On Wed, 2007-04-25 at 15:24 -0700, David Miller wrote:

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: Thu, 26 Apr 2007 08:16:44 +1000

quoted

On Wed, 2007-04-25 at 18:46 +0100, Matt Sealey wrote:

quoted

Can't the ata/ATA thing be fixed by simply fixing device trees where it
happens? strncmp seems the standards-compliant route to take.. why clutter
the common parsing routines with fixes for deviant platforms?

In that specific case, it's even fixed by having the driver have both
spellings in it's match list.

This is something I'd like to avoid.

Initially I added sparc OFW property support to the Radeon
driver by checking for both the upper-case and lower-case
variants of all the property names.

It was beyond ugly.

Well, I think we are mixing several things in the same pot here :-)

 - property names. That's what you had to deal with for Sparc. I think
it's fair enough to use strcasecmp for these. They are supposed to not
have case, while they happen to do here or there, there is almost no
chance of conflict (several properties with a name differring only by
the case).

 - property content. (That include node names which are really only the
content of the "name" property of that node). The ATA/ata case was for
property content in fact (name property and type property iirc) which is
a different thing and must be kept case sensitive I beleive. That covers
the case of of_device_is_compatible() which should use strcmp (or the
-n- version of it of course).

Do we agree on that ?

Cheers,
Ben.