[Bug record/29721] [gdb, record, aarch64] FAIL: gdb.reverse/solib-precsave.exp: reverse-next third shr1

["vries at gcc dot gnu.org" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=29721

--- Comment #16 from Tom de Vries <vries at gcc dot gnu.org> ---
First sign of trouble:
...
    [infrun] bpstat_check_breakpoint_conditions: incorrect frame
{stack=0xfffffffff320,code=0x0000fffff7f9d594,!special} not
{stack=0xfffffffff330,code=0x0000fffff7f9d594,!special}, not stopping
...

Using additional_flags=-fasynchronous-unwind-tables works around this.

On x86_64, using additional_flags=-fno-asynchronous-unwind-tables to drop the
cfi info, the problem doesn't trigger.

With x86_64, we have:
...
000000000000050a <shr1>:
 50a:   55                      push   %rbp
 50b:   48 89 e5                mov    %rsp,%rbp
 50e:   48 89 7d f8             mov    %rdi,-0x8(%rbp)
 512:   90                      nop
 513:   5d                      pop    %rbp
 514:   c3                      ret
...
and at the first insn:
...
1: /x $pc = 0x7ffff7ff150a
2: /x $sp = 0x7fffffffdad8
3: /x $rbp = 0x7fffffffdaf0
(gdb) info frame
Stack level 0, frame at 0x7fffffffdae0:
 rip = 0x7ffff7ff150a in shr1; saved rip = 0x4005df
 called by frame at 0x7fffffffdb00
 Arglist at 0x7fffffffdad0, args: 
 Locals at 0x7fffffffdad0, Previous frame's sp is 0x7fffffffdae0
 Saved registers:
  rip at 0x7fffffffdad8
...
and at the last insn:
...
1: /x $pc = 0x7ffff7ff1514
2: /x $sp = 0x7fffffffdad8
3: /x $rbp = 0x7fffffffdaf0
(gdb) info frame
Stack level 0, frame at 0x7fffffffdae0:
 rip = 0x7ffff7ff1514 in shr1; saved rip = 0x4005df
 called by frame at 0x7fffffffdb00
 Arglist at 0x7fffffffdaf0, args: 
 Locals at 0x7fffffffdaf0, Previous frame's sp is 0x7fffffffdae0
 Saved registers:
  rip at 0x7fffffffdad8
...
the same frame address (as well as the same $sp and $fp).

On aarch64 (with additional_flags=-fno-omit-frame-pointer
additional_flags=-mno-omit-leaf-frame-pointer added, to have more comparable
code), we have:
...
0000000000000594 <shr1>:
 594:   a9be7bfd        stp     x29, x30, [sp, #-32]!
 598:   910003fd        mov     x29, sp
 59c:   f9000fa0        str     x0, [x29, #24]
 5a0:   d503201f        nop
 5a4:   a8c27bfd        ldp     x29, x30, [sp], #32
 5a8:   d65f03c0        ret
...
and at the first insn:
...
1: /x $pc = 0xfffff7f9d594
2: /x $sp = 0xfffffffff320
3: /x $x29 = 0xfffffffff320
(gdb) info frame
Stack level 0, frame at 0xfffffffff320:
 pc = 0xfffff7f9d594 in shr1; saved pc = 0x400718
 called by frame at 0xfffffffff340
 Arglist at 0xfffffffff320, args: 
 Locals at 0xfffffffff320, Previous frame's sp is 0xfffffffff320
...
and at the last insn:
...
1: /x $pc = 0xfffff7f9d5a8
2: /x $sp = 0xfffffffff320
3: /x $x29 = 0xfffffffff320
(gdb) info frame
Stack level 0, frame at 0xfffffffff340:
 pc = 0xfffff7f9d5a8 in shr1; saved pc = 0xfffff7d68fa0
 called by frame at 0xfffffffff490
 Arglist at 0xfffffffff320, args: 
 Locals at 0xfffffffff320, Previous frame's sp is 0xfffffffff340
 Saved registers:
  x29 at 0xfffffffff320, x30 at 0xfffffffff328
...
so a different frame address (while having the same $sp and $fp).

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