Using the "chips" XFree86 driver on a PowerBook 3400 (almost there.)

Using the "chips" XFree86 driver on a PowerBook 3400 (almost there.)

[Michael Stephen Hanni]

[-- Attachment #1: Type: text/plain, Size: 1844 bytes --]

Hi,

Being of unsound mind and body I decided to see if the chips driver
could be made to work on my PowerBook 3400. This model of powerbook
includes the ct65550 chipset -- i.e. HiQV -- and has 1 meg of video ram.
>From reading the documentation in the source tree it would appear that
the "chips" driver fully supports this chipset and furthermore sports
full acceleration -- which would really be nice on this older powerbook!
However, in using the driver my powerbook lockedup nicely. After some
trying of various config switches I got the thing to come up by using
the "NoAccel" option. Alas, the colors were hopelessly screwed up, the
video ram wasn't detected correctly, and the hardware cursor was not an
X but rather <>. Suffice it to say, it was not usable.

After some small hacking I've gotten the colors right (though this may
just be a coincidence) and I've fixed the ram detection problem (I'm
more certain here as it was only a small fix.) Unfortunately I still can
not use the acceleration as it freezes the machine solid. Sometimes the
screen will come up, but it is frozen on arrival (FOA.)

Other issues:

1.) there seems to be a lot of visuals coming up. That is, by using the
fbdev I get 2 TrueColor visuals for this machine, with the chips driver
I get 2 TrueColor and 2 DirectColor.

2.) XVideo is hosed. When I try to use any program that makes use of
xvideo I get a BadAlloc error reporting there are insufficient
resources.

Patch:

This is against a freshly updated CVS copy of the xc module.

I'd really appreciate any help any of you could offer on making this
baby work. I'm not sure what exactly has to be fixed endian-wise to make
acceleration work (the *ToCPU* stuff maybe?) and what is going on with
xvideo.

If there is any info I can give please let me know and I'd be happy to
pass it on.

Cheers!

Michael

[-- Attachment #2: ct_driver.c.patch --]
[-- Type: text/plain, Size: 1235 bytes --]

Index: ct_driver.c
===================================================================
RCS file: /cvs/xc/programs/Xserver/hw/xfree86/drivers/chips/ct_driver.c,v
retrieving revision 1.121
diff -u -r1.121 ct_driver.c
--- ct_driver.c	2002/09/16 18:05:52	1.121
+++ ct_driver.c	2002/10/08 23:39:30
@@ -1525,7 +1525,12 @@
     /* linear base */
     if (cPtr->Flags & ChipsLinearSupport) {
 	if (cPtr->pEnt->location.type == BUS_PCI) {
+	    /* Tack on 0x800000 to access the big-endian aperture? */
+#if X_BYTE_ORDER == X_BIG_ENDIAN
+	    cPtr->FbAddress =  (cPtr->PciInfo->memBase[0] & 0xff800000) + 0x800000;
+#else
 	    cPtr->FbAddress =  cPtr->PciInfo->memBase[0] & 0xff800000;
+#endif
 	    from = X_PROBED;
 	    if (xf86RegisterResources(cPtr->pEnt->index,NULL,ResNone))
 		cPtr->Flags &= ~ChipsLinearSupport;
@@ -1786,6 +1791,8 @@
 		pScrn->videoRam = 2048;
 		break;
 	    }
+	    /* We need to break out completely, i.e. do not fall through. */
+            break;
 	default:
 	    /* XRE0: Software reg     */
 	    /* bit 3-0: memory size   */
@@ -1817,7 +1824,6 @@
 	    }
 	}
     }
-
 
     if ((cPtr->Flags & ChipsDualChannelSupport) &&
 		(xf86IsEntityShared(pScrn->entityList[0]))) {

Re: Using the "chips" XFree86 driver on a PowerBook 3400 (almost there.)

[Ani Joshi]

It seems you have changed the address space to use the big endian
aperture, this is correct.  As for the accel engine, try turning off all
acceleration (just #if them out in the driver, no need for several XaaNo*
Options), and leave on SolidFills, this is the easiset to debug.  There is
a problem with some 65550's and polling of the BR04 register which tends
to lockup the engine, you may be running into that.  I would grab the old
source from Xpmac (I believe it had c&t accel) and poke around there to
see where the engine is initialized and the proper way to BLTWAIT on this
chip, its been long since I hacked on the c&t65550 so I don't remember
sorry.

