits EFI runtime services.

    1. jekader
      jekader December 14, 2014 at 8:21 am | | Reply

      Having 64-bit for the sake of 64-bit? What other profits are there? I think the majority of BayTrail tablets don’t have more than 4 gigs of RAM, so it’s actually more efficient to stick to 32-bit for now IMHO.

  7. jekader
    jekader December 14, 2014 at 8:18 am | | Reply

    I can also confim the following for my Cube7 [1] device:
    1) it boots!
    2) installs to internal flash just fine
    3) Xorg/framebuffer starts OK only when entering BIOS before booting.
    4) Touchscreen works but often registers a single screen tap as double click which is annoying
    5) No screen rotation
    6) Battery charge not detected
    7) WiFi is RTL8723BS works after compiling the driver. I grabbed the sources in [2] and am using the Fedlet kernel.
    8) often after start some CPU timing issues occur (uptime shown as 30 seconds while actually 15+ minutes ETC), did not yet find the root cause

    So far I am happy with progress, and hopefully we can have all of it running soon.

    [1] http://www.dx.com/p/zzz-356705
    [2] https://github.com/hadess/rtl8723as

  8. tim senecal
    tim senecal December 14, 2014 at 10:02 am | | Reply

    just an fyi, for people like me that might want to get one of these specifically for fedlet testing…

    microcenter now has a whole family of atom based tablets that should run fedlet fine… starting at $59.99

    http://www.microcenter.com/search/search_results.aspx?sku_list=410605+410670+410621+468421

  9. Sam
    Sam December 14, 2014 at 7:34 pm | | Reply

    I tried the liveCD startup on Dell Venue Pro 11. It was unusable and very slow. Is this because it was a liveCD session, or the current state of FedLet on the limited hardware (The tablet has 2 GB ram and Intel Atom Z37xx processor) ? I understand things are going to be buggy, but I expect to be able to browse websites and do basic things

    1. Sam
      Sam December 14, 2014 at 7:43 pm | | Reply

      I should mention that the Wifi was working, there did not seem to be any big problems in the dmesg output.
      Thanks for all your work.

    2. jekader
      jekader December 15, 2014 at 7:22 am | | Reply

      In my case “slowness” was was caused by clock issues and I had to force the “tsc” clocksource which fixed the lag (appended clocksource=tsc to the kernel cmdline)

  10. vitaly.torshyn
    vitaly.torshyn December 15, 2014 at 2:33 am | | Reply

    Hey Adam,
    I appreciate the work you did!
    Anyway the xorg’s driver can’t be loaded on my Lenovo Mixx 2 8 tablet (Baytrail+2G Ram) due to crash of drm. Debug message contains errors about checks: [drm:intel_pipe_config_compare] *ERROR* mismatch in adjusted_mode.flags … . The same issue with video=VGA-1:800x1280e.
    I’m wondered if you have some workaround for it.
    In addition, I find that a lot of stuff related to DRI for out CPUs has been improved in kernel 3.18. Have you tried to run with this version?

    Thanks.
    PS.: Please let me know if I could help you with Fadlet release. I have some experience in development for/under Linux OS.

  11. Bobby Budnick
    Bobby Budnick December 15, 2014 at 12:03 pm | | Reply

    Hello again. I have not gotten any further at all with sound with the newer devices. I sent some e-mails out, briefly discussed the problem and haven’t heard anything further back. I did get a bugzilla link though with the problem similar to mine:
    https://bugzilla.kernel.org/show_bug.cgi?id=86581 . There has been little activity there though. They must be stumped.
    What is the easiest way to edit your live images? I would like to change the firmware to the one that actually initializes a sound device on all these devices. I realize you are using the latest firmware already but this firmware simply does not work on the newer devices. I successfully compiled a 3.18 release kernel and Mark Brown’s “for-next” branch of the kernel but I still get “baytrail-pcm-audio baytrail-pcm-audio: ipc: error DSP boot timeout” with the latest firmware. By the way, the boot error I was getting with a custom kernel was because I did not grab the old kernel config and run “make oldconfig”.
    Where can we go to discuss running Linux on these devices? The ubuntuforums are utterly garbage for this sort of thing. XDA-developers has surprisingly little going on as well.

  12. kurious
    kurious December 16, 2014 at 3:33 pm | | Reply

    Can this be used in an Asus Memo Pad 8?
    What would be the requirments? Should I root it before? Unlock the bootloader?

  13. Mario
    Mario December 17, 2014 at 4:21 am | | Reply

    Hi,

    I’m trying to start xorg on atom z3735g tablet.

    I downloaded from your repository
    https://www.happyassassin.net/fedlet/repo/SRPMS/kernel-3.18.0-1.1awb.src.rpm

    I extracted .rpm files and used linux-3.18.tar.xz.
    I built the kernel with debian tool make-kpkg (config from debian)

    Now when drm module is loaded the screen become blank, Is it baytrail issue?

    Writing here I saw in log “[drm] Memory usable by graphics device = 2048M” but my tablet have only 1GbytesRAM.
    I tested modinfo drm and i915 but there aren’t ram size setting parameter, how to adjust?

    Thanks

    ——–
    info (sorry for the disorder, it may help):
    Tablet is Mediacom iPro800 (derived from CHUWI VX8 3G) specifications:
    ATOM Z3735G
    1Gbytes RAM
    8Gbytes ROM
    DISPLAY IPS 1280×800
    touchscreen GSLX680
    wifi rtl87xx
    sensors: unknown at the moment
    battery adc: unknown at the moment

    logs after modprobe i915 (it load drm.ko):
    [ 291.821842] [drm] Memory usable by graphics device = 2048M
    [ 291.822100] checking generic (40000000 3e8000) vs hw (40000000 10000000)
    [ 291.822112] fb: switching to inteldrmfb from simple
    [ 291.822583] Console: switching to colour dummy device 80×25
    [ 291.823154] [drm] Replacing VGA console driver
    [ 291.828853] i915 0000:00:02.0: irq 88 for MSI/MSI-X
    [ 291.828910] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
    [ 291.828935] [drm] Driver supports precise vblank timestamp query.
    [ 291.829474] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=io+mem,decodes=io+mem:owns=io+mem
    [ 291.893132] ACPI: Video Device [GFX0] (multi-head: yes rom: no post: no)
    [ 291.894382] input: Video Bus as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/LNXVIDEO:00/input/input5
    [ 291.895479] [drm] Initialized i915 1.6.0 20140905 for 0000:00:02.0 on minor 0
    [ 291.953972] [drm] GMBUS [i915 gmbus dpb] timed out, falling back to bit banging on pin 5
    [ 292.037833] [drm] GMBUS [i915 gmbus dpc] timed out, falling back to bit banging on pin 4
    [ 292.041324] fbcon: inteldrmfb (fb0) is primary device
    [ 293.107840] Console: switching to colour frame buffer device 100×80
    [ 293.142077] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
    [ 293.142338] i915 0000:00:02.0: registered panic notifier

