headers
Vincent | 13 Dec 2006 11:43
Picon
Favicon

Experience with modular-x11

Hi there,
I don't know if this will be useful to anyone, but since nobody appears 
to use the modular-x11-server yet, here is what I had to suffer to get 
something sane and running.

The first thing is that the wrapper script incorrectly copies only part 
of the .pc files installed by the libX* libraries. Therefore, albeit the 
dependency check works, the configure script fails complaining about 
missing prerequisites. This can be fixed by manually copying all the .pc 
in the /usr/pkg/lib/pkgconfig files into .buildlink/lib/pkgconfig.

Next, sometimes one of the library depends on the "old" xorg. This 
dependency must be manually erased by editing the work.xxx/.depend file. 
This process must be carried out for all softwares that depend on 
xorg-libs-6.9.0, and also for other dependencies like xcursor, Xft, xpm 
and other that are now embedded in libX...

But it is then impossible to build the package, because buildlink 
reports that xorg-libs is missing during the buildlink phase. One must 
therefore create artificial .buildlink_xorg-libs_done in the 
work.xxx/.buildlink repertory, and so on, one fake file per cancelled 
dependency in the .depends file. My experience shows that practically 
only three majors fake files are to be created: xorg-libs, xcursor and xpm.

The next thing to do is to copy explicitly all the libX* libraries into 
the .buildlink/lib directory, because the script fails to do so, and the 
link phase may fail.

Once all these steps are carried out, the package compiles and installs 
normally.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jeremy C. Reed | 13 Dec 2006 16:01

Re: Experience with modular-x11

On Wed, 13 Dec 2006, Vincent wrote:

> I don't know if this will be useful to anyone, but since nobody appears 
> to use the modular-x11-server yet, here is what I had to suffer to get 
> something sane and running.

I definitely use it. I also use it with xf86-video-wsfb. (I have been 
using it for several months, before even in official pkgsrc but still via 
pkgsrc build. I have been using many of the modular components now for 
years also via pkgsrc-wip.)

> As for the impression, having an old ATI Radeon M7 video card, it seems 
> to me that the video is really faster, especially OpenGL. I can't be 
> more precise, because my feeling is only based on fluidity of some 3D 
> xlock screen and the rather vague impression that windows appear and 
> move quicker than with 6.9.0. Maybe some optimizations are carried out 
> during the build process that couldn't be achieved by the old monolithic 
> process, or this is an effet of GCC 4.1, I can't tell.

Thanks for sharing your experience here. In addition to just build 
changes, there has been ongoing improvements to the code itself.

I clipped out your steps, as they are wrong at this time, although I do 
understand that they do provide a way that will work.
Permalink | Reply | 0 comments |
headers
Blair Sadewitz | 13 Dec 2006 16:07
Picon

Re: Experience with modular-x11

Among myriad other things, I just found that the modular xorg font
packages depend upon fonts/mkfontdir, but this dependency seems not to
be listed.

--

-- 
Support WFMU-FM: free-form radio for the masses!

<http://www.wfmu.org/>
91.1 FM Jersey City, NJ
90.1 FM Mt. Hope, NY

"The Reggae Schoolroom":
<http://www.wfmu.org/playlists/RS/>
Permalink | Reply | 0 comments |
headers
Vincent | 13 Dec 2006 20:04
Picon
Favicon

Re: Experience with modular-x11 (sequel)

Jeremy C. Reed spake thus,

>> As for the impression, having an old ATI Radeon M7 video card, it seems 
>> to me that the video is really faster, especially OpenGL. I can't be 
>> more precise, because my feeling is only based on fluidity of some 3D 
>> xlock screen and the rather vague impression that windows appear and 
>> move quicker than with 6.9.0. Maybe some optimizations are carried out 
>> during the build process that couldn't be achieved by the old monolithic 
>> process, or this is an effet of GCC 4.1, I can't tell.
> Thanks for sharing your experience here. In addition to just build 
> changes, there has been ongoing improvements to the code itself.

Once again, that's more a feeling that a clearly identified and measured 
phenomenon. I'll try a newer tarball and see what gets out of it.

> I clipped out your steps, as they are wrong at this time, although I do 
> understand that they do provide a way that will work.

I am conscious what I did is nothing but tinkering, and I never hold it 
to be something else, especially a clean process others should follow. 
Let's say, as you qualified it, it provides a tough way to make it work!

By the way, I can remember something went wrong while compiling the 
server, though it is benign: a structure declared twice, and a function 
first declared extern then static, both errors in the same .c file. It 
might have been corrected in the newer versions.

I'll keep you posted.
Vincent
(Continue reading)

Permalink | Reply | 0 comments |
headers
Thomas Bieg | 14 Dec 2006 23:44
Picon

Re: ATI DRI bounty wiki page (Was: Nvidia petition ..put it on the website)

(Cross-posting to tech-x11.)