As for the hardware cursor, you will have to swap the image of the cursor.
See the hw cursor support for radeon (or r128) and apply similair
necessary the big endian bits.


ani

On 8 Oct 2002, Michael Stephen Hanni wrote:

> Hi,
>
> Being of unsound mind and body I decided to see if the chips driver
> could be made to work on my PowerBook 3400. This model of powerbook
> includes the ct65550 chipset -- i.e. HiQV -- and has 1 meg of video ram.
> >From reading the documentation in the source tree it would appear that
> the "chips" driver fully supports this chipset and furthermore sports
> full acceleration -- which would really be nice on this older powerbook!
> However, in using the driver my powerbook lockedup nicely. After some
> trying of various config switches I got the thing to come up by using
> the "NoAccel" option. Alas, the colors were hopelessly screwed up, the
> video ram wasn't detected correctly, and the hardware cursor was not an
> X but rather <>. Suffice it to say, it was not usable.
>
> After some small hacking I've gotten the colors right (though this may
> just be a coincidence) and I've fixed the ram detection problem (I'm
> more certain here as it was only a small fix.) Unfortunately I still can
> not use the acceleration as it freezes the machine solid. Sometimes the
> screen will come up, but it is frozen on arrival (FOA.)
>
> Other issues:
>
> 1.) there seems to be a lot of visuals coming up. That is, by using the
> fbdev I get 2 TrueColor visuals for this machine, with the chips driver
> I get 2 TrueColor and 2 DirectColor.
>
> 2.) XVideo is hosed. When I try to use any program that makes use of
> xvideo I get a BadAlloc error reporting there are insufficient
> resources.
>
> Patch:
>
> This is against a freshly updated CVS copy of the xc module.
>
> I'd really appreciate any help any of you could offer on making this
> baby work. I'm not sure what exactly has to be fixed endian-wise to make
> acceleration work (the *ToCPU* stuff maybe?) and what is going on with
> xvideo.
>
> If there is any info I can give please let me know and I'd be happy to
> pass it on.
>
> Cheers!
>
> Michael
>


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [Xpert]Re: Using the "chips" XFree86 driver on a PowerBook 3400 (almost there.)

[Michael Stephen Hanni]

Howdy.

Well I've got it all to work save for the "WritePixmap" acceleration as it
makes use of some fun MoveDataFromCPU/MoveDWORDS magic. With WritePixmap
active there is a small box of color distortion around the mouse cursor. I'm
sure the MoveDataFromCPU function is assuming little endian data, but I
don't know exactly what I'm looking at.

Oh, I haven't really touched this but XVideo is still fscked. I'm getting
BadAllocs all over the place.

on 10-9-2002 12:49 PM, Ani Joshi at ajoshi@unixbox.com wrote:

> It seems you have changed the address space to use the big endian
> aperture, this is correct.  As for the accel engine, try turning off all
> acceleration (just #if them out in the driver, no need for several XaaNo*
> Options), and leave on SolidFills, this is the easiset to debug.  There is
> a problem with some 65550's and polling of the BR04 register which tends
> to lockup the engine, you may be running into that.  I would grab the old
> source from Xpmac (I believe it had c&t accel) and poke around there to
> see where the engine is initialized and the proper way to BLTWAIT on this
> chip, its been long since I hacked on the c&t65550 so I don't remember
> sorry.

I had to mess around with the aperture, but I think I've got the aperture
and the MMIOBase setup correctly now. By dropping a slightly modified
ct_BltHiQV.h from the last Xpmac tarball into the latest XFree86 source
everything works nicely.

> As for the hardware cursor, you will have to swap the image of the cursor.
> See the hw cursor support for radeon (or r128) and apply similair
> necessary the big endian bits.

Ah, got ya.

I'm going to run some more tests and then I'll upload a patch somewhere for
general consumption. Maybe, just maybe, there are other people that use this
chip on a big endian machine.

Cheers,

Michael


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/