> This patch ensures that the histograms from the profile summary are streamed
> through the LTO files so that the working set can be computed for use in
> downstream optimizations.
>
> Bootstrapped and tested on x86_64-unknown-linux-gnu. Ok for trunk?
>
> Thanks,
> Teresa
>
> 2012-11-28 Teresa Johnson <tejohn...@google.com>
>
> * lto-cgraph.c (output_profile_summary): Stream out sum_all
> and histogram.
> (input_profile_summary): Stream in sum_all and histogram.
> (merge_profile_summaries): Merge sum_all and histogram.
> (input_symtab): Call compute_working_sets after merging
> summaries.
> * gcov-io.c (gcov_histo_index): Make extern for compiler.
> * gcov-io.h (gcov_histo_index): Ditto.
> * profile.c (compute_working_sets): Remove static keyword.
> * profile.h (compute_working_sets): Ditto.
OK.
>
> Index: lto-cgraph.c
> ===================================================================
> --- lto-cgraph.c (revision 193909)
> +++ lto-cgraph.c (working copy)
> @@ -46,6 +46,7 @@ along with GCC; see the file COPYING3. If not see
> #include "tree-streamer.h"
> #include "gcov-io.h"
> #include "tree-pass.h"
> +#include "profile.h"
Please update dependencies in Makefile.in
> + /* Count number of non-zero histogram entries, and fill in a bit vector
> + of non-zero indices. */
> + counters. */
> + for (bv_ix = 0; bv_ix < GCOV_HISTOGRAM_BITVECTOR_SIZE; bv_ix++)
> + histo_bitvector[bv_ix] = 0;
> + for (h_ix = 0; h_ix < GCOV_HISTOGRAM_SIZE; h_ix++)
> + {
> + if (profile_info->histogram[h_ix].num_counters > 0)
> + {
> + histo_bitvector[h_ix / 32] |= 1 << (h_ix % 32);
> + h_cnt++;
> + }
I think this would be more readable if you just produced a bitpack instead of
doing it
by hand, like into gcov-io.
> + lto_gcov_summary.sum_all = MAX (lto_gcov_summary.sum_all,
> + (file_data->profile_info.sum_all
> + * scale
> + + REG_BR_PROB_BASE / 2)
> + / REG_BR_PROB_BASE);
Use RDIV for the scaling.
> -#if IN_LIBGCOV || !IN_GCOV
> +#if !IN_GCOV
> /* Determine the index into histogram for VALUE. */
>
> -static unsigned
> +GCOV_LINKAGE unsigned
I would probably go around the trouble of declaring this static in GCOV,
so it is inlined at we do not bload libgcov more than needed.
Thanks,
Honza