There was a (short but promising) thread about DRI on tech-x11 some
weeks ago, with people who already worked on it taking part:

http://mail-index.netbsd.org/tech-x11/2006/09/12/0000.html

Regards,
Tom

David Brownlee wrote:
> On Thu, 14 Dec 2006, Mike M. Volokhov wrote:
> 
>> David Brownlee <abs <at> NetBSD.org> wrote:
>>> On Fri, 8 Dec 2006, Andrew Gillham wrote:
>>>
>>>> My $500 is still available as well. :)
>>>> http://mail-index.netbsd.org/netbsd-jobs/2003/05/07/0000.html
>>>
>>>     So... what is the scale of this project, and are there any
>>>     intermediate steps that would potentially be worth funding?
>>>     I dont have the time, but would also be willing to contribute
>>>     to the bounty...
>>
>> So can we create the Wiki page where problem will be explained and
>> total ballance calculated? Say:
>>
>>     http://wiki.netbsd.se/index.php/DRI_Support_in_NetBSD
>>
>> Also, that page can be linked into NetBSD Projects page at:
(Continue reading)

Permalink | Reply | 0 comments |
headers
Blair Sadewitz | 15 Dec 2006 21:09
Picon

80x50 text mode not restored properly upon leaving X

When I exit XFree86 or switch to another VT, my 80x50 text modes
appear as 80x25.  If my first VT is 80x25, I can switch to it and
switch back to another VT, and the 80x50 text reappears again.

If I use 80x50 for my first VT, i.e. defined in the kernel, I cannot
restore 80x50 text at all!

I am using XFree86 4.6.99.13 (latest snapshot) with some of the local
changes applied (bsd_mouse.c, disabling switching to VT #1 upon
startup, i386_video.c for amd64, and a few other things).

Does this happen with NetBSD's XFree86?  If it doesn't, what do I have
to change to produce the proper behavior?  If it does, can this be
fixed?

I have tried enabling the "VBERestore" option for the i810 driver, but
that kills my display (monitor loses sync).

Also, unrelated to this, if I use the "wskbd" protocol in XFree86, my
arrow keys do not work, but if I do not specify a protocl, they do.  I
know I could just not use wskbd, but I'd like to know what I could do
to fix this as well.

Thanks for any help,

--Blair
--

-- 
Support WFMU-FM: free-form radio for the masses!

<http://www.wfmu.org/>
(Continue reading)

Permalink | Reply | 1 comment |
headers
Pavel Cahyna | 16 Dec 2006 11:26
Picon

Re: 80x50 text mode not restored properly upon leaving X

On Fri, Dec 15, 2006 at 03:09:46PM -0500, Blair Sadewitz wrote:
> When I exit XFree86 or switch to another VT, my 80x50 text modes
> appear as 80x25.  If my first VT is 80x25, I can switch to it and
> switch back to another VT, and the 80x50 text reappears again.

I have ttyE0 with 80x25 text, and ttyE6 with 80x50 text. If I start X from
ttyE0, I can switch to ttyE6 and the 80x50 mode is properly restored. So I
would say that i can't repeat your problem, if I understand it correctly.

This is XFree86 4.5.0 from NetBSD, kernel 3.1_RC1.

Pavel
Permalink | Reply | 0 comments |
headers
Jeremy C. Reed | 21 Dec 2006 16:10

wsfb and two virtual displays and VT_ACTIVATE failed and properly aborting X???

This email has two do with two questions but related to maybe same 
problem. I am using Xorg from Xorg, but feel free to share advice or 
answers for XFree86 or Xorg from NetBSD is applicable.

I am using wsfb on wsdisplay on vesafb.

1) Does wsfb support allow starting X on two different displays (on same 
hardware)?

When I was already in X (:0), I did a "Xorg :2". It failed with:

(WW) xf86OpenConsole: VT_ACTIVATE failed

The X display I was at disappeared (but X and apps were still running).

Also the "Xorg :2" process was running (but not seen).

All I could see was now the non-X normal text console. It showed some X 
startup messages from before.

Killing the second Xorg process failed, so sent it a sigkill. I had to ssh 
in and reboot to use the keyboard and display again.

2) How can I troubleshoot the "VT_ACTIVATE failed"? I also have that 
problem when Xorg fails at startup (due to configuration error) it doesn't 
reset correctly so, next time starting does the "VT_ACTIVATE failed".

Here are some notes and kdump details:

I commented out the Mouse Protocol and Option for my InputDevice in my
(Continue reading)

Permalink | Reply | 4 comments |
headers
Michael Lorenz | 21 Dec 2006 16:33
Picon

Re: wsfb and two virtual displays and VT_ACTIVATE failed and properly aborting X???


Hello,

On Dec 21, 2006, at 10:10, Jeremy C. Reed wrote:

> This email has two do with two questions but related to maybe same
> problem. I am using Xorg from Xorg, but feel free to share advice or
> answers for XFree86 or Xorg from NetBSD is applicable.
>
> I am using wsfb on wsdisplay on vesafb.
>
> 1) Does wsfb support allow starting X on two different displays (on 
> same
> hardware)?

It should, but you're likely to run into other problems.

> When I was already in X (:0), I did a "Xorg :2". It failed with:
>
> (WW) xf86OpenConsole: VT_ACTIVATE failed

IIRC you need to tell the Xserver in which virtual console it's 
supposed to live, on i386 it will always try to grab the 5th one.

have fun
Michael
Permalink | Reply | 3 comments |
headers
Christian Biere | 21 Dec 2006 17:57
Picon
Picon
Gravatar

Re: wsfb and two virtual displays and VT_ACTIVATE failed and properly aborting X???

Michael Lorenz wrote:
> On Dec 21, 2006, at 10:10, Jeremy C. Reed wrote:
> >1) Does wsfb support allow starting X on two different displays (on 
> > same hardware)?

> IIRC you need to tell the Xserver in which virtual console it's 
> supposed to live, on i386 it will always try to grab the 5th one.

That's probably because there are 4 wscons displays by default. I've
hardcoded two unassigned terminals "vt07" and "vt08" per user here
in my startx script as workaround. This has worked for me with s3virge,
nv and mga.

--

-- 
Christian
Permalink | Reply | 2 comments |

Gmane