Corey,
On Tue, Sep 16, 2008 at 2:24 AM, Corey Ashford <[EMAIL PROTECTED]> wrote:
> Hello,
>
> I'd like to open a discussion of allowing certain architecture-specific
> functions into libpfm.
>
This is certainly a useful discussion!
> Specifically, on the powerpc arch, we have this mechanism of pre-defined
> groups of events that are allowed to be traced in counters simultaneously.
>
> However, this information is currently buried inside of libpfm and
> there's no way for a caller of libpfm to extract this information using
> any method besides trial and error (by attempting to put events in the
> same event set, and seeing if dispatch generates an error).
>
> I realize that other architectures don't use this pre-defined group
> approach, so this would have to be an arch-specific function.
>
> So, first, I'd like to ask if there's any sort of strong objection to
> adding arch-specific functions to libpfm. Of course, we'd have to make
> a strong argument that the function would be useful to callers of libpfm.
>
Model or arch-specific functions already exist for Itanium processors and
even X86 processors. The rules that I have used are:
- must be useful to caller
- cannot be retrieved using the generic libpfm API
- function naming scheme is: pfm_XXX_func() where XXX is either the
model name
(or acronym) or arch name. For instance, on Itanium2, we have
pfm_ita2_is_ear().
- does return an error code
- must have prototype defined in pfmlib_XXX.h
So feel free to define whatever you think could be useful.
Thanks.
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
perfmon2-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel