Re: [sabayon-dev] Latest Limbo update screw up tty1 cuz KMS radeon being forced

2012-07-19 Thread Lorenzo Cogotti
Hi, I was able to reproduce the issue Joost mentioned.
I was misleaded in believing that x-setup could have been the cause,
since I reproduced the issue by deleting /etc/gpu-detector.conf and
rebooted forcing x-setup to rewrite xorg.conf.

Instead it turns out that the latest linux-sabayon version on Limbo (as
of yesterday) had the: CONFIG_DRM_RADEON_KMS=y

option in its configuration, that apparently auto loaded the radeon.ko
driver, causing the conflict with fglrx.

Today I have updated my kernel and to version:
sys-kernel/linux-sabayon-3.4-r6

Which turns off that option:
zcat /proc/config.gz | grep RADEON_KMS
# CONFIG_DRM_RADEON_KMS is not set

The radeon.ko module isn't auto loaded anymore so the issue is no more.
But I now wonder if the suggested procedure to replace fglrx with radeon
for newly unsupported AMD cards, on entropy notice board, is enough to
load radeon.ko, or if the module has to be explicitly loaded by editing
/etc/conf.d/modules.

Il 18/07/2012 22:38, Joost Ruis ha scritto:
 I also noticed that both radeon and fglrx modules are loaded.
 I guess I should blacklist radeon now?

 On Wed, Jul 18, 2012 at 10:31 PM, Joost Ruis
 joost.r...@sabayonlinux.org wrote:
 Micia on IRC states that he wasn't affected by this, so let me drop in
 my info, maybe it is just my card affected here.

 xbox-360 modprobe.d # lspci | grep VGA
 01:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee
 ATI Cypress [Radeon HD 5800 Series]

 xbox-360 modprobe.d # cat /proc/cmdline
 BOOT_IMAGE=/boot/kernel-genkernel-x86_64-3.4.0-sabayon ro
 init=/linuxrc splash=silent,theme:sabayon video=vesafb:ywrap,mtrr:3
 vga=791 console=tty1 quiet dokeymap keymap=us domdadm
 root=UUID=73c78b4e-fa69-45db-abc1-2d595151f6b0 docrypt

 On Wed, Jul 18, 2012 at 10:24 PM, Joost Ruis
 joost.r...@sabayonlinux.org wrote:
 Hi all,

 Latest Limbo bump set a file in /etc/modprobe.d/ called radeon.conf
 The purpose of that file is to enable KMS for FOSS ati driver, however
 when you do this when you use FGLRX and your card is =5xxx and never
 had problems with latest FGLRX driver and new Xorg, this workaround
 actually corrupts users tty1.

 xbox-360 modprobe.d # equery belongs radeon.conf
  * Searching for radeon.conf ...
 x11-drivers/xf86-video-ati-6.14.4-r1 (/etc/modprobe.d/radeon.conf)


 IMHO the xf86-video-ati driver should NOT install this file, all users
 have this package so any modern system with a descent AMD video card
 (4xxx) is affected by this.




-- 
Lorenzo Cogotti




Re: [sabayon-dev] Latest Limbo update screw up tty1 cuz KMS radeon being forced

2012-07-19 Thread Joost Ruis
I can confirm that with latest kernel bump
(sys-kernel/linux-sabayon-3.4-r6 from Limbo) the issue is now
resolved.
The problem was exactly as pointed out by Lorenzo.


On Thu, Jul 19, 2012 at 1:30 PM, Lorenzo Cogotti miciam...@hotmail.it wrote:
 Hi, I was able to reproduce the issue Joost mentioned.
 I was misleaded in believing that x-setup could have been the cause,
 since I reproduced the issue by deleting /etc/gpu-detector.conf and
 rebooted forcing x-setup to rewrite xorg.conf.

 Instead it turns out that the latest linux-sabayon version on Limbo (as
 of yesterday) had the: CONFIG_DRM_RADEON_KMS=y

 option in its configuration, that apparently auto loaded the radeon.ko
 driver, causing the conflict with fglrx.

 Today I have updated my kernel and to version:
 sys-kernel/linux-sabayon-3.4-r6

 Which turns off that option:
 zcat /proc/config.gz | grep RADEON_KMS
 # CONFIG_DRM_RADEON_KMS is not set

 The radeon.ko module isn't auto loaded anymore so the issue is no more.
 But I now wonder if the suggested procedure to replace fglrx with radeon
 for newly unsupported AMD cards, on entropy notice board, is enough to
 load radeon.ko, or if the module has to be explicitly loaded by editing
 /etc/conf.d/modules.

 Il 18/07/2012 22:38, Joost Ruis ha scritto:
 I also noticed that both radeon and fglrx modules are loaded.
 I guess I should blacklist radeon now?

 On Wed, Jul 18, 2012 at 10:31 PM, Joost Ruis
 joost.r...@sabayonlinux.org wrote:
 Micia on IRC states that he wasn't affected by this, so let me drop in
 my info, maybe it is just my card affected here.

 xbox-360 modprobe.d # lspci | grep VGA
 01:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee
 ATI Cypress [Radeon HD 5800 Series]

 xbox-360 modprobe.d # cat /proc/cmdline
 BOOT_IMAGE=/boot/kernel-genkernel-x86_64-3.4.0-sabayon ro
 init=/linuxrc splash=silent,theme:sabayon video=vesafb:ywrap,mtrr:3
 vga=791 console=tty1 quiet dokeymap keymap=us domdadm
 root=UUID=73c78b4e-fa69-45db-abc1-2d595151f6b0 docrypt

 On Wed, Jul 18, 2012 at 10:24 PM, Joost Ruis
 joost.r...@sabayonlinux.org wrote:
 Hi all,

 Latest Limbo bump set a file in /etc/modprobe.d/ called radeon.conf
 The purpose of that file is to enable KMS for FOSS ati driver, however
 when you do this when you use FGLRX and your card is =5xxx and never
 had problems with latest FGLRX driver and new Xorg, this workaround
 actually corrupts users tty1.

 xbox-360 modprobe.d # equery belongs radeon.conf
  * Searching for radeon.conf ...
 x11-drivers/xf86-video-ati-6.14.4-r1 (/etc/modprobe.d/radeon.conf)


 IMHO the xf86-video-ati driver should NOT install this file, all users
 have this package so any modern system with a descent AMD video card
 (4xxx) is affected by this.




 --
 Lorenzo Cogotti





Re: [sabayon-dev] Latest Limbo update screw up tty1 cuz KMS radeon being forced

2012-07-19 Thread Fabio Erculiani
radeon.ko should be loaded automatically.
Quick test, is 3D working? What's inside Xorg.0.log?

-- 
Fabio Erculiani



Re: [sabayon-dev] Latest Limbo update screw up tty1 cuz KMS radeon being forced

2012-07-19 Thread Joost Ruis
On the FGLRX side everything is back to what was normal.

xbox-360 joost # lsmod | grep radeon
xbox-360 joost # lsmod | grep fglrx
fglrx2579795  123


On Thu, Jul 19, 2012 at 6:19 PM, Fabio Erculiani lx...@sabayon.org wrote:
 radeon.ko should be loaded automatically.
 Quick test, is 3D working? What's inside Xorg.0.log?

 --
 Fabio Erculiani