[9fans] Yet another VGA issue (CT65550)

[9fans] Yet another VGA issue (CT65550)

[Matt Gushee]

Hello, all--

Well, I have wanted to try out Plan9 for some time, and finally decided
to take the plunge. I have installed the distribution on my old Toshiba
Satellite Pro 420CDS, which has a Chips & Tech 65550 video card. I have
got a working system--I can log in, navigate around, execute various
commands, etc., but I can't get VGA working. Actually, this is not a
complete surprise, because I've run Linux on this box for several years,
and had a hell of a time getting XFree86 working--actually none of the
several configuration tools I tried could generate appropriate settings
for my card; I ended up having to do it manually. Anyway, here's the
output of 'aux/vga -piV' and the relevant portions of my working
XFree86Config. If anyone has an idea what to do about this, I'd greatly
appreciate your suggestions:

aux/vga -piV
------------
Oops, I guess not! The floppy disk containing that info has died, and my
9box is put away for the moment. Guess I'll have to send it later.

One somewhat unusual thing I noticed is that the base memory address for
the CT65550 on my machine is 0xE0000. But I tried adding an appropriate
entry to /lib/vgadb, e.g.

    0xE0090="TOSHIBA Video BIOS V2.00"

but that only resulted in error messages like 'hiqvideo: not found'.
(Note to Plan9 maintainers: couldn't these error messages be a little
more descriptive, explaining exactly what was not found where? It might
be helpful to confused newbies like me)

Anyway, here's the XFree86 stuff. Perhaps it will be of some use.


XF86Config
----------

Section "Monitor"

    Identifier  "Crawdad Viddy"
    VendorName  "Toshiba"
    ModelName   "800x600 LCD"

	Bandwidth 25.2
	HorizSync   25-64
	VertRefresh 40-70

	Modeline "800x600@8bpp"		40     800  840  968 1056   600  601  605  628 +hsync +vsync
	Modeline "800x600@16bpp"	35.464     800  824  896 1024   600  601  603  625
	# Modeline "800x600@24bpp"	31.5   800  824  896 1024   600  601  603  625
	Modeline "800x600@24bpp"	26.598	800 824 896 1024  600 601 603 625

EndSection


# **********************************************************************
# Graphics device section
# **********************************************************************

# Any number of graphics device sections may be present

# Standard VGA Device:

Section "Device"
    Identifier	"Generic VGA"
    VendorName	"Unknown"
    BoardName	"Unknown"
    Chipset	"generic"

#    VideoRam	256

#    Clocks	25.2 28.3

EndSection

# Device configured by xf86config:

Section "Device"
    Identifier  "My Video Card"
    VendorName  "Chips & Technologies"
    BoardName   "ct65550"
    #VideoRam    2048
Option "sw_cursor"
Textclockfreq 36
Option "STN"
# Option "no_stretch"
# Option "no_center"
# Option "use_modeline"
# Option "fix_panel_size"
    # Insert Clocks lines here if appropriate
EndSection

--
Matt Gushee                 When a nation follows the Way,
Englewood, Colorado, USA    Horses bear manure through
mgushee@havenrock.com           its fields;
http://www.havenrock.com/   When a nation ignores the Way,
                            Horses bear soldiers through
                                its streets.

                            --Lao Tzu (Peter Merel, trans.)

Re: [9fans] Yet another VGA issue (CT65550)

[rsc]

> aux/vga -piV
> ------------
> Oops, I guess not! The floppy disk containing that info has died, and my
> 9box is put away for the moment. Guess I'll have to send it later.

Also include pci.txt.  The ``not found'' message means
no hiqvideo (65550) card was found that we know how
to drive.  There's only a few PCI IDs we support, and
yours isn't in the list.

Re: [9fans] Yet another VGA issue (CT65550)

[Matt Gushee]

On Mon, May 05, 2003 at 04:24:50PM -0400, rsc@plan9.bell-labs.com wrote:
> > aux/vga -piV
> > ------------
> > Oops, I guess not! The floppy disk containing that info has died, and my
> > 9box is put away for the moment. Guess I'll have to send it later.
>
> Also include pci.txt.  The ``not found'' message means
> no hiqvideo (65550) card was found that we know how
> to drive.  There's only a few PCI IDs we support, and
> yours isn't in the list.

