xf86-video-ati 6.14.2

Hi,

Is anyone else besides me experiencing problems with the new Ati 6.14.2 driver? In my case it seems a lot slower than before.

It's looks like it sometimes has no 2D acceleration, especially in Midnight Commander. When reading a file with F3 and exiting again, you can see the window redrawing on the screen. This happens since the new ati driver so I think the new driver is the problem.

I use a Radeon 4870 on Xorg + Fluxbox btw.

Marco
 
I'm having problems, but not like you described:

ATI Radeon HD 3450 (RV620 LE):
No problems

ATI RADEON 9600/9700 Series (M10):
System locks up (mostly when exiting Xorg)

Mobility Radeon 7500 (M7 [LW]):
System locks up enough that it is unusable with the driver; switched to vesa and it's mostly stable
 
Mobility Radeon 7500 (M7 [LW]) on a Thinkpad T42 along with xf86-video-ati-6.14.2 working okay here. i386 8-STABLE as of today. Is there anything specific that causes lockups?

Oh, and a Radeon 4650 in a desktop works great. Can't recall a lockup with 6.14.2 ever.
 
I get consistent lockups on one amd64 machine with an HD4850 if I simply restart X. Within a minute or two of X restarting, the entire machine hangs. On another machine with an HD4350, no problems restarting, but I get serious artifacts when I move windows around on screen. Unfortunately, this happens with the latest 6.13.* driver, too.

Adam
 
The driver gives me problems on a lenovo x120e. It contains an ati radeon hd 6310. Xorg runs fine with the driver, but I'm unable to exit without the system locking up. Ive tried commands, hotkeys, sleep mode, killing xorg pids, and all results in the same. It appears as the brightness gets turned all the way up. You can still barely make out the contents of the screen, but its an off white and washed out. No response from anything. Sometimes it will kernel panic/reboot on its own shortly after.

Love the hardware though. Looking forward to an escape from vesa-land.
 
adamk said:
I get consistent lockups on one amd64 machine with an HD4850 if I simply restart X. Within a minute or two of X restarting, the entire machine hangs. On another machine with an HD4350, no problems restarting, but I get serious artifacts when I move windows around on screen. Unfortunately, this happens with the latest 6.13.* driver, too.

Maybe problems with 3D acceleration in the desktop (compiz)? Or maybe particular tweaks have been made for specific GPUs. I know Robert Noland made some adjustments for some cards based on my feedback from testing, but don't know exactly what he did. And he did say he had a 4650.

Anyway, I just started and stopped an ordinary xfce4.8 desktop ten times without problems on this HD4650. That's with composite on.
 
I don't typically use compiz on these machines. I can simply be using openbox, but restarting X will cause the system to lockup. This only started happening since rnoland@ disappeared from view, or I would have certainly reported it to him on IRC. He and I had a number of discussions on the fact that restarting X on machines with limited amounts of RAM would cause direct rendering to break in the X server. I don't seem to have that particular problem any more (since upgrading both boxes from 2G to 4G), but those lockups started happening.
 
wblock said:
Mobility Radeon 7500 (M7 [LW]) on a Thinkpad T42 along with xf86-video-ati-6.14.2 working okay here. i386 8-STABLE as of today. Is there anything specific that causes lockups?

It's been a few months, but I remember it locking up randomly. For example, when xscreensaver would kick in and blank the screen, I'd hit a key or move the mouse and it would be locked. Closing Xorg or a control+al+f1 from Xorg would cause it as well.
 
No problems switching to and from X with ctrl-alt-whatever. ioquake3 runs but crashes out:
Code:
drmRadeonCmdBuffer: -22. Kernel failed to parse or rejected command stream.  See dmesg for more info.

There's nothing in dmesg or /var/log/messages. But it doesn't lock up, just drops out of the game and back into X. Although the X resolution has been changed by SDL... but it's still running.

I'll install xscreensaver and give it a try.
 
With my X800GTO AGP card, I've been using XAA instead of EXA acceleration method.

I think that EXA is the default. Unlike Linux, on FreeBSD, the free ATI driver had some issues with EXA method in the past.

You need to edit /usr/local/etc/X11/xorg.conf file to specify Option "AccelMethod" "xaa" in the Section "Device"
 
Update: The T42 stayed on overnight, updating world. Switching to console or X has not caused a problem during this testing (or ever, that I can recall).

Repeatedly starting or restarting X with startx or xdm doesn't have a problem.

As before, ioquake starts and runs perfectly but quits after a few seconds. Maybe due to limited video RAM.

No AccelMethod specified, and the log shows it defaults to XAA for this old hardware.

