Re: [PATCH] drm/irq: Modify the return value type of drm_irq_uninstall

From: Daniel Vetter
Date: Mon Nov 02 2020 - 08:06:11 EST


On Mon, Nov 2, 2020 at 1:48 PM Thomas Zimmermann <tzimmermann@xxxxxxx> wrote:
>
> Hi
>
> Am 02.11.20 um 13:38 schrieb Tian Tao:
> > There is no driver to use the return value of drm_irq_uninstal,
> > so modify the return value type of drm_irq_uninstal to void.
> >
> > Signed-off-by: Tian Tao <tiantao6@xxxxxxxxxxxxx>
> > ---
> > drivers/gpu/drm/drm_irq.c | 13 ++++++-------
> > include/drm/drm_irq.h | 2 +-
> > 2 files changed, 7 insertions(+), 8 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
> > index 7537a3d..45e6471 100644
> > --- a/drivers/gpu/drm/drm_irq.c
> > +++ b/drivers/gpu/drm/drm_irq.c
> > @@ -166,14 +166,14 @@ EXPORT_SYMBOL(drm_irq_install);
> > * Returns:
> > * Zero on success or a negative error code on failure.
> > */
> > -int drm_irq_uninstall(struct drm_device *dev)
> > +void drm_irq_uninstall(struct drm_device *dev)
> > {
> > unsigned long irqflags;
> > bool irq_enabled;
> > int i;
> >
> > if (!dev->irq_enabled || !dev)
> > - return 0;
> > + return;
> >
> > irq_enabled = dev->irq_enabled;
> > dev->irq_enabled = false;
> > @@ -200,8 +200,8 @@ int drm_irq_uninstall(struct drm_device *dev)
> > spin_unlock_irqrestore(&dev->vbl_lock, irqflags);
> > }
> >
> > - if (!irq_enabled)
> > - return -EINVAL;
> > + if (!drm_WARN_ON(dev, !irq_enabled))
> > + return;
> >
> > DRM_DEBUG("irq=%d\n", dev->irq);
> >
> > @@ -213,7 +213,6 @@ int drm_irq_uninstall(struct drm_device *dev)
> >
> > free_irq(dev->irq, dev);
> >
> > - return 0;
> > }
> > EXPORT_SYMBOL(drm_irq_uninstall);
> >
> > @@ -250,10 +249,10 @@ int drm_legacy_irq_control(struct drm_device *dev, void *data,
> > return ret;
> > case DRM_UNINST_HANDLER:
> > mutex_lock(&dev->struct_mutex);
> > - ret = drm_irq_uninstall(dev);
> > + drm_irq_uninstall(dev);
>
> Oh, there actually is a user of this result! I grep'ed for this but
> didn't see it. I'm sorry for misleading you here.
>
> This is ioctl code and who which program depends on it.So we cannot
> actually drop the result code.
>
> I'll just ack your original patch, or you could add the managed
> interface that I described and convert hibmc to it. Your choice, let me
> know.

It's old UMS gunk, no one cares :-)

If you're paranoid, make an internal __drm_irq_uninstall function
which keeps the return value. But what we probably want here is to
just split this up into 2 functions, drm_legacy_irq_uninstall, which
does the validation and additional check, and then calls
drm_irq_uninstall, which is for kms drivers, and which just has a
WARN_ON(!dev->irq_installed) or so to catch driver bugs.
-Daniel

>
> Best regards
> Thomas
>
> > mutex_unlock(&dev->struct_mutex);
> >
> > - return ret;
> > + return 0;
> > default:
> > return -EINVAL;
> > }
> > diff --git a/include/drm/drm_irq.h b/include/drm/drm_irq.h
> > index d77f6e6..d9f6ec0 100644
> > --- a/include/drm/drm_irq.h
> > +++ b/include/drm/drm_irq.h
> > @@ -27,6 +27,6 @@
> > struct drm_device;
> >
> > int drm_irq_install(struct drm_device *dev, int irq);
> > -int drm_irq_uninstall(struct drm_device *dev);
> > +void drm_irq_uninstall(struct drm_device *dev);
> >
> > #endif
> >
>
> --
> Thomas Zimmermann
> Graphics Driver Developer
> SUSE Software Solutions Germany GmbH
> Maxfeldstr. 5, 90409 Nürnberg, Germany
> (HRB 36809, AG Nürnberg)
> Geschäftsführer: Felix Imendörffer



--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch