Mailinglist Archive: radeonhd (307 mails)

< Previous Next >
Re: [radeonhd] [PATCH] r6xx/r7xx Q&D implementation of wait_vblank
  • From: Alex Deucher <alexdeucher@xxxxxxxxx>
  • Date: Wed, 11 Feb 2009 18:15:36 -0500
  • Message-id: <a728f9f90902111515i464cc336xdbec3a35733dd0e5@xxxxxxxxxxxxxx>
On Wed, Feb 11, 2009 at 5:35 PM, Christian König
<deathsimple@xxxxxxxxxxx> wrote:
Am Dienstag, den 10.02.2009, 18:13 -0500 schrieb Alex Deucher:
You should have said so :)

ALWAYS (0<<0)
LT (1<<0)
LE (2<<0)
EQ (3<<0)
NE (4<<0)
GE (5<<0)
GT (6<<0)

here's ossman's patch for radeon as a reference:
http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?h=r6xx-r7xx-support&id=8e9ef8ff581892cbe1b7ea56d48b9a1abd70179d

Not knowing that there is a (undocumented) register to wait for a vline
range instead of a single line was also some part of the problem :-/

Attached are three patches:
1. The definition of the proper IT_WAIT_REG_MEM compare operators.

2. Implementation of wait_vline_range, also including a fix for a nasty
bug in "ereg", in the radeon driver source the num parameter of the
CP_PACKET0 macro isn't decremented by one, so the ereg function have to
do it, but in the radeonhd source CP_PACKET0 decrements the parameter.

3. A little bit cleanup of the register definitions in rhd_reg.h and
r600_reg_(r6xx|r7xx).h. So both files can be included at the same time.

Looks good. Pushed!

thanks,

Alex
--
To unsubscribe, e-mail: radeonhd+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: radeonhd+help@xxxxxxxxxxxx

< Previous Next >