Running i386 8.2-STABLE as of 20110812. Could the driver problems be limited to amd64?

xorg.conf
Code:
Section "ServerLayout"
        Identifier   "Manually Configured"
        Screen       0  "Screen0" 0 0
        Option       "DontZap" "Off"
EndSection

Section "Files"
        ModulePath   "/usr/local/lib/xorg/modules"
        FontPath     "/usr/local/lib/X11/fonts/misc/"
        FontPath     "/usr/local/lib/X11/fonts/TTF/"
        FontPath     "/usr/local/lib/X11/fonts/OTF"
        FontPath     "/usr/local/lib/X11/fonts/Type1/"
        FontPath     "/usr/local/lib/X11/fonts/100dpi/"
        FontPath     "/usr/local/lib/X11/fonts/75dpi/"
        FontPath     "/usr/local/lib/X11/fonts/bitstream-vera/"
EndSection

Section "DRI"
        Group 0
        Mode  0660
EndSection

Section "Extensions"
        Option       "Composite" "Enable"
EndSection

Section "Device"
        Identifier  "Card0"
        Driver      "radeon"
        VendorName  "ATI Technologies Inc"
        BoardName   "Radeon Mobility M7 LW [Radeon Mobility 7500]"
        BusID       "PCI:1:0:0"
EndSection

Section "Screen"
        Identifier "Screen0"
        Device     "Card0"
        Monitor    "Monitor0"
        SubSection "Display"
                Virtual 1024 768
                Modes "1024x768"
        EndSubSection
EndSection
 
wblock said:
Could the driver problems be limited to amd64?

Nope, I just installed the 9.0/i386 BETA on one of these boxes. Built Xorg, started it up (openbox + tint2), and launched a firefox session over ssh from another box. Closed the window, no problems. Launched FF again as I realized I wanted to download the diablo jdk, too, and then when I closed the window, the entire box hung. No kernel panic on the serial console, no reboot, just a complete hang, which is what I was getting on amd64 a few seconds/minutes after restarting X.

Adam
 
If anything, it's gotten worse for me on i386. In openbox, no compositing, I just moved a maximized terminal from one workspace to another, and the entire machine froze on me again. Which is really annoying considering the 4 builds I had going in ports.

Adam
 
wblock said:
Update: The T42 stayed on overnight... Could the driver problems be limited to amd64?

I just tried your config file, logged in to Xorg via xdm and then logged out and it went to ttyv0 and locked up before xdm could redisplay. Other details:

8.2-RELEASE i386, generic kernel, old IBM R51, xf86-video-ati-6.14.2, fluxbox-1.3.1
% pkg_version -IvL=
Code:
dri-7.6.1,2
libGL-7.6.1
libGLU-7.6.1
libdrm-2.4.17_1
libglut-7.6.1

All other ports are the latest versions.
 
This is getting interesting. fluxbox with all config options turned off works fine, I logged in via xdm and then quit several times without incident.

Just now tried openbox. No lockup, although the mouse cursor switched to a large square block sometimes (guessing a memory problem, maybe pointer problem).

Things that could be different on my system: the T42 has 1.5G, which is probably more than most notebooks that age. There could be other hardware implementation differences. I'm not running hal, and it's a custom kernel built without INET6. pkg_info, dmesg, uname, xorg.conf, kernel config, etc can be found here:
http://www.wonkity.com/~wblock/ati-lock/

If there's anything I've forgotten, please let me know.

(The desktop with 2G, an HD4650, and xfce4.8 seems fine outside of occasional border corruption on Firefox windows. I'd pefer not to mess with it right now.)
 
Changed from EXA to XAA fixes this lockup, but now flipping to another desktop or changing window sizes creates horrible flickering and lag.

Code:
FreeBSD 8.2-STABLE #0: Sun Jun 26 08:42:45 EDT 2011

xf86-video-ati-6.14.2 + libreoffice-3.3.3_2 + xorg-7.5.1 = hard lockup,
keyboard LEDs work for a bit, can't change to a console, kill X. ssh'ing in, running top shows Xorg at 100% CPU time. Nothing gets printed to the Xorg.0.log

xf86-video-vesa-2.3.0_1 + libreoffice-3.3.3_2 + xorg-7.5.1 = works fine.

