* [parisc-linux] Get your Matrox PCI cards out of the basement ...
@ 2004-06-02 22:17 Thomas Bogendoerfer
2004-06-03 3:23 ` M. Grabert
0 siblings, 1 reply; 6+ messages in thread
From: Thomas Bogendoerfer @ 2004-06-02 22:17 UTC (permalink / raw)
To: parisc-linux
....I got the matroxfb driver working on my B2600 including X on top
of fbdev (the mga driver doesn't work yet). So far I just used
a very old Matrox Millenium I. Below is the necessary patch for
2.6.7-rc1.
Have fun,
Thomas.
Index: matroxfb_base.h
===================================================================
RCS file: /home/cvs/parisc/linux-2.6/drivers/video/matrox/matroxfb_base.h,v
retrieving revision 1.3
diff -u -p -r1.3 matroxfb_base.h
--- matroxfb_base.h 18 Oct 2003 00:23:18 -0000 1.3
+++ matroxfb_base.h 2 Jun 2004 21:18:34 -0000
@@ -99,7 +99,7 @@
#endif
#endif
-#if defined(__alpha__) || defined(__mc68000__)
+#if defined(__alpha__) || defined(__mc68000__) || defined(__hppa__)
#define READx_WORKS
#define MEMCPYTOIO_WORKS
#else
@@ -165,27 +165,27 @@ typedef struct {
#ifdef READx_WORKS
static inline unsigned int mga_readb(vaddr_t va, unsigned int offs) {
- return readb(va.vaddr + offs);
+ return fb_readb(va.vaddr + offs);
}
static inline unsigned int mga_readw(vaddr_t va, unsigned int offs) {
- return readw(va.vaddr + offs);
+ return fb_readw(va.vaddr + offs);
}
static inline u_int32_t mga_readl(vaddr_t va, unsigned int offs) {
- return readl(va.vaddr + offs);
+ return fb_readl(va.vaddr + offs);
}
static inline void mga_writeb(vaddr_t va, unsigned int offs, u_int8_t value) {
- writeb(value, va.vaddr + offs);
+ fb_writeb(value, va.vaddr + offs);
}
static inline void mga_writew(vaddr_t va, unsigned int offs, u_int16_t value) {
- writew(value, va.vaddr + offs);
+ fb_writew(value, va.vaddr + offs);
}
static inline void mga_writel(vaddr_t va, unsigned int offs, u_int32_t value) {
- writel(value, va.vaddr + offs);
+ fb_writel(value, va.vaddr + offs);
}
#else
static inline unsigned int mga_readb(vaddr_t va, unsigned int offs) {
--
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea. [ Alexander Viro on linux-kernel ]
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] Get your Matrox PCI cards out of the basement ...
2004-06-02 22:17 [parisc-linux] Get your Matrox PCI cards out of the basement Thomas Bogendoerfer
@ 2004-06-03 3:23 ` M. Grabert
2004-06-03 13:35 ` Michael S. Zick
2004-06-03 13:38 ` Thomas Bogendoerfer
0 siblings, 2 replies; 6+ messages in thread
From: M. Grabert @ 2004-06-03 3:23 UTC (permalink / raw)
To: Thomas Bogendoerfer; +Cc: parisc-linux
On Thu, Jun 03, 2004 at 12:17:34AM +0200, Thomas Bogendoerfer wrote:
> ....I got the matroxfb driver working on my B2600 including X on top
> of fbdev (the mga driver doesn't work yet). So far I just used
> a very old Matrox Millenium I. Below is the necessary patch for
> 2.6.7-rc1.
I tried 2.6.7-rc2-pa1 with your patch (which applies fine),
but I wasn't successful in getting the card to work.
BTW, I'm running Debian/testing updated today.
I tried with the "VGA BIOS" dip switch set on and off,
I also tried the "init", "mem" boot parameters and different
video resolution modes. I even tried to swich it to another PCI slot,
but no success (the Voodoo2 works in the PCI slots I tried).
The result is similar to what I had with the ATI Rage XL I
also tried to get working ... the graphics card is detected,
and accessible via /dev/fb0, and I can use fbset to change
the resoltions, but I don't see any output on my monitor(s) ...
is simply stays black!
There is also a error message when the matrox is detected on
boot:
beast kernel: matroxfb: Matrox Millennium (PCI) detected
beast kernel: matroxfb: Your Matrox device does not have BIOS
beast kernel: matroxfb: 640x480x8bpp (virtual: 640x3276)
beast kernel: matroxfb: framebuffer at 0xF4800000, mapped to 0xf4800000, size 2097152
beast kernel: fb0: MATROX frame buffer device
beast kernel: fb0: initializing hardware
beast kernel: matroxfb: Pixel PLL not locked after 5 secs
beast kernel: matroxfb: Loop PLL not locked after 5 secs
N.B. the "PLL not locked after 5 secs" messages also appear when
I try to change the resolution with fbset. fbset is busy for some
seconds (at first I thought the process locked up), but then returns
printing out those messages, and another 'fbset' (without arguments)
afterwards confirms that the resolution has been changed.
