Kevin,

Please see comments below.

On Tue, 2008-04-01 at 16:45 -0700, Kevin Hilman wrote:
> In addition to Dirk's comments, I might add that patches against the
> master branch of the DaVinci git tree are preferred.
With the newly created git-tree and such, I think Steven can fix the
patches according to comments from both Dirk and you, except for this
one. 

It is an unfortunate fact that Neuros must stick a certain kernel
version (2.6.23 ATM) for well-known reasons, I am afraid Steven will not
have the opportunity to go further to make sure the patch apply to
master branch cleanly.

I take your 'might' above as an ok to this. ;-)

> Also, I understand this is a forward port of a driver from 2.6.10, so
> most of the errors/warnings come from the original TI driver, and not
> from your work.
> 
> However, for me to merge this code, at a minimum all the checkpatch
> errors/warnings should be fixed, and there are a handful of other
> whitespace issues to be cleaned up for readability as well.
Yes, they should all be fixed.

> That being said, I did compare the original driver from the TI/MV
> 2.6.10 kernel to your forward port, and you seem to have made lots of
> changes that are not scrictly forward port changes.  When doing this,
> I prefer to see separate patches for these kinds of changes.  In other
> words, I'd prefer to see one patch which is just a forward port,
> followed by cleanup patches with detailed descriptions.
I believe this was the case already with the patches Steven provided,
first there was a strict forward porting, then a patch following up to
remove the devfs, I'll let Steven speak for himself in the up-comping
new patch series.

> In addition to this, you've done several changes to the debug prints,
> as well as added braces around single statement while/if blocks, which
> is not needed, and is frowned upon by upstream maintainers.
It was me who brutally removed some of the debug printfs due to
compiling error if keeping them, I didn't know how to fix it and I still
don't. :-(

Steven, can you help bring them back and post the error messages here to
ask for help if you can't fix them yourself?

> Also, you explicitly add a NULL assignment to static variables, which
> is redundant as these end up in the .bss section which is explicitly
> zeroed. (specifically: static struct class_simple *rsz_class = NULL;)
I thought Steven has fixed this, but Steven can speak better for
himself.

Thanks,

-- 
Michael Gao
Neuros Technology-Open Consumer Electronics
http://developer.neurostechnology.com

_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source

Reply via email to