Code:
vgapci0@pci0:6:3:0:	class=0x030000 card=0x51571002 chip=0x4c571002 
rev=0x00 hdr=0x00
     vendor     = 'ATI Technologies Inc. / Advanced Micro Devices, Inc.'
     device     = 'Mobility Radeon 7500 (M7 [LW])'
     class      = display
     subclass   = VGA
vgapci0: <VGA-compatible display> port 0xb800-0xb8ff mem 
0xd0000000-0xdfffffff,0xff520000-0xff52ffff irq 19 at device 3.0 on pci6
drm0: <ATI Radeon LW RV200 Mobility 7500 M7> on vgapci0
 
wblock said:
pkg_info, dmesg, uname, xorg.conf, kernel config, etc can be found here...[/url]

The same information for my R51 can be found here. The kernel configuration is missing because it's just a generic kernel. I noticed there was a crash dump this time, so I've put that up as well.
 
Acquired two HD4850 PCIe cards today. After cleaning out the dust, this one is working okay.

One thing that may be different from nearly everybody else is I only have one workspace configured. Don't even have the xfce workspace switcher on the panel.
 
Update: console switching, xscreensaver, ioquake3, openarena-oax, nexuiz, it all runs without problems. No lockups. This is all with the same system that had the HD4650 in it. There was some messing with xorg.conf to get the dual-monitor setup working, which I suspect had to do with xfce internal settings. The HD4850 has two DVI and one S-video, where the HD4650 has DVI, VGA, and HDMI. glxinfo output for the HD4850 has been added to the web page.

The 4850 is faster, but it also runs much hotter. If there are other tests that might help, let me know and I'll run them, but I'll be switching back to the 4650 for normal use.
 
After having problems with the vesa driver with my AMD nee ATI RV350 [Mobility Radeon 9600 M10] rev 0 card over the weekend, I thought I would enter the firey hell of Xorg configuration. I think I got something working after about a year of trying, giving up, trying, giving up.... Don't ask for details; I was basically a monkey hammering on the keyboard until something worked.

xorg.conf:

Code:
Section "ServerLayout"
   Identifier "X.org Configured"
   Screen 0 "Screen0" 0 0
   InputDevice "Mouse0" "CorePointer"
   InputDevice "Keyboard0" "CoreKeyboard"
EndSection

Section "ServerFlags"
   Option "AutoAddDevices" "False"
EndSection

Section "Files"
   ModulePath "/usr/local/lib/xorg/modules"
   FontPath "/usr/local/lib/X11/fonts/misc/"
   FontPath "/usr/local/lib/X11/fonts/TTF/"
   FontPath "/usr/local/lib/X11/fonts/OTF"
   FontPath "/usr/local/lib/X11/fonts/Type1/"
   FontPath "/usr/local/lib/X11/fonts/100dpi/"
   FontPath "/usr/local/lib/X11/fonts/75dpi/"
EndSection

Section "Module"
   Load  "ddc"
   Load  "dbe"
   Load  "dri"
   Load  "dri2"
   Load  "extmod"
   Load  "glx"
   Load  "record"
EndSection

Section "InputDevice"
   Identifier  "Keyboard0"
   Driver      "kbd"
EndSection

Section "InputDevice"
   Identifier "Mouse0"
   Driver "mouse"
   Option "Protocol" "auto"
   Option "Device" "/dev/sysmouse"
   Option "ZAxisMapping" "4 5 6 7"
EndSection

Section "Monitor"
   Identifier "Monitor0"
   VendorName "Monitor Vendor"
   ModelName "Monitor Model"
EndSection

Section "Device"
   Option "AGPMode" "4"
   Option "AccelMethod" "EXA"
   Option "Colortiling" "On"
   Option "DRI" "true"
   Option "EXANoComposite" "false"
   Option "FBTexPercent" "50"
   Option "GARTSize"    "1536"
   Option "MergedFB" "off"
   Option "MigrationHeuristic" "greedy"

   Identifier "Card0"
   Driver "radeon"
   VendorName "Advanced Micro Devices [AMD] nee ATI"
   BoardName "RV350 [Mobility Radeon 9600 M10]"
   BusID "PCI:1:0:0"
EndSection

Section "Screen"
   Identifier "Screen0"
   Device "Card0"
   Monitor "Monitor0"
   SubSection "Display"
      Viewport 0 0
      Depth 24
   EndSubSection
EndSection

Section "DRI"
    Mode 0666
EndSection

Section "Extensions"
    Option "Composite" "Enable"
EndSection
 
Long time no see :)

I've been using EXA acceleration mode these days. It's the default mode and it works fine. The current open source ATI driver on FreeBSD, xf86-video-ati-6.14.3, definitely fixed some issues we had last year.
 
ahavatar said:
The current open source ATI driver on FreeBSD, xf86-video-ati-6.14.3, definitely fixed some issues we had last year.

Not for me. With each release, the amount of visual corruption I get (mostly when moving/resizing windows) just gets worse.

Adam
 
Back
Top