[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: X stopped working with 5.14 on iBook



On 10/6/21 7:39 PM, Riccardo Mottola wrote:
> Hi Stan,
> 
> On 2021-10-06 16:59:36 +0000 Stan Johnson <userm57@yahoo.com> wrote:
> 
> 
>> I can confirm that X is not working on a PowerPC G4 Cube with the stock
>> Debian kernel "vmlinux-5.14.0-2-powerpc". X does work with the stock
>> Debian kernel "vmlinux-5.10.0-8-powerpc". In both cases, I'm using the
>> latest Debian SID with Xfce. In 5.14.0-2, wdm runs but the system
>> console screen remains blank (the LCD backlight is on).
> 
> Same behaviour here.
> ...
> 
>> I'll attempt a bisect, starting with mainline v5.14 and going backwards
>> if necessary (it looks like 5.12.9 worked; I'm not sure about v5.13). If
>> there's a kernel regression, I'll confirm the regression exists on both
>> the G4 Cube and the PB G4. I don't have an iBook to test.
> 
> You have it on your Cube, two PowerBooks, mine and yours, and my iBook.
> So I guess if you bisect on your system it is enough, no need to try my
> iBook. My iMac G5 however has 5.14 running with X11. Could be a
> different HW architecture or a 32bit issue or something with the card.
> in my case, all ATI though, also the iMac.
> Did you see the error I am getting about BIOS ROM at boot? Do you get it
> too?


I see a similar error, but I think it happens because the kernel doesn't
see a BIOS on the video card.

On the Cube, I see this:

# dmesg | grep -i aty
[    0.000000] Kernel command line: root=/dev/sda12 ro
video=aty128fb:1024x768-32@60
[    0.074580] aty128fb 0000:00:10.0: enabling device (0086 -> 0087)
[    0.075045] aty128fb 0000:00:10.0: Invalid PCI ROM header signature:
expecting 0xaa55, got 0x1111
[    0.075103] aty128fb: ROM failed to map
[    0.075119] aty128fb: BIOS not located, guessing timings.
[    0.075147] aty128fb: Rage128 PF PRO AGP [chip rev 0x1]
[    0.323018] fb0: ATY Rage128 frame buffer device on Rage128 PF PRO AGP


> 
> Regards,
> Riccardo
> 


I tested these kernels from mainline Linux on the Cube:

5.14.0 works
5.15.0-rc3 works
5.15.0-rc4-00019-g5af4055fa813 works

Kernel 5.14.9 from www.kernel.org also works.

Running "apt-get install kernel-source" downloads:
  "linux-source-5.14 (5.14.9-2)".

Compiling the downloaded kernel-source using Debian's .config file
"config.powerpc_none_powerpc.xz" resulted in a 200 MB kernel (perhaps
because it included all possible options), and I couldn't get it to boot
(but I also didn't compile or install any modules). But compiling using
the same .config file (see attached) that I've been using for testing
5.13 kernels, and that I also used for the above kernels that worked,
results in a working 12 MB kernel, and X is also working:

# uname -a
Linux ppc-cube 5.14.9-pmac #2 SMP Wed Oct 6 21:38:58 MDT 2021 ppc GNU/Linux
# ls -l vmlinux
-rwxr-xr-x 1 root root 12382136 Oct  6 15:39 vmlinux

So I think the problem with the default kernel in Debian SID (5.14.0-2)
has already been fixed, or perhaps there was an issue with the options
that were selected for the .config file. Either way, it doesn't appear
that a bisect is needed.

-Stan

Attachment: dot-config-5.14-pmac.xz
Description: Binary data


Reply to: