Announcement

Collapse
No announcement yet.

Simple graphics switching through BIOS

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Bucic
    replied
    Bug submitted:

    Leave a comment:


  • Bucic
    replied
    Originally posted by Serafean View Post
    Hi, Here is what I would do in your case :
    .
    .
    .

    Serafean.
    Thanks for the outline but I don't think it's the right aproach to test this, at least initially. You focused on automation while we don't even have the system "back from AMD to Intel". IMHO we should figure out what to do in the first place to boot the system with installed fglrx back into Intel, manually. From what I see you think xorg.conf is the blocker here. I think you're wrong, as brosis noted the log file says:
    Code:
    Using config directory: "/etc/X11/xorg.conf.d"

    Originally posted by brosis View Post

    The best way would be to open a bug by fedora bugtracker.
    Won't it be outright invalid? I downgraded x server for instance (to install the catalyst legacy 13.1)... The issue was present on all unmodified distros I've tried during the past year on the other hand. I also hope they won't send me over to AMD forums, like ubuntu folks did.
    Last edited by Bucic; 28 April 2013, 06:17 PM.

    Leave a comment:


  • brosis
    replied
    According to logs:
    Code:
    [     6.184] (==) Log file: "/var/log/Xorg.0.log", Time: Sun Apr 28 15:22:13 2013
    [     6.184] (==) Using config file: "/etc/X11/xorg.conf"
    [     6.184] (==) Using config directory: "/etc/X11/xorg.conf.d"
    [     6.184] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
    [     6.184] (==) ServerLayout "aticonfig Layout"
    [     6.184] (**) |-->Screen "aticonfig-Screen[0]-0" (0)
    [     6.184] (**) |   |-->Monitor "aticonfig-Monitor[0]-0"
    [     6.184] (**) |   |-->Device "aticonfig-Device[0]-0"
    And then fglrx proceeds seeking device, finds none and Xorg aborts with screens not found.
    My understanding is that fglrx has installed own static "xorg.conf.d" Xorg session config in
    /usr/share/X11/xorg.conf.d
    and it disables autoprobing for intel driver. Given that the radeon card is disabled and only fglrx session is laid out, after fglrx aborts, Xorg is left without screens.

    When you use radeon driver+intel, they rely on Xorg automatic configuration. Unlike fglrx, they don't need static config.
    What you need, if I am correct, granted you want to keep fglrx, is to have two video card static configuration (session layouts). One for intel driver (which needs to be created), another for fglrx.
    So that, after fglrx finds no adapters, Xorg proceeds loading Intel driver, which should return a screen.

    The good news is that its just plain configuration files issue, nothing messed in your system.
    The bad news, is that I have never had configuration with multiple video cards in the system, configured via xorg.conf.d way.

    The best way would be to open a bug by fedora bugtracker.

    Open a bug containing words "Intel + fglrx, IBM T500, muxed" and attach all logs there.
    Would be also nice, if you archive and attach all files from /usr/share/X11/xorg.conf.d/
    Copy the full problem description you filled under your linked post in first thread.

    Then they will be able hopefully to give you direct instructions/configs to copy into your system.

    The majority of solutions for your T500 laptop on internet configure BIOS VGA to discrete, then use either radeon or fglrx.<
    Last edited by brosis; 28 April 2013, 06:02 PM.

    Leave a comment:


  • Serafean
    replied
    Hi, Here is what I would do in your case :
    Create 2 separate boot options, one for intel graphics and one for AMD, both of these entries have a different (random) command line parameter. in this case, let's assume intel and catalyst.
    the kernel command line can be accessed through /proc/cmdline.
    Now prepare two xorg.conf files. name one xorg.conf.intel, the other xorg.conf.catalyst. As their names imply, one should work when you want the intel card, the other with the AMD card.
    now the fun part. to switch at runtime, we must create a script. something along the lines of
    Code:
    #!/bin/bash
    if [ "$1" = "start" ];then
    grep -o catalyst /proc/cmdline && cp /etc/X11/xorg.conf.catalyst /etc/xorg.conf
    grep -o intel    /proc/cmdline && cp /etc/X11/xorg.conf.intel    /etc/xorg.conf
    fi
    if [ "$1" = "stop" ];then
    rm /etc/X11/xorg.conf
    fi
    put this script somewhere in your system. Now we must create a systemd unit file for it :
    Code:
    [Unit]
    Description=switch xorg.conf
    
    [Service]
    Type=oneshot
    ExecStart=/path/to/script/script.name.sh start
    ExecStop=/path/to/script/script.name.sh stop
    
    
    [Install]
    WantedBy=multi-user.target
    put this unit file into /etc/systemd/system named as xorg-switch.service. enable it using sytemctl enable xorg-switch.service
    note that this approach isn't very safe, nor is this a copy & paste guide. This is more of a braindump of how I would go about your issue which, from what I gather, is mostly changing the X configuration depending on the BIOS setting. If not, please be more specific.

