https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69947
Bug ID: 69947
Summary: [6 Regression] DW_OP_GNU_implicit_pointer broken on
the trunk
Product: gcc
Version: 6.0
Status: UNCONFIRMED
Keywords: wrong-debug
Severity: normal
Priority: P3
Component: debug
Assignee: jakub at gcc dot gnu.org
Reporter: jakub at gcc dot gnu.org
Target Milestone: ---
/* PR debug/ */
/* { dg-do run } */
/* { dg-options "-g" } */
#include "../nop.h"
__attribute__((noinline, noclone)) void
foo (void)
{
static const char a[] = "abcdefg";
const char *b = a;
asm (NOP : : : "memory"); /* { dg-final { gdb-test 12 "b\[4\]" "'e'" } } */
}
int
main ()
{
foo ();
return 0;
}
in form of guality testcase got broken by r231762 , before that change we were
emitting something like:
.byte 0xf2 # DW_OP_GNU_implicit_pointer
.long .Ldebug_info0+196
.sleb128 0
but after that we emit:
.byte 0xf2 # DW_OP_GNU_implicit_pointer
.long .Ldebug_info0+0
.sleb128 0
which is of course invalid, no DIE lives at offset 0 (and e.g. dwz complains
about this and gives up on optimizing it). In gcc.log dump from current trunk
one can see gdb is unhappy about that too:
Breakpoint 1, foo () at /usr/src/gcc/gcc/testsuite/gcc.dg/guality/prNNNNN.c:12
12 asm (NOP : : : "memory"); /* { dg-final { gdb-test 12 "b\[4\]"
"'e'" } } */
prNNNNN.gdb:3: Error in sourced command file:
Dwarf Error: Cannot find DIE at 0x0 referenced in module
/usr/src/gcc/obj/gcc/testsuite/gcc/prNNNNN.exe
UNSUPPORTED: gcc.dg/guality/prNNNNN.c -O1 line 12 b[4] == 'e'
