Working quite well : xorg7.2, DRI, suspend, Radeon Mobility X600

Hello everyone,
well, for once I'm not writing with a problem, and extremely
happy w/FreeBSD 

Over the last year i've been writing to this list with
snippets of info on what
worked for me and the laptop I'm currently using ( a
Thinkpad z60M ). I had
an entry for this laptop in the Compatibility list. With the
latest software
from ports (xorg, mainly) and kernel / world, there have
been some good
improvements to the whole situation. I've added more info to
the entry in the
laptop compat list :,
http:
//laptop.bsdgroup.de/freebsd/index.html?action=show_laptop_c
omment&commentid=29803&laptop=1202 .

But here goes with the lot 

Upgrade to xorg 7.2, latest kernel + World to date
(2007-05-28).

Loading agp and radeon kernel modules, enables DRI to load
fine in xorg! it now
works quite well with OpenGL:

$ kldstat
Id Refs Address    Size     Name
 1   31 0xc0400000 4de640   kernel
 2    1 0xc08df000 7a68     linprocfs.ko
 3    3 0xc08e7000 1f39c    linux.ko
 4    1 0xc0907000 13be0    snd_hda.ko
 5    2 0xc091b000 3dadc    sound.ko
 6    2 0xc0959000 17920    agp.ko
 7    2 0xc0971000 65474    acpi.ko
 8    1 0xc09d7000 4c2c     acpi_ibm.ko
 9    1 0xc09dc000 1d498    kqemu.ko
10    1 0xc09fa000 22140    radeon.ko
11    2 0xc0a1d000 10c68    drm.ko
12    1 0xc5d66000 c000     ipfw.ko
13    1 0xc5de6000 2000     rtc.ko
14    1 0xc5e5e000 7000     aio.ko


The video card is a
drm0pci1:0:0: class=0x030000 card=0x058e1014
chip=0x31501002 rev=0x00 hdr=0x00
vendor = 'ATI Technologies Inc'
device = 'M24 1P Radeon Mobility X600'
class = display
subclass = VGA

A year ago, xorg 6.x, no DRI, glxgears would show :
[betomayiin] [Wed May 10 11:22:41 2006]
~
$ glxgears
3086 frames in 5.4 seconds = 567.330 FPS
3760 frames in 5.0 seconds = 749.966 FPS
3860 frames in 5.0 seconds = 768.628 FPS
3800 frames in 5.0 seconds = 756.635 FPS
3760 frames in 5.0 seconds = 749.888 FPS
3780 frames in 5.0 seconds = 749.527 FPS
3800 frames in 5.0 seconds = 756.423 FPS
3860 frames in 5.0 seconds = 770.521 FPS
3840 frames in 5.0 seconds = 767.983 FPS
3837 frames in 5.0 seconds = 766.985 FPS


Currently:
[betomayiin] [Mon May 28 22:23:12 2007]
~
$ glxgears
8772 frames in 5.0 seconds = 1754.273 FPS
7113 frames in 5.0 seconds = 1422.555 FPS
8184 frames in 5.0 seconds = 1636.676 FPS
8841 frames in 5.0 seconds = 1768.043 FPS
9415 frames in 5.0 seconds = 1882.501 FPS
8845 frames in 5.0 seconds = 1768.976 FPS
9036 frames in 5.0 seconds = 1806.932 FPS
8637 frames in 5.0 seconds = 1727.369 FPS

So it's a DEFINITE improvement. There are still some things
that need some
fixing though:

- There are a few cases when, after logging off from X, the
video memory seems
to be stuck with the old image - GDM doesn't show up at all
(although I can type
my username and pwd and log in). the session afterwards is
the same - things
"are" there , but rather useless as they dont show
on screen. Zapping the
server (Ctrl-Alt-Backspace) works, but doesn't solve the
problem. Again,
sometimes there is no problem at all.

- When using some OpenGL apps (been testing with
graphics/mesa-demos), some
artifacts appear on the screen. Sometimes only in the area
of the GL app,
sometimes on the screen overall. For example ,

http://www.meijome.net/files/freebsd/2007-05-28-22.2
5.04.jpg

On the good side, there hasn't been any lockup that I could
attribute
to OpenGL / DRI / xorg. When using some games that use, i
believe, opengl ,
like games/warzone2100 in full screen, there is no problem
at all (other than
hours spent playing  )

Following messages from mesa-demos apps, I've added
R300_SPAN_DISABLE_LOCKING=yes
to my X Session environment. the demos worked better when
this was set, though
I need to find a test case to see whether this affects in
any way the screen
artifacts I've seen

- Composite , although working faster than before , it is
definitely slower,
even with just drop shadows enabled. I have XFCE4, default
composite settings
(I haven't played with tuning composite itself)

Sound works out of the box with the snd_hda from FreeBSD
itself (YEAY!), but I
found that it makes the sound a bit choppy under high-cpu
loads (buildworld,
for instance). Ariff's low latency binaries (sound.ko and
snd_hda.ko), build of
April 26, 2007 works without any choppiness, and no
conflicts with the kernel
either.