However the monitor stays black ...
Any hints?
Thanks,
Max
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] Get your Matrox PCI cards out of the basement ...
2004-06-03 3:23 ` M. Grabert
@ 2004-06-03 13:35 ` Michael S. Zick
[not found] ` <20040603151347.GB23488@sal.ucc.ie>
2004-06-03 13:38 ` Thomas Bogendoerfer
1 sibling, 1 reply; 6+ messages in thread
From: Michael S. Zick @ 2004-06-03 13:35 UTC (permalink / raw)
To: Thomas Bogendoerfer, parisc-linux
On Wed June 2 2004 22:23, M. Grabert wrote:
> N.B. the "PLL not locked after 5 secs" messages also appear when
> I try to change the resolution with fbset. fbset is busy for some
> seconds (at first I thought the process locked up), but then returns
> printing out those messages, and another 'fbset' (without arguments)
> afterwards confirms that the resolution has been changed.
> However the monitor stays black ...
>
> Any hints?
Monitor? What monitor? How cabled?
There are three flavors of getting sync to the monitor;
composite, sync on green, and separate...
A display card with a SVGA connector on it will produce
one, two or all three flavors - either conncurrently or selectively
depending on card (manufacturerer, age, etc)...
Similar story with monitors...
If using an older HP (or other) high resolution monitor - it
most likely expects "sync on green" (or "separate" if it uses
BNC connectors)...
While your graphics card may have to be set (either manually
or by software) to produce other than "composite".
The short answer:
Your not out of luck yet - It could be something simple.
Mike
PS: the monitor that came with my 720 (5-BNC connectors)
will run with either "separate" or "sync on green" - it doesn't
recognize "composite" (the pc-world standard).
(Once you buy or build a suitable conversion cable)
PPS: SVGA connector:
1 - red
2 - green (sync on green here)
3 - blue
4 - sense 2
5 - self test (and TTL ground)
6 - red ground
7 - green ground
8 - blue ground
9 - key
10 - sync ground
11 - sense 0
12 - sense 1
13 - Hor. Sync (separate or composite)
14 - Vert. Sync
15 - sense 3
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] Get your Matrox PCI cards out of the basement ...
2004-06-03 3:23 ` M. Grabert
2004-06-03 13:35 ` Michael S. Zick
@ 2004-06-03 13:38 ` Thomas Bogendoerfer
2004-06-05 9:47 ` Success! was Re: [parisc-linux] Get your Matrox M. Grabert
1 sibling, 1 reply; 6+ messages in thread
From: Thomas Bogendoerfer @ 2004-06-03 13:38 UTC (permalink / raw)
To: parisc-linux
On Thu, Jun 03, 2004 at 04:23:04AM +0100, M. Grabert wrote:
> beast kernel: matroxfb: Pixel PLL not locked after 5 secs
> beast kernel: matroxfb: Loop PLL not locked after 5 secs
bad, either the PLL setup write didn't reach hardware, or
reading of the locked bit of the PLL fails. It looks like
the PCI subsystem behaves differently between my B2600 and
your C3000 :-(
> N.B. the "PLL not locked after 5 secs" messages also appear when
> I try to change the resolution with fbset. fbset is busy for some
> seconds (at first I thought the process locked up), but then returns
makes sense. fbset triggers a reprogramming of the PLL and that
fails aber 5 seconds.
> Any hints?
not at the moment :-(
Can anybody describe the difference of the PCI bus between C3000 and
B2600 ?
Thomas.
--
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea. [ Alexander Viro on linux-kernel ]
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [parisc-linux] Get your Matrox PCI cards out of the basement ...
[not found] ` <20040603151347.GB23488@sal.ucc.ie>
@ 2004-06-03 16:07 ` Michael S. Zick
0 siblings, 0 replies; 6+ messages in thread
From: Michael S. Zick @ 2004-06-03 16:07 UTC (permalink / raw)
To: M. Grabert; +Cc: parisc-linux
On Thu June 3 2004 10:13, you wrote:
> On Thu, Jun 03, 2004 at 08:35:25AM -0500, Michael S. Zick wrote:
> > On Wed June 2 2004 22:23, M. Grabert wrote:
>
> SGI Multisync Monitor (NOT sync-on-green capable AFAIK, works fine
> with the Voodoo2 and also with PCs, with the 'normal' VGA cable).
