Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
* Defer register reading from remote target until needed?
@ 2004-06-03 13:08 Orjan Friberg
  2004-06-03 18:25 ` Michael Snyder
  0 siblings, 1 reply; 3+ messages in thread
From: Orjan Friberg @ 2004-06-03 13:08 UTC (permalink / raw)
  To: gdb-patches

When single-stepping a remote target, after the stop reply paket is 
sent, GDB immediately sends a request for all registers.  Is there a way 
to defer sending this request until some register other than what was 
sent with the stop reply packet (pc, stack and frame pointer) is 
actually needed?

-- 
Orjan Friberg
Axis Communications


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Defer register reading from remote target until needed?
  2004-06-03 13:08 Defer register reading from remote target until needed? Orjan Friberg
@ 2004-06-03 18:25 ` Michael Snyder
  2004-06-04  8:52   ` Orjan Friberg
  0 siblings, 1 reply; 3+ messages in thread
From: Michael Snyder @ 2004-06-03 18:25 UTC (permalink / raw)
  To: Orjan Friberg; +Cc: gdb-patches

Orjan Friberg wrote:
> When single-stepping a remote target, after the stop reply paket is 
> sent, GDB immediately sends a request for all registers.  Is there a way 
> to defer sending this request until some register other than what was 
> sent with the stop reply packet (pc, stack and frame pointer) is 
> actually needed?

It's supposed to work that way already.  Either something's broken, or
some register other than the ones you named *is* needed.  The 'g' pkt
should only go out if gdb needs the value of a register that it doesn't
already have.

This can sometimes happen 'indirectly', eg. if you try to set a 
register, gdb may try to read it first.

Try setting a breakpoint in remote_fetch_registers, then go up
the stack and see what register gdb is trying to read.



^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Defer register reading from remote target until needed?
  2004-06-03 18:25 ` Michael Snyder
@ 2004-06-04  8:52   ` Orjan Friberg
  0 siblings, 0 replies; 3+ messages in thread
From: Orjan Friberg @ 2004-06-04  8:52 UTC (permalink / raw)
  To: Michael Snyder; +Cc: gdb-patches

Michael Snyder wrote:
> 
> It's supposed to work that way already.  Either something's broken, or
> some register other than the ones you named *is* needed.  The 'g' pkt
> should only go out if gdb needs the value of a register that it doesn't
> already have.
> 
> This can sometimes happen 'indirectly', eg. if you try to set a 
> register, gdb may try to read it first.
> 
> Try setting a breakpoint in remote_fetch_registers, then go up
> the stack and see what register gdb is trying to read.

Ah; it's because we need the subroutine return pointer when unwinding 
the pc.  Thanks.


-- 
Orjan Friberg
Axis Communications


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2004-06-04  8:52 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-06-03 13:08 Defer register reading from remote target until needed? Orjan Friberg
2004-06-03 18:25 ` Michael Snyder
2004-06-04  8:52   ` Orjan Friberg

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox