Kernel 2.6.22.2 xorg problem

bgeddy · 08-14-2007, 03:33 PM

I have just compiled a custom kernel for Slackware 12 version 2.6.22.2 using make oldconfig with a working config from a previous kernel 2.6.20.4. My 2.6.21.5 standard kernels are having a problem with ACPI and I have to provide ACPI=noirq to boot so I thought I'd try a shiny new kernel !
Anyway the 2.6.22.2 boots fine with no irq probing errors and i thought all was well. This was till I tried running X when the display showed a lot of fuzzy colours and the PC hard locked - only a power off would get through. I have an onboard Intel 82815 graphics controller. So I have gone back to my generic 2.6.21.5 smp kernel for now but I am curious as to what has caused X to break. Maybe the i810 driver with this kernel doesn't agree with my hardware or something. Any ideas ? Here is a copy of the tail of my Xorg.0.log - there are some weird errors at the bottom of it. "Fatal server error - lockup".

Code:

(II) intel(0): Setting dot clock to 65.0 MHz [ 0x3f 0xa 0x30 ] [ 65 12 3 ]
(II) intel(0): chose watermark 0x2210e000: (tab.freq 65.0)
Error in I810WaitLpRing(), now is 1711378546, start is 1711376545
pgetbl_ctl: 0x17c70001 pgetbl_err: 0x14890000
ipeir: 0 iphdr: ffffffff
LP ring tail: 20 head: 0 len: f001 start 580000
eir: 0 esr: 1 emr: 3d
instdone: ff7a instpm: 0
memmode: 4 instps: 810
hwstam: 9ac7 ier: 0 imr: 9ac7 iir: 0
space: 65496 wanted 65528
(II) intel(0): [drm] removed 1 reserved context for kernel
(II) intel(0): [drm] unmapping 8192 bytes of SAREA 0xd887d000 at 0xb7ba6000

Fatal server error:
lockup

(II) AIGLX: Suspending AIGLX clients for VT switch
Error in I810WaitLpRing(), now is 1711383564, start is 1711381563
pgetbl_ctl: 0x17c70001 pgetbl_err: 0x0
ipeir: 0 iphdr: ffffffff
LP ring tail: 28 head: 0 len: f001 start 580000
eir: 0 esr: 1 emr: 3d
instdone: ff7a instpm: 0
memmode: 4 instps: 810
hwstam: 9ac7 ier: 0 imr: 9ac7 iir: 0
space: 65488 wanted 65528

FatalError re-entered, aborting
lockup

plumbum · 08-15-2007, 02:06 AM

Maybe it make sense to re-generate the xorg.conf by command 'X -configure'?
Have you try that?

bgeddy · 08-15-2007, 06:55 AM

Yes I've tried that already. It gives an error and wont load X at all. The error is :

Code:

Fatal server error:
AddScreen/ScreenInit failed for driver 0

Not the most helpful of error messages !!

bgeddy · 08-15-2007, 10:32 AM

Update

I have been adjusting settings in my xorg.conf and finally got X to work. This is what i did..

1) Take out the Highest Mode lines "1280x1024",
2) commented out "load dri"
3) uncommented VideoRam.

X now works - but try as I might I can't get DRI to work. It works fine with my 2.6.21.5 generic kernel however. Any suggestions are appreciated..

rvdboom · 08-15-2007, 02:01 PM

Maybe there's a bug in the DRI kernel driver for your graphic chip.
I also experienced recently some incompatibilities between frame buffer driver and Xorg. That's for ATI but maybe there's an issue like this for Intel too. Roughly, f I compîle in recent kernel the frame buffer driver for ATI Radeon (and use it in console, of course), my system boot but once X starts, everything goes black and I have to stop the system using power off, as even Ctr-Alt-Suppr or Del does'nt seem to work. If I just use VESA frame buffer output, I get high res console plus X.

bgeddy · 08-15-2007, 07:23 PM

Yes - I have come to that conclusion myself. I have re-compiled my kernel several times closely studying all options regarding agpart,framebuffer,character devices etc but still can't get X to work with 2.6.22.2 and DRI.

Suffice to say I have gone back to Slackware's generic 2.6.21.5 with an initrd as recommended by PV. I still have to append "acpi=noirq" to the booting of the kernel which bugs me but at least DRI works giving me a massive boost to glxgears fps and responsiveness.

Personally I like to get to the bottom of things and hate the fact that the latest kernel causes problems with my hardware. The 2.6.21.5 kernels seem to require an 'append "acpi=irq"' to successfully boot which bugs me.

What worries me is if this is the way new kernels are headed I may have a permanent problem with my hardware. Albeit this system is not that outdated by my standards (P111 celeron 1200Mhz - Intel 828** chipset).

In the past I have had no problems with the latest kernels and never had to supply boot parameters to the kernel. I just hope that my hardware is not now considered antiquated as I have always found one of the beauties of Linux is that it will run on old hardware whereas the "other" OS wont !!

Maybe I'm just being paranoid and it's just my particular chipset that is having problems.

I suppose I shouldn't consider testing 2.6.22.2 on my ancient Intel P11s that I still run !! Anyway - thanks for the reply.

plumbum · 08-15-2007, 10:25 PM

2 bgeddy:
/* X now works - but try as I might I can't get DRI to work. It works fine with my 2.6.21.5 generic kernel however. Any suggestions are appreciated.. */

Try to update driver for your videochip. I have had a very annoying problem with my old ATI card - it didn't want to work with DRI enabled. Once I've got an newer one (from /extra folder form 3rd disk) - the problem has gone.

rvdboom · 08-16-2007, 01:12 AM

Well, if you seem to have lost kernel support somewhere, you can try to report it to the kernel team, they will probably try to fix it. But, as the message above quoted, you'd better be sure it's a kernel issue before reporting to them.
Try to search the X forums and Mailing List archives, there may be other reports of such an issue with the driver of your video card.