Andreas Ehliar wrote:
>
> On Sun, Apr 29, 2001 at 10:56:42PM -0700, Frank Worsley wrote:
> > > The DRM module from the DRI CVS seems to work fine in 2.4.4.
> > > (The one in the kernel however does seem to be broken.)
> >
> > Isn't it generally best to use the module from the CVS/packages in any
> > case, no matter if the kernel module works? Just alone for the fact that we
> > then know that the driver module and kernel modules have matching versions?
>
> Speaking of this, wouldn't it make sense to add a more informative error
> message if a mismatch occurs?
>
> Take the message in the mga_dri.c for example:
> "[drm] MGADRIScreenInit failed "
> "(DRI version = %d.%d.%d, expected 4.0.x). "
> "Disabling DRI.\n",
>
> Wouldn't it be better if this read something like this:
> MGADRIScreenInit failed because of a version mismatch:
> Version of the mga kernel module: 1.0.0
> Version of the XFree86 mga module: 4.0.0
The text you suggest is even more misleading.
The code as-is is indicating that the DRI extension version isn't
what the driver is expecting. There is no connection between the
version numbers used for drivers and the DRI extension itself.
I think you're suggesting a new error message for the version
test between the kernel module and the 2D driver. Again, the
version number don't necessarily have to match identically.
Look at the i810 for example. The current 2D driver is 1.0.0, the
kernel module is 1.1.0 and libDRI is 3.0.0.
In any case, the error messages probably should be improved because
people frequently don't understand them. I'll work on that.
-Brian
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel
