[PATCH] drm/vkms: provide a parent device to drm_dev_init()

[PATCH] drm/vkms: provide a parent device to drm_dev_init()

[Emil Velikov]

From: Emil Velikov <emil.velikov@collabora.com>

Earlier commit updated the vgem driver to improve the topology, by
passing a platform device as parent to drm_dev_init(). Shortly
afterwords we updated the core function to BUG() in order to catch any
buggy drivers passing NULL as parent.

While I missed the vkms driver (as the patch predates vkms by a few
months), the BUG caught the issue within couple of hours.

Swap the drm_dev_init <> platform_device_register_simple order, to
the driver back to life.

Fixes: f08877e79485 ("drm: BUG_ON if passing NULL parent to drm_dev_init")
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Deepak Sharma <deepak.sharma@amd.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>
Cc: Sean Paul <sean@poorly.run>
Cc: Haneen Mohammed <hamohammed.sa@gmail.com>
Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
---
Rodrigo, Haneen, my plan is to push the patch to the drm-misc tree. An
Ack/Rb will be appreciated.

Thanks
Emil
---
 drivers/gpu/drm/vkms/vkms_drv.c | 21 +++++++++++----------
 1 file changed, 11 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/vkms/vkms_drv.c b/drivers/gpu/drm/vkms/vkms_drv.c
index 07cfde1b4132..a3d57e0f5ee5 100644
--- a/drivers/gpu/drm/vkms/vkms_drv.c
+++ b/drivers/gpu/drm/vkms/vkms_drv.c
@@ -108,17 +108,18 @@ static int __init vkms_init(void)
 	if (!vkms_device)
 		return -ENOMEM;
 
-	ret = drm_dev_init(&vkms_device->drm, &vkms_driver, NULL);
-	if (ret)
-		goto out_free;
-
 	vkms_device->platform =
 		platform_device_register_simple(DRIVER_NAME, -1, NULL, 0);
 	if (IS_ERR(vkms_device->platform)) {
 		ret = PTR_ERR(vkms_device->platform);
-		goto out_fini;
+		goto out_free;
 	}
 
+	ret = drm_dev_init(&vkms_device->drm, &vkms_driver,
+			   &vkms_device->platform->dev);
+	if (ret)
+		goto out_unregister;
+
 	vkms_device->drm.irq_enabled = true;
 
 	ret = drm_vblank_init(&vkms_device->drm, 1);
@@ -129,20 +130,20 @@ static int __init vkms_init(void)
 
 	ret = vkms_modeset_init(vkms_device);
 	if (ret)
-		goto out_unregister;
+		goto out_fini;
 
 	ret = drm_dev_register(&vkms_device->drm, 0);
 	if (ret)
-		goto out_unregister;
+		goto out_fini;
 
 	return 0;
 
-out_unregister:
-	platform_device_unregister(vkms_device->platform);
-
 out_fini:
 	drm_dev_fini(&vkms_device->drm);
 
+out_unregister:
+	platform_device_unregister(vkms_device->platform);
+
 out_free:
 	kfree(vkms_device);
 	return ret;
-- 
2.19.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH] drm/vkms: provide a parent device to drm_dev_init()

[Daniel Vetter]

On Fri, Oct 26, 2018 at 11:05:50AM +0100, Emil Velikov wrote:
> From: Emil Velikov <emil.velikov@collabora.com>
> 
> Earlier commit updated the vgem driver to improve the topology, by
> passing a platform device as parent to drm_dev_init(). Shortly
> afterwords we updated the core function to BUG() in order to catch any
> buggy drivers passing NULL as parent.
> 
> While I missed the vkms driver (as the patch predates vkms by a few
> months), the BUG caught the issue within couple of hours.
> 
> Swap the drm_dev_init <> platform_device_register_simple order, to
> the driver back to life.
> 
> Fixes: f08877e79485 ("drm: BUG_ON if passing NULL parent to drm_dev_init")
> Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
> Cc: Deepak Sharma <deepak.sharma@amd.com>
> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> Cc: Sean Paul <sean@poorly.run>
> Cc: Haneen Mohammed <hamohammed.sa@gmail.com>
> Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
> Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
> ---
> Rodrigo, Haneen, my plan is to push the patch to the drm-misc tree. An
> Ack/Rb will be appreciated.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

btw just read through this stuff again, and noticed how vgem is using
drm_gem_prime_import_dev(). I think now that your patches have landed we
could get rid of that?
-Daniel

