Skip to contentSkip to conversation

amd8111e endian & barrier fixes

PATCH Andi Kleen 2 participants 2 messages Oct 8, 2004 – Oct 9, 2004 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Andi Kleen

author
#1 lore raw
Quoted 5 lines · Benjamin Herrenschmidt

Hi Jeff !

This patch against the amd8111e makes it work on some about-to-be-released piece of PPC hardware. It does:

Quoted 7 lines

- Fix endian
- Search for the PHY on MII instead of hard coding the ID
- Add a couple of wmb's where needed on descriptor updates

I must appologize for having re-indented one of the rx functions, but I just couldn't read/understand it without doing so, it was going back leftward in the middle of a { } block ...

It's basically impossible to review the patch properly because of that change. Can you please separate the arbitary white space change into a different patch?

Also I would suggest you send the patch to the driver maintainers for review first (cc'ed)

Quoted 1 line

From a quick look the change to clear the ring rx flags completely

instead of clearing the bit looks bogus. Why did you not just add a endian conversion there?

I can test it when it's properly reviewd.

-Andi

#2 lore raw
Quoted 3 lines · Andi Kleen

It's basically impossible to review the patch properly because of that change. Can you please separate the arbitary white space change into a different patch?

Hrm... I think you are doing too much here ;) The change concerns only a single function and is still quite readable :) But anyways, I'll eventually do that next week.

Quoted 2 lines

Also I would suggest you send the patch to the driver maintainers for review first (cc'ed)

Well, I would have done if any maintainer email was in the source code, which isn't the case, but thanks for providing them :)

Quoted 3 lines
Quoted 1 line

From a quick look the change to clear the ring rx flags completely

instead of clearing the bit looks bogus. Why did you not just add a endian conversion there?

Because the bits mask was ... 0 :)

Quoted 1 line

I can test it when it's properly reviewd.

Sure. No hurry.

Ben