http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54699
Bug #: 54699
Summary: [4.8 Regression] [SH] gfortran.dg/class_array_9.f03
ICEs
Classification: Unclassified
Product: gcc
Version: 4.8.0
Status: UNCONFIRMED
Keywords: ice-on-valid-code
Severity: normal
Priority: P3
Component: target
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: kkoj...@gcc.gnu.org
CC: olege...@gcc.gnu.org
Target: sh*-*-*
gfortran.dg/class_array_9.f03 ICEs for a while on sh4-unknown-linux-gnu.
FAIL: gfortran.dg/class_array_9.f03 -O0 (internal compiler error)
gfortran.log says:
internal compiler error: in change_address_1, at emit-rtl.c:2006
Here is a gdb's backtrace for f951:
(gdb) bt
#0 internal_error (gmsgid=0x8abd243 "in %s, at %s:%d")
at ../../ORIG/trunk/gcc/diagnostic.c:951
#1 0x0897228f in fancy_abort (
file=0x89fedb8 "../../ORIG/trunk/gcc/emit-rtl.c", line=2006,
function=0x89feba1 "change_address_1")
at ../../ORIG/trunk/gcc/diagnostic.c:1011
#2 0x0832006b in change_address_1 (memref=0xb7db49f0, mode=SFmode,
addr=0xb7dcfbe8, validate=1) at ../../ORIG/trunk/gcc/emit-rtl.c:2006
#3 0x08323c9e in adjust_address_1 (memref=0xb7db49f0, mode=SFmode, offset=4,
validate=1, adjust_address=1, adjust_object=0)
at ../../ORIG/trunk/gcc/emit-rtl.c:2111
#4 0x0836ea01 in alter_subreg (xp=0xb7dcf850)
at ../../ORIG/trunk/gcc/final.c:3004
#5 0x0836ec50 in cleanup_subreg_operands (insn=0xb7dd0384)
at ../../ORIG/trunk/gcc/final.c:2951
#6 0x0852f941 in reload (first=0xb7dad680, global=0)
at ../../ORIG/trunk/gcc/reload1.c:1241
#7 0x08468e49 in do_reload () at ../../ORIG/trunk/gcc/ira.c:4328
#8 rest_of_handle_reload () at ../../ORIG/trunk/gcc/ira.c:4419
#9 0x084e2305 in execute_one_pass (pass=0x8ba03e0)
at ../../ORIG/trunk/gcc/passes.c:2199
#10 0x084e26a5 in execute_pass_list (pass=0x8ba03e0)
at ../../ORIG/trunk/gcc/passes.c:2254
---Type <return> to continue, or q <return> to quit---q
Quit
(gdb) fr 2
#2 0x0832006b in change_address_1 (memref=0xb7db49f0, mode=SFmode,
addr=0xb7dcfbe8, validate=1) at ../../ORIG/trunk/gcc/emit-rtl.c:2006
2006 gcc_assert (memory_address_addr_space_p (mode, addr, as));
(gdb) p addr
$1 = (rtx_def *) 0xb7dcfbe8
(gdb) call debug_rtx(addr)
(plus:SI (reg/f:SI 14 r14)
(const_int 4 [0x4]))
(gdb) fr 4
#4 0x0836ea01 in alter_subreg (xp=0xb7dcf850)
at ../../ORIG/trunk/gcc/final.c:3004
3004 *xp = adjust_address (y, GET_MODE (x), offset);
(gdb) p xp
$1 = (rtx *) 0xb7dcf850
(gdb) call debug_rtx(*xp)
(subreg:SF (mem/c:DI (reg/f:SI 14 r14) [0 %sfp+-200 S8 A32]) 4)
(gdb) fr 5
#5 0x0836ec50 in cleanup_subreg_operands (insn=0xb7dd0384)
at ../../ORIG/trunk/gcc/final.c:2951
2951 recog_data.operand[i] = alter_subreg (recog_data.operand_loc[i]);
(gdb) call debug_rtx(insn)
(insn 1124 661 1125 67 (parallel [
(set (subreg:SF (mem/c:DI (reg/f:SI 14 r14) [0 %sfp+-200 S8 A32])
4)
(reg:SF 0 r0))
(use (reg/v:PSI 151 ))
(clobber (scratch:SI))
]) ch9.f03:38 269 {movsf_ie}
(nil))
Looks that something wrong happens during reload and it requests
an illegal memory address r14+4 for SFmode on SH.
It started to fail between trunk revision 185498 and 185552.
Perhaps related to the changes for PR target/50751.
