ATI Radeon HD 3650 AGP boot problem

English Install/Boot/Login
1

My problem is next:
I have installed OpenSuse 11.3 . System doesn’t boot in NORMAL mode, system boots only in Fail-safe mode. In NORMAL mode i have kernel panic (may be i am wrong, but Caps Lock + Scroll Lock are blinking synchronously during NORMAL boot).

ATI Radeon HD 3650 AGP figured on thread Subject because i am sure that the problem is in this video card.
2 weeks ago i have installed ATI Proprietary driver for this card and system successfully boot in NORMAL mode without any kernel panic. You may ask me, “What the problem, install this driver again and have fun!”. The problem is that when I install this driver and boot in NORMAL mode video sub system works very slowly. For example one mouse scroll in Internet browser takes 2 seconds to redraw the window. It is not only browser window problem. It refers to all windows. I even can see the wave from up to down on the screen when any window redraws. Very poor performance of 2D video. So , installation of this driver is not the solution.

In the Fail-safe mode 2D video performance is not so cool as at Windows XP but is much more better than in NORMAL boot with proprietary driver installed.

Now I have clean installation of system without any drivers installed over distributive.

Okey, now system configuration:

  1. Motherboard: Intel 845D(has on board AC’97 sound, but this sound is disabled in BIOS currently)
  2. CPU:Intel Celeron 1.7 GHz
  3. Main memory: 1,5 GB DDR 266MHz
  4. Video card: PowerColor ATI Radeon HD 3650 AGP (512 MB, 64 bit)
  5. Sound card: CMI8738
  6. Hard drive1: SATAII Western Digital 500GB 32MB cache(connected to IDE slot on motherboard through SATAII-IDE converter. This converter is separate module inserted into IDE slot on motherboard)
  7. Hard drive2: IDE Seagate 40 GB
  8. DVD-RW: IDE LG DVD-RW device(without converter)
  9. Network card: D-Link DFE-520TX 10&100Mbps Fast Ethernet PCI Adapter

I have installed OpenSuse 11.3 from DVD burned from ISO openSUSE-11.3-DVD-i586.iso, downloaded from file sharing site.

Now additional info:

zigo@linux-n7ip:~> /sbin/lspci
00:00.0 Host bridge: Intel Corporation 82845 845 [Brookdale] Chipset Host Bridge (rev 04)
00:01.0 PCI bridge: Intel Corporation 82845 845 [Brookdale] Chipset AGP Bridge (rev 04)
00:1e.0 PCI bridge: Intel Corporation 82801 PCI Bridge (rev 05)
00:1f.0 ISA bridge: Intel Corporation 82801BA ISA Bridge (LPC) (rev 05)
00:1f.1 IDE interface: Intel Corporation 82801BA IDE U100 Controller (rev 05)
00:1f.2 USB Controller: Intel Corporation 82801BA/BAM USB Controller #1 (rev 05)
00:1f.3 SMBus: Intel Corporation 82801BA/BAM SMBus Controller (rev 05)
00:1f.4 USB Controller: Intel Corporation 82801BA/BAM USB Controller #1 (rev 05)
01:00.0 VGA compatible controller: ATI Technologies Inc Mobility Radeon HD 3600 Series
02:0c.0 Multimedia audio controller: C-Media Electronics Inc CM8738 (rev 10)
02:0f.0 Ethernet controller: VIA Technologies, Inc. VT6105/VT6106S [Rhine-III] (rev 8b)

zigo@linux-n7ip:~> uname -r
2.6.34-12-default

output of dmesg: http://myfhost.ucoz.com/dmesg_output.txt
/var/log/messages: http://myfhost.ucoz.com/var_log_messages.txt
/var/log/syslog: http://myfhost.ucoz.com/var_log_syslog.txt
/var/log/Xorg.0.log: http://myfhost.ucoz.com/var_log_Xorg.0.log.txt

2

i determined that my system needs boot option nomodeset . Kernel panic happens always without this option. With this option windows redraw very slowly.

3

Can you also provide the output of:

rpm -qa '*kernel*'
rpm -qa '*fglrx*'

and post to the web site SUSE Paste the content of the file /var/log/Xorg.0.log and press ‘create’ on that side and post here the website URL/address it provides. That enables us to look at the Xorg.0.log file without cluttering this thread.

Note also per the openSUSE release notes (and per 4th thing to try in this guide I wrote: SDB:Configuring graphics cards - openSUSE ) that it may also be desirable to to run “yast” (you can run yast in text mode with root permissions if X window not available) and navigate to yast > System > /etc/sysconfig Editor > System > Kernel > NO_KMS_IN_INITRD and change it to “yes”. This takes a minute or two to save once changed is submitted.

4

What is more interesting is the output of:

/sbin/lspci -nnk

for only 01:00.0 VGA device. It provides more salient detail.

5

/sbin/lspci -nnk: SUSE Paste

Guys , what do you think about “Mobility” ? My computer is desktop, not the notebook. Why mobility?

6

Thanks. The salient part is:


#01:00.0 VGA compatible controller [0300]: ATI Technologies Inc Mobility Radeon HD 3600 Series [1002:9598]
#        Subsystem: ATI Technologies Inc Mobility Radeon HD 3600 [1002:9598]

Please also do not forget to post output of:


rpm -qa '*kernel*'
rpm -qa '*fglrx*'

and also copy the contents of the /var/log/Xorg.0.log file as requested to the opensuse paste site and share here the URL/address so we can examine that file.

7

*linux-n7ip:~ # rpm -qa 'kernel’
kernel-default-2.6.34-12.3.i586
kernel-firmware-20100617-2.2.noarch
linux-n7ip:~ # rpm -qa 'fglrx’
linux-n7ip:~ # *

** /var/log/Xorg.0.log**: SUSE Paste

Current problem is very slow video playback and slow windows redrawing. Adding boot option nomodeset to NORMAL boot options list allows to boot in NORMAL mode, but 2D video sub system is too slow.

8

Ok, thanks. The problem as I see it is by specifying ‘nomodeset’ you are forcing X to fall back to the ‘radeonhd’ driver, which is mostly not as good as the ‘radeon’ driver.

I’m surprised that ‘nomodeset’ is necessary, as it should not be necessary. For some reason the ‘radeon’ open source graphic driver is struggling with your hardware.

What you could do is try the ‘fglrx’ proprietary ATI graphic driver. There is guidance here for that: SDB:ATI drivers - openSUSE

I confess I never follow that method, and hence I am unclear what the requirements are. I have a Radeon HD3450 and what I do to install it is ensure in YaST > Software > Software Management that the “base development pattern” is selected (and associated apps installed providing C, C++ and make) and I also install kernel-source and kernel-syms (and associated dependencies).

Once those are installed, and ONLY after those are installed, then I go to the ATI web site and download the appropriate .run file for my graphic card (which in your case and also in my case will be the file: ati-driver-installer-11-2-x86.x86_64.run . Put the file under /home/yourusername or somewhere that you can easily navigate to from a terminal.

Then after downloading the file “ati-driver-installer-11-2-x86.x86_64.run” from the ATI web site, I then rebooted to run level 3 (do that by rebooting and pressing ‘3’ (no quotes) in the 1st boot menu that comes up (where you choose between openSUSE, openSUSE failsafe, Windows … ) , and also after the 3 (with a space between) type ‘nomodeset’ (no quotes), and then boot and log in as a regular user. Then type ‘su’ to obtain root permissions. Since you have already indacted you do NOT have any ‘fglrx’ rpms installed, then you do not need to remove such an rpm (normally one must 1st remove such an rpm).

Then rebuild a new rpm using the Catalyst-11.2 .run file that was downloaded previous, using the command to rebuild (for a 32-bit user):

sh ./ati-driver-installer-11-2-x86.x86_64.run --buildpkg SuSE/SUSE113-IA32

Note a 64-bit user would instead use:

sh ./ati-driver-installer-11-2-x86.x86_64.run --buildpkg SuSE/SUSE113-AMD64

but you are NOT a 64-bit user, so use the 32-bit user command

That will build the new rpm file “fglrxsomething_xpic_SUSE113-8.821-1.something.rpm” (fill in the exact text under ‘something’) which can then installed with:

zypper install fglrx*something*_xpic_SUSE113-8.821-1.*something*.rpm

(fill in the exact text under ‘something’).

the above steps can be very slow … so give them time to run. Also, you may need to configure “YES” to NO KMS with YaST as I noted in my previous post: (navigate to yast > System > /etc/sysconfig Editor > System > Kernel > NO_KMS_IN_INITRD and change it to “yes”. This takes a minute or two to save once changed is submitted)

Then reboot the PC with the nominal command ‘shutdown -r now’ and then after reboot log on to the desktop.

