Few fixes for intel driver

Maxim Levitsky maximlevitsky at gmail.com
Sat Mar 22 07:01:02 PDT 2008


Hi there,

I was offline for two weeks, thus I didn't reply on my patch
for overlay, good to see it is accepted.

Also I want to say big thanks to developers of intel driver, since you fixed
two most annoying bugs, which were blank vesafb VTs, which forced me to use plain VGA,
and all 3D applications did crash when starting them after suspend to ram - not anymore.


I have fixed the most annoying to me bug, and it turns out that this bug
it a xserver bug:

Almost always an attempt to change the screen resolution ended with randomly blanking screen, and
no way other that reboot to fix this.

Amazingly this isn't connected to modesetting code, but rather this is a result
of hardware going mad when the hardware cursor is placed offscreen, and this is the 'Almost'
from above:

If resolution is decreased, and cursor is off the new screen, the xserver code does virtually nothing
to correct cursor position.

The intel driver does correct cursor position, but does this too late, in i830_crtc_commit
when destination mode is already set, and hardware is mad.

To fix this I decided to put a call to xf86_reload_cursors in xf86CrtcSetMode
just before setting the mode, and now modesetting is bulletproof.

(an additional call to xf86_reload_cursors in i830_crtc_commit won't hut I think, but it is unnecessary now)

On top of that I fixed another this time small bug, a xserver crash triggered by running the overlay,
and putting the screen on DPMS save mode.

And I send a small patch to make overlay first XV adapter is user chooses to do so.


Best regards,
	Maxim Levitsky



More information about the xorg mailing list