    Serafean.

    Leave a comment:


  • Bucic
    replied
    Here are the logs

    Leave a comment:


  • Bucic
    replied
    It looks like it's /var/log/messages* files in Fedora.

    Leave a comment:


  • brosis
    replied
    Originally posted by Bucic View Post
    I'm on it but are you sure you got the part with 'dmesg' right? First - AFAIK it's a command, not a set of files. Second - there are no such files in /var/log even on my mature system.
    Hi, yes, dmesg is a command and also log files (dmesg.1, dmesg.2, dmesg.tar.gz) that are autosaved by logrotate I think. Thats what on my Ubuntu and Debian machines.
    They are in "/var/log/.." and are text files/archives.
    Dmesg-command is a binary in "/bin/dmesg", on contrast. Do not confuse.

    I remember that if you use Fedora, there is systemd and now there should be different way/location. You can ask Google or IRC of Fedora how to capture your kernel and Xorg logs.
    Thats basically what step above did - clean them, do steps leading to the problem, and then submit the logs with crash info.

    Leave a comment:


  • Bucic
    replied
    Originally posted by brosis View Post
    A solution what you are talking about is called workaround... If you want fix, opensource fix for your trouble would be available much faster than in catalyst, because its a matter of applying patch and resintalling few things (dozen of commands in CLI). Catalyst would only release fix with next release.

    Could you please switch to radeon card, install catalyst, then remove files as root
    /var/log/dmesg*
    and
    /var/log/Xorg*

    Copypaste oneliner:
    sudo rm /var/log/dmesg* && sudo rm /var/log/Xorg*

    THEN, immediately reboot, switch in BIOS to Intel, let system boot and hang.

    THEN, switch back to radeon, boot up, and attach those files to your post, for example as archive (zip,7z,gz etc)

    This way, you will attach relevant logs in the moment of crash. Because currently, there are no logs. No logs - no real help.
    I'm on it but are you sure you got the part with 'dmesg' right? First - AFAIK it's a command, not a set of files. Second - there are no such files in /var/log even on my mature system.

    Leave a comment:


  • brosis
    replied
    Originally posted by Bucic View Post
    I need the solution -now-. I'm not linux-savvy enough to conclude anything from xorg logs but I know enough that there's little more silly things than hoping for a release of some open-source software in a given timeframe. Grey hair guaranteed!
    A solution what you are talking about is called workaround... If you want fix, opensource fix for your trouble would be available much faster than in catalyst, because its a matter of applying patch and resintalling few things (dozen of commands in CLI). Catalyst would only release fix with next release.

    Could you please switch to radeon card, install catalyst, then remove files as root
    /var/log/dmesg*
    and
    /var/log/Xorg*

    Copypaste oneliner:
    sudo rm /var/log/dmesg* && sudo rm /var/log/Xorg*

    THEN, immediately reboot, switch in BIOS to Intel, let system boot and hang.

    THEN, switch back to radeon, boot up, and attach those files to your post, for example as archive (zip,7z,gz etc)

    This way, you will attach relevant logs in the moment of crash. Because currently, there are no logs. No logs - no real help.

    Leave a comment:


  • Bucic
    replied
    Originally posted by brosis View Post
    My suggestion would be for you to KICKstart the funding for working dynpm for radeon opensource driver.
    Because its just a matter of few years till your card will be unsupported, so why invest money there?
    Xorg log when you install catalyst and switch to intel will help you. Probably some module or some library replaced and loading with intel hardware.
    I need the solution -now-. I'm not linux-savvy enough to conclude anything from xorg logs but I know enough that there's little more silly things than hoping for a release of some open-source software in a given timeframe. Grey hair guaranteed!

    Originally posted by Sonadow View Post
    By the way, just 1 issue. That notebook of yours is rather old and last I remembered the mainline Catalyst doesn't even support that the 3xxx series hardware anymore. And Catalyst legacy has not been updated to work with anything after, what, xserver 1.11? Mind telling me how in the world are you even going to install Catalyst Legacy on F19?
    Please refer to post #1. It's all there. I'm currently using F18 with Catalyst legacy 13.1.
    Last edited by Bucic; 28 April 2013, 12:07 PM.

    Leave a comment:

Working...
X