Suspend and resume work fine, with DRI inclusive - it is a
single core CPU ( I
believe there are still issues with suspend/resume with dual
cores.

One down side of using xorg7.2 - i can't seem to be able to
have 2 keyboard
layouts and be able to use the Alt key. I'll put more info
into another email
when I can -i've gone full 'us' for now..

that's it 
B
_________________________
{Beto|Norberto|Numard} Meijome

Q. How do you make God laugh?
A. Tell him your plans.

I speak for myself, not my employer. Contents may be hot.
Slippery when wet.
Reading disclaimers makes you go blind. Writing them is
worse. You have been
Warned.
_______________________________________________
freebsd-mobilefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-mobile

To unsubscribe, send any mail to
"freebsd-mobile-unsubscribefreebsd.org"

On Mon, 28 May 2007 10:32:02 -0500
Bruce Burden <brucegbrealtime.net> wrote:

> 
> 
>    Hello Norberto,

Hi Bruce,
please keep the list in CC:

> 
[...]

> 
> 	I have a HP ZD8000 with the X600 in it, and while DRI
is
>     better under 6.2-Stable and Xorg 7.2, the display
is still not
>     correct. And, yes, the world is up to date, as of
last week.

what do you mean, is not correct? If you see artifacts, you
may want to grab a
screenshot so we can see what it is...

> 
> 	So, I am curious as to what modules you have loaded?
As I
>     say, if I uncomment DRI in the Modules section, the
display
>     does not render correctly. I have
"radeon" as the driver in 
>     the card section. I do see drm.ko and radeon.ko in
the kldstat
>     output when I uncomment DRI, but it is not useful.
at least in my case:
1) I  need to have AGP support in your kernel (or load it as
module, for some
reason I've built it in...)
2) I loaded radeon by hand, drm got loaded automatically
3) i started X and it didnt die.
4) I then enabled DRI and tried again... 

I then spent some more time re-running xorgcfg to get a
clean config (in case
the options available in xorg6.x were different to the ones
in xorg7.2

> 
> 	Perhaps you can tell me what I am overlooking?

not sure, i've put some of hte info I think is relevant here
: 


http://www.meijome.net/files/freebsd/ayiin_20070529/

Beto
_________________________
{Beto|Norberto|Numard} Meijome

"A problem cannot be solved with the same type of
thinking that created it."
  Albert Einstein

I speak for myself, not my employer. Contents may be hot.
Slippery when wet.
Reading disclaimers makes you go blind. Writing them is
worse. You have been
Warned.
_______________________________________________
freebsd-mobilefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-mobile

To unsubscribe, send any mail to
"freebsd-mobile-unsubscribefreebsd.org"

On Mon, 28 May 2007 22:38:33 +1000
Norberto Meijome <freebsdmeijome.net> wrote:

> Suspend and resume work fine, with DRI inclusive - it
is a single core CPU ( I
> believe there are still issues with suspend/resume with
dual cores.

Ian Smith reminded me of some previous postings I made on
this (august 2006).
with xorg 7.2 I made some slight changes to the my suspend
script, basically i
dont switch back to VT1 before sending zzz - it works fine.
Resuming on text
mode seemed to be corrupting video RAM too (more tests
needed).

relevant sysctls that are working ok currently

---
# Should prevent freezing in X when doing suspend
## for TOSHIBA A2, NOT THINKPAD
## hw.syscons.sc_no_suspend_vtswitch=1

# Fixes resume on RELENG_6
hw.acpi.reset_video=1
---


_________________________
{Beto|Norberto|Numard} Meijome

Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"

I speak for myself, not my employer. Contents may be hot.
Slippery when wet.
Reading disclaimers makes you go blind. Writing them is
worse. You have been
Warned.
_______________________________________________
freebsd-mobilefreebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-mobile

To unsubscribe, send any mail to
"freebsd-mobile-unsubscribefreebsd.org"