NV43 (6600 Go) Report

NV43 (6600 Go) Report

[Pascal Schoenhardt]

Hi everyone,

My name is Pascal Schoenhardt. I am working on a SOC project to add
XRandR 1.2 support to GNOME, and I use a laptop with an Nvidia card.
Therefore, I also use Nouveau.

First of all, great work guys. I am extremely excited about this
project, and I'm amazed by the progress you are making with such a
difficult task. I am posting this message to report my success with
the nouveau driver (GIT version from the 12'th of May), and to ask for
support on one thing. So, here we go:

Driver compiled just fine, and everything worked on the first try. The
"Start X and run for cover" was not required! glxgears runs at approx.
350 fps; this is according to the console output, however the
animation looks very jumpy... According to the wiki, this is the only
3D application that runs, however I am also able to use every one of
the GL screensavers included in Gnome (2.20, I believe - running
Feisty). They run like slideshows, but they render perfectly! Nice
work. The only other GL app I've tried is Google Earth, which resulted
in X crashing, but I know that advanced apps aren't meant to work yet
anyways.

Now for the problems: My laptop is an Asus Z71v, which has a 1680x1050
LCD panel. While I am running Feisty Fawn, which has x.org 7.2 by
default, I have upgraded this to 7.3 from Burtoni's repo's, which were
create with the intent of testing display hotplugging.

When I first started up nouveau, I only had 1024x768. xrandr reported
that screen 0 had a minimum of 640x480, a max of 1024x768, and was
connected at 1024x768. My output was labeled "default" (not LVDS) and
was connected at 1024x768. Trying to change the screen size with the
--fb option was not successful.

A bit of Googling lead me to adding:
    Virtual  1680  1050
to the Screen -> Display subsection in my xorg.conf. This increased
the screen size to 1680x1050, however my LCD was still at 1024x768,
allowing me to pan around. The strange thing was that the "default"
output was registering itself to be running at 1680x1050. It seems
nouveau tried changing my LCD to its native res, and thought it
succeeded, when it really didn't. At this point, I could use the --fb
command to lower the resolution, but not increase it. Also, changing
the resolution like that caused X to crash sometimes, and some other
GNOME related issues other times (nautilus crashing).

My next step was to try some modelines. After 4 attempts, I found this
one to work:
   ModeLine "1680x1050" 144.02 1680 1784 1968 2256 1050 1051 1054 1064
Now I have full resolution. The output of running 'xrandr' right now is:

Screen 0: minimum 640 x 480, current 1680 x 1050, maximum 1680 x 1050
default connected 1680x1050+0+0 0mm x 0mm
   1680x1050      60.0*
   1024x768       60.0
   800x600        60.0     56.0
   640x480        60.0
   1280x1024      60.0
   1440x900       60.0
   1280x960       60.0
   1280x800       60.0
   1280x768       60.0
   1152x768       55.0

Now, here is the real issue: My laptop has a VGA port and an svideo
port. To start working on smooth multi-head support for gnome, I
obviously need use of at least the VGA port. As it stands, nouveau
does not detect any of these outputs. As I understand it, xrandr
should be reporting something like this:

Screen 0: minimum 640 x 480, current 1680 x 1050, maximum 1680 x 1050
LVDS connected 1680x1050+0+0 0mm x 0mm
VGA not connected
SVideo not connected
...

Plugging in a monitor in the VGA port does nothing at all. I did
notice that there are some other heads in the GIT repository for
xrandr1.2. Do I maybe need to get a special version of nouveau? Has
the randr stuff been merged into the main driver yet? If I need a
special version, I would be forever grateful if someone could just
post the exact git command I need to run to get what I need. :)

So, I appologize for the length of this report. I hope it provides one
of you developers with something useful!

Cheers,

Pascal

Re: NV43 (6600 Go) Report

[Chris Ball]

Hi Pascal,

On Tue, May 15, 2007 at 11:23:50AM +0100, Pascal Schoenhardt wrote:
> Now, here is the real issue: My laptop has a VGA port and an svideo
> port. To start working on smooth multi-head support for gnome, I
> obviously need use of at least the VGA port. As it stands, nouveau
> does not detect any of these outputs. As I understand it, xrandr
> should be reporting something like this:

Have you tried "xrandr --auto"?  It's usually required to run this
command before xrandr picks up on a VGA port connection, because 
usually (at least, on Intel) we don't get an interrupt when a 
monitor is plugged in.

