On Tue, Dec 9, 2014 at 4:52 PM, Martin Liška <mli...@suse.cz> wrote: > On 11/21/2014 01:23 PM, Richard Biener wrote: >> >> On Fri, Nov 21, 2014 at 12:52 PM, Martin Liška <mli...@suse.cz> wrote: >>> >>> On 11/20/2014 05:41 PM, Richard Biener wrote: >>>> >>>> >>>> On Thu, Nov 20, 2014 at 5:30 PM, Martin Liška <mli...@suse.cz> wrote: >>>>> >>>>> >>>>> Hello. >>>>> >>>>> Following patch fixes ICE in IPA ICF. Problem was that number of >>>>> non-debug >>>>> statements in a BB can >>>>> change (for instance by IPA split), so that the number is recomputed. >>>> >>>> >>>> >>>> Huh, so can it get different for both candidates? I think the stmt >>>> compare >>>> loop should be terminated on gsi_end_p of either iterator and return >>>> false for any remaining non-debug-stmts on the other. >>>> >>>> Thus, not walk all stmts twice here. >>> >>> >>> >>> Hello. >>> >>> Sorry for the previous patch, you are right it can be fixed in purer way. >>> Please take a look at attached patch. >>> >>>> >>>> As IPA split is run early I don't see how it should affect a real IPA >>>> pass though? >>> >>> >>> >>> >>> >>> Sorry for non precise information, the problematic BB is changed here: >>> #0 gsi_split_seq_before (i=0x7fffffffd550, pnew_seq=0x7fffffffd528) at >>> ../../gcc/gimple-iterator.c:429 >>> #1 0x0000000000b95a2a in gimple_split_block (bb=0x7ffff6c41548, >>> stmt=0x0) >>> at ../../gcc/tree-cfg.c:5707 >>> #2 0x00000000007563cf in split_block (bb=0x7ffff6c41548, i=i@entry=0x0) >>> at >>> ../../gcc/cfghooks.c:508 >>> #3 0x0000000000756b44 in split_block_after_labels (bb=<optimized out>) >>> at >>> ../../gcc/cfghooks.c:549 >>> #4 make_forwarder_block (bb=<optimized out>, >>> redirect_edge_p=redirect_edge_p@entry=0x75d4e0 >>> <mfb_keep_just(edge_def*)>, >>> new_bb_cbk=new_bb_cbk@entry=0x0) at ../../gcc/cfghooks.c:842 >>> #5 0x000000000076085a in create_preheader (loop=0x7ffff6d56948, >>> flags=<optimized out>) at ../../gcc/cfgloopmanip.c:1563 >>> #6 0x0000000000760aea in create_preheaders (flags=1) at >>> ../../gcc/cfgloopmanip.c:1613 >>> #7 0x00000000009bc6b0 in apply_loop_flags (flags=15) at >>> ../../gcc/loop-init.c:75 >>> #8 0x00000000009bc7d3 in loop_optimizer_init (flags=15) at >>> ../../gcc/loop-init.c:136 >>> #9 0x0000000000957914 in estimate_function_body_sizes >>> (node=0x7ffff6c47620, >>> early=false) at ../../gcc/ipa-inline-analysis.c:2480 >>> #10 0x000000000095948b in compute_inline_parameters (node=0x7ffff6c47620, >>> early=false) at ../../gcc/ipa-inline-analysis.c:2907 >>> #11 0x000000000095bd88 in inline_analyze_function (node=0x7ffff6c47620) >>> at >>> ../../gcc/ipa-inline-analysis.c:3994 >>> #12 0x000000000095bed3 in inline_generate_summary () at >>> ../../gcc/ipa-inline-analysis.c:4045 >>> #13 0x0000000000a70b71 in execute_ipa_summary_passes (ipa_pass=0x1dcb9e0) >>> at >> >> >> So inline_summary is generated after IPA-ICF does its job? >> >> But the bug is obviously that an IPA analysis phase does a code transform >> (here initializes loops without AVOID_CFG_MANIPULATIONS). >> Honza - if that is really needed then I think we should make sure >> loops are initialized at the start of the IPA analysis phase, not randomly >> inbetween. >> >> Thanks, >> Richard. > > > Hello. > > Even thought the root of problem is hidden somewhere in loop creation, I > would > like to apply the patch which makes iteration of non-debug gimple statement > more clearly? > > What do you think Richard?
Works for me - but we have to address the underlying issue. Richard. > Thanks, > Martin > > >> >>> ../../gcc/passes.c:2137 >>> #14 0x0000000000777a15 in ipa_passes () at ../../gcc/cgraphunit.c:2074 >>> #15 symbol_table::compile (this=this@entry=0x7ffff6c3a000) at >>> ../../gcc/cgraphunit.c:2187 >>> #16 0x0000000000778bcd in symbol_table::finalize_compilation_unit >>> (this=0x7ffff6c3a000) at ../../gcc/cgraphunit.c:2340 >>> #17 0x00000000006580ee in c_write_global_declarations () at >>> ../../gcc/c/c-decl.c:10777 >>> #18 0x0000000000b5bb8b in compile_file () at ../../gcc/toplev.c:584 >>> #19 0x0000000000b5def1 in do_compile () at ../../gcc/toplev.c:2041 >>> #20 0x0000000000b5e0fa in toplev::main (this=0x7fffffffdc9f, argc=20, >>> argv=0x7fffffffdd98) at ../../gcc/toplev.c:2138 >>> #21 0x000000000063f1d9 in main (argc=20, argv=0x7fffffffdd98) at >>> ../../gcc/main.c:38 >>> >>> >>> Patch can bootstrap on x86_64-linux-pc and no regression has been seen. >>> Ready for trunk? >>> >>> >>> Thanks, >>> Martin >>> >>>> >>>> Thanks, >>>> Richard. >>>> >>>>> Patch can bootstrap on x86_64-linux-pc and no regression has been seen. >>>>> Ready for trunk? >>>>> >>>>> Thanks, >>>>> Martin >>> >>> >>> >
