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

Re: Report on Debian 8 on Late 2005 G5 with custom kernel, Graphics Issues

Hi Rick,
Interesting..... Someone else has got it running on jessie using a Quadro FX4500 as well.
I have a 6600 with 256Mb VRAM in my possesion and it seemed to work with this as well.
Regarding detection of the nVidia DCB, the main thing to be concerned for now is that you see something like this in dmesg:

[   15.056167] nouveau  [   VBIOS][0000:0a:00.0] using image from OpenFirmware
[   15.056193] nouveau  [   VBIOS][0000:0a:00.0] BIT signature found
[   15.056199] nouveau  [   VBIOS][0000:0a:00.0] version

I suppose this means that nouveau has detected a valid nVidia DCB block from the OpenFirmware device tree, i.e. within the FCODE ROM of the nVidia GPU ROM for now i feel this is all you should be concerned with.

The DCB block describes the nVidia GPU card to the driver, which configures it accordingly i suppose.
Google is your friend, read up on it if interested.

If i recall, your card was a 6600?
So it probably has a different version number, but as long as you see in dmesg something to the effect of :

nouveau  [   VBIOS][0000:0a:00.0] BIT signature found

along with a corresponding version number below, you should be ok.

My hunch is that the nouveau driver has multiple methods of extracting the DCB block from the VGA BIOS / FCODE ROM. Refer to the link below regarding the NvBIOS module parameters:


NvBios: Specify VBIOS source as one of OpenFirmware/PRAMIN/PROM/ACPI/PCIROM/PLATFORM, or a filename passed to request_firmware

As you can see, nouveau goes through the various methods automatically trying to detect a valid DCB block, in our case it OpenFirmware is the correct method.
As it goes through the other various methods automatically, it wont find it on our machines.... And thats why you get an invalid rom message!

Dont know what kernel version is used in sid, but in newer vanilla kernels, i.e. 4.0.0+, i recall even trying to force the detection of the NvBIOS to use OpenFirmware method fails. so it cant find a DCB Block.
So something has changed in the nouveau driver or in the powerpc arch to stop it from working anymore.

Ill tell you more in another thread to work around this... note, not a fix, just a hack!

Please give full dmesg log dumps next time, grepping stuff doesnt give me all the stuff id like to know.
For starters, i dont know exactly what sort of H/W you got!



On Thu, Jul 9, 2015 at 7:42 AM, Rick Thomas <rbthomas@pobox.com> wrote:
Thanks Peter!

I’m sure I’ve got the 3.18 kernel.  I confirmed that with “uname -a” after the boot.  Your yaboot stanza looks exactly (in the relevant particulars) like mine.  And, yes, I did “ybin -v” and checked to make sure the output looked right.

The ppdev and parport_pc anomalies have been there all along.  I don’t think they have anything to do with this problem, though.

The only thing I can think of that might be related (and possibly causing the “Invalid ROM” message) is that the base system is Sid, not Jessie.  The default kernel is

> rbthomas@bigal:~$ uname -a
> Linux bigal 4.0.0-2-powerpc64 #1 SMP Debian 4.0.5-1 (2015-06-16) ppc64 GNU/Linux

So there may be remnants of that lying around that confuse the issue when running with your kernel…

Didn’t you mention you had some trouble with the firmware blob in Sid?

I’ll be interested to look at your dmesg and xorg logs.  I’m not getting to anything involving xorg — the black screen happens before I get to the login prompt.  I have /etc/X11/default-display-manager set to “/bin/false" to prevent X from starting up automatically.  When I want to start X, I do a manual “startx” from a text console.


PS: I’m planning to install a plain-vanilla, fresh-out-of-the-box, Jessie then add your kernel, and I’ll report back on that.

Re, the FCODE ROM:  Where can I read-up (RTFM, but what FM?) on the nouveau module parameters?


On Jul 8, 2015, at 1:04 PM, Peter Saisanas <psaisanas@gmail.com> wrote:

> Hi Rick,
> I should probably start a new thread for reports....
> Seems odd to me that systemd is attempting to load modules ppdev, parport_pc along with snd-powermac...
> Are you sure you are booting up with my actual compiled kernel and initrd?
> Please confirm your /etc/yaboot.conf file has entries for the image and initrd, see below.
> image=/boot/vmlinux-3.18.16-powerpc64
>     label=linux
>     read-only
>     initrd=/boot/initrd.img-3.18.16-powerpc64
>     append=""
> Ensure you have saved the yaboot.conf and then run "ybin -v" to apply the changes to your boot loader.
> I have attached a dmesg log from my system. note my audio hardware isn't working (hardware fault), so disregard the issues with snd-aoa-onyx.
> I have also attached my xorg log for reference.
> The nouveau driver attempts to scan for the nVidia DCB block (within a typical VGA bios or FCODE ROM via alternate methods, it seems to perform this scan automatically, but if curious, refer to the "NvBios" nouveau module parameter.
> Let me know how you go.
> Cheers,
> Peter
> On 08/07/15 19:40, Rick Thomas wrote:
>> On Jun 30, 2015, at 3:48 AM, Rick Thomas <rbthomas@pobox.com>
>>  wrote:
>>> I’ll definitely report on how things work on my hardware with your kernel, but it will have to wait til after the weekend.  4th of July == US Independence Day == lots of neighborhood/family/etc parties and celebrations == no time for playing with computers.
>>> I’m looking forward to sending a report soon!
>>> Rick
>> Hi Peter,
>> I installed your 3.18… kernel.  I get a black screen.
>> Here’s what “journalctl -b | grep -C5 nouveau ” says.  I hope it help.  Note the error from nouveau about “Invalid ROM contents”.  Is that expected?
>> Thanks!
>> Rick
> <dmesg.log><Xorg.0.log>

Reply to: