Re: G5 Video issue

2020-06-19 Thread Jeroen Diederen

Hi all,

I tested ScottĀ“s kernel image on my iMac G5 and it did indeed not solve 
the nouveau issues.

ScottĀ± thanks for sharing your newly built image!

Regards,
Jeroen




Admittedly I am a kernel compilation newbie however this may not be
the only issue.  After changing the page size and recompiling the
kernel:


scott@debian-10-DG5:~$ uname -a
Linux debian-10-DG5 5.6.14 #1 SMP Mon Jun 15 18:21:13 CDT 2020 ppc64 
GNU/Linux

scott@debian-10-DG5:~$ getconf PAGESIZE
4096

[   23.626150] u3msi: allocated virq 0x20 (hw 0x8) addr 0xfee0
[   23.682413] u3msi: allocated virq 0x21 (hw 0x9) addr 0xfee0
[   25.343462] nouveau :0a:00.0: DRM: 0x14CC: Parsing digital
output script table
[   25.413056] nouveau :0a:00.0: DRM: 0x14CC: Parsing digital
output script table
[   26.899243] tg3 0001:05:04.1 enP1p5s4f1: Link is up at 1000 Mbps, 
full duplex

[   26.899248] tg3 0001:05:04.1 enP1p5s4f1: Flow control is on for TX
and on for RX
[   26.899273] IPv6: ADDRCONF(NETDEV_CHANGE): enP1p5s4f1: link becomes 
ready

[   30.638637] fuse: init (API version 7.31)
[ 2895.662905] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource
0002 [DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000
Xorg[470]] subc 7 class 4097 mthd 0208 data 02000248
[ 2975.214714] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource
0002 [DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000
Xorg[470]] subc 7 class 4097 mthd 0208 data 02000248
[ 2979.828455] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource
0002 [DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000
Xorg[470]] subc 7 class 4097 mthd 0208 data 02000248

The display is still quite problematic.

Scott Thompson




Re: G5 Video issue

2020-06-17 Thread Scott Thompson



> On Jun 5, 2020, at 1:33 AM, John Paul Adrian Glaubitz 
>  wrote:
> 
> On 6/5/20 8:15 AM, Mathieu Malaterre wrote:
>> On Thu, Jun 4, 2020 at 10:50 PM Rui Salvaterra  wrote:
>>> 
>>> Hi, Adrian,
>>> 
>>> A quinta, 4/06/2020, 21:17, John Paul Adrian Glaubitz 
>>>  escreveu:
 
 
 We should revert to 4k pages on ppc64be then. There is no reason to enable
 64k pages for us. That may be well the case for ppc64el though.
>>> 
>>> 
>>> To be honest, I believe there should be two ppc64 kernel flavours. 4 kiB 
>>> pages for smaller, mostly desktop systems (say, Power Mac G5 and IBM 
>>> IntelliStation systems) and 64 kiB for bigger iron. Would that be feasible?
>> 
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=826796
> 
> Let's just switch back to 4k pages. All the hassle is not worth the little
> gain.
> 
> Anyone with big iron is going to run the little-endian port anyway. I will
> open a pull request over the weekend to revert the 64k change for ppc64.
> 
> Adrian
> 

Admittedly I am a kernel compilation newbie however this may not be the only 
issue.  After changing the page size and recompiling the kernel:


scott@debian-10-DG5:~$ uname -a
Linux debian-10-DG5 5.6.14 #1 SMP Mon Jun 15 18:21:13 CDT 2020 ppc64 GNU/Linux
scott@debian-10-DG5:~$ getconf PAGESIZE
4096

[   23.626150] u3msi: allocated virq 0x20 (hw 0x8) addr 0xfee0
[   23.682413] u3msi: allocated virq 0x21 (hw 0x9) addr 0xfee0
[   25.343462] nouveau :0a:00.0: DRM: 0x14CC: Parsing digital output script 
table
[   25.413056] nouveau :0a:00.0: DRM: 0x14CC: Parsing digital output script 
table
[   26.899243] tg3 0001:05:04.1 enP1p5s4f1: Link is up at 1000 Mbps, full duplex
[   26.899248] tg3 0001:05:04.1 enP1p5s4f1: Flow control is on for TX and on 
for RX
[   26.899273] IPv6: ADDRCONF(NETDEV_CHANGE): enP1p5s4f1: link becomes ready
[   30.638637] fuse: init (API version 7.31)
[ 2895.662905] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource 0002 
[DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000 Xorg[470]] subc 7 
class 4097 mthd 0208 data 02000248
[ 2975.214714] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource 0002 
[DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000 Xorg[470]] subc 7 
class 4097 mthd 0208 data 02000248
[ 2979.828455] nouveau :0a:00.0: gr: intr 0010 [ERROR] nsource 0002 
[DATA_ERROR] nstatus 0200 [BAD_ARGUMENT] ch 1 [000d3000 Xorg[470]] subc 7 
class 4097 mthd 0208 data 02000248

The display is still quite problematic.

Scott Thompson



Re: G5 Video issue

2020-06-05 Thread John Paul Adrian Glaubitz
On 6/5/20 8:15 AM, Mathieu Malaterre wrote:
> On Thu, Jun 4, 2020 at 10:50 PM Rui Salvaterra  wrote:
>>
>> Hi, Adrian,
>>
>> A quinta, 4/06/2020, 21:17, John Paul Adrian Glaubitz 
>>  escreveu:
>>>
>>>
>>> We should revert to 4k pages on ppc64be then. There is no reason to enable
>>> 64k pages for us. That may be well the case for ppc64el though.
>>
>>
>> To be honest, I believe there should be two ppc64 kernel flavours. 4 kiB 
>> pages for smaller, mostly desktop systems (say, Power Mac G5 and IBM 
>> IntelliStation systems) and 64 kiB for bigger iron. Would that be feasible?
> 
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=826796

Let's just switch back to 4k pages. All the hassle is not worth the little
gain.

Anyone with big iron is going to run the little-endian port anyway. I will
open a pull request over the weekend to revert the 64k change for ppc64.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaub...@debian.org
`. `'   Freie Universitaet Berlin - glaub...@physik.fu-berlin.de
  `-GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913



Re: G5 Video issue

2020-06-05 Thread Mathieu Malaterre
On Thu, Jun 4, 2020 at 10:50 PM Rui Salvaterra  wrote:
>
> Hi, Adrian,
>
> A quinta, 4/06/2020, 21:17, John Paul Adrian Glaubitz 
>  escreveu:
>>
>>
>> We should revert to 4k pages on ppc64be then. There is no reason to enable
>> 64k pages for us. That may be well the case for ppc64el though.
>
>
> To be honest, I believe there should be two ppc64 kernel flavours. 4 kiB 
> pages for smaller, mostly desktop systems (say, Power Mac G5 and IBM 
> IntelliStation systems) and 64 kiB for bigger iron. Would that be feasible?

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=826796



> Cheers,
> Rui



Re: G5 Video issue

2020-06-04 Thread Rui Salvaterra
Hi, Adrian,

A quinta, 4/06/2020, 21:17, John Paul Adrian Glaubitz <
glaub...@physik.fu-berlin.de> escreveu:

>
> We should revert to 4k pages on ppc64be then. There is no reason to enable
> 64k pages for us. That may be well the case for ppc64el though.
>

To be honest, I believe there should be two ppc64 kernel flavours. 4 kiB
pages for smaller, mostly desktop systems (say, Power Mac G5 and IBM
IntelliStation systems) and 64 kiB for bigger iron. Would that be feasible?

