[Intel-gfx] [PATCH i-g-t 5/7] kms_cursor_crc: Add reference software rendering

Ville Syrjälä ville.syrjala at linux.intel.com
Wed Apr 2 13:38:11 CEST 2014


On Wed, Apr 02, 2014 at 02:27:40PM +0300, Antti Koskipää wrote:
> On 04/02/2014 02:21 PM, Ville Syrjälä wrote:
> > On Wed, Apr 02, 2014 at 02:06:28PM +0300, Antti Koskipaa wrote:
> > <snip>
> >> @@ -184,9 +192,6 @@ static void test_crc_offscreen(test_data_t *test_data)
> >>  	do_test(test_data, left - (cursor_w+512), right + (cursor_w+512), top                 , bottom                 );
> >>  	do_test(test_data, left                 , right                 , top - (cursor_h+512), bottom + (cursor_h+512));
> >>  	do_test(test_data, left - (cursor_w+512), right + (cursor_w+512), top - (cursor_h+512), bottom + (cursor_h+512));
> >> -
> >> -	/* go nuts */
> >> -	do_test(test_data, INT_MIN, INT_MAX, INT_MIN, INT_MAX);
> > 
> > Why are you dropping the go nuts test? We had an actual bug in the
> > driver that this managed to hit. We should keep this test to avoid
> > regressions.
> 
> Link please. Cairo went nuts with that one, but if there is a case for
> the test, I'll work around Cairo to keep it.

commit 92e76c8c7e436a07af3e0b594480ff8689add078
Author: Ville Syrjälä <ville.syrjala at linux.intel.com>
Date:   Mon Oct 21 19:01:58 2013 +0300

    drm/i915: Clamp cursor coordinates to int16_t range

is the commit that fixed the bug.

-- 
Ville Syrjälä
Intel OTC



More information about the Intel-gfx mailing list