View previous topic :: View next topic |
Author |
Message |
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
Posted: Sun Nov 22, 2009 4:28 pm Post subject: [Solved] Any ati radeon ppc64 success story? |
|
|
I have radeon 9600 in power mac g5 and installed 64-bit userspace. Looks like graphics is not very reliable, and I observed two types of faults so far.
The first one sometimes happens on video playback and causes X server to get stuck, burning cpu:
Code: |
#0 0x00000fffb43f371c in __ioctl (fd=0, request=536896580) at ../sysdeps/unix/sysv/linux/powerpc/ioctl.c:57
#1 0x00000fffb4161410 in drmIoctl (fd=6, request=536896580, arg=0x0) at xf86drm.c:187
#2 0x00000fffb4161950 in drmCommandNone (fd=<value optimized out>, drmCommandIndex=<value optimized out>)
at xf86drm.c:2313
#3 0x00000fffb40b2440 in RADEONWaitForIdleCP (pScrn=0x1f8a6bf0) at radeon_commonfuncs.c:722
#4 0x00000fffb40c3b90 in RADEONDisplayVideo (pScrn=0x1f8a6bf0, crtc=<value optimized out>, pPriv=0x1f8ce750,
id=0, base_offset=<value optimized out>, offset1=2777088, offset2=3145729, offset3=3237889, offset4=2777088,
offset5=3145729, offset6=3237889, width=720, height=480, pitch=768, left=0, right=47185920, top=0,
dstBox=0xffff3d2cca8, src_w=720, src_h=480, drw_w=<value optimized out>, drw_h=<value optimized out>,
deinterlacing_method=0) at radeon.h:1429
#5 0x00000fffb40c4354 in RADEONPutImage (pScrn=0x1f8a6bf0, src_x=<value optimized out>, src_y=360,
drw_x=<value optimized out>, drw_y=<value optimized out>, src_w=720, src_h=0, drw_w=1920, drw_h=1080,
id=<value optimized out>,
buf=0xfff9f10d000 "xyzzz{{|{|}}}~}}~\177\177\200~\200\201\200\202\202\202\203\203\203\204\204\203\203\203\205\205\205\206\206\207\207\210\210\207\207\210\211\210\211\211\212\212\212\214\214\214\213\215\215\215\216\216\216\216\220\220\217\220\220\220\221\221\221\221\223\223\221\221\222\222\223\225\226\226\226\227\227\226\225\227\230\230\230\231\231\231\232\232\232\232\233\233\234\234\233\233\235\235\235\235\235\235\236\236\235\236\236\237\237���������", '�' <repeats 11 times>, "������������������������������������������������������������"..., width=720,
height=480, Sync=0, clipBoxes=0xffff3d2ce60, data=0x1f8ce750, pDraw=0x1ff37910) at radeon_video.c:3076
#6 0x0000000010086abc in xf86XVPutImage (client=<value optimized out>, pDraw=0x1ff37910, pPort=0x1f8dd210,
pGC=<value optimized out>, src_x=0, src_y=0, src_w=0, src_h=0, drw_x=0, drw_y=0, drw_w=1920,
drw_h=<value optimized out>, format=0x1f8dcae0,
data=0xfff9f10d000 "xyzzz{{|{|}}}~}}~\177\177\200~\200\201\200\202\202\202\203\203\203\204\204\203\203\203\205\205\205\206\206\207\207\210\210\207\207\210\211\210\211\211\212\212\212\214\214\214\213\215\215\215\216\216\216\216\220\220\217\220\220\220\221\221\221\221\223\223\221\221\222\222\223\225\226\226\226\227\227\226\225\227\230\230\230\231\231\231\232\232\232\232\233\233\234\234\233\233\235\235\235\235\235\235\236\236\235\236\236\237\237���������", '�' <repeats 11 times>, "������������������������������������������������������������"..., sync=0,
width=<value optimized out>, height=<value optimized out>) at xf86xv.c:1767
#7 0x00000fffb4207bc8 in XvdiPutImage (client=0x20037120, pDraw=0x1ff37910, pPort=0x1f8dd210, pGC=0x1ff591e0,
src_x=0, src_y=0, src_w=720, src_h=480, drw_x=0, drw_y=0, drw_w=1920, drw_h=<value optimized out>,
image=0x1f8dcae0,
data=0xfff9f10d000 "xyzzz{{|{|}}}~}}~\177\177\200~\200\201\200\202\202\202\203\203\203\204\204\203\203\203\205\205\205\206\206\207\207\210\210\207\207\210\211\210\211\211\212\212\212\214\214\214\213\215\215\215\216\216\216\216\220\220\217\220\220\220\221\221\221\221\223\223\221\221\222\222\223\225\226\226\226\227\227\226\225\227\230\230\230\231\231\231\232\232\232\232\233\233\234\234\233\233\235\235\235\235\235\235\236\236\235\236\236\237\237���������", '�' <repeats 11 times>, "������������������������������������������������������������"..., sync=0,
width=720, height=480) at xvmain.c:718
|
The second type of problem involves 3D graphics (glxgears) plus some other activity (like video playback). In this case everything locks up completely and the machine even does not respond to ping.
I guess I have a long trial and error process ahead trying to mix and match different kernel verisons, versions of xf86-video-ati driver, configuration options, etc. But just before starting, I wonder if anybody has similar setup running flawlessly already? I'm just a bit afraid that defective hardware may be one of the probabilities.
Last edited by ssvb on Wed Dec 02, 2009 10:47 pm; edited 1 time in total |
|
Back to top |
|
|
ursusca Apprentice
Joined: 10 Sep 2008 Posts: 284 Location: Toronto ON, CANADA
|
|
Back to top |
|
|
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
Posted: Mon Nov 23, 2009 8:40 pm Post subject: |
|
|
ursusca wrote: | Did you try to install ATI's closed source driver? |
Not really. Do they provide binary drivers for ppc64 arch? In any case, binary blobs are the worst possible option because we can't fix bugs in them, do we? |
|
Back to top |
|
|
mattst88 Developer
Joined: 28 Oct 2004 Posts: 422
|
Posted: Tue Nov 24, 2009 2:45 am Post subject: |
|
|
They don't provide binary drivers for Linux/PPC[64].
ssvb: what version of xorg-server, xf86-video-ati, libdrm, and mesa are you using? What kernel version?
It all _should_ work, but if it in fact does not, I'll help you file a bug at FreeDesktop and CC the correct person.
The first order of business is to try to reproduce whatever problems you're experiencing with
- =xorg-server-1.7.1
- xf86-video-ati from git
- =libdrm-2.4.15
- =mesa-7.6
_________________ My Wiki page |
|
Back to top |
|
|
xyz n00b
Joined: 29 Oct 2003 Posts: 67
|
Posted: Tue Nov 24, 2009 9:02 am Post subject: |
|
|
My ppc64 of 64-bit userland works well with Radeon 9800XT card:
xorg-server-7.5.1
xf86-video-ati-6.12.4
libdrm-2.4.15
mesa-7.5.2
gentoo-sources-2.6.31-r6 |
|
Back to top |
|
|
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
Posted: Tue Nov 24, 2009 10:16 am Post subject: |
|
|
OK, thank you very much for the information (especially for the list of relevant packages). Looks like I may just need to move from current ppc64 stable to ~ppc64 for this set of packages to solve the graphics reliability problems. I'll report later about the results. |
|
Back to top |
|
|
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
Posted: Wed Nov 25, 2009 10:03 pm Post subject: |
|
|
|mattst88| wrote: |
The first order of business is to try to reproduce whatever problems you're experiencing with
- =xorg-server-1.7.1
- xf86-video-ati from git
- =libdrm-2.4.15
- =mesa-7.6
|
Upgrading to these versions and also gentoo-sources-2.6.31-r6 did not help much.
The same test case hangs everything pretty fast. It's start of glxgears and mplayer video playback in a window (so that both glxgears and video windows are visible on screen at the same time). This looks like some race condition and improper sharing of GPU HW.
Setting
Code: | Option "DMAForXv" "false" |
in xorg.conf (all the rest is default) makes video look a bit distorted (apparently wrong endian issue) but hangs disappeared, or at least got much harder to reproduce.
Also there are some interesting lines in Xorg log:
Code: | (WW) RADEON(0): Video BIOS not detected, using default clock settings!
(II) RADEON(0): Probed PLL values: xtal: 27.000000 Mhz, sclk: 324.000000 Mhz, mclk: 199.125000 Mhz |
Looks like the card is a bit underclocked (in my case it's "ATI Radeon 9600 AP (AGP)" (ChipID = 0x4150)).
Could this result in worse stability? Are clock frequencies set correctly for your cards? |
|
Back to top |
|
|
xyz n00b
Joined: 29 Oct 2003 Posts: 67
|
Posted: Sat Nov 28, 2009 4:51 am Post subject: |
|
|
My xorg.conf. Don't know if it could help.
# This file was generated by Xorgautoconfig v0.2.2.
Section "ServerLayout"
Identifier "X.Org Configured"
Screen 0 "Screen0" 0 0
InputDevice "Mouse0" "CorePointer"
InputDevice "Keyboard0" "CoreKeyboard"
Option "AIGLX" "true" #"false"
EndSection
Section "Files"
# For X Font Server support, uncomment this and comment the other FontPaths
# Note that this is not required for most configurations
FontPath "/usr/share/fonts/75dpi/"
FontPath "/usr/share/fonts/75dpi/:unscaled"
FontPath "/usr/share/fonts/100dpi/"
FontPath "/usr/share/fonts/100dpi/:unscaled"
FontPath "/usr/share/fonts/kochi-substitute/"
FontPath "/usr/share/fonts/Type1/"
FontPath "/usr/share/fonts/ttf-bitstream-vera/"
FontPath "/usr/share/fonts/TrueType/VeraSansYuanTi/"
FontPath "/usr/share/fonts/baekmuk-fonts/"
EndSection
Section "Module"
Load "dbe"
Load "extmod"
Load "glx"
Load "dri"
Load "fbdevhw"
EndSection
Section "Extensions"
Option "Composite" "true" #"false"
EndSection
Section "ServerFlags"
Option "BlankTime" "10" # 10 minutes
Option "StandbyTime" "20"
Option "SuspendTime" "30"
Option "OffTime" "60"
Option "AllowEmptyInput" "false"
Option "DontZap" "false"
