On Fri, Feb 18, 2011 at 09:30:55AM +0100, Sascha Hauer wrote:

On Fri, Feb 18, 2011 at 01:24:20PM +0800, Shawn Guo wrote:

quoted

Sorry, I did not catch up with v1 of the patch set.

I have a overall comment on the driver.  There is many occurrences
of mx23, mx28 etc. throughout the file.  IMHO, this is not a good
idea.  It may be better to use the IP version to handle the
differences.  In this way, if we have another SoC coming later
using the same LCDIF revision as i.MX28.  The driver could
immediately fit in without code change, ideally.

The only problem with version register is that the offset of LCDIF
version register is different on i.MX28 from i.MX23.

Can opener inside can. I love it ;)

Well, that's the situation we have to deal with.

quoted

You still need
cpu_is_mx23 to read the correct version.

BTW, I would try to use cpu_is_mx23 than cpu_is_mx28, as the 'else'
of cpu_is_mx23 could _possibly_ cover later SoC as well as i.MX28.

There is only one cpu_is_mx28 in the driver without else. I can switch
MXSFB_MX23 and MXSFB_MX28 to MXSFB_V3 and MXSFB_V4 if you like it
better.

Yes, I like it better.  With it, if we have a new SoC using LCDIF v4,
the driver needs no change. 

-- 
Regards,
Shawn