This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: GDB speed using Remote Serial Protocol
- From: Daniel Jacobowitz <drow at mvista dot com>
- To: David Mc Kenna <mckennad at esatclear dot ie>
- Cc: gdb at sources dot redhat dot com
- Date: Mon, 2 Dec 2002 13:55:43 -0500
- Subject: Re: GDB speed using Remote Serial Protocol
- References: <3debabab.66df.0@esatclear.ie>
No, you missed my gist. With debug info printing, GDB takes a little
bit longer (because it has to do the printing). Something is being
sent too quickly, and being dropped and then eventually retried.
On Mon, Dec 02, 2002 at 06:51:23PM +0000, David Mc Kenna wrote:
> I would agree with this except that with debug info switched on it is much faster
> than with it switched off. What is the difference in code execution within GDB
> with debug info switched off?
>
> Cheers,
> Dave
>
> >On Mon, Dec 02, 2002 at 05:27:04PM +0000, David Mc Kenna wrote:
> >> Hi all,
> >>
> >> At present I have an outline for a stub ( GDB RSP -> uController -> Jtag
> ) for
> >> an Arm7TDMi. I have successfully implemented the GDB -> uController side
> of
> >> the equation. Here I have two questions.
> >>
> >> Is there another way of seeing what GDB sends to the serial port without
> enabling
> >> debug info within GDB ?, i.e. without using the commands :
> >> "set debug remote 1" & " set debug serial 1"
> >>
> >> My second questions relates to the first. With debug info switched on ( "set
>
> >> debug remote 1" & " set debug serial 1"), a single step talks < 1 second
> to
> >> complete, including all the debug info being written to the screen. Yet when
>
> >> I switch off this info the single step time starts at < 1 second but soon
> mushrooms
> >> to ~ 5 seconds.
> >>
> >> I have tried this on gdb v5 and gdb v5.2.1 with the same results.
> >> GDB configured as arm-elf
> >>
> >>
> >> Any ideas??
> >
> >Sounds like something is being sent more quickly than your
> >stub/controller/whatever is ready to handle it.
> >
> >What I recommend in this case is writing (or getting; rproxy may be
> >able to do this?) a stub which just connects a TCP port and a serial
> >port; forwards data from one to the other, and logs.
> >
> >--
> >Daniel Jacobowitz
> >MontaVista Software Debian GNU/Linux Developer
> >
> >
> --
> http://www.iol.ie
>
--
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer