DRM IOCTLS come regularly(every minute) still even after screen off

Marek Olšák maraeo at gmail.com
Sat Sep 18 07:19:38 UTC 2021


Hi,

Printing the backtrace from si_flush_gfx_cs while /etc/environment contains
GALLIUM_THREAD=0 at boot should show which GL call and X call caused the
flush.

Marek

On Thu, Sep 16, 2021 at 10:58 PM Quan, Evan <Evan.Quan at amd.com> wrote:

> [Public]
>
>
>
> > -----Original Message-----
> > From: Michel Dänzer <michel at daenzer.net>
> > Sent: Thursday, September 16, 2021 4:12 PM
> > To: Christian König <ckoenig.leichtzumerken at gmail.com>; Quan, Evan
> > <Evan.Quan at amd.com>
> > Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian
> > <Christian.Koenig at amd.com>; 'amd-gfx at lists.freedesktop.org' <amd-
> > gfx at lists.freedesktop.org>
> > Subject: Re: DRM IOCTLS come regularly(every minute) still even after
> screen
> > off
> >
> > On 2021-09-16 08:27, Christian König wrote:
> > > Good morning,
> > >
> > > for the background I advised to contact you Michel since we couldn't
> find
> > an explanation.
> > >
> > > Alex explanation might be correct, but the CS seems to come from the X
> > server and is always roughly 60 seconds after going into power save.
> > >
> > > Any idea what that could be?
> >
> > For the DRM_IOCTL_MODE_* ioctls, attaching gdb to Xorg and setting a
> > breakpoint on drmIoctl should show where they're coming from.
> >
> > The AMDGPU_CS ioctls are probably from glamor calling OpenGL, most likely
> > for X11 drawing protocol requests. Note that in this case drmIoctl/ioctl
> will
> > likely be called from a separate thread. Pierre / Marek should be able
> to help
> > you find out where the OpenGL calls are coming from.
> >
> >
> > In general, Xorg only ever does anything in response to:
> >
> > * X11 protocol requests, i.e. on behalf of clients.
> > * Input events, i.e. on behalf of the user.
> > * Other kernel events, e.g. hotplug events.
> >
> Thanks for sharing these.
> >
> > FWIW, with only xfwm4 & xterm running on Xorg, I'm not seeing any
> drmIoctl
> > calls during DPMS off. So the ioctls you're seeing are most likely
> triggered by
> > clients.
> >
> Got it. Thanks!
>
> Evan
> >
> > --
> > Earthling Michel Dänzer               |
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fredh
> > at.com%2F&data=04%7C01%7CEvan.Quan%40amd.com%7C296d316427
> > b4469f8eb908d978e9b46a%7C3dd8961fe4884e608e11a82d994e183d%7C0%7
> > C0%7C637673767426116632%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4w
> > LjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&am
> > p;sdata=gBj4idZGCuaPdyaH1g1mfuDqPCcKhFMjPCbqpM8drlg%3D&res
> > erved=0
> > Libre software enthusiast             |             Mesa and X developer
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/amd-gfx/attachments/20210918/c4cd2c7c/attachment-0001.htm>


More information about the amd-gfx mailing list