https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63432
--- Comment #5 from Teresa Johnson <tejohnson at google dot com> --- On Fri, Oct 3, 2014 at 10:21 AM, hjl.tools at gmail dot com <gcc-bugzi...@gcc.gnu.org> wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63432 > > --- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> --- > r215830 introduced: > > /* Scale up the frequency by REG_BR_PROB_BASE, to avoid rounding > errors applying the edge probability when the frequencies are very > small. */ > epath->src->count = epath->src->frequency * REG_BR_PROB_BASE; > FOR_EACH_EDGE (esucc, ei, epath->src->succs) > esucc->count = apply_probability (esucc->src->count, > esucc->probability); > > Can it guarantee that esucc->src->count has been scaled up by > REG_BR_PROB_BASE from esucc->src->frequency? I believe so. Since esucc are successor edges of epath->src, esucc->src == epath->src, and we have just set epath->src->count = epath->src->frequency * REG_BR_PROB_BASE. BTW, I am having a little trouble doing an LTO profiledbootstrap on my linux/x86_64 machine. Here's what I did: $ /usr/local/google/home/tejohnson/gcc_trunk_7/configure --prefix=/usr/local/google/home/tejohnson/gcc_trunk_7_validate_lto/bld-gcc/install --with-build-config=bootstrap-lto $ make -j8 profiledbootstrap I first got an error: --- /usr/local/google/home/tejohnson/gcc_trunk_7/gcc/gengtype.c: In function ‘type* create_user_defined_type(const char*, fileloc*)’: /usr/local/google/home/tejohnson/gcc_trunk_7/gcc/gengtype.c:583:29: error: ‘next’ may be used uninitialized in this function [-Werror=maybe-uninitialized] str += strspn (str, delim); ^ /usr/local/google/home/tejohnson/gcc_trunk_7/gcc/gengtype.c:636:13: note: ‘next’ was declared here char *next; ^ cc1plus: all warnings being treated as errors make[3]: *** [gengtype.o] Error 1 --- which I made a simple change in my gcc client to address. After restarting the "make -j8 profiledbootstrap" I got a bunch of other errors about undefined references in ltrans.o files. In case the first error left things in a bad state I have wiped everything out and restarted the process. Will let you know once I have reproduced. Thanks, Teresa > > -- > You are receiving this mail because: > You are on the CC list for the bug.