General Rule -
Anything recent enough to call itself "multisync" can sense and use
"sync-on-green" (My limited experience is that anything properly
named "multisync" with an SVGA connector will accept any of the three.)
>
> This monitor doesn't display a "out-of-sync" (you will just hear
> a clicking noise), so just to get sure I also connected it to a
> analogue SVGA LCD display, that tells you if it is out-of-sync.
Ah...
I had that sort of monitor - the noise is relays used to select
sync ranges.
So the monitor is sensing sync pulses (however they get there)
but can't find a matching range (or the pulses aren't stable
enough).
So suspect software setup of the card first -
Sounds like you have (unusable) sync to the monitor.
>
> BTW, I thought that the 'seperate' is actually using the normal
> VGA cable signalling, but seperates each signal/wire physically and
> uses BNC connectors.
I don't have any handy reference for VGA cable/connector - SVGA
connector has them all.
> Eg. I have a PC MultiSync (not here in Ireland) that accepts two
> inputs, VGA connector and 5 BNC.
If you ever get the chance - disconnect the H & V BNCs - it will probably
still work (VGA, 5-BNC and 3-BNC)
> Yeah, I know. I had a SUN Monitor (using sync-on-green), for which
> I built a VGA-to-SyncOnGreen adapter. That was pretty easy to do.
>
> Unfortunately the old HP monitors/graphics cards are using a much more
> complicated, proprietary sync-on-green signalling (and voltage)
> much unlike the SUNs, so building a VGA converter is much more
> complicated. IIRC. (I remember that I found a schematic for those
> HP VGA converters some time ago and I was scared).
I didn't know that - thanks for the info.
I have driven several different Sony multisync monitors with the 720
(and a custom - wire only - cable) - they survived the experience;
but based on the above - it might be good that I didn't do it for
very long.
>
> However Milleniums have (undocumented) hardware SyncOnGreen support,
> and most OSS drivers (eg. XFree86) can enable it, but it is
> disabled by default, at least on the PC version (which I have).
Those messages about not getting the cards PLLs to lock are more
likely to be the lead to getting a picture - but something else might
not be getting its setup also.
Mike
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
* Success! was Re: [parisc-linux] Get your Matrox ...
2004-06-03 13:38 ` Thomas Bogendoerfer
@ 2004-06-05 9:47 ` M. Grabert
0 siblings, 0 replies; 6+ messages in thread
From: M. Grabert @ 2004-06-05 9:47 UTC (permalink / raw)
To: Thomas Bogendoerfer; +Cc: parisc-linux
On Thu, Jun 03, 2004 at 03:38:19PM +0200, Thomas Bogendoerfer wrote:
> On Thu, Jun 03, 2004 at 04:23:04AM +0100, M. Grabert wrote:
> >
> > beast kernel: matroxfb: Pixel PLL not locked after 5 secs
> > beast kernel: matroxfb: Loop PLL not locked after 5 secs
I finally was able to put the Matrox in a desktop PC.
It worked without problems! But just to be sure I performed a
graphic card BIOS upgrade (Setup351.exe from Matrox.com).
And then I put it back into my PA machine and ...
... now it works! No more PLL errors!
Now I'm running 1280x1024, 16bit @ 75Hz (32bit also works,
but visibly slower, since it's not accelerated and twice as much
pixel data has to be moved if you move a window. 16bit is fine).
Well, I could run higher resolutions, but I just have a 17" monitor ;)
> bad, either the PLL setup write didn't reach hardware, or
> reading of the locked bit of the PLL fails. It looks like
> the PCI subsystem behaves differently between my B2600 and
> your C3000 :-(
>
> Can anybody describe the difference of the PCI bus between C3000 and
> B2600 ?
I thought there is almost no difference between the B2xxx and C3xxx
series ...
Greetings,
Max
PS: In case your PA box crashes, the Matrox RAM/config/whatever can
come into a inconsistent state and will refuse to be initialized
properly if you simply reboot (but it will not print out error msgs).
You *must* power cycle the computer (and I believe even disconnect it
physically from the power line for a couple of seconds) in order
to reset the Matrox properly, and be able to use it again.
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-06-05 9:47 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-06-02 22:17 [parisc-linux] Get your Matrox PCI cards out of the basement Thomas Bogendoerfer
2004-06-03 3:23 ` M. Grabert
2004-06-03 13:35 ` Michael S. Zick
[not found] ` <20040603151347.GB23488@sal.ucc.ie>
2004-06-03 16:07 ` Michael S. Zick
2004-06-03 13:38 ` Thomas Bogendoerfer
2004-06-05 9:47 ` Success! was Re: [parisc-linux] Get your Matrox M. Grabert
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox