[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #14 from CVS Commits --- The releases/gcc-8 branch has been updated by Harald Anlauf : https://gcc.gnu.org/g:e0c7ff6237c78d99b40869afbdcdbe27fe98b698 commit r8-10521-ge0c7ff6237c78d99b40869afbdcdbe27fe98b698 Author: Tobias Burnus Date: Thu Sep 17 14:01:09 2020 +0200 Fortran: Avoid double-free with parse error (PR96041, PR93423) gcc/fortran/ PR fortran/96041 PR fortran/93423 * decl.c (gfc_match_submod_proc): Avoid later double-free in the error case. (cherry picked from commit c12facd22881517127ebbe213d7ecc7fc1fcea4e)
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #13 from CVS Commits --- The releases/gcc-9 branch has been updated by Harald Anlauf : https://gcc.gnu.org/g:0e442063a0adf01e6348e9fc55cc3e9869974769 commit r9-8921-g0e442063a0adf01e6348e9fc55cc3e9869974769 Author: Tobias Burnus Date: Thu Sep 17 14:01:09 2020 +0200 Fortran: Avoid double-free with parse error (PR96041, PR93423) gcc/fortran/ PR fortran/96041 PR fortran/93423 * decl.c (gfc_match_submod_proc): Avoid later double-free in the error case. (cherry picked from commit c12facd22881517127ebbe213d7ecc7fc1fcea4e)
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #12 from CVS Commits --- The releases/gcc-10 branch has been updated by Harald Anlauf : https://gcc.gnu.org/g:13421890f81844acb134a460eda7132db3e504ed commit r10-8775-g13421890f81844acb134a460eda7132db3e504ed Author: Tobias Burnus Date: Thu Sep 17 14:01:09 2020 +0200 Fortran: Avoid double-free with parse error (PR96041, PR93423) gcc/fortran/ PR fortran/96041 PR fortran/93423 * decl.c (gfc_match_submod_proc): Avoid later double-free in the error case. (cherry picked from commit c12facd22881517127ebbe213d7ecc7fc1fcea4e)
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 Tobias Burnus changed: What|Removed |Added Resolution|--- |FIXED Status|NEW |RESOLVED --- Comment #11 from Tobias Burnus --- FIXED on trunk (GCC 11), which is the only one affected. Note: The patch causing this issue is for PR93423. The latter was a 8/9/10/11 regression of type ICE-on-invalid. If the PR93423 patch gets cherry-picked for a branch, you may likely also see cherry-picks commits in PR.
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #10 from CVS Commits --- The master branch has been updated by Tobias Burnus : https://gcc.gnu.org/g:c12facd22881517127ebbe213d7ecc7fc1fcea4e commit r11-3259-gc12facd22881517127ebbe213d7ecc7fc1fcea4e Author: Tobias Burnus Date: Thu Sep 17 14:01:09 2020 +0200 Fortran: Avoid double-free with parse error (PR96041, PR93423) gcc/fortran/ PR fortran/96041 PR fortran/93423 * decl.c (gfc_match_submod_proc): Avoid later double-free in the error case.
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 Tobias Burnus changed: What|Removed |Added CC||burnus at gcc dot gnu.org --- Comment #9 from Tobias Burnus --- Patch: https://gcc.gnu.org/pipermail/gcc-patches/2020-September/553799.html
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041
--- Comment #8 from anlauf at gcc dot gnu.org ---
(In reply to Dominique d'Humieres from comment #7)
> f951`gfc_resolve_formal_arglist(proc=0x00014301fbb0) at resolve.c:313:18
> frame #2: 0x0001000eb283
Setting a breakpoint here and inspecting the arguments of
313 if (strcmp (proc->name, sym->name) == 0)
one can see that sym->name can be junk after the error message was emitted.
The place where sym is determined is a couple of lines above:
290 for (f = proc->formal; f; f = f->next)
291 {
292 gfc_array_spec *as;
293
294 sym = f->sym;
Either proc got corrupted or has not been set up properly.
