Message-ID: <20190626142652.GL12905@phenom.ffwll.local>
Date:   Wed, 26 Jun 2019 16:26:52 +0200
From:   Daniel Vetter <daniel@...ll.ch>
To:     Rodrigo Siqueira <rodrigosiqueiramelo@...il.com>
Cc:     Daniel Vetter <daniel@...ll.ch>,
        Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
        Maxime Ripard <maxime.ripard@...tlin.com>,
        Sean Paul <sean@...rly.run>, David Airlie <airlied@...ux.ie>,
        dri-devel <dri-devel@...ts.freedesktop.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        kernel-janitors@...r.kernel.org,
        intel-gfx <intel-gfx@...ts.freedesktop.org>
Subject: Re: [Intel-gfx] [PATCH V4] drm/drm_vblank: Change EINVAL by the
 correct errno

On Wed, Jun 26, 2019 at 10:37:11AM -0300, Rodrigo Siqueira wrote:
> On Wed, Jun 26, 2019 at 4:53 AM Daniel Vetter <daniel@...ll.ch> wrote:
> >
> > On Wed, Jun 26, 2019 at 4:00 AM Rodrigo Siqueira
> > <rodrigosiqueiramelo@...il.com> wrote:
> > >
> > > On 06/19, Daniel Vetter wrote:
> > > > On Wed, Jun 19, 2019 at 09:48:56AM +0200, Daniel Vetter wrote:
> > > > > On Tue, Jun 18, 2019 at 11:07:50PM -0300, Rodrigo Siqueira wrote:
> > > > > > For historical reason, the function drm_wait_vblank_ioctl always return
> > > > > > -EINVAL if something gets wrong. This scenario limits the flexibility
> > > > > > for the userspace make detailed verification of the problem and take
> > > > > > some action. In particular, the validation of “if (!dev->irq_enabled)”
> > > > > > in the drm_wait_vblank_ioctl is responsible for checking if the driver
> > > > > > support vblank or not. If the driver does not support VBlank, the
> > > > > > function drm_wait_vblank_ioctl returns EINVAL which does not represent
> > > > > > the real issue; this patch changes this behavior by return EOPNOTSUPP.
> > > > > > Additionally, some operations are unsupported by this function, and
> > > > > > returns EINVAL; this patch also changes the return value to EOPNOTSUPP
> > > > > > in this case. Lastly, the function drm_wait_vblank_ioctl is invoked by
> > > > > > libdrm, which is used by many compositors; because of this, it is
> > > > > > important to check if this change breaks any compositor. In this sense,
> > > > > > the following projects were examined:
> > > > > >
> > > > > > * Drm-hwcomposer
> > > > > > * Kwin
> > > > > > * Sway
> > > > > > * Wlroots
> > > > > > * Wayland-core
> > > > > > * Weston
> > > > > > * Xorg (67 different drivers)
> > > > > >
> > > > > > For each repository the verification happened in three steps:
> > > > > >
> > > > > > * Update the main branch
> > > > > > * Look for any occurrence "drmWaitVBlank" with the command:
> > > > > >   git grep -n "drmWaitVBlank"
> > > > > > * Look in the git history of the project with the command:
> > > > > >   git log -SdrmWaitVBlank
> > > > > >
> > > > > > Finally, none of the above projects validate the use of EINVAL which
> > > > > > make safe, at least for these projects, to change the return values.
> > > > > >
> > > > > > Change since V3:
> > > > > >  - Return EINVAL for _DRM_VBLANK_SIGNAL (Daniel)
> > > > > >
> > > > > > Change since V2:
> > > > > >  Daniel Vetter and Chris Wilson
> > > > > >  - Replace ENOTTY by EOPNOTSUPP
> > > > > >  - Return EINVAL if the parameters are wrong
> > > > > >
> > > > >
> > > > > Reviewed-by: Daniel Vetter <daniel.vetter@...ll.ch>
> > > > >
> > > > > Apologies for the confusion on the last time around. btw if someone tells
> > > > > you "r-b (or a-b) with these changes", then just apply the r-b/a-b tag
> > > > > next time around. Otherwise people will re-review the same thing over and
> > > > > over again.
> > > >
> > > > btw when resending patches it's good practice to add anyone who commented
> > > > on it (or who commented on the igt test for the same patch and other way
> > > > round) onto the explicit Cc: list of the patch. That way it's easier for
> > > > them to follow the patch evolution and do a quick r-b once they're happy.
> > >
> > > Thanks for these valuable tips.
> > > Do you think that is a good idea to resend this patch CC's everybody? Or
> > > is it ok if I just apply it?
> >
> > Hm I thought I answered that on irc ... but just today I realized that
> > we missed 2 ioctls. There's also drm_crtc_get_sequence_ioctl and
> > drm_crtc_queue_sequence_ioctl which have the same dev->irq_enabled
> > check and I think should be treated the same.
> 
> Hi,
> 
> I reexamined all the composers described in the commit message (latest
> versions) to check if any project use and validate the return value
> from  drm_crtc_get_sequence_ioctl and drm_crtc_queue_sequence_ioctl. I
> noticed that mesa and xserver use them. FWIU replace EINVAL by
> EOPNOTSUPP is harmless for mesa project, however it is not the same
> for xserver.
> 
> Take a look at line 189 and 238 of hw/xfree86/drivers/modesetting/vblank.c
> 
> * https://gitlab.freedesktop.org/xorg/xserver/blob/master/hw/xfree86/drivers/modesetting/vblank.c#L238
> * https://gitlab.freedesktop.org/xorg/xserver/blob/master/hw/xfree86/drivers/modesetting/vblank.c#L189
> 
> A little bit below the above lines, you can see a validation like that:
> 
>   if (ret != -1 || (errno != ENOTTY && errno != EINVAL))
> 
> In other words, if we change the EINVAL by EOPNOTSUPP
> drm_crtc_[get|queue]_sequence_ioctl we could break xserver. I noticed
> that Keith Packard introduced these ioctls to the kernel and also to
> the xserver, I will prepare a new version and CC Keith. Should I do
> another thing to notify xserver developers?

If you want cc: xorg-devel or so, but I think they all moved to gitlab and
the m-l is pretty dead. Cc'ing Keith should be enough.

I looked at the code and I think we're fine. Better than fine actually,
because if dev->irq_enabled == false then we really shouldn't use vblank
ioctl, no matter whether the new or old version.

So for drivers without vblank support, all that will happen is that we
leave ->has_queue_sequence as false and then fail a bit later on with the
legacy ioctl. Should be all harmless.

Note that the idea behind filtering out EINVAL is that if you do a
QueueSequence on a crtc that's currently off, then it'll fail with EINVAL.
But the ioctl still works, hence why we want to accept that. Setting
has_queue_sequence = TRUE for the case where there's actually no vblank
might trigger a bug somewhere later on.
-Daniel

> 
> Thanks
> 
> > Can you pls resend with those addressed too? Then you can also resend
> > with the cc's all added.
> > -Daniel
> >
> > >
> > > > If you don't do that then much bigger chances your patch gets ignored.
> > > > -Daniel
> > > > >
> > > > > > Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@...il.com>
> > > > > > ---
> > > > > >  drivers/gpu/drm/drm_vblank.c | 2 +-
> > > > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > >
> > > > > > diff --git a/drivers/gpu/drm/drm_vblank.c b/drivers/gpu/drm/drm_vblank.c
> > > > > > index 603ab105125d..bed233361614 100644
> > > > > > --- a/drivers/gpu/drm/drm_vblank.c
> > > > > > +++ b/drivers/gpu/drm/drm_vblank.c
> > > > > > @@ -1582,7 +1582,7 @@ int drm_wait_vblank_ioctl(struct drm_device *dev, void *data,
> > > > > >   unsigned int flags, pipe, high_pipe;
> > > > > >
> > > > > >   if (!dev->irq_enabled)
> > > > > > -         return -EINVAL;
> > > > > > +         return -EOPNOTSUPP;
> > > > > >
> > > > > >   if (vblwait->request.type & _DRM_VBLANK_SIGNAL)
> > > > > >           return -EINVAL;
> > > > > > --
> > > > > > 2.21.0
> > > > >
> > > > > --
> > > > > Daniel Vetter
> > > > > Software Engineer, Intel Corporation
> > > > > http://blog.ffwll.ch
> > > >
> > > > --
> > > > Daniel Vetter
> > > > Software Engineer, Intel Corporation
> > > > http://blog.ffwll.ch
> > >
> > > --
> > > Rodrigo Siqueira
> > > https://siqueira.tech
> > > _______________________________________________
> > > Intel-gfx mailing list
> > > Intel-gfx@...ts.freedesktop.org
> > > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> >
> >
> >
> > --
> > Daniel Vetter
> > Software Engineer, Intel Corporation
> > +41 (0) 79 365 57 48 - http://blog.ffwll.ch
> 
> 
> 
> -- 
> 
> Rodrigo Siqueira
> https://siqueira.tech

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

Powered by blists - more mailing lists