[mpiwg-tools] "read, write, and reset" of pvar

[Martin Schulz]

Hi Junchao,

On Sep 6, 2013, at 3:37 PM, Junchao Zhang <jczhang at mcs.anl.gov> wrote:

> P581~582 of the Standard says
> 
> "Advice to implementors: Sampling-based tools rely on the ability to call the MPI tool information interface, in particular routines to start, stop, read, write and reset performance variables, from any program context, including asynchronous contexts such as signal handlers. MPI implementations should strive, if possible in their particular environment, to enable these usage scenarios for all or a subset of the routines mentioned above. If implementing only a subset, the read, write, and reset routines are typically the most critical for sampling based tools."
> 
> Does the standard really want to say "read, write, reset" are critical or should it be "start, stop, read, reset"?
> 

The signal-safetly "requirement" was part of quite a debate, since this is hard for some platforms. We therefore wanted to be as specific as possible and only ask for a minimal set of functions. The idea was that a tool can always start/stop a variable outside of a signal handler (start of the tool or start of a code segment) and only needs to read the variable during an interrupt. Sure, having all routines signal-safe would be ideal, but that was not acceptable to the forum.

Martin


> 
> -- Junchao Zhang
> 
> _______________________________________________
> mpiwg-tools mailing list
> mpiwg-tools at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-tools

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-tools/attachments/20130907/b74c9e54/attachment-0001.html>