headers
Garrett D'Amore | 11 May 2006 01:32

looking for definitive VESA DMT specs

(Sorry for the broad cross-post.)

I'm looking for the definitive list of VESA "Discrete Monitor Timings",
so that I can incorporate them into the automatic EDID support that I'm
adding to NetBSD.  I've taken some modes from X11 mode lines, but some
of the modes that EDID refers to are missing, and I think others might
have deviations from the VESA specs.

I'd like to get the "official list" of VESA DMT modes (including all the
timing data needed to set up monitor modes -- e.g. the kind of data that
would normally be in an X11 ModeLine.)

If anyone has this information available to them, or has a copy of the
spec that they can loan me, I'd appreciate it.  (The spec is available
from VESA, but it costs like ~$350, and I'd rather not have to fork over
just to get the official list.)  The upshot of this will be improved
monitor detection and mode selection by framebuffer drivers like
radeonfb (to be integrated later), and probably also machfb, voodoofb,
and vesafb.

Thanks!

--

-- 
Garrett D'Amore, Principal Software Engineer
Tadpole Computer / Computing Technologies Division,
General Dynamics C4 Systems
http://www.tadpolecomputer.com/
Phone: 951 325-2134  Fax: 951 325-2191
(Continue reading)

Permalink | Reply | 0 comments |
headers
David Hopper | 12 May 2006 06:41

That old bugbear alpha_video.c

Apologies for the cross-post, but X11 on Alpha is awfully esoteric,  
it seems.

For the past week I've been hacking at compiling three codebases:   
the in-tree xsrc (XFree86 4.5.0), the X.org modular 7.0 release  
(haven't tried it from pkgsrc yet), and the brand shiny new XFree86  
4.6.0 that was released Wednesday.  They all three fail to compile on  
NetBSD Alpha in various ways-- where in the past our in-tree copy of  
XFree86 4.5.0 compiled fine on Alpha using 'make World' (see PR  
29882), it now takes a number of gross manual hacks to get it to  
produce a server binary (PR TBD).  [Our in-tree 4.5.0 codebase still  
wedges the multi-hose AlphaServer 1200, my main impetus to all this  
madness...]  I digress.

A number of brand-new NetBSD Alpha-specific fixes exist in hw/xfree86/ 
os-support/bsd/alpha_video.c for XFree86 4.6.0-- I have no idea on  
whose machine they function, or if they have ever been tested, or are  
simply theoretical code.  Well, I'm testing.  I get the following:

alpha_video.c:712: error: conflicting types for `xf86WriteMmio16'
../../../../../../programs/Xserver/hw/xfree86/common/compiler.h:1718:  
error: previous declaration of `xf86WriteMmio16'
alpha_video.c:714: error: conflicting types for `xf86WriteMmio32'
../../../../../../programs/Xserver/hw/xfree86/common/compiler.h:1719:  
error: previous declaration of `xf86WriteMmio32'
alpha_video.c:716: error: conflicting types for `xf86WriteMmioNB8'
../../../../../../programs/Xserver/hw/xfree86/common/compiler.h:1720:  
error: previous declaration of `xf86WriteMmioNB8'
alpha_video.c:718: error: conflicting types for `xf86WriteMmioNB16'
../../../../../../programs/Xserver/hw/xfree86/common/compiler.h:1721:
(Continue reading)

Permalink | Reply | 8 comments |
headers
Daniel Carosone | 12 May 2006 12:36
Picon

Re: That old bugbear alpha_video.c

On Thu, May 11, 2006 at 09:41:29PM -0700, David Hopper wrote:
> where in the past our in-tree copy of  
> XFree86 4.5.0 compiled fine on Alpha using 'make World' (see PR  
> 29882), it now takes a number of gross manual hacks to get it to  
> produce a server binary (PR TBD). 

Hm. It's very likely the case that it's been quite some time
since anyone tried building in-tree X11 this way, let alone on
alpha.

However, the code builds, and cross-builds, just fine via the
src/x11 build infrastructure, and build.sh with MKX11=yes.
Autobuilds and releases are done this way all the time.

That should at least get you to your next hurdle/hose.

--
Dan.
Permalink | Reply | 6 comments |
headers
Pavel Cahyna | 12 May 2006 12:51
Picon

Re: That old bugbear alpha_video.c

On Fri, May 12, 2006 at 08:36:21PM +1000, Daniel Carosone wrote:
> On Thu, May 11, 2006 at 09:41:29PM -0700, David Hopper wrote:
> > where in the past our in-tree copy of  
> > XFree86 4.5.0 compiled fine on Alpha using 'make World' (see PR  
> > 29882), it now takes a number of gross manual hacks to get it to  
> > produce a server binary (PR TBD). 
> 
> Hm. It's very likely the case that it's been quite some time
> since anyone tried building in-tree X11 this way, let alone on
> alpha.
> 
> However, the code builds, and cross-builds, just fine via the
> src/x11 build infrastructure, and build.sh with MKX11=yes.

It does not, at least not in 3.0 . Where "the code" means the XFree86
server that you need to support modern graphic cards.

> Autobuilds and releases are done this way all the time.

Autobuilds and releases do not include the XFree86 server.

Pavel
Permalink | Reply | 4 comments |
headers
David Hopper | 14 May 2006 14:11

XFree86 4.6.0 on Alpha - success

Hi there,

After a good deal of trial-and-error and learning more than I wanted  
to about volatile void pointers, I have a functional XFree86 4.6.0  
installation on my PWS 600au with a Radeon 9200 SE, running NetBSD  
from March of 2005 (3.99.1).  Anything >=3.0 should work, I imagine:   
current (3.99.19) also compiles the tree cleanly after patching.

XFree86 still wedges my AlphaServer 1200, but it has been doing this  
since 2002, so I'm not surprised by that.

XFree86 4.6.0 was released May 10, 2006.  If it is imported, then PCI- 
based Alphas shouldn't be marooned by the move.  X.org is still a no- 
op on Alpha, but apparently there is a proposal to rip out the old  
PCI cruft for 7.1 (cf. http://wiki.x.org/wiki/PciReworkProposal); I  
hope NetBSD-Alpha is represented there.

There are three files to patch, and one of them is, squeamishly, <sys/ 
types.h>.  Two simple #ifndef's bracket dev_t and key_t-- these must  
be handled by xf86_libc.h, but we need all of pio.h-->bus.h-->types.h  
and can't simply junk the #include.  It would be nice to shove this  
off onto xf86_libc.h but I couldn't figure out how (interestingly,  
this here Darwin on OS X brackets #ifndef's around dev_t and key_t in  
<sys/types.h>, too).

Download XFree86 4.6.0 from ftp.xfree86.org, apply the diffs, and do  
'make World && make install'.  Some card drivers may still have  
compilation issues; I haven't tested them all, only the ones I use.

Please-- I'm not a programmer, just a scavenger, so no points for
(Continue reading)

Permalink | Reply | 2 comments |
headers
David Brownlee | 14 May 2006 18:15
Gravatar

xorg or screen & more than 96 ptys

 	I've created /dev/ptm{,x} many ptys, and now have around 280
 	ptys for norm use (ssh etc), but xorg and screen only seem to
 	be able to see the first 96. Has anyone got any ideas on this?

--

-- 
 			   David Brownlee -- abs <at> absd.org
Permalink | Reply | 10 comments |
headers
David Brownlee | 14 May 2006 18:15
Gravatar

xorg or screen & more than 96 ptys

 	I've created /dev/ptm{,x} many ptys, and now have around 280
 	ptys for norm use (ssh etc), but xorg and screen only seem to
 	be able to see the first 96. Has anyone got any ideas on this?

--

-- 
 			   David Brownlee -- abs <at> absd.org
Permalink | Reply | 0 comments |
headers
Christos Zoulas | 14 May 2006 22:50

Re: xorg or screen & more than 96 ptys

In article <24633.8918725543$1147636141 <at> news.gmane.org>,
David Brownlee  <abs <at> absd.org> wrote:
> 	I've created /dev/ptm{,x} many ptys, and now have around 280
> 	ptys for norm use (ssh etc), but xorg and screen only seem to
> 	be able to see the first 96. Has anyone got any ideas on this?

Are you running 3.0 or current?

christos
Permalink | Reply | 8 comments |
headers
David Brownlee | 15 May 2006 17:25
Picon

Re: xorg or screen & more than 96 ptys

On Sun, 14 May 2006, Christos Zoulas wrote:

> In article <24633.8918725543$1147636141 <at> news.gmane.org>,
> David Brownlee  <abs <at> absd.org> wrote:
>> 	I've created /dev/ptm{,x} many ptys, and now have around 280
>> 	ptys for norm use (ssh etc), but xorg and screen only seem to
>> 	be able to see the first 96. Has anyone got any ideas on this?
>
> Are you running 3.0 or current?

 	sorry, should have stated - 3.0_STABLE from about two weeks ago.
--

-- 
 		David/absolute       -- www.NetBSD.org: No hype required --
Permalink | Reply | 4 comments |
headers
David Brownlee | 15 May 2006 17:25
Picon

Re: xorg or screen & more than 96 ptys

On Sun, 14 May 2006, Christos Zoulas wrote:

> In article <24633.8918725543$1147636141 <at> news.gmane.org>,
> David Brownlee  <abs <at> absd.org> wrote:
>> 	I've created /dev/ptm{,x} many ptys, and now have around 280
>> 	ptys for norm use (ssh etc), but xorg and screen only seem to
>> 	be able to see the first 96. Has anyone got any ideas on this?
>
> Are you running 3.0 or current?

 	sorry, should have stated - 3.0_STABLE from about two weeks ago.
--

-- 
 		David/absolute       -- www.NetBSD.org: No hype required --
Permalink | Reply | 0 comments |

Gmane