# Option "AutoAddDevices" "false"
### Uncomment this Server flag if X should not change resolutions.
### This is useful if mode changing crashes / corrupts the X server
Option "DisableVidModeExtensions"
EndSection
Section "InputDevice"
Identifier "Keyboard0"
Driver "kbd"
# Change "XkbModel" to "macintosh_old" if you are using
# the deprecated adb keycodes.
Option "XkbModel" "pc105"
Option "XkbLayout" "us"
Option "XkbRules" "xorg"
Option "XkbOptions" "compose:rwin"
### Uncomment this if your layout uses nodeadkeys
#Option "XkbVariant" "nodeadkeys"
EndSection
Section "InputDevice"
Identifier "Mouse0"
Driver "mouse"
Option "CorePointer"
Option "Protocol" "IMPS/2"
Option "Device" "/dev/input/mice"
Option "ZAxisMapping" "4 5"
EndSection
Section "Monitor"
Identifier "Monitor0"
UseModes "Modes0"
ModelName "Monitor Model"
Option "DPMS"
HorizSync 30-92
VertRefresh 40-60
EndSection
Section "Modes"
Identifier "Modes0"
# Generated
# D: 146.263 MHz, H: 65.296 kHz, V: 59.960 Hz
Modeline "1680x1050" 146.263 1680 1784 1960 2240 1050 1053 1059 1089 +HSync -VSync
EndSection
Section "Device"
Identifier "Card0"
Driver "radeon"
### Available Driver options:
Option "ColorTiling" "true" # "false" # True has trouble with latest xorg
Option "BackingStore" "true" # Increases performance
Option "EnablePageFlip" "true" # Increases performance
Option "SWcursor" "false" # Use Software Cursor
Option "NoAccel" "false" # Disable Hardware Acceleration
Option "AGPMode" "8" # Set AGP data transfer rate, DRI only
Option "AccelMethod" "EXA" # "XAA" "EXA" # New RenderAccel
Option "RenderAccel" "true" # Hardware acceleration
Option "DRI" "True" #"False"
Option "AGPFastWrite" "true" # "false" # Use AGP Fast Write, DRI only
Option "AccelDFS" "True"
### See "man radeon" for additional options
EndSection
Section "Screen"
Identifier "Screen0"
Device "Card0"
Monitor "Monitor0"
DefaultDepth 24
SubSection "Display"
Depth 8
Modes "1680x1050"
EndSubSection
SubSection "Display"
Depth 16
Modes "1680x1050"
EndSubSection
SubSection "Display"
Depth 24
Modes "1680x1050"
EndSubSection
EndSection
Section "DRI"
Group 0
Mode 0666
EndSection |
|
Back to top |
|
|
mattst88 Developer
Joined: 28 Oct 2004 Posts: 422
|
Posted: Sat Nov 28, 2009 11:14 pm Post subject: |
|
|
Firstly, please put code tags around your xorg.conf.
If updating versions didn't help, then I think it's probably time to file a bug report.
Head over to https://bugs.freedesktop.org/index.cgi and create an account. When you file the bug report, be sure to include relevant version numbers, hardware information, and the steps you take to reproduce this bug.
You'll want to file the bug in Product: xorg; Component: Driver/Radeon. Make sure to CC me at mattst88 at gmail dot com, and the PPC/Big Endian guy, Michel Dänzer at michel at daenzer dot net. _________________ My Wiki page |
|
Back to top |
|
|
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
Posted: Wed Dec 02, 2009 10:47 pm Post subject: |
|
|
Seems like everything works more or less stable if "DRI" is disabled in xorg.conf
As for reporting bugs, I see that there are actually too many of them getting discovered and fixed. And the issues are not just ppc/ppc64 related. Maybe I'll try to make some attempts to get at least EXA usable on my box when I have more time for this.
Thanks a lot. Problem solved. |
|
Back to top |
|
|
mattst88 Developer
Joined: 28 Oct 2004 Posts: 422
|
Posted: Thu Dec 03, 2009 4:04 am Post subject: |
|
|
ssvb wrote: | Seems like everything works more or less stable if "DRI" is disabled in xorg.conf
As for reporting bugs, I see that there are actually too many of them getting discovered and fixed. And the issues are not just ppc/ppc64 related. Maybe I'll try to make some attempts to get at least EXA usable on my box when I have more time for this. |
If it works without DRI, this narrows the problem down a bit, I think.
ssvb wrote: | Thanks a lot. Problem solved. |
I'm glad your problem isn't affecting you badly, but until you report it, it'll probably crop up with other users. Do the right thing and report it. _________________ My Wiki page |
|
Back to top |
|
|
ssvb Tux's lil' helper
Joined: 06 Nov 2003 Posts: 96
|
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|