Re: Try to address the drm_debugfs issues

[Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>]

On Tue, Feb 14, 2023 at 10:28:24AM +0100, Christian König wrote:
> Am 14.02.23 um 09:59 schrieb Stanislaw Gruszka:
> > On Thu, Feb 09, 2023 at 09:18:35AM +0100, Christian König wrote:
> > > Hello everyone,
> > > 
> > > the drm_debugfs has a couple of well known design problems.
> > > 
> > > Especially it wasn't possible to add files between initializing and registering
> > > of DRM devices since the underlying debugfs directory wasn't created yet.
> > > 
> > > The resulting necessity of the driver->debugfs_init() callback function is a
> > > mid-layering which is really frowned on since it creates a horrible
> > > driver->DRM->driver design layering.
> > > 
> > > The recent patch "drm/debugfs: create device-centered debugfs functions" tried
> > > to address those problem, but doesn't seem to work correctly. This looks like
> > > a misunderstanding of the call flow around drm_debugfs_init(), which is called
> > > multiple times, once for the primary and once for the render node.
> > > 
> > > So what happens now is the following:
> > > 
> > > 1. drm_dev_init() initially allocates the drm_minor objects.
> > > 2. ... back to the driver ...
> > > 3. drm_dev_register() is called.
> > > 
> > > 4. drm_debugfs_init() is called for the primary node.
> > > 5. drm_framebuffer_debugfs_init(), drm_client_debugfs_init() and
> > >     drm_atomic_debugfs_init() call drm_debugfs_add_file(s)() to add the files
> > >     for the primary node.
> > > 6. The driver->debugfs_init() callback is called to add debugfs files for the
> > >     primary node.
> > > 7. The added files are consumed and added to the primary node debugfs directory.
> > > 
> > > 8. drm_debugfs_init() is called for the render node.
> > > 9. drm_framebuffer_debugfs_init(), drm_client_debugfs_init() and
> > >     drm_atomic_debugfs_init() call drm_debugfs_add_file(s)() to add the files
> > >     again for the render node.
> > > 10. The driver->debugfs_init() callback is called to add debugfs files for the
> > >      render node.
> > > 11. The added files are consumed and added to the render node debugfs directory.
> > > 
> > > 12. Some more files are added through drm_debugfs_add_file().
> > > 13. drm_debugfs_late_register() add the files once more to the primary node
> > >      debugfs directory.
> > > 14. From this point on files added through drm_debugfs_add_file() are simply ignored.
> > > 15. ... back to the driver ...
> > > 
> > > Because of this the dev->debugfs_mutex lock is also completely pointless since
> > > any concurrent use of the interface would just randomly either add the files to
> > > the primary or render node or just not at all.
> > > 
> > > Even worse is that this implementation nails the coffin for removing the
> > > driver->debugfs_init() mid-layering because otherwise drivers can't control
> > > where their debugfs (primary/render node) are actually added.
> > > 
> > > This patch set here now tries to clean this up a bit, but most likely isn't
> > > fully complete either since I didn't audit every driver/call path.
> > > 
> > > Please comment/discuss.
> > What is end goal here regarding debugfs in DRM ? My undersigning is that
> > the direction is get rid of debugfs_init callback as described in:
> > https://cgit.freedesktop.org/drm/drm-misc/tree/Documentation/gpu/todo.rst#n511
> > and also make it driver/device-centric instead of minor-centric as
> > described here:
> > https://cgit.freedesktop.org/drm/drm-misc/commit/?id=99845faae7099cd704ebf67514c1157c26960a	
> 
> Well my main goal is to get rid of the debugfs_init() mid-layering in the
> mid term, everything else is just nice to have.
> 
> > I'm asking from accel point of view. We can make things there as they
> > should look like at the end for DRM, since currently no drivers have
> > established their interfaces and they can be changed.
> > 
> > Is drivers/device-centric mean we should use drm_dev->unique for debugfs
> > dir entry name instead of minor ?
> 
> Oh, good idea! That would also finally make it a bit less problematic to
> figure out which PCI or platform device corresponds to which debugfs
> directory.
> 
> Only potential problem I see is that we would need to rename the directory
> should a driver every decide to set drm_dev->unique to something else than
> the default. But a quick check shows no users of drm_dev_set_unique(), so we
> could potentially just unexport the function
>
> > Or perhaps we should have 2 separate dir entries: one (old dri/minor/)
> > for device drm debugfs files and other one for driver specific files ?
> 
> How about we just create symlinks between the old and the new directory for
> now which we remove after everything has settled again?

Yes, that would make perfect sense. 

However my idea was a bit different, that we have separate directories
one for drm specific debugfs files (i.e. clints, framebuffer, gem, ... )
and another one for driver specific files (registers, whatever
individual needs for debugging). I'm just considering different options.

> > Also what regarding sysfs ? Should we do something with accel_sysfs_device_minor ?
> 
> I see sysfs as a different and probably even more complicated topic.

I wish to have some clear guidance how things should be done regarding
sysfs. But I guess we can stick with accel_sysfs_device_minor for accel
as it is currently. And make changes along with whole DRM.

Regards
Stanislaw