You know, I don't think the machine in question even has a PCI bus. It's
a 1996 model. At any rate, pci.txt is empty. Here's the vga stuff.


aux/vga -piV
------------
main->snarf
vga->snarf
vga->dump
vmf 25175000 vmdf 0 vf1 0 vbw 0
vga->init
vga->dump
main->exits
aux/vga: controller not in /lib/vgadb
0xE0000 55 AA 80 EB 3F 37 34 30 30 30 30 30 30 30 30 30  U...?74000000000
0xE0010 30 30 30 30 30 30 30 23 78 01 90 01 30 30 49 42  0000000#x...00IB
0xE0020 4D 20 56 47 41 20 43 6F 6D 70 61 74 69 62 6C 65  M VGA Compatible
0xE0030 20 42 49 4F 53 2E 20 34 1A 05 81 00 B2 01 C2 01   BIOS. 4........
0xE0040 D0 01 F6 01 E9 F2 8F 86 D0 D2 E0 E6 40 C2 DC C8  ............@...
0xE0050 40 A8 CA C6 D0 DC DE D8 DE CE D2 CA E6 58 40 92  @............X@.
0xE0060 DC C6 5C 00 FF 01 50 55 2D 10 83 00 50 55 54 00  ..\...PU-...PUT.
0xE0070 00 00 00 07 12 CC 07 0A 18 07 0A 00 A0 7F 00 47  ...............G
0xE0080 00 00 E0 00 A0 00 00 47 00 00 00 91 16 00 00 00  .......G........
0xE0090 54 4F 53 48 49 42 41 20 56 69 64 65 6F 20 42 49  TOSHIBA Video BI
0xE00A0 4F 53 20 56 32 2E 30 30 0D 0A 00 00 00 00 00 00  OS V2.00........
0xE00B0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
0xE00C0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
0xE00D0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
0xE00E0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
0xE00F0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
main->snarf
vga->snarf
vga->dump
vga misc             67
vga feature          01
vga sequencer        03 01 03 00 02
vga crt              5F 4F 50 82 55 81 BF 1F - 00 4F 0D 0E 00 00 07 80
                     9C 0E 8F 28 1F 96 B9 A3 - FF
vga graphics         00 00 00 00 00 10 0E 00 - FF
vga attribute        00 01 02 03 04 05 14 07 - 38 39 3A 3B 3C 3D 3E 3F
                     0C 00 0F 00 00
vga virtual         0 0
vga panning         off
vga apz                     0
vga linear                  0

vmf 25175000 vmdf 0 vf1 0 vbw 0
vga->init
dbdumpmode
type=vga, size=640x480x1
frequency=25175000
x=640 (0x280), y=480 (0x1E0), z=1 (0x1)
ht=800 (0x320), shb=664 (0x298), ehb=760 (0x2F8)
shs=664 (0x298), ehs=760 (0x2F8)
vt=525 (0x20D), vrs=491 (0x1EB), vre=493 (0x1ED)
hsync=0, vsync=0, interlace=0
vga->dump
vga flag             Fdump|Finit|Fsnarf
vga misc             E3
vga feature          01
vga sequencer        03 01 0F 00 06
vga crt              5F 4F 52 9F 53 1F20B 3E - 00 40 00 00 00 00 00 00
                    1EB 2D1DF 28 001EB1EC C3 -7FF
vga graphics         00 00 00 00 00 00 05 0F - FF
vga attribute        00 01 02 03 04 05 06 07 - 08 09 0A 0B 0C 0D 0E 0F
                     01 FF 0F 00 00
vga virtual         640 480
vga panning         off
vga apz                     0
vga linear                  0

main->exits


--
Matt Gushee                 When a nation follows the Way,
Englewood, Colorado, USA    Horses bear manure through
mgushee@havenrock.com           its fields;
http://www.havenrock.com/   When a nation ignores the Way,
                            Horses bear soldiers through
                                its streets.

                            --Lao Tzu (Peter Merel, trans.)

Re: [9fans] Yet another VGA issue (CT65550)

[rsc]

Are you sure it's a CT65550 and not a CT65540 or CT65545?
If it's one of the latter, then you need to move your ID string
into the ct65540 section of the vgadb.

Re: [9fans] Yet another VGA issue (CT65550)

[Matt Gushee]

On Mon, May 05, 2003 at 06:01:53PM -0400, rsc@plan9.bell-labs.com wrote:
> Are you sure it's a CT65550 and not a CT65540 or CT65545?

About as sure as one could be without physically inspecting the card.
The specs from Toshiba say that's what it is*; there was also a time a
few years ago when the XFree86 developers were having trouble with the
driver for this card, and I did a register dump for them; they seemed to
think it was the right stuff. But as I suggested, there's something
weird about the way it's set up on this machine, so that XFree86
configuration tools just don't work.

One thing that caught my eye as I was sending an earlier message: the
XF86Config had

  Option "sw_cursor"

and I think that may have been required in order to get X working
(other than FontPath settings, I haven't touched that config for over 5
years, so I'm not sure). Of course, I have no idea if that's relevant to
Plan9.


  * I bought the computer in Japan, but I couldn't find specs for it at
    Toshiba's Japanese Web site, so I got them from the US site. It's
    the same model number, but I suppose it's conceivable that the US
    and Japanese models used different components.

--
Matt Gushee                 When a nation follows the Way,
Englewood, Colorado, USA    Horses bear manure through
mgushee@havenrock.com           its fields;
http://www.havenrock.com/   When a nation ignores the Way,
                            Horses bear soldiers through
                                its streets.

                            --Lao Tzu (Peter Merel, trans.)

Re: [9fans] Yet another VGA issue (CT65550)

[jmk]

Since there's no PCI info, I'm guessing the CT65550 (if it is such)
is connected via a Vesa Local Bus (VLB). The aux/vga code doesn't handle
that, neither does the kernel part of the driver. Code needs to be
added to recognise the chip revision and to set up the framebuffer
in an unallocated memory range, etc. It's probably not hard, but awkward
without the machine to hand.

Re: [9fans] Yet another VGA issue (CT65550)

[Matt Gushee]

On Mon, May 05, 2003 at 10:11:44PM -0400, jmk@plan9.bell-labs.com wrote:
> Since there's no PCI info, I'm guessing the CT65550 (if it is such)
> is connected via a Vesa Local Bus (VLB).

Yes, that's what the specs say.

> The aux/vga code doesn't handle
> that, neither does the kernel part of the driver. Code needs to be
> added to recognise the chip revision and to set up the framebuffer
> in an unallocated memory range, etc. It's probably not hard, but awkward
> without the machine to hand.

:-((

Maybe I should finally learn C, so I can do it.

Oh well, I guess that machine's going back to Linux. I'm not
discouraged, but I want at least a windowing system and a usable text
editor on that box by tomorrow, so ...

I'll be back someday, as time allows. Thanks to you and 'rsc' for the
prompt attention.

--
Matt Gushee                 When a nation follows the Way,
Englewood, Colorado, USA    Horses bear manure through
mgushee@havenrock.com           its fields;
http://www.havenrock.com/   When a nation ignores the Way,
                            Horses bear soldiers through
                                its streets.

                            --Lao Tzu (Peter Merel, trans.)

[9fans] Re: Yet another VGA issue (CT65550)

[Jim Choate]

On Mon, 5 May 2003, Matt Gushee wrote:

> Hello, all--
>
> Well, I have wanted to try out Plan9 for some time, and finally decided
> to take the plunge. I have installed the distribution on my old Toshiba
> Satellite Pro 420CDS, which has a Chips & Tech 65550 video card. I have
> got a working system--I can log in, navigate around, execute various
> commands, etc., but I can't get VGA working. Actually, this is not a
> complete surprise, because I've run Linux on this box for several years,
> and had a hell of a time getting XFree86 working

Interesting in that I've got a 445CDX (if memory serves it uses the same
chipset) and had zero problem with X under Linux, though I gave up on
Plan 9 windowing a long time ago (this was before Toshiba actually
'supported' Linux even). Have you tried another 420CDS to make sure yours
is all hunkey dorey? The general experience of the 4xx Sat Pro's was
positive under Linux for the vast majority of users.

If you work on it and need somebody to test a mod I'll help.

I've recently come into access (though not strict possession) of an iBook
(2 ed, slate & white) anybody know if it works once a PPC version of Plan
9 is compiled? To that end, I'd be interested in hearing of any experience
doing cross-compiles x86 -> PPC, I assume you don't need the --host and
--target like with the gcc tools...


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] Yet another VGA issue (CT65550)

[Jim Choate]

On Mon, 5 May 2003 rsc@plan9.bell-labs.com wrote:

> Are you sure it's a CT65550 and not a CT65540 or CT65545?
> If it's one of the latter, then you need to move your ID string
> into the ct65540 section of the vgadb.

I believe the 445 uses 65554.


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] Yet another VGA issue (CT65550)

[M.Stein]

Hello All,

this is only for your information:

the 65555 chip works too, when adding 'case 0x00E5' in
vgahiqvideo.c/hiqvideo.c and setting the frame buffer
size to 4Mb (tested with Kontron IP LiteII).

					matthias

On Mon, 5 May 2003 rsc@plan9.bell-labs.com wrote:

> > aux/vga -piV
> > ------------
> > Oops, I guess not! The floppy disk containing that info has died, and my
> > 9box is put away for the moment. Guess I'll have to send it later.
>
> Also include pci.txt.  The ``not found'' message means
> no hiqvideo (65550) card was found that we know how
> to drive.  There's only a few PCI IDs we support, and
> yours isn't in the list.
>
>

Re: [9fans] Re: Yet another VGA issue (CT65550)

[Charles Forsyth]

>>doing cross-compiles x86 -> PPC, I assume you don't need the --host and
>>--target like with the gcc tools...

no, as you suggested, that part's trivial with plan 9:  a cross-compiler suite
is simply a compiler suite compiled for some other architecture than its target.
thus, compile qa, qc, ql on x86 (or anything else), run them, and they generate code for powerpc.
for commands and libraries, use
	objtype=power
	mk install
in /sys/src and it will make and install /power versions of everything
(except to the extent that it runs into dented mkfiles or source files).

Re: [9fans] Re: Yet another VGA issue (CT65550)

[Charles Forsyth]

>>is simply a compiler suite compiled for some other architecture than its target.

of course i meant, `compiled on some architecture other than ...', not `compiled for ...'.

[9fans] Cross-Building (was: Yet another VGA issue (CT65550))

[Jim Choate]

On Tue, 6 May 2003, Charles Forsyth wrote:

> for commands and libraries, use
> 	objtype=power
> 	mk install
> in /sys/src and it will make and install /power versions of everything
> (except to the extent that it runs into dented mkfiles or source files).

Thanks for the feedback Charles.

And would anybody have actually tried this and identified 'dented' files?

And speaking of alternate targets, I've got an old Tadpole 3GX
SparcStation 5 equiv. laptop, I assume the ISDN adapter is -not-
supported but should otherwise work pretty much as a regular SparcStation
5 would behave?

I've also got a IBM (Tadpole) N40 laptop that is currently using AIX
3.2.5, anybody try this particular model?


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] Yet another VGA issue (CT65550)

[splite]

On Mon, May 05, 2003 at 02:08:10PM -0600, Matt Gushee wrote:
>
> Well, I have wanted to try out Plan9 for some time, and finally decided
> to take the plunge. I have installed the distribution on my old Toshiba
> Satellite Pro 420CDS, which has a Chips & Tech 65550 video card. I have

Dunno if this will help you, but here are the changes I made to a 3rd Ed.
system to get video working on an old Toshiba (Libretto, I think) with
a CT65550 (HiQV32) chipset in VLB mode.  (Sorry for the format; I don't
have a Plan 9 machine running at the moment.  I can make the full source
files available, if that would be more useful.)