> 
> Thanks
> Emil
> ---
>  drivers/gpu/drm/vkms/vkms_drv.c | 21 +++++++++++----------
>  1 file changed, 11 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/gpu/drm/vkms/vkms_drv.c b/drivers/gpu/drm/vkms/vkms_drv.c
> index 07cfde1b4132..a3d57e0f5ee5 100644
> --- a/drivers/gpu/drm/vkms/vkms_drv.c
> +++ b/drivers/gpu/drm/vkms/vkms_drv.c
> @@ -108,17 +108,18 @@ static int __init vkms_init(void)
>  	if (!vkms_device)
>  		return -ENOMEM;
>  
> -	ret = drm_dev_init(&vkms_device->drm, &vkms_driver, NULL);
> -	if (ret)
> -		goto out_free;
> -
>  	vkms_device->platform =
>  		platform_device_register_simple(DRIVER_NAME, -1, NULL, 0);
>  	if (IS_ERR(vkms_device->platform)) {
>  		ret = PTR_ERR(vkms_device->platform);
> -		goto out_fini;
> +		goto out_free;
>  	}
>  
> +	ret = drm_dev_init(&vkms_device->drm, &vkms_driver,
> +			   &vkms_device->platform->dev);
> +	if (ret)
> +		goto out_unregister;
> +
>  	vkms_device->drm.irq_enabled = true;
>  
>  	ret = drm_vblank_init(&vkms_device->drm, 1);
> @@ -129,20 +130,20 @@ static int __init vkms_init(void)
>  
>  	ret = vkms_modeset_init(vkms_device);
>  	if (ret)
> -		goto out_unregister;
> +		goto out_fini;
>  
>  	ret = drm_dev_register(&vkms_device->drm, 0);
>  	if (ret)
> -		goto out_unregister;
> +		goto out_fini;
>  
>  	return 0;
>  
> -out_unregister:
> -	platform_device_unregister(vkms_device->platform);
> -
>  out_fini:
>  	drm_dev_fini(&vkms_device->drm);
>  
> +out_unregister:
> +	platform_device_unregister(vkms_device->platform);
> +
>  out_free:
>  	kfree(vkms_device);
>  	return ret;
> -- 
> 2.19.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH] drm/vkms: provide a parent device to drm_dev_init()

[Emil Velikov]

On Fri, 26 Oct 2018 at 11:27, Daniel Vetter <daniel@ffwll.ch> wrote:
>
> On Fri, Oct 26, 2018 at 11:05:50AM +0100, Emil Velikov wrote:
> > From: Emil Velikov <emil.velikov@collabora.com>
> >
> > Earlier commit updated the vgem driver to improve the topology, by
> > passing a platform device as parent to drm_dev_init(). Shortly
> > afterwords we updated the core function to BUG() in order to catch any
> > buggy drivers passing NULL as parent.
> >
> > While I missed the vkms driver (as the patch predates vkms by a few
> > months), the BUG caught the issue within couple of hours.
> >
> > Swap the drm_dev_init <> platform_device_register_simple order, to
> > the driver back to life.
> >
> > Fixes: f08877e79485 ("drm: BUG_ON if passing NULL parent to drm_dev_init")
> > Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
> > Cc: Deepak Sharma <deepak.sharma@amd.com>
> > Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> > Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> > Cc: Sean Paul <sean@poorly.run>
> > Cc: Haneen Mohammed <hamohammed.sa@gmail.com>
> > Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
> > Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> > Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
> > ---
> > Rodrigo, Haneen, my plan is to push the patch to the drm-misc tree. An
> > Ack/Rb will be appreciated.
>
> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
>
Thanks, pushed via dim. Since this way breaking the 0day buildbot,
I've decided to not wait too long.

> btw just read through this stuff again, and noticed how vgem is using
> drm_gem_prime_import_dev(). I think now that your patches have landed we
> could get rid of that?

AFAICT the exynos is also using drm_gem_prime_import_dev. Having a
slightly closer look, it uses a distinct component driver for DMA.
One that varies across HW generation.

It should be doable to rework/remove that, since my knowledge in
exynos is limited, I'd refrain for now.

Thanks
-Emil
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel