Re: [remote protocol] support for disabling packet acknowledgement

[Daniel Jacobowitz <drow@false.org>]

On Thu, Jul 10, 2008 at 06:23:13PM +0100, Pedro Alves wrote:
> Hi,
> 
> This patch adds support to the remote protocol to disable packet
> acknowlegment.  This is useful when the transport being used is itself
> reliable, e.g., TCP/IP.  In these cases, by removing the acking
> we reduce the amount of roundtrips, and decrease the communication
> latency.
> 
> While we were discussing implementing this, I noticed that Apple
> already had something of the sort, so we ended up reusing Apple's
> QStartNoAckMode packet name, and basing our implementation on
> their version, modernizing it, adding qSupport support
> to it, tweaking the user interface so it is integrated with
> the "set remote xxx-packet" set of commands, and documenting it.
> 
> The docs changes below explain more in detail the approach taken.
> 
> I'm including in the patch GDB changes to support the feature,
> and adding the support to gdbserver.  Regtesting against a native
> gdbserver on x86_64-unknown-linux-gnu doesn't show any regression.

This patch is OK, with the two texinfo corrections from Eli, and
without this:

> Index: src/gdb/gdbserver/remote-utils.c
> ===================================================================
> --- src.orig/gdb/gdbserver/remote-utils.c	2008-07-10 17:37:11.000000000 +0100
> +++ src/gdb/gdbserver/remote-utils.c	2008-07-10 17:38:08.000000000 +0100
> @@ -90,7 +90,7 @@ static struct sym_cache *symbol_cache;
>     failures.  */
>  int all_symbols_looked_up;
>  
> -int remote_debug = 0;
> +int remote_debug = 1;
>  struct ui_file *gdb_stdlog;
>  
>  static int remote_desc = INVALID_DESCRIPTOR;

Also, could you please add a NEWS entry (there's a new packet and new
associated commands).

-- 
Daniel Jacobowitz
CodeSourcery