They call that ‘the hardway’ although its not hard.

9

My video card is AGP
I have visited AMD Graphics Drivers & Software
My choice was:

  1. Desktop Graphics
  2. Radeon HD Series
  3. All Radeon HD Series AGP
  4. … no linux driver here

What should I do ?

10

Good point. I did not realize that AMD had a limitation for proprietary HD Series graphics driver for AGP cards for Linux.

Thats not so good.

Looks like the proprietary graphic driver is not an option for you.

You could be out of luck here …

One option is to try go get your PC functioning correct with the ‘radeon’ opensource driver, … try and to figure out why the ‘nomodeset’ option is necessary to use this hardware in Linux (which gives the less than satisfactory) ‘radeonhd’ driver, as opposed to the ‘radeon’ driver. Unfortunately, I have no specific answers for that.

There is a Phoronix thread here on this subject, but no real answers come out: hd 3650 AGP … although in that thread I note one post there claiming a newer kernel (such as 2.6.35) is needed. You could explore that possibility by trying an openSUSE-11.4 RC1 liveCD (which has a 2.6.37 kernel) and see how it functions. If that works, then you could either switch to openSUSE-11.4 in March, or try a tumbleweed update to your 11.3 graphics/kernel.

If its any consolation, I see Ubuntu users are having similar problems

11

‘radeon’ opensource driver problem is next…
http://savepic.org/1313651.jpg
p.1
http://savepic.org/1344370.jpg
p.2

12

‘radeon’ opensource driver problem is next…
http://savepic.org/1325938.jpg
p.1
http://savepic.org/1315698.jpg
p.2

13

I can’t make much out of that other than the error message of a ‘tainted 2q.6.34-12 kernel’.

Did you try my openSUSE-11.4 RC1 liveCD suggestion to see if a newer kernel works better?

14

Please, next time use a site such as ImageBam - Fast, Free Image Hosting and Photo Sharing, instead of posting massive images in this thread and making it practically unreadable.

15

sidebar:

haven’t completely digested all the posts in this thread, but i did notice that the possibility of using the proprietary driver had been abandoned since the webpage selection of drivers did not list anything.

BUT,

i have a 3850HD AGP card using the Catalyst 11.2 driver (2.6.31.14 kernel) which is functioning well albeit some flaky scrolling, and based on my usage of the radeon driver, the proprietary driver should perform much better if it can be made to work.

i have a bookmark that takes me directly to the linux driver download page, so i had no idea that it was not available from the driver selection webpage, there may be an error there.

my 2 cents, good luck.

16

Removed post looks like a hijacked ATI page.

17

This is content of that image above:
0 x FF FF E4 30
sysenter_do_call
sys_init_module
do_one_initcall
__pci_register_driver
driver_register
bus_add_driver
driver_attach
bus_for_each_dev
__driver_attach
driver_probe_device
really_probe
pci_device_probe
drm_get_dev [drm]
radeon_driver_load_kms [radeon]
radeon_modeset_init [radeon]
drm_helper_initial_config [drm_kms_helper]
drm_helper_disable_unused_functions [drm_kms_helper]
atombios_crtc_dpms [radeon]
atombios_blanc_crtc [radeon]
atom_execute_table [radeon]
atom_execute_table_locked [radeon]
common_interrupt
do_IRQ
handle_irq
handle_level_irq
handle_IRQ_event
radeon_driver_irq_handler_kms [radeon]
r600_irq_process [radeon]
error_code
do_page_fault
bad_area_nosemaphore
no_context
__die
panic
dump_stack
show_trace
show_trace_log_lvl
dump_trace

