[Bug tdep/29409] [gdb, tdep/aarch64] FAIL: gdb.opt/inline-small-func.exp: info breakpoints

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

https://sourceware.org/bugzilla/show_bug.cgi?id=29409

--- Comment #11 from Tom de Vries <vries at gcc dot gnu.org> ---
Another regression, test-case gdb.base/store.exp, insns:
...
00000000004005e4 <add_charest>:
  4005e4:       13001c02        sxtb    w2, w0
  4005e8:       13001c20        sxtb    w0, w1
  4005ec:       12001c41        and     w1, w2, #0xff
  4005f0:       12001c00        and     w0, w0, #0xff
  4005f4:       0b000020        add     w0, w1, w0
  4005f8:       12001c00        and     w0, w0, #0xff
  4005fc:       13001c00        sxtb    w0, w0
  400600:       d65f03c0        ret
...
and lines:
...
CU: store.c:
File name                   Line number    Starting address    View    Stmt
store.c                              14            0x4005e4               x
store.c                              15            0x4005ec               x
store.c                              16            0x400600               x
...

Without the patch, we use skip_prologue_using_sal and return 0x4005ec and
consequently line 15, which is the expected one.

With the patch, we return 0x4005e4 and consequently line 14.

Intesting question: are the two first insns part of the prologue or not?

In GCC terms, they come after NOTE_INSN_PROLOGUE_END but before
NOTE_INSN_FUNCTION_BEG and consequently have line info markers for the opening
brace '{' at line 14.

You could argue that this is a gcc bug and that the insn at 0x4005e4 should
have line 15, so something like this:
...
store.c                              14            0x4005e4               x
store.c                              15            0x4005e4               x
store.c                              16            0x400600               x
...
AFAIU, this would get us line 15 both with and without the patch.

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