Quoting Amit Kucheria (2013-01-25 03:18:05)
On Fri, Jan 25, 2013 at 9:08 AM, Sanjay Singh Rawat sanjay.rawat@linaro.org wrote:
On Thursday 24 January 2013 11:24 PM, Mike Turquette wrote:
Quoting Sanjay Singh Rawat (2013-01-24 08:01:33)
You're missing a changelog describing why this patch is required. :)
Signed-off-by: Sanjay Singh Rawat sanjay.rawat@linaro.com
Sanjay,
Have you seen the patch to dump the CCF tree from a single sysfs file, in JSON format?
http://git.linaro.org/gitweb?p=people/mturquette/linux.git%3Ba=commitdiff%3B...
This is much nicer than parsing a directory structure. It is faster and most importantly the data is atomic. A lock is held across the clock framework while the data is accessed so you don't have to worry about data changing in between reads like you do with the directory method.
Regards, Mike
Thanks Mike, yes i remember your point. This change is for addressing the bug of clock entries not getting populated on CCF enabled kernel.
Will check for clock_tree_dump file implementation separately.
I think what Mike is saying is that you could get rid of all the directory parsing code and just read this one file for all of your clock data.
I think leaving the legacy code in is fine. The current powerdebug code already checks to see if this is a CCF or legacy clk tree, so we could leave the legacy code in and change the CCF method to use the clk-dump sysfs file when CCF is detected.
But it wouldn't work for platforms that haven't converted over to CCF. Mike, all member platforms converted over now?
I'll have to check, but as stated above we don't need to remove the legacy clk code.
Regards, Mike