> > I think you are also not removing the common_target and > > common_target_probability from cgraph_edge. > > There is now code in ipa-utils merging the histograms. You will need to > > update that to your representation. It should not be hard - it either > > copies all the values from target function or merges them with given > > probability scales. > > This seems a bit tricky for me. ipa-icf runs after ipa-profile, the reason > why > remove <common_target_id, common_target_probability> is they are never used in > later passes during previous discussion.
There is also merging done from lto-symtab. If you have COMDAT inline which is late inlined at -fprofile-generate time you will end up with multiple profiles for it and they needs to be merged. This is done before IPA passes including ipa-profile. > > I added a new variable target_prob to save it in direct edge when calling > make_speculative, since the target_prob is assigned in ipa-profile, so no > need to stream out and in through LTO like speculative_id. > > Even though, scaling in ipa-icf with multiple direct edges is more complicate > then > single direct edge, need double circulation to find each src to dst edge map > and do > scaling one by one. > > BTW, the breaking patch r279373 "Fix merging of common traget info." only > mentioned > firefox but doesn't include a testcase, I tried to construct some test > cases, but fail to cover all the circumstance (I just implemented the N:N map > yet, > please refer to the code in below patch.), do you have cases other than > firefox to > cover it? Or how to test it as r279373 required? As this piece of code is > not quite > related to this patch, can we just leave it for future refine? Thanks :) You need testcase with a comdat that is used & inlined in two units, contains indirect call and build with -fno-early-inlining. However we can handle this incrementally. > 2019-12-26 Xiong Hu Luo <luo...@linux.ibm.com> > > PR ipa/69678 > * cgraph.c (symbol_table::create_edge): Init speculative_id and > target_prob. > (cgraph_edge::make_speculative): Add param for setting speculative_id > and target_prob. > (cgraph_edge::speculative_call_info): Update comments and find reference > by speculative_id for multiple indirect targets. > (cgraph_edge::resolve_speculation): Decrease the speculations > for indirect edge, drop it's speculative if not direct target > left. Update comments. > (cgraph_edge::redirect_call_stmt_to_callee): Likewise. > (cgraph_node::dump): Print num_speculative_call_targets. > (cgraph_node::verify_node): Don't report error if speculative > edge not include statement. > (cgraph_edge::num_speculative_call_targets_p): New function. > * cgraph.h (int common_target_id): Remove. > (int common_target_probability): Remove. > (num_speculative_call_targets): New variable. > (make_speculative): Add param for setting speculative_id. > (cgraph_edge::num_speculative_call_targets_p): New declare. > (target_prob): New variable. > (speculative_id): New variable. > * ipa-fnsummary.c (analyze_function_body): Create and duplicate > call summaries for multiple speculative call targets. > * cgraphclones.c (cgraph_node::create_clone): Clone speculative_id. > * ipa-profile.c (struct speculative_call_target): New struct. > (class speculative_call_summary): New class. > (class speculative_call_summaries): New class. > (call_sums): New variable. > (ipa_profile_generate_summary): Generate indirect multiple targets > summaries. > (ipa_profile_write_edge_summary): New function. > (ipa_profile_write_summary): Stream out indirect multiple targets > summaries. > (ipa_profile_dump_all_summaries): New function. > (ipa_profile_read_edge_summary): New function. > (ipa_profile_read_summary_section): New function. > (ipa_profile_read_summary): Stream in indirect multiple targets > summaries. > (ipa_profile): Generate num_speculative_call_targets from > profile summaries. > * ipa-ref.h (speculative_id): New variable. > * ipa-utils.c (ipa_merge_profiles): Update with target_prob. > * lto-cgraph.c (lto_output_edge): Remove indirect common_target_id and > common_target_probability. Stream out speculative_id and > num_speculative_call_targets. > (input_edge): Likewise. > * predict.c (dump_prediction): Remove edges count assert to be > precise. > * symtab.c (symtab_node::create_reference): Init speculative_id. > (symtab_node::clone_references): Clone speculative_id. > (symtab_node::clone_referring): Clone speculative_id. > (symtab_node::clone_reference): Clone speculative_id. > (symtab_node::clear_stmts_in_references): Clear speculative_id. > * tree-inline.c (copy_bb): Duplicate all the speculative edges > if indirect call contains multiple speculative targets. > * value-prof.h (check_ic_target): Remove. > * value-prof.c (gimple_value_profile_transformations): > Use void function gimple_ic_transform. > * value-prof.c (gimple_ic_transform): Handle topn case. > Fix comment typos. Change it to a void function. > > gcc/testsuite/ChangeLog > > 2019-12-26 Xiong Hu Luo <luo...@linux.ibm.com> > > PR ipa/69678 > * gcc.dg/tree-prof/indir-call-prof-topn.c: New testcase. > * gcc.dg/tree-prof/crossmodule-indir-call-topn-1.c: New testcase. > * gcc.dg/tree-prof/crossmodule-indir-call-topn-1a.c: New testcase. > * gcc.dg/tree-prof/crossmodule-indir-call-topn-2.c: New testcase. > * lib/scandump.exp: Dump executable file name. > * lib/scanwpaipa.exp: New scan-pgo-wap-ipa-dump. OK. You will need to do the obvious updates for Martin's patch which turned some member functions into static functions. Honza
