On Mon, Oct 28, 2013 at 2:13 PM, Justin Bogner <[email protected]> wrote: > Bob Wilson <[email protected]> writes: >> On Oct 28, 2013, at 7:23 AM, Diego Novillo <[email protected]> wrote: >>> Ping? Bob, Chandler, does this sound reasonable to you? I'm thinking >>> of changing the current patch to use -fprofile-sample-use=... >>> >>> This partially keeps the symmetry with -fprofile-generate / >>> -fprofile-use and avoids the autodetection logic that Chandler >>> dislikes. >> >> I don't have a strong opinion on this. Justin, how does this line up >> with what you were considering for our profiling options? > > Using -fprofile-sample-use= seems simplest for now. We'll probably need > to figure out exactly what to do with -fprofile-use= and > -fprofile-generate= in the near future, as the current state is actually > fairly strange: > > - profile-use and profile-use= are ignored > - profile-generate= isn't accepted > - profile-generate causes libprofile_rt to be linked in, but does > nothing else > > In GCC, -fprofile-generate implies -fprofile-arcs, which we currently > use to turn on the GCOVProfilerPass, and the =path versions of these > options take a directory, whereas both Diego's and Bob's profiling > approaches deal with a file.
Thanks. I've renamed the flag and moved the patch over to phabricator: http://llvm-reviews.chandlerc.com/D2066 Diego. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
