Bug 111128

Summary: Impossible to change pstate on GT610M
Product: Mesa Reporter: Gaga <gzgagz>
Component: Drivers/DRI/nouveauAssignee: Nouveau Project <nouveau>
Status: RESOLVED NOTABUG QA Contact: Nouveau Project <nouveau>
Severity: normal    
Priority: medium    
Version: 19.1   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:

Description Gaga 2019-07-14 15:20:48 UTC
Hello,
I have an asus UX32VD, with a GT610M inside, which works with the Optimus architecture, handle by PRIME. The bug is currently with the kernel 5.2.0, but it have always been like that as far as I remember.

uname -a :
Linux ga-laptop 5.2.0-arch2-1-ARCH #1 SMP PREEMPT Mon Jul 8 18:18:54 UTC 2019 x86_64 GNU/Linux

lspci :
01:00.0 3D controller: NVIDIA Corporation GF117M [GeForce 610M/710M/810M/820M / GT 620M/625M/630M/720M] (rev a1)

While it is not in use, nouveau shutdown the chipset as expected :
cat /sys/kernel/debug/dri/1/pstate
07: core 270 MHz memory 405 MHz
0f: core 625 MHz memory 900 MHz
DC: core 0 MHz memory 0 MHz

While an app is launched with DRI_PRIME=1, we can see that the chipset is set to 07 :
DRI_PRIME=1 glxgears &
cat /sys/kernel/debug/dri/1/pstate
07: core 270 MHz memory 405 MHz
0f: core 625 MHz memory 900 MHz
DC: core 270 MHz memory 324 MHz

But when I tried to reclock, I have an error :
echo 0f > /sys/kernel/debug/dri/1/pstate
-bash: echo: write error: Function not implemented

And I have this in dmesg:
[  351.283183] nouveau 0000:01:00.0: bus: MMIO write of ffff8c1f FAULT at 6013d4 [ IBUS ]
[  458.362570] nouveau 0000:01:00.0: bus: MMIO write of 0000001f FAULT at 6013d4 [ IBUS ]

Thanks
Comment 1 Karol Herbst 2019-07-14 17:55:42 UTC
we don't support reclocking on fermi GPUs yet. And yours is a  GF117. There has some work be done towards this but progress is rather slow. There might be some branches you can try out, but I don't know the details.

Best would be to join #nouveau on freenode and ask skeggsb or RSpliet.

Closing now as there isn't anything we can do right now besides implementing this feature. Sorry.
Comment 2 Gaga 2019-07-14 19:34:19 UTC
I am sorry, I was sure it was a Kepler GPU, I don't why.
Thank you for your response and your work anyway !

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.