Cheers,
Rui

>


Re: G5 Video issue

2020-06-04 Thread John Paul Adrian Glaubitz
On 6/4/20 10:11 PM, John Paul Adrian Glaubitz wrote:
> On 6/4/20 9:45 PM, Mathieu Malaterre wrote:
>> On Thu, Jun 4, 2020 at 5:36 PM Scott Thompson  wrote:
>>>
>>> I have a Power Mac G5 dual 2.3GHz with 8GB of ram model PowerMac11,2
>>> with a Nvidia GeForce 6600.  The machine is triple booted with OS X
>>> Leopard, Ubuntu 16.04.6, and Debian booting via GRUB.  Video works fine
>>> under Leopard and Ubuntu but is basically unusable under Debian once the
>>> desktop is loaded.  As one can see in the dmesg file there seems to be
>>> nouveau interrupt issues.
>>
>> [0.00] Linux version 5.6.0-2-powerpc64
>> (debian-ker...@lists.debian.org) (gcc version 9.3.0 (Debian 9.3.0-12))
>> #1 SMP Debian 5.6.14-1 (2020-05-23)
>>
>> Most likely:
>>
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=767969
> 
> Why hasn't this been upstreamed?

Ok, having read through everything, it seems non-trivial.

We should revert to 4k pages on ppc64be then. There is no reason to enable
64k pages for us. That may be well the case for ppc64el though.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaub...@debian.org
`. `'   Freie Universitaet Berlin - glaub...@physik.fu-berlin.de
  `-GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913



Re: G5 Video issue

2020-06-04 Thread John Paul Adrian Glaubitz
On 6/4/20 9:45 PM, Mathieu Malaterre wrote:
> On Thu, Jun 4, 2020 at 5:36 PM Scott Thompson  wrote:
>>
>> I have a Power Mac G5 dual 2.3GHz with 8GB of ram model PowerMac11,2
>> with a Nvidia GeForce 6600.  The machine is triple booted with OS X
>> Leopard, Ubuntu 16.04.6, and Debian booting via GRUB.  Video works fine
>> under Leopard and Ubuntu but is basically unusable under Debian once the
>> desktop is loaded.  As one can see in the dmesg file there seems to be
>> nouveau interrupt issues.
> 
> [0.00] Linux version 5.6.0-2-powerpc64
> (debian-ker...@lists.debian.org) (gcc version 9.3.0 (Debian 9.3.0-12))
> #1 SMP Debian 5.6.14-1 (2020-05-23)
> 
> Most likely:
> 
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=767969

Why hasn't this been upstreamed?

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaub...@debian.org
`. `'   Freie Universitaet Berlin - glaub...@physik.fu-berlin.de
  `-GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913



Re: G5 Video issue

2020-06-04 Thread Mathieu Malaterre
On Thu, Jun 4, 2020 at 5:36 PM Scott Thompson  wrote:
>
> I have a Power Mac G5 dual 2.3GHz with 8GB of ram model PowerMac11,2
> with a Nvidia GeForce 6600.  The machine is triple booted with OS X
> Leopard, Ubuntu 16.04.6, and Debian booting via GRUB.  Video works fine
> under Leopard and Ubuntu but is basically unusable under Debian once the
> desktop is loaded.  As one can see in the dmesg file there seems to be
> nouveau interrupt issues.

[0.00] Linux version 5.6.0-2-powerpc64
(debian-ker...@lists.debian.org) (gcc version 9.3.0 (Debian 9.3.0-12))
#1 SMP Debian 5.6.14-1 (2020-05-23)

Most likely:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=767969

> Thanks
>
> Scott Thompson
>
>
>


-- 
Mathieu



G5 Video issue

2020-06-04 Thread Scott Thompson
I have a Power Mac G5 dual 2.3GHz with 8GB of ram model PowerMac11,2 
with a Nvidia GeForce 6600.  The machine is triple booted with OS X 
Leopard, Ubuntu 16.04.6, and Debian booting via GRUB.  Video works fine 
under Leopard and Ubuntu but is basically unusable under Debian once the 
desktop is loaded.  As one can see in the dmesg file there seems to be 
nouveau interrupt issues. ThanksScott Thompson 

dmesg.Debian
Description: Binary data