[PATCH] drm: Don't override active mode.

[Forest Bond]

Hi Ray,

On Sun, Jun 13, 2010 at 11:35:48PM -0400, Ray Strode wrote:
> On Sun, Jun 13, 2010 at 1:27 PM, Forest Bond <forest at alittletooquiet.net> wrote:
> > The drm renderer calls drmModeSetCrtc to tell the CRTC to use the
> > allocated scan out buffer.  This should not result in setting a
> > new mode for the CRTC, but can do exactly that since plymouth saves
> > the wrong mode when creating heads, and inadvertently overrides the
> > initial mode set by the kernel.
> >
> > The root of the problem is the assumption that the first mode in the
> > connector's mode list is the current active mode.  This will be true
> > if the active mode is set by automatic detection, as the kernel will
> > have selected the first mode in the list as the initial mode, but it
> > will not be true if the user has selected a different mode via the
> > video= kernel boot parameter.
> >
> > To fix this, the saved mode is copied from the controller itself,
> > unless that mode is for some reason not valid, in which case we fall
> > back to the old behavior (which could result in a mode switch).
> Ah, very clear description of the problem.  Thanks for taking the time
> to write such a detailed commit message.

After spending a few hours figuring out why my video= mode was not being
respected, it was a pleasure to get it all off my chest. ;)

> Your patch seems basically fine, but I don't really like the idea of
> calling drmModeFreeModeInfo on memory we allocate ourselves with
> calloc.

Thanks for taking a look.  C is not my primary language, so you have a much
better sense of these things than I do.  In thinking about it, I agree that
pairing malloc with drmModeFreeModeInfo doesn't quite feel right.

> I've done three commits here:
> 
> http://cgit.freedesktop.org/plymouth/commit/?id=abfda7550ac2fca6d816862e271126c2f44fa3cc
> http://cgit.freedesktop.org/plymouth/commit/?id=b5b5f081bae003fd2d8065586249085f11d1c4a7
> http://cgit.freedesktop.org/plymouth/commit/?id=9f25189bc8c6cb77cc37e3325e1089e23addf9a0
> 
> that restructure things, so we don't keep a duplicated copy of the
> mode around in the head strucutre and attempts to fix yoru problems
> I haven't given them any testing yet.  Do they look okay to you?

These look fine to me.  I can test them today or tomorrow and report back.

Thanks,
Forest
-- 
Forest Bond
http://www.alittletooquiet.net
http://www.pytagsfs.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.freedesktop.org/archives/plymouth/attachments/20100614/a141d186/attachment.pgp>