Which brings me to a question:  do we get one on nVidia?  Could nouveau
be able to kick off xrandr as soon as a monitor's connected?

(I don't know how well Xrandr actually works on nouveau, I'm afraid.)

- Chris.
-- 
Chris Ball   <cjb-2X9k7bc8m7Mdnm+yROfE0A@public.gmane.org>

Re: NV43 (6600 Go) Report

[Pascal Schoenhardt]

Yes, I have tried "xrandr --auto". Unfortunately, this doesn't help.

Pascal

On 5/15/07, Chris Ball <cjb-2X9k7bc8m7Mdnm+yROfE0A@public.gmane.org> wrote:
> Hi Pascal,
>
> On Tue, May 15, 2007 at 11:23:50AM +0100, Pascal Schoenhardt wrote:
> > Now, here is the real issue: My laptop has a VGA port and an svideo
> > port. To start working on smooth multi-head support for gnome, I
> > obviously need use of at least the VGA port. As it stands, nouveau
> > does not detect any of these outputs. As I understand it, xrandr
> > should be reporting something like this:
>
> Have you tried "xrandr --auto"?  It's usually required to run this
> command before xrandr picks up on a VGA port connection, because
> usually (at least, on Intel) we don't get an interrupt when a
> monitor is plugged in.
>
> Which brings me to a question:  do we get one on nVidia?  Could nouveau
> be able to kick off xrandr as soon as a monitor's connected?
>
> (I don't know how well Xrandr actually works on nouveau, I'm afraid.)
>
> - Chris.
> --
> Chris Ball   <cjb-2X9k7bc8m7Mdnm+yROfE0A@public.gmane.org>
>

Re: NV43 (6600 Go) Report

[Pekka Paalanen]

On Tue, 15 May 2007 11:23:50 +0100
"Pascal Schoenhardt" <stoanhart-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:

> the nouveau driver (GIT version from the 12'th of May)
> 
> Driver compiled just fine, and everything worked on the first try. The
> "Start X and run for cover" was not required! glxgears runs at approx.
> 350 fps; this is according to the console output, however the
> animation looks very jumpy... According to the wiki, this is the only
> 3D application that runs, however I am also able to use every one of
> the GL screensavers included in Gnome (2.20, I believe - running
> Feisty). They run like slideshows, but they render perfectly! Nice
> work. The only other GL app I've tried is Google Earth, which resulted
> in X crashing, but I know that advanced apps aren't meant to work yet
> anyways.

Yes, Mesa software rendering is nice, isn't it? ;-)
I suppose you did not install Mesa from git, did you?
Might be worth to check glxinfo, if you really got DRI called properly.
That glxgears behaviour sounds like software rendering to me.

> Now for the problems: My laptop is an Asus Z71v, which has a 1680x1050
> LCD panel. While I am running Feisty Fawn, which has x.org 7.2 by
> default, I have upgraded this to 7.3 from Burtoni's repo's, which were
> create with the intent of testing display hotplugging.
...
> Now, here is the real issue: My laptop has a VGA port and an svideo
> port. To start working on smooth multi-head support for gnome, I
> obviously need use of at least the VGA port. As it stands, nouveau
> does not detect any of these outputs. As I understand it, xrandr
...
> Plugging in a monitor in the VGA port does nothing at all. I did
> notice that there are some other heads in the GIT repository for
> xrandr1.2. Do I maybe need to get a special version of nouveau? Has
> the randr stuff been merged into the main driver yet? If I need a
> special version, I would be forever grateful if someone could just
> post the exact git command I need to run to get what I need. :)

I am not sure what the current state of Nouveau randr-1.2 is, but the
following is my impression.
randr-1.2 stuff is not in the master branch, but in the randr-1.2
branch only. This requires xorg-server 1.3, like master branch now does.
I have the same kind of nv28 card as Airlied who works on the
Nouveau randr-1.2. The last I tried I got, with one analog and one DFP,
something different on both heads, but not really usable (at least DFP
screen was geometrically screwed, and colors too). Some how it managed
to make my mouse all weird, too, swapping and inversing x and y
coordinates or something. AFAIK, trouble with basic mode setting for
multiple heads. Since then, I see one or two batches of commits from
Airlied to randr-1.2 branch, so things may have changed.

I still do not believe randr-1.2 features are yet at a usable state.

-- 
Pekka "PQ" Paalanen
http://www.iki.fi/pq/

Re: NV43 (6600 Go) Report

[Pascal Schoenhardt]

OK, so I have compiled and installed the randr1.2 branch of nouveau.
The problem I am having is the following: with just the laptop
flatpanel display connected, x won't start; with an external device
connected, x starts but all displays remain blank.

With just the laptop panel on its own, X fails with:
  (EE) NOUVEAU(0): No valid modes.
  (EE) Screen(s) found, but none have a usable configuration.

I have tried the following settings for xorg.conf:
  -Same settings as before (non-randr1.2 version)
  -Removed all "Display" subsections, to let modes be autodetected
  -All monitor options removed, except the identifier
  -No modes in "Display" subsection, just one large virtual, big
enough to span all my display devices

None of these have helped. Here is a full log output, with non-nouveau
stuff stripped out:

(II) LoadModule: "nouveau"
(II) Loading /usr/lib/xorg/modules/drivers//nouveau_drv.so
(II) Module nouveau: vendor="X.Org Foundation"
	compiled for 1.3.0, module version = 1.2.0
	Module class: X.Org Video Driver
	ABI class: X.Org Video Driver, version 1.2
(II) NOUVEAU driver Date:   Sun May 6 17:30:30 2007 +1000
(II) NOUVEAU driver for NVIDIA chipset families :
	RIVA 128    (NV03)
	RIVA TNT    (NV04)
	RIVA TNT2   (NV05)
	GeForce 256 (NV10)
	GeForce 2   (NV11, NV15)
	GeForce 4MX (NV17, NV18)
	GeForce 3   (NV20)
	GeForce 4Ti (NV25, NV28)
	GeForce FX  (NV3x)
	GeForce 6   (NV4x)
	GeForce 7   (G7x)
(II) Primary Device is: PCI 01:00:0
(--) Chipset GeForce Go 6600 found
(II) Setting vga for screen 0.
(II) NOUVEAU(0): Initializing int10
(WW) NOUVEAU(0): Bad V_BIOS checksum
(II) NOUVEAU(0): Primary V_BIOS segment is: 0xc000
(--) NOUVEAU(0): Chipset: "GeForce Go 6600"
(**) NOUVEAU(0): Depth 24, (--) framebuffer bpp 32
(==) NOUVEAU(0): RGB weight 888
(==) NOUVEAU(0): Default visual is TrueColor
(==) NOUVEAU(0): Using HW cursor
(==) NOUVEAU(0): Using EXA acceleration method
(--) NOUVEAU(0): Linear framebuffer at 0xC0000000
(--) NOUVEAU(0): MMIO registers at 0xFD000000
(II) Loading sub module "dri"
(II) LoadModule: "dri"
(II) Reloading /usr/lib/xorg/modules/extensions//libdri.so
(II) NOUVEAU(0): Loaded DRI module
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 7, (OK)
drmOpenByBusid: Searching for BusID pci:0000:01:00.0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 7, (OK)
drmOpenByBusid: drmOpenMinor returns 7
drmOpenByBusid: drmGetBusid reports pci:0000:01:00.0
(II) NOUVEAU(0): [dri] Found DRI library version 1.3.0 and kernel
module version 0.0.6
(II) NOUVEAU(0): BIOS signature not found!
(II) NOUVEAU(0): vtowner is 0
(II) Loading sub module "i2c"
(II) LoadModule: "i2c"(II) Module already built-in
(II) Loading sub module "ddc"
(II) LoadModule: "ddc"(II) Module already built-in
(II) NOUVEAU(0): I2C bus "DDC1" initialized.
(II) NOUVEAU(0): I2C bus "DDC2" initialized.
(II) NOUVEAU(0): I2C bus "TV" initialized.
(II) NOUVEAU(0): Output Analog-0 using monitor section Laptop FlatPanel
(--) NOUVEAU(0): Adding output Analog-0
(II) NOUVEAU(0): Output Analog-1 has no monitor section
(--) NOUVEAU(0): Adding output Analog-1
(II) NOUVEAU(0): Output Digital-0 has no monitor section
(--) NOUVEAU(0): Adding output Digital-0
(II) NOUVEAU(0): I2C device "DDC1:ddc2" registered at address 0xA0.
(II) NOUVEAU(0): I2C device "DDC1:ddc2" removed.
(--) NOUVEAU(0): CRT detect returned 0
(II) NOUVEAU(0): Output Analog-0 disconnected
(II) NOUVEAU(0): EDID for output Analog-0
(II) NOUVEAU(0): I2C device "DDC2:ddc2" registered at address 0xA0.
(II) NOUVEAU(0): I2C device "DDC2:ddc2" removed.
(II) NOUVEAU(0): Output Analog-1 disconnected
(II) NOUVEAU(0): EDID for output Analog-1
(II) NOUVEAU(0): I2C device "DDC1:ddc2" registered at address 0xA0.
(II) NOUVEAU(0): I2C device "DDC1:ddc2" removed.
(II) NOUVEAU(0): Output Digital-0 disconnected
(II) NOUVEAU(0): EDID for output Digital-0
(II) NOUVEAU(0): Output Analog-0 disconnected
(II) NOUVEAU(0): Output Analog-1 disconnected
(II) NOUVEAU(0): Output Digital-0 disconnected
(EE) NOUVEAU(0): No valid modes.
(II) UnloadModule: "nouveau"
(II) UnloadModule: "dri"
(EE) Screen(s) found, but none have a usable configuration.

