https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82045
--- Comment #12 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org>
---
(In reply to r...@cebitec.uni-bielefeld.de from comment #11)
> Unfortunately, the patch breaks x86 bootstrap (e.g. for the 32-bit
> _multc3.o), both in i386-pc-solaris2.* and x86_64-pc-linux-gnu compilers:
>
> $ cc1 -fpreprocessed libgcc2.i -quiet -o libgcc2.s
>
> Thread 2 hit Breakpoint 1, fancy_abort (
> file=0x98dbb80 "/vol/gcc/src/hg/trunk/local/gcc/calls.c", line=4571,
> function=0x98dc691 <emit_library_call_value(int, rtx_def*, rtx_def*,
> libcall_type, machine_mode, int, std::pair<rtx_def*,
> machine_mode>*)::__FUNCTION__> "emit_library_call_value") at
> /vol/gcc/src/hg/trunk/local/gcc/diagnostic.c:1488
> 1488 internal_error ("in %s, at %s:%d", function, trim_filename (file),
> line);
>
> #0 fancy_abort (file=0x98dbb80 "/vol/gcc/src/hg/trunk/local/gcc/calls.c",
> line=4571,
> function=0x98dc691 <emit_library_call_value(int, rtx_def*, rtx_def*,
> libcall_type, machine_mode, int, std::pair<rtx_def*,
> machine_mode>*)::__FUNCTION__> "emit_library_call_value") at
> /vol/gcc/src/hg/trunk/local/gcc/diagnostic.c:1488
> #1 0x085fa6de in emit_library_call_value (retval=1, orgfun=0xfe7894a0,
> value=0x0, fn_type=LCT_CONST, outmode=E_TFmode, nargs=3, args=0x8046970)
> at /vol/gcc/src/hg/trunk/local/gcc/calls.c:4570
> #2 0x085e8409 in emit_library_call_value (fun=0xfe7894a0, value=0x0,
> fn_type=LCT_CONST, outmode=E_TFmode, arg1=0xfe789460,
> arg1_mode=E_TFmode,
> arg2=0xfe789480, arg2_mode=E_TFmode)
> at /vol/gcc/src/hg/trunk/local/gcc/rtl.h:3873
> #3 0x089ed51e in expand_binop (mode=E_TFmode, binoptab=smul_optab,
> op0=0xfe789460, op1=0xfe789480, target=0xfe789440, unsignedp=0,
> methods=OPTAB_LIB_WIDEN) at /vol/gcc/src/hg/trunk/local/gcc/optabs.c:1763
> #4 0x08738653 in expand_mult (mode=E_TFmode, op0=0xfe789460,
> op1=0xfe789480,
> target=0xfe789440, unsignedp=0)
> at /vol/gcc/src/hg/trunk/local/gcc/expmed.c:3455
> #5 0x0875ddb3 in expand_expr_real_2 (ops=0x8047190, target=0xfe789440,
> tmode=E_TFmode, modifier=EXPAND_NORMAL)
> at /vol/gcc/src/hg/trunk/local/gcc/expr.c:8827
> #6 0x08611992 in expand_gimple_stmt_1 (stmt=0xfe610104)
> at /vol/gcc/src/hg/trunk/local/gcc/cfgexpand.c:3691
> #7 0x08611bbc in expand_gimple_stmt (stmt=0xfe610104)
> at /vol/gcc/src/hg/trunk/local/gcc/cfgexpand.c:3751
> #8 0x0861898c in expand_gimple_basic_block (bb=0xfe784080,
> disable_tail_calls=false)
> at /vol/gcc/src/hg/trunk/local/gcc/cfgexpand.c:5750
> #9 0x0861a160 in (anonymous namespace)::pass_expand::execute
> (this=0xa31ea30,
> fun=0xfe758068) at /vol/gcc/src/hg/trunk/local/gcc/cfgexpand.c:6357
> #10 0x08a21b8e in execute_one_pass (pass=0xa31ea30)
> at /vol/gcc/src/hg/trunk/local/gcc/passes.c:2495
> #11 0x08a21ec9 in execute_pass_list_1 (pass=0xa31ea30)
> at /vol/gcc/src/hg/trunk/local/gcc/passes.c:2584
> #12 0x08a21f43 in execute_pass_list (fn=0xfe758068, pass=0xa31c300)
> at /vol/gcc/src/hg/trunk/local/gcc/passes.c:2595
> #13 0x0865bdec in cgraph_node::expand (this=0xfe777000)
> at /vol/gcc/src/hg/trunk/local/gcc/cgraphunit.c:2054
> #14 0x0865c7c6 in output_in_order ()
> at /vol/gcc/src/hg/trunk/local/gcc/cgraphunit.c:2296
> #15 0x0865cf5a in symbol_table::compile (this=0xfe6070c8)
> at /vol/gcc/src/hg/trunk/local/gcc/cgraphunit.c:2537
> #16 0x0865d1b2 in symbol_table::finalize_compilation_unit (this=0xfe6070c8)
> at /vol/gcc/src/hg/trunk/local/gcc/cgraphunit.c:2631
> #17 0x08b327b5 in compile_file ()
> at /vol/gcc/src/hg/trunk/local/gcc/toplev.c:496
> #18 0x08b34b12 in do_compile ()
> at /vol/gcc/src/hg/trunk/local/gcc/toplev.c:2037
> #19 0x08b34dd8 in toplev::main (this=0x80475fe, argc=6, argv=0x804762c)
> at /vol/gcc/src/hg/trunk/local/gcc/toplev.c:2172
> #20 0x0963505a in main (argc=6, argv=0x804762c)
> at /vol/gcc/src/hg/trunk/local/gcc/main.c:39
>
> (gdb) p mode
> $12 = E_BLKmode
> (gdb) p val
> $13 = (rtx) 0x0
Yeah, I've been testing a fixed version today. I hope to post soon
once I've got a bit more coverage.
