This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v4 4/6] Support software single step on ARM in GDBServer.
- From: Yao Qi <qiyaoltc at gmail dot com>
- To: Antoine Tremblay <antoine dot tremblay at ericsson dot com>
- Cc: Yao Qi <qiyaoltc at gmail dot com>, <gdb-patches at sourceware dot org>
- Date: Thu, 03 Dec 2015 14:49:03 +0000
- Subject: Re: [PATCH v4 4/6] Support software single step on ARM in GDBServer.
- Authentication-results: sourceware.org; auth=none
- References: <1449062264-18565-1-git-send-email-antoine dot tremblay at ericsson dot com> <1449062264-18565-5-git-send-email-antoine dot tremblay at ericsson dot com> <86a8pru6hf dot fsf at gmail dot com> <56604A85 dot 5030805 at ericsson dot com>
Antoine Tremblay <antoine.tremblay@ericsson.com> writes:
> byte_order fields seemed like a good idea at first and I liked your
> suggested change for read_memory_unsigned_integer.
>
> However GDB is using 2 byte orders : byte_order (for data) and
> byte_order_for_code to support BE8 endianness.
>
> This complicates things a bit since in common code I can't call:
>
> self->ops->read_memory_unsigned_integer (self, loc, 2)
>
> I would have no way to specify if it should read with byte_order or
> with byte_order_for_code.
We can use two methods for reading data and reading instructions
respectively, and we can hide byte_order and byte_order_for_code still.
Anyway, I strong opinion on this, either way is OK to me.
> const gdb_byte *arm_thumb2_breakpoint: This one needs to stay, since
> while on GDB's side it could be computed through regcache/gdbarch, on
> GDBServer's side it's directly a variable.
OK.
--
Yao (éå)