[m-rev.] for post-commit review: Coverage Profiling completed.
pbone at csse.unimelb.edu.au
Sat Sep 27 18:00:32 AEST 2008
On Sat, Sep 27, 2008 at 11:01:29AM +1000, Paul Bone wrote:
> On Fri, Sep 26, 2008 at 05:06:19PM +1000, Zoltan Somogyi wrote:
> > I did make two significant changes. First, I fixed a bug in the code that
> > created the foreign_proc invocation of increment_coverage_pointer_counter:
> > it used to set the instmap delta of the foreign_proc to say that it bound
> > both its arguments, even though they are both input. Second, I added a
> > new implementor-only option, --coverage-profiling-use-calls, that tells
> > deep_profiling.m to generate a call to increment_coverage_pointer_counter
> > instead of a foreign_proc. This should be useful if assertions are turned
> > on.
> Thanks Zoltan,
> I'm not sure what the benefit of --coverage-profiling-use-calls is.
> What assertions are you referring to that might prefer the increment
> coverage point code to be called rather than inlined? Are these the
> runtime assertions within the instrumentation?
Ah, Now that I've read the whole diff I understand as a comment in the
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 189 bytes
Desc: Digital signature
More information about the reviews