diff hiqvideo.c.orig hiqvideo.c
50a51,52
> 	enum {DUNNO, CT65550, CT65554, CT69000};
> 	int model=DUNNO;
55,58c57,86
< 		if((p = pcimatch(nil, 0x102C, 0)) == nil)
< 			error("%s: not found\n", ctlr->name);
< 		switch(p->did){
< 		case 0x00C0:		/* 69000 HiQVideo */
---
> 		if((p = pcimatch(nil, 0x102C, 0))){
> 			switch(p->did){
> 			case 0x00C0:		/* 69000 HiQVideo */
> 				model=CT69000;
> 				break;
> 			case 0x00E0:		/* 65550 HiQV32 */
> 				model=CT65550;
> 				break;
> 			case 0x00E4:		/* 65554 HiQV64 */
> 				model=CT65554;
> 				break;
> 			default:
> 				error("%s: PCI DID %4.4uX unsupported\n",
> 					ctlr->name, p->did);
> 			}
> 		}
> 		else{
> 			switch(hiqvideoxi(Xrx, 0x02)){
> 			case 0xE0:
> 			case 0xE4:
> 				model=CT65550;
> 				print("%s: found CT65550 in VLB mode\n", ctlr->name);
> 				break;
> 			default:
> 				error("%s: not found\n", ctlr->name);
> 			}
> 		}
> 		hqv->pci = p;
> 		switch(model){
> 		case CT69000:
62,63c90,91
< 		case 0x00E0:		/* 65550 HiQV32 */
< 		case 0x00E4:		/* 65554 HiQV32 */
---
> 		case CT65550:
> 		case CT65554:
67,69d94
< 		default:
< 			error("%s: DID %4.4uX unsupported\n",
< 				ctlr->name, p->did);
71d95
< 		hqv->pci = p;
84,85c108,109
< 	switch(hqv->pci->did){
< 	case 0x00C0:			/* 69000 HiQVideo */
---
> 	switch(model){
> 	case CT69000:			/* 69000 HiQVideo */
89,90c113,114
< 	case 0x00E0:			/* 65550 HiQV32 */
< 	case 0x00E4:			/* 65554 HiQV32 */
---
> 	case CT65550:			/* 65550 HiQV32 */
> 	case CT65554:			/* 65554 HiQV64 */


diff vgahiqvideo.c.orig vgahiqvideo.c
15a16,17
> 	Pos102	= 0x102,	/* Global Enable */
> 	Vse		= 0x3C3,	/* Video Subsystem Enable */
53c55
< 		case 0x00E4:		/* 65554 HiQV32 */
---
> 		case 0x00E4:		/* 65554 HiQV64 */
60a63,72
> 	else{					/* check for 65550 in VLB mode */
> 		switch(hiqvideoxi(Xrx, 0x02)){
> 		case 0xE0:
> 		case 0xE4:
> 			*size = *align = 8*1024*1024;
> 			break;
> 		default:
> 			break;
> 		}
> 	}
80a93,100
> 	if (aperture && !p){		/* set aperature by hand if in VLB mode */
> 		int v;
> 		v = hiqvideoxi(Xrx, 0x0A);
> 		hiqvideoxo(Xrx, 0x0A, v | 0x02);
> 		hiqvideoxo(Xrx, 0x20, 0x00);
> 		hiqvideoxo(Xrx, 0x05, (aperture>>16) & 0xFF);
> 		hiqvideoxo(Xrx, 0x06, (aperture>>24) & 0xFF);
> 	}
103c123
< 		case 0x00E4:		/* 65554 HiQV32 */
---
> 		case 0x00E4:		/* 65554 HiQV64 */
104a125,128
> 			case 0:
> 				vmsize = 1*1024*1024;
> 				break;
> 			case 1:
105a130,144
> 				vmsize = 2*1024*1024;
> 				break;
> 			}
> 			break;
> 		default:
> 			return;
> 		}
> 		size = p->mem[0].size;
> 		align = 0;
> 	}
> 	else{					/* maybe in VLB mode? */
> 		switch(hiqvideoxi(Xrx, 0x02)){
> 		case 0xE0:
> 		case 0xE4:
> 			switch((hiqvideoxi(Xrx, 0x43)>>1) & 0x03){
109a149
> 			default:
112a153
> 			size = align = 8*1024*1024;
118,119d158
< 	else
< 		return;
121,122d159
< 	size = p->mem[0].size;
< 	align = 0;