Re: [PATCH 2/2 V2] MMC: cleanup endianity conversions and style problems

[Pierre Ossman <drzeus-list@drzeus.cx>]

[-- Attachment #1: Type: text/plain, Size: 1349 bytes --]

On Sun, 6 Jul 2008 02:54:56 +0300
"Winkler, Tomas" <tomas.winkler@intel.com> wrote:

> >
> >NAK in its current form. The data needs to be DMA:able, which means
> >that the stack cannot be used.
> >
> I see, anyway only the last hunk is problematic in this matter.
>  

All of them actually. The other registers are possible DMA victims when
on an SPI host (which is also the only code path your patch modifies).

> So we need to define raw_src as __be32 and translate it in
> mmc_decode_src 

I'd prefer if raw_scr is in native endian. The conversions should be
done at the entry/exit points IMO.

> I will get to the MMC spec only tomorrow but it sound strange that SDIO
> registers are in little endian and these are in big.

I'd say that it's SDIO that's the odd man. Protocols tend to use big
endian after all. A reason for the difference is probably that SDIO
should be regarded as a register interface, not as a bus protocol.
Different worlds, different norms...

Rgds
-- 
     -- Pierre Ossman

  Linux kernel, MMC maintainer        http://www.kernel.org
  rdesktop, core developer          http://www.rdesktop.org

  WARNING: This correspondence is being monitored by the
  Swedish government. Make sure your server uses encryption
  for SMTP traffic and consider using PGP for end-to-end
  encryption.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]