Hi Stephane, On Fri June 2 2006 7:41 am, Stephane Eranian wrote: > > Why is there a get_num_counters() API, instead of simply a numerical > > field in the pfm_pmu_support_t structure like there is for pmc_count and > > pmd_count? Again, what's the distinction between PMCs, PMDs, and > > counters? > > Ok, I fixed that now.
Great! > > Another note about the arch-specific module APIs - there seems to be a > > lot of inconsistency in how parameters and return values are passed. For > > instance, > > You have to be careful here. Some routines are below the generic > layer, e.g., pfm_ita2_get_event_name(). Others are user interface that > are arch-specific, e.g., pfm_ita2_get_event_maxincr(). I was actually only referring to the routines defined in pfm_pmu_support_t (the "generic" interface as you refer to it below). > In general you > can tell by verifying whether the function is declared static or not. > Not the greatest test, I agree. I wanted to avoid re-using the same name > in each file to avoid confusion when debugging. But I think it would make > it easier to distinguish between arch-private interface and implementation > of the generic interface. We could say pfm_arch_get_event_name() for > something needed for the generic interface and keep the pfm_ita2_ prefix > pattern. I actually haven't looked through much of the ia64 code yet, so I wasn't aware it had its own "external" APIs. So are these exposed to the user of the library? Are there any restrictions or guidelines on what APIs can/should be exported? I can definitely think of some possibilities for "private" APIs for Pentium4 (such as event-tagging and event-filtering). Thanks! -- Kevin Corry [EMAIL PROTECTED] http://www.ibm.com/linux/ http://evms.sourceforge.net/ _______________________________________________ perfmon mailing list [email protected] http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/
