Re: [PATCH v2] gdb: Check bounds before reading DWARF expression operands

[Jielun Wu <firmiana402@gmail.com>]

Thank you for the review.

> It's conventional in C for the end pointer to be one-past-the-end.
> [...]
> So using "<" seems more correct than "<=" here?

For reference, the helper is:

> static const gdb_byte *
> safe_skip_bytes (const gdb_byte *buf, const gdb_byte *buf_end,
>                  ULONGEST bytes)
> {
>   gdb_assert (buf <= buf_end);
>   if (bytes > (ULONGEST) (buf_end - buf))
>     error (_("DWARF expression error: ran off end of buffer reading bytes"));
>   return buf + bytes;
> }

BUF_END is indeed one-past-the-end, but BUF can equal it when an
operand is entirely missing.

For example, the existing truncated_const1u case contains only
DW_OP_const1u.  After reading the opcode, op_ptr == op_end, and
safe_skip_bytes is called with bytes == 1.  The <= assertion lets
the subsequent length check report the expected DWARF error;
a < assertion would instead trigger an internal error first.

Equality also allows a zero-byte skip at the end, so I think we
should keep <= here.

Jielun