This is the mail archive of the
mailing list for the GDB project.
Re: [RFC] Re: Profiling gdb?
- From: Andrew Cagney <ac131313 at redhat dot com>
- To: Daniel Jacobowitz <drow at mvista dot com>
- Cc: gdb-patches at sources dot redhat dot com
- Date: Mon, 06 Jan 2003 09:14:41 -0500
- Subject: Re: [RFC] Re: Profiling gdb?
- References: <20020120210343.A22638@nevyn.them.org> <3C4B8A6B.firstname.lastname@example.org> <20030105222713.GA22511@nevyn.them.org>
(Unfortunatly, I've also leant a few things about how profiling works
Perhaps some other time. I do _use_ monstartup, however.
This patch should address all comments from the last three times this
came up. I chose to use monstartup/_mcleanup instead of using
moncontrol; when someone wants to use this on a system without those
functions, _then_ we can decide how to handle it. That's just my
opinion though. What I've done should work for at least GNU/Linux and
FreeBSD, and that's enough to be useful. It only comes in if you
configure it on, anyway.
It works like a charm; absolutely beautifully. Thoughts, all?
I think the commands should always be present - conditional on the
corresponding function being present. Enabling / disabling the profiler
is orthogonal to compiling with -pg. I'm not sure about the unpublished
_moncleanup() call. Are you sure of monstartup()'s function signature
(ari doesn't like the extern declarations but there isn't a header file
I also suspect that more explict commands such as:
monstartup [ <start> <end> ]
would be better. The person using this feature will need to know how it
is implemented anyway.
(Did Jason Molenda, have something to do with the original patch?)