See patch.
After my last patch from http://www.coreboot.org/pipermail/coreboot/2009-October/053987.html which makes ram_check() at least partially configurable at build-time via menuconfig didn't get much love, here's a different one.
It drops all ram_check() calls from all boards for now.
However, I still do believe that it is very useful to have an easy way in menuconfig to enable such a check. This allows us to tell a user "Please enable the RAM check in menuconfig and send a boot log" without having to make/send patches or explain what to add where etc.
I'm fine with making it only visible if CONFIG_EXPERT is set, but IMHO the option should be there. My last implementation did admittedly suck though, but this can be implemented in a much nicer way.
Uwe.
The reason I like leaving that ram_check code in there is as a reminder to new people that, if they are having trouble, there is a good place to put a ram check.
It's a form of documentation, with some typical examples.
So I am afraid I don't see the need for the patch -- what problem does it solve?
ron
ron minnich wrote:
It's a form of documentation, with some typical examples.
I think it would be better if it was somewhere else.
So I am afraid I don't see the need for the patch -- what problem does it solve?
I think it reduces clutter AKA noise in the source code.
In general I'm inclined to Ack, but my only reservation is that some of the calls to ram_check() might point to particularly strategic addresses for some boards, which were only known to the person who did the port. If so, it would be nice to keep that information somewhere..
//Peter
On Thu, Jan 28, 2010 at 9:21 AM, Peter Stuge peter@stuge.se wrote:
In general I'm inclined to Ack, but my only reservation is that some of the calls to ram_check() might point to particularly strategic addresses for some boards, which were only known to the person who did the port. If so, it would be nice to keep that information somewhere..
Which is my point. Where else would you put it? And, if it changes for some reason, you've now got to fix it in a document which people may not catch up with. That's a problem-generator.
ron
On Thu, 28 Jan 2010 10:13:40 -0800, ron minnich rminnich@gmail.com wrote:
On Thu, Jan 28, 2010 at 9:21 AM, Peter Stuge peter@stuge.se wrote:
In general I'm inclined to Ack, but my only reservation is that some of the calls to ram_check() might point to particularly strategic addresses for some boards, which were only known to the person who did the port. If so, it would be nice to keep that information somewhere..
Which is my point. Where else would you put it? And, if it changes for some reason, you've now got to fix it in a document which people may not catch up with. That's a problem-generator.
Sorry Uwe, I am going to have to agree with Ron here. The ram_check() only really needs to be used when developing or troubleshooting raminit. For a regular build it is not needed, and in my opinion just complicates kconfig.
-
Joseph Smith -
Peter Stuge -
ron minnich -
Uwe Hermann