Problem with radeon new ROM probe code

Problem with radeon new ROM probe code

[Benjamin Herrenschmidt]

Hi Jon !

Your new code for looking at the card's ROM is causing a problem for
me. On some cards, typically the "Mac" cards, the ROM doesn't contain
an x86 BIOS, but rather an OF driver. The current code will blindly
tap that, though it contains none of the expected tables, and will
use completely wrong values.

Do you have ideas how to make some sanity checking on the ROM content ?

Ben.




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problem with radeon new ROM probe code

[Michel Dänzer]

On Wed, 2003-09-24 at 14:06, Benjamin Herrenschmidt wrote:
> 
> Your new code for looking at the card's ROM is causing a problem for
> me. On some cards, typically the "Mac" cards, the ROM doesn't contain
> an x86 BIOS, but rather an OF driver. The current code will blindly
> tap that, though it contains none of the expected tables, and will
> use completely wrong values.
> 
> Do you have ideas how to make some sanity checking on the ROM content ?

The X radeon driver has this in RADEONGetBIOSParameters():

if (info->VBIOS[0] != 0x55 || info->VBIOS[1] != 0xaa) {
	xfree(info->VBIOS);
	info->FPBIOSstart = 0;
	info->VBIOS = NULL;
	info->BIOSAddr = 0x00000000;
	xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
		   "Video BIOS not found!\n");
    }


-- 
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \     http://svcs.affero.net/rm.php?r=daenzer



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problem with radeon new ROM probe code

[Benjamin Herrenschmidt]

On Wed, 2003-09-24 at 17:18, Michel Dänzer wrote:
> On Wed, 2003-09-24 at 14:06, Benjamin Herrenschmidt wrote:
> > 
> > Your new code for looking at the card's ROM is causing a problem for
> > me. On some cards, typically the "Mac" cards, the ROM doesn't contain
> > an x86 BIOS, but rather an OF driver. The current code will blindly
> > tap that, though it contains none of the expected tables, and will
> > use completely wrong values.
> > 
> > Do you have ideas how to make some sanity checking on the ROM content ?
> 
> The X radeon driver has this in RADEONGetBIOSParameters():
> 
> if (info->VBIOS[0] != 0x55 || info->VBIOS[1] != 0xaa) {

We check that, but that isn't enough, an OF ROM will trigger that
as well

Ben.




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problem with radeon new ROM probe code

[Sven Luther]

On Wed, Sep 24, 2003 at 11:12:57AM -0700, Jon Smirl wrote:
> Is the 'Mac' card being used on a x86 machine?
> I thought the ROM code was compiled out on non-X86 machines.

Notice that not all powerpc boxes are in this case. The pegasos at least
includes an x86 emulator in the OF which runs the cards bios. I know
that the amigaone does the same. So testing for powerpc may well not be
enough for detecting OF cards.

Friendly,

Sven Luther


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problem with radeon new ROM probe code

[Stefan Reinauer]

* Sven Luther <sven.luther@wanadoo.fr> [030925 07:39]:
> On Wed, Sep 24, 2003 at 11:12:57AM -0700, Jon Smirl wrote:
> > Is the 'Mac' card being used on a x86 machine?
> > I thought the ROM code was compiled out on non-X86 machines.
> 
> Notice that not all powerpc boxes are in this case. The pegasos at least
> includes an x86 emulator in the OF which runs the cards bios. I know
> that the amigaone does the same. So testing for powerpc may well not be
> enough for detecting OF cards.

Whether it's an Open Firmware card or not can be read from the PCI
headers stored in the ROM. Check the romheaders utility from OpenBIOS
CVS ( http://www.openbios.org/ )

Look at the code here:
http://cvs.openbios.org/cgi-dom/viewcvs.cgi/utils/romheaders.c?rev=1.2&content-type=text/vnd.viewcvs-markup

The program prints whether the rom is OpenFirmware, x86 or HP PA-Risc.
Note that graphics cards can theoretically contain ROM images for both
x86 and OF in one chip.

  Stefan
-- 
Architecture Team
    SuSE Linux AG


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problem with radeon new ROM probe code

[Benjamin Herrenschmidt]

> Whether it's an Open Firmware card or not can be read from the PCI
> headers stored in the ROM. Check the romheaders utility from OpenBIOS
> CVS ( http://www.openbios.org/ )
> 
> Look at the code here:
> http://cvs.openbios.org/cgi-dom/viewcvs.cgi/utils/romheaders.c?rev=1.2&content-type=text/vnd.viewcvs-markup
> 
> The program prints whether the rom is OpenFirmware, x86 or HP PA-Risc.
> Note that graphics cards can theoretically contain ROM images for both
> x86 and OF in one chip.

Yup, "Mac" cards can contain in theory both OF and x86 BIOS drivers, though
I yet have to find a vendor doing that....

Thanks for the pointers to code, I'll dig that out.

Regards,
Ben.




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf