[Bug testsuite/29161] New: [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0

[Bug testsuite/29161] New: [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29161

            Bug ID: 29161
           Summary: [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp:
                    print operand0
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: testsuite
          Assignee: unassigned at sourceware dot org
          Reporter: vries at gcc dot gnu.org
  Target Milestone: ---

On openSUSE Tumbleweed, with gcc-12, I run into:
...
FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0
FAIL: gdb.opt/clobbered-registers-O2.exp: print operand1
...
while these pass when using gcc-11 instead.

Also fail with gcc-12 when forcing dwarf-4 as in gcc-11.

In more detail, with gcc-11:
...
print operand0^M
$1 = (unsigned int *) 0x7fffffffcd50^M
(gdb) print *operand0^M
$2 = 13^M
(gdb) PASS: gdb.opt/clobbered-registers-O2.exp: print operand0
print operand1^M
$3 = (unsigned int *) 0x7fffffffcd54^M
(gdb) print *operand1^M
$4 = 14^M
(gdb) PASS: gdb.opt/clobbered-registers-O2.exp: print operand1
...
and with gcc-12:
...
print operand0^M
$1 = (unsigned int *) 0x7fffffffcfb0^M
(gdb) print *operand0^M
$2 = 4198725^M
(gdb) FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0
print operand1^M
$3 = (unsigned int *) 0x7fffffffcfb4^M
(gdb) print *operand1^M
$4 = 0^M
(gdb) FAIL: gdb.opt/clobbered-registers-O2.exp: print operand1
...

A significant difference between the two execs is that the assignments to x and
y are optimized away: gcc-11:
...
0000000000401020 <main>:
  401020:       48 83 ec 10             sub    $0x10,%rsp
  401024:       48 8d 74 24 0c          lea    0xc(%rsp),%rsi
  401029:       48 8d 7c 24 08          lea    0x8(%rsp),%rdi
  40102e:       c7 44 24 08 0d 00 00    movl   $0xd,0x8(%rsp)
  401035:       00 
  401036:       c7 44 24 0c 0e 00 00    movl   $0xe,0xc(%rsp)
  40103d:       00 
  40103e:       e8 1d 01 00 00          call   401160 <gen_movsd>
  401043:       48 83 c4 10             add    $0x10,%rsp
  401047:       c3                      ret    
...
vs gcc-12:
...
0000000000401020 <main>:
  401020:       48 8d 74 24 fc          lea    -0x4(%rsp),%rsi
  401025:       48 8d 7c 24 f8          lea    -0x8(%rsp),%rdi
  40102a:       e9 11 01 00 00          jmp    401140 <gen_movsd>
...

Another significant difference is the application of sibcall optimization, but
switching that off doesn't fix the fail.

This fixes the FAIL for me:
...
diff --git a/gdb/testsuite/gdb.opt/clobbered-registers-O2.c
b/gdb/testsuite/gdb.opt/cl
obbered-registers-O2.c
index 7776024eb90..83cf2267d1e 100644
--- a/gdb/testsuite/gdb.opt/clobbered-registers-O2.c
+++ b/gdb/testsuite/gdb.opt/clobbered-registers-O2.c
@@ -33,7 +33,7 @@ gen_movsd (unsigned * operand0, unsigned * operand1)

 int main(void)
 {
-  unsigned x, y;
+  volatile unsigned x, y;

   x = 13;
   y = 14;
...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/29161] [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29161

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |luis.machado at arm dot com

--- Comment #1 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #0)
>  int main(void)
>  {
> -  unsigned x, y;
> +  volatile unsigned x, y;
>  
>    x = 13;
>    y = 14;
> ...


AFAICT, although the test-case is supposed to test optimized code, and the fix
disables an optimization, the focus is on debug info in functions other than
main, so disabling this specific optimization should be ok.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/29161] [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29161

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #1)
> (In reply to Tom de Vries from comment #0)
> >  int main(void)
> >  {
> > -  unsigned x, y;
> > +  volatile unsigned x, y;
> >  
> >    x = 13;
> >    y = 14;
> > ...
> 
> 
> AFAICT, although the test-case is supposed to test optimized code, and the
> fix disables an optimization, the focus is on debug info in functions other
> than main, so disabling this specific optimization should be ok.

Submitted: https://sourceware.org/pipermail/gdb-patches/2022-May/189364.html

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/29161] [gcc-12] FAIL: gdb.opt/clobbered-registers-O2.exp: print operand0

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=29161

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |13.1
         Resolution|---                         |FIXED

--- Comment #3 from Tom de Vries <vries at gcc dot gnu.org> ---
https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=a0ae328a266d2b01ff640be9277e01592b08971a

-- 
You are receiving this mail because:
You are on the CC list for the bug.