On my x86_64 system I have:
(gdb) p *proc->formal->sym->name
$23 = 0 '\000'
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #7 from Dominique d'Humieres --- Reduced test (provided t.mod exists) submodule (t) ts contains module procedure bp(s) ! end procedure bp end submodule ts end pr93423_red.f90:5:19: 5 | module procedure bp(s) | 1 Error: MODULE PROCEDURE at (1) must be in a generic module interface f951: internal compiler error: Segmentation fault: 11 * thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0) * frame #0: 0x7fff6bd198b5 libsystem_platform.dylib`_platform_strcmp + 181 frame #1: 0x0001000c8856 f951`gfc_resolve_formal_arglist(proc=0x00014301fbb0) at resolve.c:313:18 frame #2: 0x0001000eb283 f951`::do_traverse_symtree(st=, st_func=0x, sym_func=(f951`::find_arglists(gfc_symbol *) at resolve.c:546:3))(gfc_symtree *), void (*)(gfc_symbol *)) at symbol.c:4170:18 frame #3: 0x0001000c9098 f951`::resolve_types(gfc_namespace *) [inlined] resolve_formal_arglists(ns=) at resolve.c:563:19 frame #4: 0x0001000c9089 f951`::resolve_types(gfc_namespace *) [inlined] resolve_contained_functions(ns=0x000143839200) frame #5: 0x0001000c9089 f951`::resolve_types(ns=0x000143839200) frame #6: 0x0001000bb9a3 f951`gfc_resolve(ns=0x000143839200) at resolve.c:17326:17 frame #7: 0x0001000ae701 f951`gfc_parse_file() at parse.c:6448:15 frame #8: 0x0001001032dc f951`::gfc_be_parse_file() at f95-lang.c:212:18 frame #9: 0x000100e8fddb f951`::compile_file() at toplev.c:458:25 frame #10: 0x0001015dcc7b f951`toplev::main(int, char**) at toplev.c:2307:24 frame #11: 0x0001015dc8e0 f951`toplev::main(this=0x7ffeefbff26e, argc=, argv=) frame #12: 0x0001015dfb61 f951`main(argc=2, argv=0x7ffeefbff2a8) at main.c:39:22 frame #13: 0x7fff6bb23cc9 libdyld.dylib`start + 1 frame #14: 0x7fff6bb23cc9 libdyld.dylib`start + 1
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #6 from anlauf at gcc dot gnu.org --- On x86_64 I never reach symbol.c:2662, since the code returns from line 2659. This the closest I get to your backtrace: Breakpoint 2, free_st_labels (label=0x0) at ../../gcc-trunk/gcc/fortran/symbol.c:2659 2659 if (label == NULL) (gdb) bt #0 free_st_labels (label=0x0) at ../../gcc-trunk/gcc/fortran/symbol.c:2659 #1 0x0073ad9a in gfc_free_namespace (ns=0x27c7c60) at ../../gcc-trunk/gcc/fortran/symbol.c:4050 #2 0x0073af45 in gfc_free_symbol (sym=0x27c7b00) at ../../gcc-trunk/gcc/fortran/symbol.c:3091 #3 0x0073b09f in free_sym_tree (sym_tree=0x275f280) at ../../gcc-trunk/gcc/fortran/symbol.c:3902 #4 0x0073b096 in free_sym_tree (sym_tree=0x27c5a00) at ../../gcc-trunk/gcc/fortran/symbol.c:3900 #5 0x0073b08d in free_sym_tree (sym_tree=0x275fb90) at ../../gcc-trunk/gcc/fortran/symbol.c:3899 #6 0x0073b08d in free_sym_tree (sym_tree=0x27c79a0) at ../../gcc-trunk/gcc/fortran/symbol.c:3899 #7 0x0073ad12 in gfc_free_namespace (ns=0x27c2670) at ../../gcc-trunk/gcc/fortran/symbol.c:4041 #8 0x0073b34b in gfc_symbol_done_2 () at ../../gcc-trunk/gcc/fortran/symbol.c:4101 #9 0x006d9379 in gfc_done_2 () at ../../gcc-trunk/gcc/fortran/misc.c:358 #10 0x006f2053 in clean_up_modules (gsym=0x275e880) at ../../gcc-trunk/gcc/fortran/parse.c:6267 #11 0x006f201d in clean_up_modules (gsym=0x2823580) at ../../gcc-trunk/gcc/fortran/parse.c:6259 #12 0x006fd73c in translate_all_program_units (gfc_global_ns_list=) at ../../gcc-trunk/gcc/fortran/parse.c:6330 #13 gfc_parse_file () at ../../gcc-trunk/gcc/fortran/parse.c:6546 #14 0x00749f00 in gfc_be_parse_file () at ../../gcc-trunk/gcc/fortran/f95-lang.c:212 #15 0x00d74e8f in compile_file () at ../../gcc-trunk/gcc/toplev.c:458 #16 0x006621c9 in do_compile () at ../../gcc-trunk/gcc/toplev.c:2307 #17 toplev::main (this=this@entry=0x7fffd4ae, argc=, argc@entry=2, argv=, argv@entry=0x7fffd5a8) at ../../gcc-trunk/gcc/toplev.c:2446 #18 0x0066602b in main (argc=2, argv=0x7fffd5a8) at ../../gcc-trunk/gcc/main.c:39 (gdb) p label $1 = (gfc_st_label *) 0x0
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #5 from Bill Seurer --- It hits that breakpoints many times and for the first many it was always 0. Then I just let it run until it ICEd and backed up the call stack a bit. Program received signal SIGSEGV, Segmentation fault. 0x106dcbd4 in free_st_labels (label=0x981) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 2662 free_st_labels (label->left); (gdb) where #0 0x106dcbd4 in free_st_labels (label=0x981) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 #1 0x106dcbe0 in free_st_labels (label=0x13492760) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 #2 0x106dcbe0 in free_st_labels (label=0x3fffb7e50a58 ) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 #3 0x106dff28 in gfc_free_namespace (ns=0x13495ae0) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:4050 #4 0x106dda60 in gfc_free_symbol (sym=0x13493e40) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:3084 #5 0x106ddc14 in gfc_release_symbol (sym=0x13493e40) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:3125 #6 0x106dfa10 in free_sym_tree (sym_tree=0x13493e00) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:3902 #7 0x106dfa00 in free_sym_tree (sym_tree=0x13493ac0) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:3900 #8 0x106df9f0 in free_sym_tree (sym_tree=0x13493fa0) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:3899 #9 0x106dfe90 in gfc_free_namespace (ns=0x1348f6b0) at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:4041 #10 0x106e0144 in gfc_symbol_done_2 () at /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:4101 #11 0x1062536c in gfc_done_2 () at /home/seurer/gcc/git/gcc-test/gcc/fortran/misc.c:358 #12 0x1066cec8 in clean_up_modules (gsym=0x13490330) at /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:6267 #13 0x1066d22c in translate_all_program_units (gfc_global_ns_list=0x13477bf0) at /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:6330 #14 0x1066da84 in gfc_parse_file () at /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:6546 #15 0x106f0488 in gfc_be_parse_file () at /home/seurer/gcc/git/gcc-test/gcc/fortran/f95-lang.c:212 #16 0x11248514 in compile_file () at /home/seurer/gcc/git/gcc-test/gcc/toplev.c:458 #17 0x1124d87c in do_compile () at /home/seurer/gcc/git/gcc-test/gcc/toplev.c:2307 #18 0x1124dda4 in toplev::main (this=0x3fffe910, argc=20, argv=0x3fffed78) at /home/seurer/gcc/git/gcc-test/gcc/toplev.c:2446 #19 0x124facf4 in main (argc=20, argv=0x3fffed78) at /home/seurer/gcc/git/gcc-test/gcc/main.c:39 up a few times... 4050 free_st_labels (ns->st_labels); (gdb) print ns->st_labels $8 = (gfc_st_label *) 0x3fffb7e50a58
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #4 from anlauf at gcc dot gnu.org --- (In reply to Dominique d'Humieres from comment #3) > From several runs > > frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) > [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 > frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) > [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 > > It seems that some trees are not built properly. I set breakpoints in gdb in these routines but see nothing special on x86_64. Maybe some stack/heap corruption issue affecting ref counts?
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #3 from Dominique d'Humieres --- >From several runs frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 frame #0: 0x0001000f0b27 f951`::free_sym_tree(gfc_symtree *) [inlined] free_sym_tree(sym_tree=0x00ce) at symbol.c:3899:17 frame #0: 0x0001000f11ed f951`gfc_free_namespace(gfc_namespace*) [inlined] free_uop_tree(uop_tree=0x00ce) at symbol.c:3881:17 It seems that some trees are not built properly.
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 Dominique d'Humieres changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2020-07-04 Ever confirmed|0 |1 --- Comment #2 from Dominique d'Humieres --- Confirmed on darwin.
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 --- Comment #1 from anlauf at gcc dot gnu.org --- (In reply to Bill Seurer from comment #0) > f951: internal compiler error: Segmentation fault > 0x10be671b crash_signal > /home/seurer/gcc/git/gcc-test/gcc/toplev.c:328 > 0x103a6484 free_st_labels > /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 > 0x103a648f free_st_labels > /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 > 0x103a648f free_st_labels > /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:2662 > 0x103afd63 gfc_free_namespace(gfc_namespace*) > /home/seurer/gcc/git/gcc-test/gcc/fortran/symbol.c:4044 Bill, can you set a breakpoint at symbol.c:4044 and print ns->st_labels? On x86_64 I always have a NULL pointer here, so that symbol.c:2662 is never reached.
[Bug fortran/96041] [11 regression] ICE in gfortran.dg/pr93423.f90 after r11-1792
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96041 Richard Biener changed: What|Removed |Added Target Milestone|--- |11.0
