[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 Nathan Sidwell changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #11 from Nathan Sidwell --- Fixed r248687.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #10 from Nathan Sidwell --- Author: nathan Date: Tue May 30 14:43:45 2017 New Revision: 248687 URL: https://gcc.gnu.org/viewcvs?rev=248687=gcc=rev Log: PR c++/80913 * name-lookup.c (add_decl_to_level): Assert not making a circular chain. (update_binding): Don't prematurely slide artificial decl. * g++.dg/lookup/pr80913.C: New. Added: trunk/gcc/testsuite/g++.dg/lookup/pr80913.C Modified: trunk/gcc/cp/ChangeLog trunk/gcc/cp/name-lookup.c trunk/gcc/testsuite/ChangeLog
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #9 from ro at CeBiTec dot Uni-Bielefeld.DE --- > --- Comment #7 from Nathan Sidwell --- > It doesn't appear to be the stathack patch at fault here. I still get the > infinite loop with my reduced testcase when I revert it. (Which is good, > because I can't see how that patch could cause this behaviour.) This is weird since just reverting the stathack patch was enough to allow the bootstraps to succeed for me. Rainer
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 David Edelsohn changed: What|Removed |Added Target|i386-pc-solaris2.12,|i386-pc-solaris2.12, |sparc-sun-solaris2.12 |sparc-sun-solaris2.12, ||powerpc-ibm-aix* CC||dje at gcc dot gnu.org --- Comment #8 from David Edelsohn --- Probably also a cause of AIX bootstrap hang.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #7 from Nathan Sidwell --- It doesn't appear to be the stathack patch at fault here. I still get the infinite loop with my reduced testcase when I revert it. (Which is good, because I can't see how that patch could cause this behaviour.) I think it was my do_pushdecl change that didn't include another change I had thought was unrelated.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913
--- Comment #6 from Nathan Sidwell ---
I have a small reproducer, regardless of bitsize:
extern int meminfo ();
struct meminfo {};
void frob ()
{
meminf ();
}
stat hack when the struct decl is last. It looks like a patch I have on the
modules branch is not as orthogonal to this as I thought.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #5 from ro at CeBiTec dot Uni-Bielefeld.DE --- > --- Comment #4 from Nathan Sidwell --- > yes, must be something with a 32-bit target. I can reproduce it on x86_64 > linux host targeting either i386-linux or i386-solaris. > > We have a TYPE_DECL whose TREE_CHAIN points to itself. So when the spelling > corrector iterates over the decls in a namespace it gets stuck at that point. Good, so no point in me trying the 64-bit route. > It's end of day here, so don't expect more progress until tomorrow. No worries: same here, and I can always revert the patch locally until the bug is resolved. Thanks, Rainer
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #4 from Nathan Sidwell --- yes, must be something with a 32-bit target. I can reproduce it on x86_64 linux host targeting either i386-linux or i386-solaris. We have a TYPE_DECL whose TREE_CHAIN points to itself. So when the spelling corrector iterates over the decls in a namespace it gets stuck at that point. It's end of day here, so don't expect more progress until tomorrow.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #3 from ro at CeBiTec dot Uni-Bielefeld.DE --- > --- Comment #1 from Nathan Sidwell --- [...] > But x86_64-linux is not being affected in the way you describe. The backtrace > is something borking in the spelling correcting stuff. Maybe it's a 64-bit vs. 32-bit thing then? I could try amd64-pc-solaris2.12 and sparcv9-sun-solaris2.12 bootstraps with your patch included once the current ones have finished. Rainer
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #2 from Nathan Sidwell --- oh, you said that in the title :)
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 --- Comment #1 from Nathan Sidwell --- My money's on: Stat hack representation * name-lookup.c (STAT_HACK_P, STAT_TYPE, STAT_DECL, MAYBE_STAT_DECL, MAYBE_STAT_TYPE): New. (stat_hack): New. (find_namespace_binding): Replace with ... (find_namespace_slot): ... this. (find_namespace_value): New. (name_lookup::search_namespace_only, name_lookup::adl_namespace_only): Adjust. (update_binding): Add SLOT parameter, adjust. (check_local_shadow): Use find_namespace_value. (set_local_extern_decl_linkage): Likewise. (do_pushdecl): Adjust for namespace slot. (push_local_binding): Assert not a namespace binding. (check_for_out_of_scope_variable): Use find_namespace_value. (set_identifier_type_value_with_scope): Likewise. (get_namespace_binding): Likewise. (set_namespace_binding): Delete. (set_global_binding): Directly update the binding. (finish_namespace_using_decl): Likewise. (lookup_type_scope_1): Use find_namespace_slot and update. (do_push_nested_namespace): Use find_namespace_value. But x86_64-linux is not being affected in the way you describe. The backtrace is something borking in the spelling correcting stuff.
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 Nathan Sidwell changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2017-05-29 Assignee|unassigned at gcc dot gnu.org |nathan at gcc dot gnu.org Ever confirmed|0 |1
[Bug c++/80913] [8 regression] Infinite loop in cc1plus with stat hack patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80913 Rainer Orth changed: What|Removed |Added Target Milestone|--- |8.0
