https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90928
Thomas Schwinge <tschwinge at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed|2019-10-14 00:00:00 |2020-5-7
--- Comment #3 from Thomas Schwinge <tschwinge at gcc dot gnu.org> ---
Already since r231651 (if my old notes are to be believed...), we've been
seeing this issue for 'g++.dg/eh/alias1.C' (before: 'sorry, unimplemented:
target cannot support nonlocal goto'), and still do now:
FAIL: g++.dg/eh/alias1.C -std=gnu++98 (internal compiler error)
[etc.]
during RTL pass: vregs
[...]/gcc/testsuite/g++.dg/eh/alias1.C: In function 'void g(int)':
[...]/gcc/testsuite/g++.dg/eh/alias1.C:20:1: internal compiler error: in
instantiate_virtual_regs_in_insn, at function.c:1739
0xb5214b instantiate_virtual_regs_in_insn
[...]/gcc/function.c:1739
0xb5214b instantiate_virtual_regs
[...]/gcc/function.c:1977
0xb5214b execute
[...]/gcc/function.c:2026
(Ever since they got added?) we'd also seen this issue for
'g++.dg/ext/pr84724-1.C', 'g++.dg/ext/pr84724-2.C', but the ICE then
disappeared in (r265343, r265570].
Since (r259629 ,r261792] we also saw this issue for 'gcc.dg/graphite/pr82321.c'
(before: 'PASS'), then the ICE disappeared in (r271346, r272643], and now in
cff5a231480526aa73094a22fe8b2f39b63735a4..c416c52bcdb120db5e8c53a51bd78c4360daf79b
it has re-appeared:
[-PASS:-]{+FAIL: gcc.dg/graphite/pr82321.c (internal compiler error)+}
{+FAIL:+} gcc.dg/graphite/pr82321.c (test for excess errors)
during RTL pass: vregs
[...]/gcc/testsuite/gcc.dg/graphite/pr82321.c: In function 'dm':
[...]/gcc/testsuite/gcc.dg/graphite/pr82321.c:36:1: internal compiler
error: in instantiate_virtual_regs_in_insn, at function.c:1739
0x8ffedb instantiate_virtual_regs_in_insn
[...]/gcc/function.c:1739
0x8ffedb instantiate_virtual_regs
[...]/gcc/function.c:1977
0x8ffedb execute
[...]/gcc/function.c:2026
These are all the instances of this ICE, per my records.
