[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556 Patrick Palka changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #9 from Patrick Palka --- Fixed for GCC 13.1, thanks for the bug report. FWIW a potential workaround for this bug, which seems to work for me at least, is to remove the Magnitude constraint from to_prefix_base and to_base_scaled_unit.
[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556 --- Comment #8 from CVS Commits --- The releases/gcc-13 branch has been updated by Patrick Palka : https://gcc.gnu.org/g:90361bc6f4edb444e86380b6d1e08475fa74 commit r13-7227-g90361bc6f4edb444e86380b6d1e08475fa74 Author: Patrick Palka Date: Wed Apr 19 13:07:46 2023 -0400 c++: bad ggc_free in try_class_unification [PR109556] Aside from correcting how try_class_unification copies multi-dimensional 'targs', r13-377-g3e948d645bc908 also made it ggc_free this copy as an optimization. But this is wrong since the call to unify within might've captured the args in persistent memory such as the satisfaction cache (as part of constrained auto deduction). PR c++/109556 gcc/cp/ChangeLog: * pt.cc (try_class_unification): Don't ggc_free the copy of 'targs'. gcc/testsuite/ChangeLog: * g++.dg/cpp2a/concepts-placeholder13.C: New test. (cherry picked from commit 5e284ebbc3082c5a8974d24e3a0977aa48f3cc60)
[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556 --- Comment #7 from CVS Commits --- The master branch has been updated by Patrick Palka : https://gcc.gnu.org/g:5e284ebbc3082c5a8974d24e3a0977aa48f3cc60 commit r14-91-g5e284ebbc3082c5a8974d24e3a0977aa48f3cc60 Author: Patrick Palka Date: Wed Apr 19 13:07:46 2023 -0400 c++: bad ggc_free in try_class_unification [PR109556] Aside from correcting how try_class_unification copies multi-dimensional 'targs', r13-377-g3e948d645bc908 also made it ggc_free this copy as an optimization. But this is wrong since the call to unify within might've captured the args in persistent memory such as the satisfaction cache (as part of constrained auto deduction). PR c++/109556 gcc/cp/ChangeLog: * pt.cc (try_class_unification): Don't ggc_free the copy of 'targs'. gcc/testsuite/ChangeLog: * g++.dg/cpp2a/concepts-placeholder13.C: New test.
[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556 --- Comment #6 from Jakub Jelinek --- Indeed, the reduction got stuck at around 1.4M size and didn't reduce further until I've killed it.
[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556
Patrick Palka changed:
What|Removed |Added
Keywords|needs-bisection,|
|needs-reduction |
--- Comment #5 from Patrick Palka ---
template
concept C = (N != 0);
template
struct A { };
template auto M>
void f(A);
int main() {
f(A<1, 42>{});
f(A<2, 42>{});
}
Bisection seems unreliable probably due to the use-after-free nature of the
bug, but removing those calls to ggc_free fixes the issue so
r13-377-g3e948d645bc908 is almost certainly the culprit.
[Bug c++/109556] [13/14 Regression] internal compiler error: in iterative_hash_template_arg, at cp/pt.cc:1927
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109556 Andrew Pinski changed: What|Removed |Added Last reconfirmed||2023-04-19 Status|UNCONFIRMED |ASSIGNED Target Milestone|--- |13.0 Keywords||GC, ice-on-valid-code Ever confirmed|0 |1 Summary|internal compiler error: in |[13/14 Regression] internal |iterative_hash_template_arg |compiler error: in |, at cp/pt.cc:1927 |iterative_hash_template_arg ||, at cp/pt.cc:1927