Fatal server error:
no screens found

=========================

Now, when I connect another screen, X actually starts. Reading the log
files, nouveau is getting the EDID info from the external screen
(1024x768@70 LCD), and determining the correct modeline (after quite a
lot of probing). As for the internal panel, it seems not to provide
any EDID info. However, nouveau does seem to want to choose 1680x1050,
provided I have specified a virtual large enough to fit that and the
external screen in the xorg.conf (2704x1050).

Once X is started, both monitors are black. However, I can hear the
"please enter your loging credentials" tone you get with ubuntu, and I
can blindly log in to a normal session.

I am guessing that this is just because this branch of nouvea is even
more experimental than the whole project is, so there probably isn't
much to do but wait for it to improve. However, hopefully this is
useful information to some of the devs. And if someone has a tip on
getting this to work, I'm all ears.

Here is a full log of X starting, using a modeline for the internal
panel, defining a virtual screen large enough for the two displays,
and with the external display connected and turned on. This is too
large to post as text, so here's a link:

http://www.pascals.ca/files/xorg-modeline-virt2704x1050-VGAConnectedOn.log

Thanks!

Pascal

Re: NV43 (6600 Go) Report

[Pekka Paalanen]

On Thu, 17 May 2007 10:15:14 +0100
"Pascal Schoenhardt" <stoanhart-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:

> OK, so I have compiled and installed the randr1.2 branch of nouveau.
> The problem I am having is the following: with just the laptop
> flatpanel display connected, x won't start; with an external device
> connected, x starts but all displays remain blank.

I don't know how to help you, but...

> Now, when I connect another screen, X actually starts. Reading the log
> files, nouveau is getting the EDID info from the external screen
> (1024x768@70 LCD), and determining the correct modeline (after quite a
> lot of probing). As for the internal panel, it seems not to provide
> any EDID info. However, nouveau does seem to want to choose 1680x1050,
> provided I have specified a virtual large enough to fit that and the
> external screen in the xorg.conf (2704x1050).
> 
> Once X is started, both monitors are black. However, I can hear the
> "please enter your loging credentials" tone you get with ubuntu, and I
> can blindly log in to a normal session.

Try hitting Ctrl+Alt+Numpad- multiple times, slowly. That gave me
"something" on screens.

> I am guessing that this is just because this branch of nouvea is even
> more experimental than the whole project is, so there probably isn't
> much to do but wait for it to improve.

I think this is very much true, because mode setting on nv28 is also
not totally correct yet. Airlied did get two analog heads going for
himself, but for me analog+DFP has always been broken. I should
probably try again now, there are some commits that might have an
effect.

-- 
Pekka "PQ" Paalanen
http://www.iki.fi/pq/