Kevin, Carl,

I have applied both patches as they make perfect sense. Thanks for fixing
that sread() bug.


On Wed, May 02, 2007 at 04:20:01PM -0500, Kevin Corry wrote:
> 
> Here are two quick fixes we found for the pfm_pmu_register() routine during 
> our testing.
> 
> 1/2
> In pfm_pmu_register(), the check for virtual read-only PMDs is not very
> intuitive. Since we've already checked if the PMD is virtual, just add a
> check under that if statement to see if the PMD is also read-only.
> 
> Obviously this is just a coding style issue, but Carl and I both got tripped
> up a couple times trying to understand when the pmd_sread and pmd_swrite
> routines are required. I think this helps make that a bit clearer. And it 
> also 
> eliminates a few instructions (not that this code path is performance 
> sensitive).
> 
> 
> 2/2
> The pfm_read_pmd() routine in include/linux/perfmon.h always attempts to
> call pmd_sread if the register is marked as virtual. Therefore, in
> pfm_pmu_register(), the PMU is required to have a pmd_sread function if
> there are any virtual PMDs, not just if there are read-only virtual PMDs.
> 
> Thanks,
> -- 
> Kevin Corry
> [EMAIL PROTECTED]
> http://www.ibm.com/linux/
> _______________________________________________
> perfmon mailing list
> [email protected]
> http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/

-- 

-Stephane
_______________________________________________
perfmon mailing list
[email protected]
http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/

Reply via email to