    log of X (screen remains blank, exited with ctrl-c)
    [ 404.560]
    X.Org X Server 1.12.4
    Release Date: 2012-08-27
    [ 404.560] X Protocol Version 11, Revision 0
    [ 404.560] Build Operating System: Linux 3.11-2-amd64 x86_64 Debian
    [ 404.561] Current Operating System: Linux abox 3.18.0 #3 SMP Wed Dec 17 08:52:34 CET 2014 x86_64
    [ 404.561] Kernel command line: \boot\vmlinuz-3.18.0 ro root=UUID=3852a863-8b01-4bf1-a45a-1eb88cd9b242 initrd=boot\initrd.img-3.18.0
    [ 404.561] Build Date: 17 December 2013 07:37:58PM
    [ 404.561] xorg-server 2:1.12.4-6+deb7u2 (Julien Cristau )
    [ 404.562] Current version of pixman: 0.26.0
    [ 404.562] Before reporting problems, check http://wiki.x.org
    to make sure that you have the latest version.
    [ 404.562] Markers: (–) probed, (**) from config file, (==) default setting,
    (++) from command line, (!!) notice, (II) informational,
    (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
    [ 404.563] (==) Log file: “/var/log/Xorg.0.log”, Time: Wed Dec 17 12:56:38 2014
    [ 404.564] (==) Using system config directory “/usr/share/X11/xorg.conf.d”
    [ 404.565] (==) No Layout section. Using the first Screen section.
    [ 404.565] (==) No screen section available. Using defaults.
    [ 404.565] (**) |–>Screen “Default Screen Section” (0)
    [ 404.565] (**) | |–>Monitor “”
    [ 404.567] (==) No device specified for screen “Default Screen Section”.
    Using the first device section listed.
    [ 404.567] (**) | |–>Device “Intel Graphics”
    [ 404.567] (==) No monitor specified for screen “Default Screen Section”.
    Using a default monitor configuration.
    [ 404.567] (==) Automatically adding devices
    [ 404.567] (==) Automatically enabling devices
    [ 404.567] (WW) The directory “/usr/share/fonts/X11/cyrillic” does not exist.
    [ 404.567] Entry deleted from font path.
    [ 404.567] (WW) The directory “/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType” does not exist.
    [ 404.568] Entry deleted from font path.
    [ 404.568] (==) FontPath set to:
    /usr/share/fonts/X11/misc,
    /usr/share/fonts/X11/100dpi/:unscaled,
    /usr/share/fonts/X11/75dpi/:unscaled,
    /usr/share/fonts/X11/Type1,
    /usr/share/fonts/X11/100dpi,
    /usr/share/fonts/X11/75dpi,
    built-ins
    [ 404.568] (==) ModulePath set to “/usr/lib/xorg/modules”
    [ 404.568] (II) The server relies on udev to provide the list of input devices.
    If no devices become available, reconfigure udev or disable AutoAddDevices.
    [ 404.568] (II) Loader magic: 0x7f1f861c9ae0
    [ 404.568] (II) Module ABI versions:
    [ 404.568] X.Org ANSI C Emulation: 0.4
    [ 404.568] X.Org Video Driver: 12.1
    [ 404.568] X.Org XInput driver : 16.0
    [ 404.568] X.Org Server Extension : 6.0
    [ 404.570] (–) PCI:*(0:0:2:0) 8086:0f31:8086:7270 rev 15, Mem @ 0x50400000/4194304, 0x40000000/268435456, I/O @ 0x00001000/8
    [ 404.570] (II) Open ACPI successful (/var/run/acpid.socket)
    [ 404.570] (II) LoadModule: “extmod”
    [ 404.571] (II) Loading /usr/lib/xorg/modules/extensions/libextmod.so
    [ 404.573] (II) Module extmod: vendor=”X.Org Foundation”
    [ 404.573] compiled for 1.12.4, module version = 1.0.0
    [ 404.573] Module class: X.Org Server Extension
    [ 404.573] ABI class: X.Org Server Extension, version 6.0
    [ 404.573] (II) Loading extension SELinux
    [ 404.573] (II) Loading extension MIT-SCREEN-SAVER
    [ 404.574] (II) Loading extension XFree86-VidModeExtension
    [ 404.574] (II) Loading extension XFree86-DGA
    [ 404.574] (II) Loading extension DPMS
    [ 404.574] (II) Loading extension XVideo
    [ 404.574] (II) Loading extension XVideo-MotionCompensation
    [ 404.574] (II) Loading extension X-Resource
    [ 404.574] (II) LoadModule: “dbe”
    [ 404.574] (II) Loading /usr/lib/xorg/modules/extensions/libdbe.so
    [ 404.575] (II) Module dbe: vendor=”X.Org Foundation”
    [ 404.575] compiled for 1.12.4, module version = 1.0.0
    [ 404.575] Module class: X.Org Server Extension
    [ 404.575] ABI class: X.Org Server Extension, version 6.0
    [ 404.575] (II) Loading extension DOUBLE-BUFFER
    [ 404.575] (II) LoadModule: “glx”
    [ 404.576] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
    [ 404.576] (II) Module glx: vendor=”X.Org Foundation”
    [ 404.577] compiled for 1.12.4, module version = 1.0.0
    [ 404.577] ABI class: X.Org Server Extension, version 6.0
    [ 404.577] (==) AIGLX enabled
    [ 404.577] (II) Loading extension GLX
    [ 404.577] (II) LoadModule: “record”
    [ 404.577] (II) Loading /usr/lib/xorg/modules/extensions/librecord.so
    [ 404.578] (II) Module record: vendor=”X.Org Foundation”
    [ 404.578] compiled for 1.12.4, module version = 1.13.0
    [ 404.578] Module class: X.Org Server Extension
    [ 404.578] ABI class: X.Org Server Extension, version 6.0
    [ 404.578] (II) Loading extension RECORD
    [ 404.578] (II) LoadModule: “dri”
    [ 404.579] (II) Loading /usr/lib/xorg/modules/extensions/libdri.so
    [ 404.580] (II) Module dri: vendor=”X.Org Foundation”
    [ 404.580] compiled for 1.12.4, module version = 1.0.0
    [ 404.580] ABI class: X.Org Server Extension, version 6.0
    [ 404.580] (II) Loading extension XFree86-DRI
    [ 404.580] (II) LoadModule: “dri2″
    [ 404.580] (II) Loading /usr/lib/xorg/modules/extensions/libdri2.so
    [ 404.581] (II) Module dri2: vendor=”X.Org Foundation”
    [ 404.581] compiled for 1.12.4, module version = 1.2.0
    [ 404.581] ABI class: X.Org Server Extension, version 6.0
    [ 404.581] (II) Loading extension DRI2
    [ 404.581] (II) LoadModule: “intel”
    [ 404.582] (II) Loading /usr/lib/xorg/modules/drivers/intel_drv.so
    [ 404.583] (II) Module intel: vendor=”X.Org Foundation”
    [ 404.583] compiled for 1.12.4, module version = 2.99.916
    [ 404.583] Module class: X.Org Video Driver
    [ 404.583] ABI class: X.Org Video Driver, version 12.1
    [ 404.583] (II) intel: Driver for Intel(R) Integrated Graphics Chipsets:
    i810, i810-dc100, i810e, i815, i830M, 845G, 854, 852GM/855GM, 865G,
    915G, E7221 (i915), 915GM, 945G, 945GM, 945GME, Pineview GM,
    Pineview G, 965G, G35, 965Q, 946GZ, 965GM, 965GME/GLE, G33, Q35, Q33,
    GM45, 4 Series, G45/G43, Q45/Q43, G41, B43
    [ 404.586] (II) intel: Driver for Intel(R) HD Graphics: 2000-6000
    [ 404.586] (II) intel: Driver for Intel(R) Iris(TM) Graphics: 5100, 6100
    [ 404.586] (II) intel: Driver for Intel(R) Iris(TM) Pro Graphics: 5200, 6200, P6300
    [ 404.586] (–) using VT number 7