=========================================
fb: conflicting fb hw usage radeondrmfb vs VESA VGA - removing generic driver
Console: switching to colour dummy device 80x25

My vision of situation is next. When I boot system without nomodeset boot option, then happens this:

  1. INIT loads radeon driver.
  2. Initialization function radeon_driver_load_kms of radeon driver calls radeon_modeset_init function.
  3. This radeon_modeset_init function executes some code in ** atom_execute_table_locked **
  4. Then interrupt comes from video card
  5. System executes radeon interrupt handler (because interrupt come from video card)
  6. radeon_driver_irq_handler_kms calls ** r600_irq_process **
  7. And inside this handler happens page fault … do_page_fault
  8. After this page fault we have kernel panic no_context => __die => panic

My supposition A about cause of such behavior:

  1. Interrupt handler called by system has name ** r600_irq_process **
  2. My video card has GPU name RV635, not RV600
  3. So, may be driver mismatches GPU ?

My supposition B about cause of such behavior:

  1. r600_irq_process is good for RV635 GPU. May be it uses some additional modules that extend functionality of basic R600 driver.
  2. But we have software bug in video card’s interrupt handler of radeon driver.
  3. So programmers of radeon driver should debug it and update their release.

My supposition C about cause of such behavior:

  1. All is fine with driver
  2. Something is not fine with the motherboard or BIOS
  3. May be IRQ’s numbers configured incorrect.

My supposition D about cause of such behavior:

  1. My motherboard has been released in 2002 year.
  2. My video card has been released in 2008 year.
  3. These two parts of system may be not 100% compatible due to big age difference.

VERY IMPORTANT, READ THIS
My notes that may help you to help me…

  1. At Windows I had similar problem. At Windows 7 I had video card working perfectly but I had no sound. At Windows Vista and at Windows XP I had sound but I can’t install video driver.
  2. When I installed Windows XP without drivers for video card I had very similar behavior of 2D graphics, i.e. very slow video playback, slow windows redrawing.
  3. Installation of official AMD driver at Windows XP lead to system hung up at boot. (some analogy to kernel panic in Linux that is using open – source ‘radeon’ driver)
  4. The solution was next:
  5. Disable in BIOS AC’97 audio device that is built in the motherboard.
  6. Install video card driver
  7. ** Enable** in BIOS AC’97 audio
  8. After that driver had been installed successfully. And after that I can enable AC’97 audio in BIOS, BUT AC’97 IS NO MORE VISIBLE IN WINDOWS AFTER VIDEO CARD DRIVER INSTALLATION >>> LOOKS LIKE HARDWARE RESOURCE CONFLICT AC’97 AUDIO vs Radeon AGP.
  9. It looks like video card driver has captured some resource and AC’97 cannot be used. Before video driver installation situation was opposite: AC’97 captured this resource and video card driver cannot be installed until AC’97 will be disabled.
  10. I had go to shop and buy PCI sound card because I had no sound in Windows after video card installation…
  11. **My conclusion: it is not driver problem. It even is not OS problem. It is motherboard <<>>video card “compatibility/wrong configuration” problem. **

But Linux open – source ‘radeon’ driver don’t want boot regardless of AC’ 97 is enabled or disabled.
And one more point: lspci detects my video card as ** “Mobility” ** but my computer is Desktop, not notebook. It can be one more argument of **hardware configuration ** problem.

18

Along these lines there have been a LOT of updates to the Radeon driver between openSUSE-11.3 and 11.4 RC1, most of which are due to an updated X, an updated Mesa, and a newer kernel. Which is why I suggested you boot to 11.4 RC1 liveCD and see how it functions. If you notice an improvement there, then it may be worth the risk to update 11.3 to the factory directory used by tumbledweed project.

19

All your problems could be from a bad install. Did you run the media check function on the install media?

Mobility refers to a class of video chip. So the MB maker put a a mobile chip on the board maybe.

20

I have Updated my OpenSuse 11.3 with this DVD (11.4 RC1).
Nothing has changed. Only graphics interface. System still boots only with nomodeset, i.e. only with radeonhd.
Windows redraw slowly, video playback is slow…