[PATCH] Permedia2 fb on PPC

[PATCH] Permedia2 fb on PPC

[Marc ZYNGIER]

[I'm forwarding this from linux-kernel, per Tom Rini request. Could
anyone with a previously working Permedia2 test this patch and tell me
if this breaks something. You can also scream if it works ;-)]

Hi all,

The following patchlet allows the Permedia2 framebuffer to work
properly on power-pc (instead of crashing the machine...).

This has been tested with a PCI Formac Promedia3D (aka formacGA8) on a
PowerMac 7500 (with a G3 card, if that matters...).

Patch is against 2.4.16, but should apply to any recent 2.[45] kernel.

        M.

--- drivers/video/pm2fb.c.orig	Fri Sep 14 22:44:49 2001
+++ drivers/video/pm2fb.c	Tue Dec  4 20:44:18 2001
@@ -48,7 +48,7 @@
 #error	"The endianness of the target host has not been defined."
 #endif

-#if defined(__BIG_ENDIAN) && !defined(__sparc__)
+#if defined(__BIG_ENDIAN) && !defined(__sparc__) && (!defined(CONFIG_PPC) || defined(CONFIG_FB_PM2_CVPPC))
 #define PM2FB_BE_APERTURE
 #endif


--
Places change, faces change. Life is so very strange.

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

Re: [PATCH] Permedia2 fb on PPC

[Michel Dänzer]

On Wed, 2001-12-05 at 11:29, Marc ZYNGIER wrote:

> The following patchlet allows the Permedia2 framebuffer to work
> properly on power-pc (instead of crashing the machine...).

Why is this needed? And does it work with the XFree86 glint driver with
Option "UseFBDev"?


--
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast

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

Re: [PATCH] Permedia2 fb on PPC

[Marc ZYNGIER]

>>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:

Michel> Why is this needed?

Because it simply crashes the system without it ;-).

This prevent the driver from using a non-existant (at least on the
board I own) big endian aperture for registers.

Michel> And does it work with the XFree86 glint driver with Option
Michel> "UseFBDev"?

Don't know yet. _Now_ that I have a console (thanks to this patch),
I'll have a look at the X11 thingy.

Regards,

        M.
--
Places change, faces change. Life is so very strange.

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

Re: [PATCH] Permedia2 fb on PPC

[Michel Dänzer]

On Wed, 2001-12-05 at 23:52, Marc ZYNGIER wrote:
> >>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:
>
> Michel> Why is this needed?
>
> Because it simply crashes the system without it ;-).
>
> This prevent the driver from using a non-existant (at least on the
> board I own) big endian aperture for registers.

Oh my...


> Michel> And does it work with the XFree86 glint driver with Option
> Michel> "UseFBDev"?
>
> Don't know yet. _Now_ that I have a console (thanks to this patch),
> I'll have a look at the X11 thingy.

Well at least when I last worked on the glint driver, it relied on the
register aperture being big endian.


--
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast

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

Re: [PATCH] Permedia2 fb on PPC

[Marc ZYNGIER]

>>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:

Michel> And does it work with the XFree86 glint driver with Option
Michel> "UseFBDev"?
>>
>> Don't know yet. _Now_ that I have a console (thanks to this patch),
>> I'll have a look at the X11 thingy.

Michel> Well at least when I last worked on the glint driver, it
Michel> relied on the register aperture being big endian.

Indeed, glint driver doesn't work with UseFBDev.

Could this be a chipset issue (3DLabs vs TI) ? Mine is a Texas
Instrument (aka ti_pm2) :

00:0d.0 Display controller: Texas Instruments TVP4020 [Permedia 2] (rev 01)
        Subsystem: Texas Instruments TVP4020 [Permedia 2]
        Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
        Status: Cap- 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
        Latency: 32 (48000ns min, 48000ns max)
        Interrupt: pin A routed to IRQ 23
        Region 0: Memory at 80840000 (32-bit, non-prefetchable) [size=128K]
        Region 1: Memory at 81000000 (32-bit, non-prefetchable) [size=8M]
        Region 2: Memory at 90000000 (32-bit, non-prefetchable) [size=8M]
        Expansion ROM at 80800000 [disabled] [size=64K]

Any idea ?

        M.
--
Places change, faces change. Life is so very strange.

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

Re: [PATCH] Permedia2 fb on PPC

[Michel Dänzer]

On Sat, 2001-12-08 at 23:41, Marc ZYNGIER wrote:
> >>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:
>
> Michel> And does it work with the XFree86 glint driver with Option
> Michel> "UseFBDev"?
> >>
> >> Don't know yet. _Now_ that I have a console (thanks to this patch),
> >> I'll have a look at the X11 thingy.
>
> Michel> Well at least when I last worked on the glint driver, it
> Michel> relied on the register aperture being big endian.
>
> Indeed, glint driver doesn't work with UseFBDev.
>
> Could this be a chipset issue (3DLabs vs TI) ? Mine is a Texas
> Instrument (aka ti_pm2) :

No, IIRC that's exactly the one it assumes you have if you're running
pm2fb.

As I said, the driver assumes the register aperture is big endian with
Option "UseFBDev". And without, it will also try to use the big endian
aperture.


If there's really no way to use the big endian aperture on some boards,
the only solution I see is to change both pm2fb and the glint driver to
always use the little endian one.


--
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast

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

Re: [PATCH] Permedia2 fb on PPC

[Marc ZYNGIER]

>>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:

Michel> As I said, the driver assumes the register aperture is big
Michel> endian with Option "UseFBDev". And without, it will also try
Michel> to use the big endian aperture.

Well, with UseFBDev set to false, XFree works ok (although switching
between consoles seems to be broken...).

Michel> If there's really no way to use the big endian aperture on
Michel> some boards, the only solution I see is to change both pm2fb
Michel> and the glint driver to always use the little endian one.

Indeed. What are the sparc people doing about that ? They are already
using the little indian aperture in pm2gb.

        M.
--
Places change, faces change. Life is so very strange.

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

Re: [PATCH] Permedia2 fb on PPC

[Michel Dänzer]

On Sun, 2001-12-09 at 00:08, Marc ZYNGIER wrote:
>
> >>>>> "Michel" == Michel Dänzer <michdaen@iiic.ethz.ch> writes:
>
> Michel> As I said, the driver assumes the register aperture is big
> Michel> endian with Option "UseFBDev". And without, it will also try
> Michel> to use the big endian aperture.
>
> Well, with UseFBDev set to false, XFree works ok (although switching
> between consoles seems to be broken...).
>
> Michel> If there's really no way to use the big endian aperture on
> Michel> some boards, the only solution I see is to change both pm2fb
> Michel> and the glint driver to always use the little endian one.
>
> Indeed. What are the sparc people doing about that ? They are already
> using the little indian aperture in pm2gb.

I have only been talking about what glint does on PPC, maybe on sparc
it's already using the little endian aperture.


--
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast

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