    [ 404.737] (II) intel(0): Using Kernel Mode Setting driver: i915, version 1.6.0 20140905
    [ 404.737] (II) intel(0): SNA compiled from 2.99.916-177-g9953708
    [ 404.972] (–) intel(0): Integrated Graphics Chipset: Intel(R) HD Graphics
    [ 404.972] (–) intel(0): CPU: x86-64, sse2, sse3, ssse3, sse4.1, sse4.2
    [ 404.972] (II) intel(0): Creating default Display subsection in Screen section
    “Default Screen Section” for depth/fbbpp 24/32
    [ 404.972] (==) intel(0): Depth 24, (–) framebuffer bpp 32
    [ 404.972] (==) intel(0): RGB weight 888
    [ 404.972] (==) intel(0): Default visual is TrueColor
    [ 404.972] (**) intel(0): Option “DRI” “False”
    [ 404.974] (II) intel(0): Output HDMI1 has no monitor section
    [ 404.974] (II) intel(0): Enabled output HDMI1
    [ 404.974] (II) intel(0): Output DP1 has no monitor section
    [ 404.974] (II) intel(0): Enabled output DP1
    [ 404.974] (II) intel(0): Output HDMI2 has no monitor section
    [ 404.974] (II) intel(0): Enabled output HDMI2
    [ 404.974] (II) intel(0): Output DP2 has no monitor section
    [ 404.975] (II) intel(0): Enabled output DP2
    [ 404.975] (II) intel(0): Output DSI1 has no monitor section
    [ 404.975] (II) intel(0): Enabled output DSI1
    [ 404.975] (–) intel(0): Using a maximum size of 256×256 for hardware cursors
    [ 404.975] (II) intel(0): Output VIRTUAL1 has no monitor section
    [ 404.975] (II) intel(0): Enabled output VIRTUAL1
    [ 404.975] (–) intel(0): Output DSI1 using initial mode 800×1280 on pipe 0
    [ 404.976] (==) intel(0): TearFree disabled
    [ 404.976] (==) intel(0): DPI set to (96, 96)
    [ 404.977] (==) Depth 24 pixmap format is 32 bpp
    [ 405.420] (II) intel(0): SNA initialized with Baytrail (gen7) backend
    [ 405.420] (==) intel(0): Backing store disabled
    [ 405.420] (==) intel(0): Silken mouse enabled
    [ 405.491] (II) intel(0): HW Cursor enabled
    [ 405.491] (II) intel(0): RandR 1.2 enabled, ignore the following RandR disabled message.
    [ 405.492] (==) intel(0): DPMS enabled
    [ 405.493] (–) RandR disabled
    [ 405.493] (II) Initializing built-in extension Generic Event Extension
    [ 405.493] (II) Initializing built-in extension SHAPE
    [ 405.493] (II) Initializing built-in extension MIT-SHM
    [ 405.493] (II) Initializing built-in extension XInputExtension
    [ 405.493] (II) Initializing built-in extension XTEST
    [ 405.493] (II) Initializing built-in extension BIG-REQUESTS
    [ 405.493] (II) Initializing built-in extension SYNC
    [ 405.493] (II) Initializing built-in extension XKEYBOARD
    [ 405.493] (II) Initializing built-in extension XC-MISC
    [ 405.493] (II) Initializing built-in extension SECURITY
    [ 405.493] (II) Initializing built-in extension XINERAMA
    [ 405.493] (II) Initializing built-in extension XFIXES
    [ 405.494] (II) Initializing built-in extension RENDER
    [ 405.494] (II) Initializing built-in extension RANDR
    [ 405.494] (II) Initializing built-in extension COMPOSITE
    [ 405.494] (II) Initializing built-in extension DAMAGE
    [ 405.494] (II) SELinux: Disabled on system
    [ 405.634] (II) AIGLX: Screen 0 is not DRI2 capable
    [ 405.634] (II) AIGLX: Screen 0 is not DRI capable
    [ 413.184] (II) AIGLX: Loaded and initialized swrast
    [ 413.184] (II) GLX: Initialized DRISWRAST GL provider for screen 0
    [ 413.334] (II) intel(0): switch to mode 800×1280@60.1 on DSI1 using pipe 0, position (0, 0), rotation normal, reflection none
    [ 413.469] (II) intel(0): Setting screen physical size to 211 x 338
    [ 415.034] (II) config/udev: Adding input device Power Button (/dev/input/event2)
    [ 415.034] (**) Power Button: Applying InputClass “evdev keyboard catchall”
    [ 415.034] (II) LoadModule: “evdev”
    [ 415.035] (II) Loading /usr/lib/xorg/modules/input/evdev_drv.so
    [ 415.089] (II) Module evdev: vendor=”X.Org Foundation”
    [ 415.089] compiled for 1.12.1, module version = 2.7.0
    [ 415.089] Module class: X.Org XInput Driver
    [ 415.089] ABI class: X.Org XInput driver, version 16.0
    [ 415.089] (II) Using input driver ‘evdev’ for ‘Power Button’
    [ 415.090] (**) Power Button: always reports core events
    [ 415.090] (**) evdev: Power Button: Device: “/dev/input/event2”
    [ 415.090] (–) evdev: Power Button: Vendor 0 Product 0x1
    [ 415.090] (–) evdev: Power Button: Found keys
    [ 415.090] (II) evdev: Power Button: Configuring as keyboard
    [ 415.090] (**) Option “config_info” “udev:/sys/devices/LNXSYSTM:00/LNXPWRBN:00/input/input2/event2”
    [ 415.090] (II) XINPUT: Adding extended input device “Power Button” (type: KEYBOARD, id 6)
    [ 415.090] (**) Option “xkb_rules” “evdev”
    [ 415.091] (**) Option “xkb_model” “pc105”
    [ 415.091] (**) Option “xkb_layout” “us”
    [ 415.094] (II) config/udev: Adding input device Video Bus (/dev/input/event5)
    [ 415.094] (**) Video Bus: Applying InputClass “evdev keyboard catchall”
    [ 415.094] (II) Using input driver ‘evdev’ for ‘Video Bus’
    [ 415.095] (**) Video Bus: always reports core events
    [ 415.095] (**) evdev: Video Bus: Device: “/dev/input/event5”
    [ 415.095] (–) evdev: Video Bus: Vendor 0 Product 0x6
    [ 415.095] (–) evdev: Video Bus: Found keys
    [ 415.095] (II) evdev: Video Bus: Configuring as keyboard
    [ 415.095] (**) Option “config_info” “udev:/sys/devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/LNXVIDEO:00/input/input5/event5”
    [ 415.095] (II) XINPUT: Adding extended input device “Video Bus” (type: KEYBOARD, id 7)
    [ 415.095] (**) Option “xkb_rules” “evdev”
    [ 415.096] (**) Option “xkb_model” “pc105”
    [ 415.096] (**) Option “xkb_layout” “us”
    [ 415.099] (II) config/udev: Adding input device Barcode Reader (/dev/input/event0)
    [ 415.099] (**) Barcode Reader : Applying InputClass “evdev keyboard catchall”
    [ 415.100] (II) Using input driver ‘evdev’ for ‘Barcode Reader ‘
    [ 415.100] (**) Barcode Reader : always reports core events
    [ 415.100] (**) evdev: Barcode Reader : Device: “/dev/input/event0”
    [ 415.100] (–) evdev: Barcode Reader : Vendor 0x13ba Product 0x18
    [ 415.100] (–) evdev: Barcode Reader : Found keys
    [ 415.100] (II) evdev: Barcode Reader : Configuring as keyboard
    [ 415.100] (**) Option “config_info” “udev:/sys/devices/pci0000:00/0000:00:14.0/usb1/1-1/1-1.4/1-1.4:1.0/0003:13BA:0018.0001/input/input0/event0”
    [ 415.101] (II) XINPUT: Adding extended input device “Barcode Reader ” (type: KEYBOARD, id 8)
    [ 415.101] (**) Option “xkb_rules” “evdev”
    [ 415.101] (**) Option “xkb_model” “pc105”
    [ 415.101] (**) Option “xkb_layout” “us”
    [ 415.105] (II) config/udev: Adding input device Barcode Reader (/dev/input/event1)
    [ 415.105] (**) Barcode Reader : Applying InputClass “evdev pointer catchall”
    [ 415.105] (**) Barcode Reader : Applying InputClass “evdev keyboard catchall”
    [ 415.105] (II) Using input driver ‘evdev’ for ‘Barcode Reader ‘
    [ 415.105] (**) Barcode Reader : always reports core events
    [ 415.105] (**) evdev: Barcode Reader : Device: “/dev/input/event1”
    [ 415.105] (–) evdev: Barcode Reader : Vendor 0x13ba Product 0x18
    [ 415.106] (–) evdev: Barcode Reader : Found 9 mouse buttons
    [ 415.106] (–) evdev: Barcode Reader : Found scroll wheel(s)
    [ 415.106] (–) evdev: Barcode Reader : Found relative axes
    [ 415.106] (–) evdev: Barcode Reader : Found x and y relative axes
    [ 415.106] (–) evdev: Barcode Reader : Found absolute axes
    [ 415.106] (II) evdev: Barcode Reader : Forcing absolute x/y axes to exist.
    [ 415.106] (–) evdev: Barcode Reader : Found keys
    [ 415.106] (II) evdev: Barcode Reader : Configuring as mouse
    [ 415.106] (II) evdev: Barcode Reader : Configuring as keyboard
    [ 415.106] (II) evdev: Barcode Reader : Adding scrollwheel support
    [ 415.106] (**) evdev: Barcode Reader : YAxisMapping: buttons 4 and 5
    [ 415.106] (**) evdev: Barcode Reader : EmulateWheelButton: 4, EmulateWheelInertia: 10, EmulateWheelTimeout: 200
    [ 415.107] (**) Option “config_info” “udev:/sys/devices/pci0000:00/0000:00:14.0/usb1/1-1/1-1.4/1-1.4:1.1/0003:13BA:0018.0002/input/input1/event1”
    [ 415.107] (II) XINPUT: Adding extended input device “Barcode Reader ” (type: KEYBOARD, id 9)
    [ 415.107] (**) Option “xkb_rules” “evdev”
    [ 415.107] (**) Option “xkb_model” “pc105”
    [ 415.107] (**) Option “xkb_layout” “us”
    [ 415.109] (II) evdev: Barcode Reader : initialized for relative axes.
    [ 415.109] (WW) evdev: Barcode Reader : ignoring absolute axes.
    [ 415.110] (**) Barcode Reader : (accel) keeping acceleration scheme 1
    [ 415.110] (**) Barcode Reader : (accel) acceleration profile 0
    [ 415.110] (**) Barcode Reader : (accel) acceleration factor: 2.000
    [ 415.110] (**) Barcode Reader : (accel) acceleration threshold: 4
    [ 415.112] (II) config/udev: Adding input device Barcode Reader (/dev/input/mouse0)
    [ 415.112] (II) No input driver specified, ignoring this device.
    [ 415.112] (II) This device may have been added with another device file.
    [ 415.118] (II) config/udev: Adding input device PC Speaker (/dev/input/event3)
    [ 415.118] (II) No input driver specified, ignoring this device.
    [ 415.118] (II) This device may have been added with another device file.
    [ 415.122] (II) config/udev: Adding input device ACPI Virtual Keyboard Device (/dev/input/event4)
    [ 415.122] (**) ACPI Virtual Keyboard Device: Applying InputClass “evdev keyboard catchall”
    [ 415.122] (II) Using input driver ‘evdev’ for ‘ACPI Virtual Keyboard Device’
    [ 415.122] (**) ACPI Virtual Keyboard Device: always reports core events
    [ 415.122] (**) evdev: ACPI Virtual Keyboard Device: Device: “/dev/input/event4”
    [ 415.122] (–) evdev: ACPI Virtual Keyboard Device: Vendor 0 Product 0
    [ 415.122] (–) evdev: ACPI Virtual Keyboard Device: Found keys
    [ 415.122] (II) evdev: ACPI Virtual Keyboard Device: Configuring as keyboard
    [ 415.123] (**) Option “config_info” “udev:/sys/devices/virtual/input/input4/event4”
    [ 415.123] (II) XINPUT: Adding extended input device “ACPI Virtual Keyboard Device” (type: KEYBOARD, id 10)
    [ 415.123] (**) Option “xkb_rules” “evdev”
    [ 415.123] (**) Option “xkb_model” “pc105”
    [ 415.123] (**) Option “xkb_layout” “us”
    [ 420.424] (II) evdev: Power Button: Close
    [ 420.424] (II) UnloadModule: “evdev”
    [ 420.440] (II) evdev: Video Bus: Close
    [ 420.440] (II) UnloadModule: “evdev”
    [ 420.468] (II) evdev: Barcode Reader : Close
    [ 420.468] (II) UnloadModule: “evdev”
    [ 420.500] (II) evdev: Barcode Reader : Close
    [ 420.500] (II) UnloadModule: “evdev”
    [ 420.532] (II) evdev: ACPI Virtual Keyboard Device: Close
    [ 420.532] (II) UnloadModule: “evdev”
    [ 420.599] Server terminated successfully (0). Closing log file.

    1. vitaly.torshyn
      vitaly.torshyn December 17, 2014 at 5:18 am | | Reply

      Hi Mario
      The size of video mem. for Intel’s GMA can’t be configured with kernel parameters. Check your UEFI firmware/BIOS for such options if present.
      If your screen is blank for X, try to disable modeset by appending ‘nomodeset’ to kernel parameters. In this case, X will try to detect correct mode. If it still not working, additionally add ‘video=…’ with correct values for your device.

      In addition, you could try to play with “VideoRam [integer]” property in xorg.conf inside device section.

      I can confirm that Xorg can be started (but slow as hell) with Adams kernel only if ‘nomodeset video=VGA-1:800x1280e’ is passed on Lenovo Miix 2 8 (64Gb) device.

    2. vitaly.torshyn
      vitaly.torshyn December 17, 2014 at 5:26 am | | Reply

      Also, I would pay attention on the following lines:

      [ 404.972] (**) intel(0): Option “DRI” “False”

      [ 405.634] (II) AIGLX: Screen 0 is not DRI2 capable
      [ 405.634] (II) AIGLX: Screen 0 is not DRI capable

      In addition your configuration is slightly outdated. See https://01.org/linuxgraphics/downloads/2014/2014q3-intel-graphics-stack-release for BayTrail graphics stack release notes.

  14. Luis
    Luis December 17, 2014 at 10:30 am | | Reply

    Made the experience in toshiba Encore 2 wt10-A32 works well, but the wifi does not work. Do you have any suggestion to solve the problem? will have to do with anything in the bios boot EUFI?

  15. Mario
    Mario December 17, 2014 at 10:01 pm | | Reply

    With nomodeset X does not start:

    [ 101.895] (EE) No devices detected.
    Fatal server error:
    [ 101.895] no screens found

    VideoRam 100000 doesn’t change anything.
    I add, when X is running (blank screen) emits following error:
    [intel_init_bufmgr:584] Error initializing buffer manager.

    Considered:
    menu UEFI/BIOS doesn’t have setting for graphics ram size.
    graphics patch should already be in the upstream kernel 3.18.0.

    I’m thinking the output video is wrong or framebuffer don’t work or binaries are too old.

    Thanks for help. If I have time will try to install fedlet on usb hdd and test (for sure touchscreen differ)

  16. Verrm
    Verrm December 19, 2014 at 5:42 am | | Reply

    Hey, did you take a notice of a fresh new product MeeGo Pad T01? It is a ultra-mini PC and rocks on Quad Core Z3735F. You can check it out on geekbuying.com. I have a dream of having such mini-pcs being workstations in small-medium sized offices. Of course centralized with FreeIPA or Zentyal. What do you think? Will the Fedlet work on this? I think it should, but I am worried about the Wifi/BT: The RTL8723BS combo chipset is reported to have wifi working for someone, but I’m not sure. BT is also an issue there, but not many people utilize it anyway.

    1. jekader
      jekader December 22, 2014 at 2:29 am | | Reply

      RTL8723BS worked for me after I built the driver from here: https://github.com/hadess/rtl8723as/
      As for Bluetooth – who would even use it in an office environment?

      The device you mentioned does not have a touchscreen, accelerometer, backlight control. So it should mostly work, just ACPI may be a problem, so sleep functions better be disabled on it.

  17. Mikhail
    Mikhail December 22, 2014 at 10:00 pm | | Reply

    Adam, is correct to write about fedlet problems at the official bugzilla Fedora?

  18. boxed
    boxed December 23, 2014 at 6:58 am | | Reply

    Hi, thanks for building a tablet distribution. Seems no other distribution has anything out at all.

    I am using Gentoo as a base.

    I’m trying to build a custom kernel for my Nextbook 8 Windows8.1 (1GB RAM, Atom Z3735). I was able to get into FW setup and verified secure boot is off. I was able to build Grub with UEFI32 support and have verified that my i386-efi grub-mkimage output seems to be able to run on the tablet though I don’t think I have all the disk options set quite right yet. In the meantime I am using Fedlet’s grub as that I know is working now (but my tablet seems to not like my USB DVDRW drive).

    All my attempts at building a Linux kernel from 3.18.1 stock down to 3.12.35 all result in the tablet hanging. I built with EFI Stub support. When I run this in grub:

    linuxefi /efi/my-32bit-3.18.1.efi
    boot

    it just hangs there with no additional terminal output.

    If I do the same experiment with the Fedlet kernel, it actually boots (and hangs because I did not specify the root partition or initramfs) – this is what I expect with my custom built kernel too. Any ideas why the Fedlet kernel would get penguins to show up but not my kernel? Is this still a signing/secureboot problem?

    1. boxed
      boxed December 28, 2014 at 7:05 pm | | Reply

      I finally got the stock 3.18.1 kernel to get to a command prompt on my tablet. KMS is the problem. When KMS sets up the display, the screen goes blank but machine continues to run. I had my userland to boot Gnome through gdm and autologin but the screen remains dark.

      I also have an RTL8723BS and was able to get wifi working, and able to ssh in remotely. Using this, I was able to try xrandr on the running X11 session. It reports that DSI1 is at 800×1280… so at least X/xrandr appears to detect the right display but it’s not using it properly.

      Amongst the continuing challenges thoug wifi is up, I can disconnect the USB, but when I try to plug it in to a USB port, it appears to not want to charge. Anyone have this working? Granted getting USB-gadget working would be nice, but charging while the machine is running would be good …

      1. boxed
        boxed December 29, 2014 at 1:37 pm | | Reply

        I think now I got my distribution to be about set up as any other. I’m using 3.18.1-stock Linux kernel, Gentoo Userland (systemd)

        Working on my Nextbook8 Windows 8.1:

        RTL8723BS 802.11b/g/n wifi
        X11 eventually starts up with video=DSI-1
        USB2, using xHCI-HCD driver with EHCI-HCD. Building a kernel without xHCI-HCD results in usb dead.
        eMMC internal disk and eMMC MicroSD port (very slow!)

        Problems I see on my Nextbook8 Windows 8.1:

        Initial EFI FB does not work.
        Screen goes black when starting KMS, but eventually it starts into X11 and then Gnome. Seems to work with external mouse and keyboard.
        HPET does not work! (see one warning and that’s the end of it) – clock runs slow, timing is slow due to clocksource=refined-jiffies. Setting to TSC returns very unpredictable timing behavior.
        Touchscreen does not work. No touchscreen identified in Xorg.log
        Sound is not working but I must get HPET or clocksource working first anyway as media players need that working.
        Battery reports 100% all of the time.
        Battery does not charge after booting USB as OTG/host mode even after disconnecting and hooking to charger, also need to set up device/gadget mode.
        Bluetooth/FM radio has no driver (RTL8723BS side functions)
        Accelerometer has not been detected.

        It looks most of these peripherals are I2C/SPI/??? devices. Anyone know of a way to identify these devices?

        1. boteium
          boteium April 24, 2015 at 9:25 pm | | Reply

          Charging works now.
          Build linux-next-git 20140423 (linux 4.1)

          CONFIG_AXP288_FUEL_GAUGE=y

          I didn’t test if the driver is fully working.
          But if the power cable is plugged in before booting and stay plugged-in, it works

          These PMIC related parameter seems functioning :
          /sys/devices/platform/80860F41:04/i2c-4/i2c-INT33F4:00/axp288_adc/iio:device0/

          “in_current3_input” seems to be current drawn from battery.
          “in_current4_input” seems to be current drawn form power cable.

          If it’s charging, current3 would be zero.

          Battery status isn’t working yet.

          1. boteium
            boteium April 24, 2015 at 10:36 pm | | Reply

            correction : current3 should be current drawn from power cable and vice versa

  19. Niranjan
    Niranjan December 23, 2014 at 11:50 am | | Reply

    hey do you know of intel tablets which have 64-bit UEFI on it so that i could install fedora on it ?

  20. Calvin Walton
    Calvin Walton December 23, 2014 at 2:22 pm | | Reply

    Just want to join in; I tried this out on my Acer Iconia W4 and, well, it booted. Generally quite slow, and a bunch of mmc errors in the kernel log; the installer crashes at startup saying it couldn’t scan disks or something to that effect.

    At the moment, I’m trying to get my own kernel to boot on it, loading a 64bit kernel via grub. I haven’t gotten any console output yet, so still poking at it.

  21. Kevo
    Kevo December 25, 2014 at 10:30 pm | | Reply

    the download doesn’t report its size, so its hard to tell how long it will take. for the record, its 1.26 GB. search tags: gb mb size filesize. hope this helps.

  22. Andreas Lorensen
    Andreas Lorensen December 29, 2014 at 12:43 pm | | Reply

    Lenovo Miix2 tablet:

    Boots fine, but after 2-3 min screen goes black and nothing works.

    Have tried lots of different boot parameters whitout any result. Did also make a Fedora LiveCD on a USB and tweaked it to boot uefi32bit, and same happend.

    If i shutdown gdm same result.

    Anyone have any idea for what is wrong or how I can locate what is going on?

    1. Markus Larsson
      Markus Larsson December 29, 2014 at 1:26 pm | | Reply

      I have the same problems, not sure why. Could be related to the problems with the eMMC. Not sure though. I’m doing some fiddling about when I have time and have concluded that I see different behavior with different kernels (same problem different intervals and such) and with different versions of Gnome3. Still haven’t narrowed it down though.

  23. Andreas Lorensen
    Andreas Lorensen December 29, 2014 at 1:57 pm | | Reply

    Wow – i think I found it – it seems that systemd-login is suspending because it thinks a botton is pressed….So I edited /etc/systemd/login.conf and tried to change what could be the reason.

    Now I boot and can do install whitout black screen.

    /etc/systemd/login.conf
    [Login]

    NAutoVTs=6

    ReserveVT=6

    KillUserProcesses=no

    KillOnlyUsers=

    KillExcludeUsers=root

    Controllers=

    ResetControllers=cpu

    InhibitDelayMaxSec=5

    HandlePowerKey=poweroff
    HandleSuspendKey=ignore
    HandleHibernateKey=ignore
    HandleLidSwitch=ignore
    PowerKeyIgnoreInhibited=yes
    SuspendKeyIgnoreInhibited=yes
    HibernateKeyIgnoreInhibited=yes
    LidSwitchIgnoreInhibited=yes

    IdleAction=ignore

    IdleActionSec=30min

    1. Markus Larsson
      Markus Larsson December 29, 2014 at 5:06 pm | | Reply

      Confirmed! My hat is off to you good sir, well played.
      Seems I was barking up the wrong tree all along.

  24. Andreas Lorensen
    Andreas Lorensen December 30, 2014 at 7:54 am | | Reply

    Well I think thre is lots of trees to barking up in….

    Lenovo Miix2 tablet:

    Boots and install whit changing /etc/systemd/login.conf
    But still after some time it freezes – maybe when doing network trafic as yum update.

    Powerindicator dosnt show correct time left.

    Hasnt tryed sound yet

    1. Markus Larsson
      Markus Larsson December 30, 2014 at 9:05 am | | Reply

      I see the freezes too – it seems yum is triggering it as you said but I believe it has to do with the eMMC problems of not being able to write to disk from time to time. I’ve been watching the logs when going in to a freeze a couple of times and the two things complaining is the network not being able to access data and write operations not getting the correct response (or any response) from the mmc.
      I’ll dig further later tonight when the old wifey and son goes to bed.

  25. boxed
    boxed January 1, 2015 at 11:09 am | | Reply

    Just something amusing with my TSC and HPET not working, and stuck with clocksource=refined-jiffies which appears also to be wrong:

    busybox hdparm -tT /dev/mmcblk0

    /dev/mmcblk0:
    Timing buffer-cache reads: 6538 MB in 0.51 seconds = 13042857 kB/s
    Timing buffered disk reads: 3185 MB in 3.00 seconds = 1086010 kB/s

    busybox hdparm -tT /dev/mmcblk1

    /dev/mmcblk1:
    Timing buffer-cache reads: 6902 MB in 0.51 seconds = 13769012 kB/s
    Timing buffered disk reads: 352 MB in 3.00 seconds = 119757 kB/s

    I dont think so.

    The mmcblk1 is a Class 4 MicroSD card, mmcblk0 is the onboard eMMC.

    1. jekader
      jekader January 2, 2015 at 10:04 am | | Reply

      Just force tsc by adding “clocksource=tsc” to the kernel cmdline. This worked for me and it stopped falling back to refined-jiffies.

      1. boxed
        boxed January 2, 2015 at 10:40 am | | Reply

        Doesn’t work for me, behavior is even worse with clocksource=tsc on my tablet. There was a warning

        [ 5.527929] Clocksource tsc unstable (delta = 433033193 ns)
        [ 5.839999] Switched to clocksource refined-jiffies

        Last time I tried forcing clocksource to tsc, it hung every so often, behavior seemed worse. Another thing I noticed with refined-jiffies is that the clock ticks start out fast when freshly booted but by the time it starts Gnome the ticks come very slowly.

