On Sat, Nov 14, 2009 at 8:08 PM, Corey Ashford <cjash...@linux.vnet.ibm.com> wrote: > On 11/12/2009 03:30 PM, stephane eranian wrote: >> >> On Fri, Nov 13, 2009 at 12:09 AM, Corey Ashford >> <cjash...@linux.vnet.ibm.com> wrote: >>> >>> Hi Stephane, >>> >>> I'm currently doing a "first cut" at porting Power to libpfm4 and I >>> noticed >>> one thing that's a little confusing about this new API. >>> >>> The two terms "umask" and "attr" are used interchangeably, but without >>> really mentioning that anywhere. >>> >> attr is the generic term, it encapsulates unit masks and register filters >> such >> as priv level, invert and so on. >> >>> I think the API should stick to using one word, either umask or attr. >>> Personally, I prefer "attr" because it's a bit more general purpose. >>> >> Agreed. >> >>> Is there some fine difference that I'm not perceiving between the two in >>> libpfm4? >>> >>> Otherwise, so far it looks very clean and easy to work with. >>> >> Thanks. Yes, I spent a lot of time cleaning the code as much as I could. >> I think the user-visible API is much simpler now. > > One other question in this area: as I was doing the port of libpfm3.x to > libpfm4 for Power, I noticed that there's another thing called "modifiers", > which appears to be a bitmask sort of thing. I didn't need it for Power, > but I'm wondering how these differ from a umask or attribute... > attribute = umask + modifiers
modifers = filters based on the register (not the event) ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ perfmon2-devel mailing list perfmon2-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
