Re: [RFA/tracepoint] Make GDB can work with some old GDB server

[Hui Zhu <teawater@gmail.com>]

Ping.

On Fri, Jun 24, 2011 at 16:46, Hui Zhu <teawater@gmail.com> wrote:
> Hi,
>
> I got some bug report about ppa in https://lkml.org/lkml/2011/6/4/65
> It is:
> kgdb_breakpoint () at kernel/kgdb.c:1721
> 1721            wmb(); /* Sync point after breakpoint */
> Sending packet: $qSymbol::#5b...Ack
> Packet received:
> Packet qSymbol (symbol-lookup) is NOT supported
> Sending packet: $qTStatus#49...Ack
> Packet received: E22
> trace API error 0x2.
> (gdb) bt
> No stack.
> (gdb)
>
> This is because this kgdb(I think trunk have fixed this bug) don't
> support qtstatus and reply -0x22.
> So gdb throw a error.  Then all connect process stop.
>
> This is a bug of kgdb, but I think make gdb just output a warning is
> not affect anything else.  So I make this patch.
>
> Please help me review it.
>
> Thanks,
> Hui
>
>
> 2011-06-24  Hui Zhu  <teawater@gmail.com>
>
>        * remote.c (remote_start_remote): Add TRY_CATCH for
>        remote_get_trace_status.
>        * tracepoint.c (disconnect_tracing): Ditto.
> ---
>  remote.c     |   13 ++++++++++++-
>  tracepoint.c |   14 +++++++++++++-
>  2 files changed, 25 insertions(+), 2 deletions(-)
>
> --- a/remote.c
> +++ b/remote.c
> @@ -3146,6 +3146,8 @@ remote_start_remote (int from_tty, struc
>   struct remote_state *rs = get_remote_state ();
>   struct packet_config *noack_config;
>   char *wait_status = NULL;
> +  int ret = 0;
> +  volatile struct gdb_exception ex;
>
>   immediate_quit++;            /* Allow user to interrupt it.  */
>
> @@ -3389,7 +3391,16 @@ remote_start_remote (int from_tty, struc
>
>   /* Possibly the target has been engaged in a trace run started
>      previously; find out where things are at.  */
> -  if (remote_get_trace_status (current_trace_status ()) != -1)
> +  TRY_CATCH (ex, RETURN_MASK_ERROR)
> +    {
> +      ret = remote_get_trace_status (current_trace_status ());
> +    }
> +  if (ex.reason < 0)
> +    {
> +      warning(_("%s"), ex.message);
> +      ret = -1;
> +    }
> +  if (ret != -1)
>     {
>       struct uploaded_tp *uploaded_tps = NULL;
>       struct uploaded_tsv *uploaded_tsvs = NULL;
> --- a/tracepoint.c
> +++ b/tracepoint.c
> @@ -1939,11 +1939,23 @@ trace_status_mi (int on_stop)
>  void
>  disconnect_tracing (int from_tty)
>  {
> +  int ret = 0;
> +  volatile struct gdb_exception ex;
> +
>   /* It can happen that the target that was tracing went away on its
>      own, and we didn't notice.  Get a status update, and if the
>      current target doesn't even do tracing, then assume it's not
>      running anymore.  */
> -  if (target_get_trace_status (current_trace_status ()) < 0)
> +  TRY_CATCH (ex, RETURN_MASK_ERROR)
> +    {
> +      ret = target_get_trace_status (current_trace_status ());
> +    }
> +  if (ex.reason < 0)
> +    {
> +      warning(_("%s"), ex.message);
> +      ret = -1;
> +    }
> +  if (ret < 0)
>     current_trace_status ()->running = 0;
>
>   /* If running interactively, give the user the option to cancel and
>