        On a side note, trying Linux kernel 3.19-rc2 fixed my graphics problems – it seems to change video modes with KMS much better than 3.18.1 vanilla (I had to manually (and remotely through wifi) run systemctl stop gdm; wait 10 minutes; systemctl start gdm on 3.18.1 and it will get into graphics mode. On 3.19-rc2 it still takes 5 minutes but will get into Gnome on its own.)

        My touch panel however is the #1 problem I have with the tablet now. No driver seems to activate it, I have to use an external usb keyboard and mouse. I’m at a loss to even find what driver to use, I just loaded all of the ones in the kernel to no avail…

        1. jekader
          jekader January 2, 2015 at 11:57 am | | Reply

          Well let’s hope someone fixes ACPI and we get a working HPET on this chipset.

          1. boxed
            boxed January 2, 2015 at 4:31 pm | | Reply

            Yeah hope there’s a fix to it, likely will need to work around the ACPI problem. Reading the Z3600/Z3700 data sheet, it should implement a full HPET like any other x86 machine…

            The other possibility is that I should disable tickless timer in the kernel. Though this will suck up battery life, likely this will be sufficient to allow accurate timing.

            Also something interesting while glossing over the datasheet: While poking /proc and /sys I found something weird: this Z3600/3700 does not support S1/S3 suspend! However, if the low power S0 (with the CPU in C7 state) is also as low power as S3, then maybe it’s fine if it just needs to transition from/to C7 which should be fast (and according to the datasheet if it hits C7 the backlight should turn off). Getting in/out of S3 on my full sized machines are pretty slow (~5 seconds is slow) – I was really wondering about this anyway, you can’t really use S1/S3 suspend on a phone chip to conserve power as 5 seconds on an incoming call is bad, and having S0/C7 makes sense…

            and the more I can’t wait to see a full, non-Android Linux running on this thing!

  26. jekader
    jekader January 2, 2015 at 11:17 am | | Reply

    OFFTOPIC

    For everybody out there I just tried android-x84 4.4-R2 and it booted successfully (only thing I had to do is set the clocksource so that it doesn’t freeze soon after start). It’s got the 3.18 kernel so hardware support is pretty much the same: wifi does not work, screen does not rotate. Otherwise it looks surprisingly usable!

    Sure, it’s not Fedora, but still it’s the first build of Android which ran successfully on my tablet.

    Release notes here:
    http://www.android-x86.org/releases/releasenote-4-4-r2

    1. boxed
      boxed January 3, 2015 at 11:52 am | | Reply

      First off I retract the 3.19-rc2 kernel comment on it working on my tablet. On another boot attempt, KMS and X11 blanked the screen once more. The workaround of restarting gdm/Xorg after letting it initialize the first time around still works – but the key issue is that Xorg must completely die before restarting.

      For the lack of any additional leads on current issues, I tried downloading android-x86, which seems to run fine in graphics mode. Alas this is not X so not the same thing, but curiously it should still be using KMS and draw on the FB directly. Android ended up crashing on me anyway after the “Welcome” screen, and it seems the OOM killer got its kill despite having 1GB RAM on my tablet. I think it also has problem with clocksource on my tablet as well just like my custom built kernels, so likely I need to get HPET working. But one thing that I got a quick glimpse of – my touch screen worked? Or did a timeout click the button to select a region for me…?

      1. jekader
        jekader January 3, 2015 at 12:32 pm | | Reply

        I used unetbootin to unpack the ISO to the flash drive, after which I booted it and it also died after Welocome. I then edited grub.cfg to include clocksource=tsc which is known to work best on my tablet. This made it boot, go through initial setup and show me the desktop which was quite usable. And yes, it also never went into the blank screen mode for me either.

        1. boxed
          boxed January 3, 2015 at 5:43 pm | | Reply

          And I think I end up have to rescind saying that Android works. It worked once and every subsequent boot I can’t seem to reproduce the console. Android blanks out the display as well.

          So the only piece of data I have for the Intel maintainers is that something in the Xorg exit routine does reset the Bay Trail graphics subsystem in such a way the next time KMS is invoked, it actually works.

          Alas I don’t know of any workaround for Android as I can’t ssh into it and peek at what’s going on… though I suppose I can nomodeset android too…

  27. jockyw2001
    jockyw2001 January 4, 2015 at 6:22 am | | Reply

    boxed: nomodeset doesn’t work for my voyo a1 mini with android, though it works fine with fedlet. Both with a 3.18 kernel. But I found a workaround to also boot android on my voyo which may also work for you: boot your tablet into firmware (“bios”) and then select boot from your usb device from the boot override menu. My voyo then boots the kernel in lowres resolution and the kms switch to native res then succeeds and android boots up in full glory. Changing boot order or rebooting from win 8.1 don’t work. Goodluck!

  28. Russell Senior
    Russell Senior January 5, 2015 at 5:06 am | | Reply

    I have a Toshiba encore 2 wt8 b264pro. I managed to boot fedlet but: a) have no Wi-Fi; and b) don’t see the NTFS filesystem or a recognizable block device beyond the live USB device. The windows storage device is described as a Toshiba 064GE. The Wi-Fi device is described as a Realtek RTL8723BS 802.11n SDIO Network Adapter.

  29. Russell Senior
    Russell Senior January 5, 2015 at 5:43 am | | Reply

    Aha! I found the windows storage, it’s on a mmcblock device. When the battery recharges, I will try installing to a microSD card.

  30. boxed
    boxed January 6, 2015 at 6:17 am | | Reply

    Anyone with a micro-usb “OTG” charge port (nonindependent charge port) have their tablets charging while running Linux?
    Should this work?

    I wish I could leave the tablet on to work on it remotely but if it won’t charge, this won’t work…

    1. boteium
      boteium February 13, 2015 at 9:10 am | | Reply

      it’s very possible and depend on the bios of your device.
      BUT it’s very likely that charging while using otg would only give you 2.5 watts. ( 5V * 500 mA ).
      it will only slower battery drain unless you drew less than 2.5 watts.

      1. boxed
        boxed February 13, 2015 at 2:23 pm | | Reply

        The problem is that once Linux is loaded, unless there’s some ACPI help, this doesn’t work. As far as I can tell if I have OTG enabled during boot by connecting an external device to boot off of, then disconnect the OTG cable and connect the charging cable, the table does not charge when in Linux.

        Only 2.5W would be great. The tablet “should” use less than 2.5W when idle and should net charge the battery.

        But there’s no reason why it couldn’t hit fast charge I would think, other than Linux not having the support?

  31. Russell Senior
    Russell Senior January 6, 2015 at 6:51 pm | | Reply

    I need clocksource=tsc also for the Toshiba Encore 2, also fedlet sees my 8gig microSD card as read-only, and the installer won’t recognize it.

  32. Pierre Bäfwer Balkhed
    Pierre Bäfwer Balkhed January 8, 2015 at 1:36 am | | Reply

    Unknown (please let me know!)

    Hardware support (wireless, bluetooth etc) on devices other than Venue 8 Pro

    Toshiba Encore 2, no wifi or bluetooth.

    Very responsive from live usb, but after a while slows down. Will try full install and report back when I’ve gotten the wifi and bluetooth to work (kinda pointless without).

    Greetings from Sweden and thanks for the all the work of putting fedora on bay tablets.

    \Pierre

  33. Russell Senior
    Russell Senior January 8, 2015 at 5:41 am | | Reply

    On Toshiba encore 2, after install to mmcblk0p3, with automatic partitioning, it boots and sort of works, except: a) my usb-wifi doesn’t work (the interface is present, but it doesn’t see networks, maybe a firmware blob is missing? curiously it worked from the live-usb). Another problem is that recovering the windows installation doesn’t seem to work. Maybe because the linux install is using the partition that windows recovery wants to use. Maybe manually restoring the windows partition would work. I ran out of battery, so I’ll have to let it recharge again to try more.

  34. Russell Senior
    Russell Senior January 8, 2015 at 12:57 pm | | Reply

    Hmm. after a recharge, the wifi dongle works again. Maybe just a power problem. The battery was pretty close to exhausted when I was having trouble with it.

