Re: [PATCH] drm/radeon: don't add files at control minor debugfs directory

From: Daniel Vetter
Date: Mon Dec 05 2016 - 04:52:38 EST

On Mon, Dec 05, 2016 at 09:48:02AM +0100, Christian König wrote:
> Am 05.12.2016 um 09:39 schrieb Nicolai Stange:
> > Christian König <christian.koenig@xxxxxxx> writes:
> >
> > > Am 05.12.2016 um 08:27 schrieb Daniel Vetter:
> > > > On Sat, Dec 03, 2016 at 03:47:00PM +0100, Nicolai Stange wrote:
> > > > > Since commit 8a357d10043c ("drm: Nerf DRM_CONTROL nodes"), a
> > > > > struct drm_device's ->control member is always NULL.
> > > > >
> > > > > In the case of CONFIG_DEBUG_FS=y, radeon_debugfs_add_files() accesses
> > > > > ->control->debugfs_root though. This results in the following Oops:
> > > > >
> > > > > BUG: unable to handle kernel NULL pointer dereference at 0000000000000018
> > > > > IP: radeon_debugfs_add_files+0x90/0x100 [radeon]
> > > > > PGD 0
> > > > > Oops: 0000 [#1] SMP
> > > > > [...]
> > > > > Call Trace:
> > > > > ? work_on_cpu+0xb0/0xb0
> > > > > radeon_fence_driver_init+0x120/0x150 [radeon]
> > > > > si_init+0x122/0xd50 [radeon]
> > > > > ? _raw_spin_unlock_irq+0x2c/0x40
> > > > > ? device_pm_check_callbacks+0xb3/0xc0
> > > > > radeon_device_init+0x958/0xda0 [radeon]
> > > > > radeon_driver_load_kms+0x9a/0x210 [radeon]
> > > > > drm_dev_register+0xa9/0xd0 [drm]
> > > > > drm_get_pci_dev+0x9c/0x1e0 [drm]
> > > > > radeon_pci_probe+0xb8/0xe0 [radeon]
> > > > > [...]
> > > > >
> > > > > Fix this by omitting the drm_debugfs_create_files() call for the
> > > > > control minor debugfs directory which is now non-existent anyway.
> > > > >
> > > > > Fixes: 8a357d10043c ("drm: Nerf DRM_CONTROL nodes")
> > > > > Signed-off-by: Nicolai Stange <nicstange@xxxxxxxxx>
> > > > Applied to drm-misc with Dave's irc ack, thanks for your patch.
> > > If it's still worth it the patch is Reviewed-by: Christian König
> > > <christian.koenig@xxxxxxx>.
> > >
> > > On the other hand when ->control is always NULL, why do we still have
> > > ->control anyway?
> > Yes, I was wondering about that, too.
> >
> > Quoting from 8a357d10043c ("drm: Nerf DRM_CONTROL nodes"):
> >
> > Since I don't like dead uabi, let's remove it. But since this would be
> > a really big change I think it's better to start out small by simply
> > not registering anything. We can garbage-collect the dead code later
> > on, once we're sure it's really not used anywhere.
> >
> > I'd too prefer compile time errors by purging ->control here. Daniel?
>
> Seconded.

We're super-late for 4.10, I think that'd would need to be postponened for
4.11. Not need to wait with creating the patches (drm-misc is always
open), but wee need to plug the oopses first.

> > > And BTW: Please double check the other drivers as well.
> > # git grep '\->control' -- drivers/gpu/
> > drivers/gpu/drm/amd/amdgpu/amdgpu_device.c: adev->ddev->control->debugfs_root,
> > drivers/gpu/drm/amd/amdgpu/amdgpu_device.c: adev->ddev->control);
> > drivers/gpu/drm/amd/amdgpu/amdgpu_device.c: adev->ddev->control);
> >
> > Oops.
>
> Yeah, that's what I expected as well but Daniel said it would only affect
> qxl.

No idea why I've missed that, I guess coffee wasn't working.

> > drivers/gpu/drm/amd/amdkfd/kfd_dbgdev.c: ib_packet->control = (1 << 23) | (1 << 31) |
> > drivers/gpu/drm/drm_drv.c: return &dev->control;
> >
> > That's drm_minor_get_slot(dev, type), but grepping for DRM_MINOR_CONTROL
> > doesn't yield anything -> dead code.
> >
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: switch (sdvo->controlled_output) {
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS0;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS1;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= type;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB0;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB1;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS0;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS1;
> > drivers/gpu/drm/gma500/psb_intel_sdvo.c: psb_intel_sdvo->controlled_output = 0;
> > drivers/gpu/drm/i915/intel_sdvo.c: switch (sdvo->controlled_output) {
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS0;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS1;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= type;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB0;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB1;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS0;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS1;
> > drivers/gpu/drm/i915/intel_sdvo.c: intel_sdvo->controlled_output = 0;
> > drivers/gpu/drm/msm/msm_debugfs.c: ret = late_init_minor(dev->control);
> >
> > Not an oops but dead code.
> >
> > drivers/gpu/drm/qxl/qxl_debugfs.c: qdev->ddev->control->debugfs_root,
> > drivers/gpu/drm/qxl/qxl_debugfs.c: qdev->ddev->control);
> > drivers/gpu/drm/qxl/qxl_debugfs.c: qdev->ddev->control);
> >
> > Oops.
> >
> >
> > I'll send compile-only tested patches either tonight or tomorrow. Shall
> > they cover the oopses only or the dead code as well?
>
> Please start with the ops, cause we certainly will get complains about that
> rather fast.

I sent out qxl already, and yes oops patches first pls.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch