Re: [Linux-fbdev-devel] [PATCH 1/2] fb: add support for foreign endianness

[Clemens Koller <clemens.koller@anagramm.de>]

Benjamin Herrenschmidt schrieb:
> On Tue, 2008-02-19 at 00:35 +0100, Clemens Koller wrote:
>> Valdis.Kletnieks@vt.edu schrieb:
>>> On Mon, 18 Feb 2008 08:18:47 +0100, Krzysztof Helt said:
>>>> I know two fb drivers which use endianess information (pm2fb and s3c2410fb).
>>>> Both resolve endianess at driver level. Actually, both handle it by setting special
>>>> bits so the graphics chip itself reorder bytes to transform foreign endianess. 
>>>> I understand that this patch is for chips which cannot reorder bytes by themselves.
>>> Does anybody know of such a chip that's actually available in the wild?  Or are
>>> we writing drivers for speculative possible chips?
>>>
>> I had troubles with the Silicon Motion SM501/SM502 endianess on PowerPC PCI vs. LocalBus.
>> The chip also has a register to swap endianess, but that seems to only affect some
>> LocalBus modes.
>> The current fb and X drivers are working, but when it comes to font
>> aliasing and hw-acceleration, the problems start to rise again...
> 
> Most "sane" gfx chips nowadays provide configurable surfaces that allow
> to perform the swap when writing/reading from regions of the
> framebuffer, with the ability to set a different swapper setting (based
> on bit depth) per region.

Most! But not the SM50x. I still hope I would be wrong here. :-(

> Then there is also the risk that your PCI<->Localbus has been wired
> improperly :-)

That's not an issue in my case. The SM50x can be connected to
either an PCI or some Local/CPU-whateverbus IF.
I.e. on the MPC85xx PowerPC, PCI and LocalBus are separate bussses.
If the sm501 is attached to the MPC85xx' PCI like any other video card,
the PCI config-space is can be accessed as usual, whereas the framebuffer
memory area is byte-swapped compared to other common video cards.

So, to get back on topic:
I would welcome endianess swapping in SW. Some architectures (PowerPC)
should also be able to do swapped-endian mmapping. I just haven't
had time for a closer look but it looks also interesting way to do it
that way.

Regards,

Clemens