  35. Lare
    Lare January 8, 2015 at 2:15 pm | | Reply

    I tried this on HP Pavilion x2 10″. The screen goes black when the KMS kicks in and I couldn’t get it to show anything even if I tried booting through different options

  36. Lare
    Lare January 8, 2015 at 2:45 pm | | Reply

    Follow up. I added ‘nomodeset’ to kernel commandline and the system booted. The gnome felt very sluggish. Shouldn’t the intel graphics drivers load even though the kernel skips the loading in the boot? Inside gnome, accelerometer, wifi, and display brightness options weren’t available and they didn’t work from the keyboard shortcut keys.

    Very nice work over all.

  37. Russell Senior
    Russell Senior January 8, 2015 at 9:00 pm | | Reply

    Looks like removing /dev/mmcblk0p3 and /dev/mmcblk0p5 was the wrong thing to do to test win8 recovery. Now in a boot loop, gives me the grub2 menu, selecting Windows chainloader shows a message that suggests it’s trying to recover/repair or something, then a long pause, and I get grub again. How do I get back to EFI USB boot from grub2?

    Any other unbricking suggestions?

  38. Russell Senior
    Russell Senior January 8, 2015 at 9:31 pm | | Reply

    From the grub2 commandline:

    set root=(hd0)
    chainloader /EFI/BOOT/BOOTIA32.efi
    boot

    let me boot from USB again.

  39. boxed
    boxed January 9, 2015 at 1:21 am | | Reply

    For fun I hacked my 3.19-rc3 kernel to ignore all methods to ignore HPET (both the quirk and the interrupt mismatch) — DO AT YOUR OWN RISK!!! Just like anything else on the Internet!

    Lo and behold, it used the HPET! I wonder what the accuracy problems are… this thing works much better with HPET than TSC or especially jiffies – the clock seems to be updating at the correct rate now, and CPU benchmarks seems to now match my other Atom.

    The behavior of the machine is still not great, not sure if it’s due to a crappy touch screen or what, but animations seem to work better and touch screen is actually considerably more usable.

    The RTL8723BS seems to go dead after a while, not sure why. No system crashes so far.

    1. Russell Senior
      Russell Senior January 14, 2015 at 5:47 am | | Reply

      Can you share your kernel .config? I’m looking for a starting place.

  40. Andreas Lorensen
    Andreas Lorensen January 9, 2015 at 3:07 am | | Reply

    I would love to test your kernel or description on how to do it myself.

    I have a Lenovo MIIX2 and defnetly have problems whith clocksource.

    1. boxed
      boxed January 9, 2015 at 9:39 am | | Reply

      I’ll get a diff patch out later today, I don’t have my code hacks handy but i thought it was fairly trivial – comment out the code that disables HPET when it detects (problems with) the hardware. Recompile is needed. I don’t know the differences between all the BYT SoCs but the one I’m using is the Z3735G, and YMMV.

      Is there some touch screen calibration tool? Perhaps it’s just an artifact of the cheap parts on my particular tablet (Nextbook 8 Windows) but it almost seems I have to press hard on spots to get my on screen keyboard to detect ‘key’ presses in Gnome. Definitely can’t type as fast on other touchscreen implementations I’ve tried.

      On a different note I was looking at Bluetooth. I did end up having a /dev/ttyS0 show up, could it be possible that bluetooth in RTL8723BS (which I think I saw somewhere has a “serial” or “UART” connection to BT instead of SDIO/I2C) is connected to the SoC serial port and thus use HCI-UART as the driver? Anyone got this working? Maybe this is enough ideas that someone could get theirs working? When I tried this hail-mary pass, it didn’t detect, but maybe there’s something simple I missed here or perhaps I’m just completely overthrowing the ball… Of course this would be machine specific for my tablet, but perhaps the reference is to hook it up this way?

      1. boxed
        boxed January 10, 2015 at 4:12 am | | Reply

        This is the patch I used… again, WARNING, do at your own risk. No idea why hpet was disabled to begin with.
        Perhaps the boot option

        hpet=force

        would make this unnecessary but I liked hacking the kernel…

        diff -ur linux-3.19-rc3-orig/arch/x86/kernel/early-quirks.c linux-3.19-rc3/arch/x86/kernel/early-quirks.c
        — linux-3.19-rc3-orig/arch/x86/kernel/early-quirks.c 2015-01-05 18:05:20.000000000 -0700
        +++ linux-3.19-rc3/arch/x86/kernel/early-quirks.c 2015-01-08 11:08:46.000000000 -0700
        @@ -583,8 +583,11 @@
        static void __init force_disable_hpet(int num, int slot, int func)
        {
        #ifdef CONFIG_HPET_TIMER
        +/*
        boot_hpet_disable = 1;
        pr_info(“x86/hpet: Will disable the HPET for this platform because it’s not reliable\n”);
        +*/
        + pr_info(“x86/hpet: WARNING: HPET was deemed unstable on this platform but hacked out to experiment with.\n”);
        #endif
        }

        diff -ur linux-3.19-rc3-orig/drivers/char/hpet.c linux-3.19-rc3/drivers/char/hpet.c
        — linux-3.19-rc3-orig/drivers/char/hpet.c 2015-01-05 18:05:20.000000000 -0700
        +++ linux-3.19-rc3/drivers/char/hpet.c 2015-01-07 01:12:59.000000000 -0700
        @@ -886,10 +886,13 @@
        ntimer = ((cap & HPET_NUM_TIM_CAP_MASK) >> HPET_NUM_TIM_CAP_SHIFT) + 1;

        if (hpetp->hp_ntimer != ntimer) {

        printk(KERN_WARNING “hpet: number irqs doesn’t agree”
        ” with number of timers\n”);
        kfree(hpetp);
        return -ENODEV;
        +// printk(KERN_WARNING “hpet: number irqs doesn’t agree”
        +// ” with number of timers\n”);
        printk(KERN_WARNING “hpet: number irqs (%d) doesn’t agree”
        ” with number of timers (%d)\n”,hpetp->hp_ntimer,ntimer);
        printk(KERN_WARNING “Dangerously Ignoring :-)\n”);
        +// kfree(hpetp);
        +// return -ENODEV;
        }

        if (last)

        ~

        1. Andreas Lorensen
          Andreas Lorensen January 10, 2015 at 12:17 pm | | Reply

          Super

          Can you or anybody point me in direction for reading about how to compile custom kernel on fedora? Last time I made a kernel was in late nineties on FreeBSD.

  41. Vaidotas
    Vaidotas January 10, 2015 at 6:56 am | | Reply

    Finally, working accelerated video on Lenovo Miix 2 8 tablet: https://groups.google.com/d/msg/android-x86/-i-D2C7hdZk/Oqw9SlscMXsJ
    Not really sure about details yet, but I think this hack might be used in Fedlet too.

    1. Vaidotas
      Vaidotas January 11, 2015 at 1:11 am | | Reply

      Scratch that – today video just refuses to work. Looks like that it is random – and usually video just won’t work in accelerated mode.

      1. boxed
        boxed January 12, 2015 at 8:09 am | | Reply

        I think I see the same behavior. About 1 in 20 tries the screen works on my tablet. Smells like incomplete initialization: see if this matches if you have network working, or see if you can script doing this:

        Once X11 fully starts up, tell the machine to shut down X completely so that X stops running. Then restart X/gdm.

        For me, I have my rtl8723bs working so I can ssh into my tablet, and systemctl stop gdm; <wait until X completely exits, it takes a while for my tablet for whatever reason…check ps -ax and look for X and wait until it’s gone>; systemctl start gdm .

        100% of the time, for me on my Z3735G/DSI-1 it will start X the second time around. As far as I can tell this is completely repeatable, it will always boot into the GUI on this second attempt.

  42. boxed
    boxed January 10, 2015 at 8:10 am | | Reply

    Ugh this is what’s annoying: More info of what’s going on with BT with the RTL8723BS.

    Supposedly it should use the “hci_h5.c” uart 3-wire driver released by Intel/GPL based on some webpages I saw. The stock kernel release BT HCI UART driver protocol “h5” apparently is different than another “rtk_h5” driver I found that was part of someones Android kernel (sunxi) but has Broadcomm copyright(Apache) – also “hci_h5.c” … Now I don’t know why there are two h5 drivers out there or whether they are supposed to be the same or not. In any case the Intel h5 protocol in the mainline kernel does not appear to work with the rtl8723bs. I wonder how much effort it will be to port the sunxi h5 back to mainline.

    Now why “rtk” and “broadcom” and “intel” – all different companies… unless realtek decided to rip off broadcom?

    1. boxed
      boxed January 10, 2015 at 11:38 am | | Reply

      Correction: https://github.com/menghang/a31_422_v33_android_external/tree/master/bluetooth/bluedroid seems to have the “alternate” h5 protocol driver… hci/src/hci_h5.c

    2. jekader
      jekader January 15, 2015 at 12:05 am | | Reply

      I must say I was also not expecting this many unsupported and non-working devices when bying an intel tablet. And the thing is that drivers are there but in incompatible ARM forks of the kernel.

      Let’s take the goodix touch panel I have on my device. Here’s the mainline driver, just one file:
      https://github.com/torvalds/linux/blob/master/drivers/input/touchscreen/goodix.c
      And here’s the sunxi tree for it:
      http://dl.linux-sunxi.org/SDK/A80/A80_SDK_20140728_stripped/lichee/linux-3.4/drivers/input/touchscreen/gt9xx/
      Drivers are completely different as it seems. In fact there are other folders like “gt9xxl” and “gt9xxnew” in this tree and I don’t even know if these are different devices or just different drivers for the same stuff. And of course they’re all linked to ARM and cannot be easily ported…

      On the other hand what did I expect from a noname Chinese device? However I must admit it’s branded counterparts work just as bad with Linux.

  43. Mario
    Mario January 15, 2015 at 11:06 am | | Reply

    I discovered on my tablet X11 is working although the screen is blank.

    With “xrandr –output DSI1 –rotate right” I unblank the screen, and it works fine.

    Other issue is backlight, completely missing in linux-3.19.0-rc4.

    For the moment I’m using original kernel 3.10.20 from android. It let me: start X, use gnome iceweasel(firefox) icedove(thunderbird), use wifi 8723bs, use modem 3g in /dev/ttyACMx, backlight regulation, view battery status charge. On the other hand, Android kernel misses: console on vt, audio doesn’t work, no swap (no hibernate), no ip-filter, and it use modules with authentication so can’t build external modules.

    My knowledge (I flashed both EFI firmware) without warranty.
    CHUWI VX8 3G android version (BIOS HAMPOO) boot EFI 64bit (I’m using this one).
    CHUWI VX8 3G W8.1 boot EFI 32bit.

    1. boxed
      boxed January 16, 2015 at 7:58 am | | Reply

      Interesting and thanks, that’s a lot faster than restarting X11. This leads to another hack that I did to my tablet while the accelerometer is still MIA and modeset does not work properly on startup most of the time: Create a file /etc/xdg/autostart/autoxrandr.desktop that contains:

      [Desktop Entry]
      Name=autoxrandr
      GenericName=Automatically run an XRandR on startup
      Comment=Some description about your script
      Exec=xrandr –output DSI1 –rotate right
      Terminal=false
      Type=Application
      X-GNOME-Autostart-enabled=true

      This will make it rotate the screen on Gnome3 bootup so even if you have a black screen on start, it will still run the xrandr and reenable the display once it finishes starting Gnome… This way I don’t get a dead display when I start the tablet (my nextbook 8) though it may not be in the orientation I wanted initially.

      I haven’t tried if I just used –rotate normal in that startup script, which would keep the portrait orientation versus “right” meaning landscape orientation.

  44. hakdunn
    hakdunn January 15, 2015 at 11:12 pm | | Reply

    I tested the newest img on a winbook tw700 with a goodix touchscreen. First it will boot to a black screen with nomodeset in grub. But when you get it to boot the touchscreen work. But No the WiFi or sound don’t work.

  45. Vaidotas
    Vaidotas January 17, 2015 at 4:33 am | | Reply

    I got dmesg from the moment when screen blanks out on boot. Maybe someone will be able to identify the problem? Device is Lenovo Miix 2 8 tablet with Z3740 CPU.

    [ 961.483435] [drm] Initialized drm 1.1.0 20060810
    [ 961.544659] [drm] Memory usable by graphics device = 2048M
    [ 961.544668] checking generic (80000000 3e8000) vs hw (80000000 10000000)
    [ 961.544672] fb: switching to inteldrmfb from EFI VGA
    [ 961.544729] Console: switching to colour dummy device 80×25
    [ 961.544798] [drm] Replacing VGA console driver
    [ 961.546297] i915 0000:00:02.0: irq 267 for MSI/MSI-X
    [ 961.546314] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
    [ 961.546318] [drm] Driver supports precise vblank timestamp query.
    [ 963.133512] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=io+mem,decodes=io+mem:owns=io+mem
    [ 963.657071] ACPI: Video Device [GFX0] (multi-head: yes rom: no post: no)
    [ 963.657841] acpi device:1a: registered as cooling_device4
    [ 963.657984] input: Video Bus as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/LNXVIDEO:00/input/input2
    [ 963.658078] [drm] Initialized i915 1.6.0 20140905 for 0000:00:02.0 on minor 0
    [ 964.951932] [drm] GMBUS [i915 gmbus dpb] timed out, falling back to bit banging on pin 5
    [ 966.719278] Raw EDID:
    [ 966.719284] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719288] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719291] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719294] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719296] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719299] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719302] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719305] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    [ 966.719312] i915 0000:00:02.0: HDMI-A-2: EDID block 0 invalid.
    [ 968.309347] fbcon: inteldrmfb (fb0) is primary device
    [ 985.193660] Console: switching to colour frame buffer device 100×80
    [ 985.274287] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
    [ 985.274292] i915 0000:00:02.0: registered panic notifier

  46. Kyle
    Kyle January 18, 2015 at 6:53 pm | | Reply

    FYI, Ubuntu 14.10 32-bit (using Fedlet’s EFI directory for initial USB booting) is running well on the HP Stream Bay Trail tablet ($99 US currently) with RTL8723BS Wifi and Goodix Touchscreen working. More information is available here: http://h30434.www3.hp.com/t5/Windows-Tablets/Linux-on-the-Stream-tablet/m-p/4831534.

    1. Lare
      Lare January 19, 2015 at 12:09 pm | | Reply

      I hope these changes get merged and released as a new fedlet image

  47. James
    James January 20, 2015 at 5:00 am | | Reply

    Anybody tried booting an Acer Iconia W511 with Fedlet? I’m having trouble booting from a USB key.

    As soon as the RAM disk loads the unit drops all USB devices connected, including the USB key and keyboard connected through the OTG adapter. The kernel then cannot load the filesystem from squashfs since the /live/ filesystem is no longer connected and available.

    Any tips to how I can keep USB available through the boot sequence?

    1. Torfinn Ingolfsen
      Torfinn Ingolfsen March 31, 2015 at 1:18 pm | | Reply

      Yes, I tried to boot the latest Fedlet image on a Iconia Tab W511; volume up + power, then I get a glimpse of the (grub?) boot menu, then a blank screen. This happens so quickly that I can’t press any keys on the (usb attached) keyboard. Didn’t try more than that.

  48. Lare
    Lare January 22, 2015 at 2:18 am | | Reply

    Just out of curiosity, does the TPM module work for anyone and how hard would it be to make it work?

  49. n. ghoul
    n. ghoul January 24, 2015 at 6:39 am | | Reply

    Hi,

    I tried latest release on Lenovo Miix 2 8″ tablet, it’s working fine in first boot, auto-rotating works fine, touch screen works but a bit slow, wifi networks are visible but while I tried to connect my wireless network, (the password input comes, I touched it, keyboard opened) system freezed. And then I can’t boot Fedlet again with same USB stick.

    I wish you’ll release a new version with Fedora 21’s source. Thanks.

  50. Pawel
    Pawel January 25, 2015 at 2:53 am | | Reply

    Hello

    Guys did you try to install android for intel atom?

    https://01.org/android-ia/downloads

    I tried but it wont boot, I do not have option to start from usb with this android os. I assume that it support only 64bit UEFI, my tablet support only 32bit. Is there some way to fix this and it will start on 32 UEFI tablet?

Leave a Reply

Your email address will not be published. Required fields are marked *