From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22582 invoked by alias); 5 Feb 2008 08:33:45 -0000 Received: (qmail 22348 invoked by uid 48); 5 Feb 2008 08:33:02 -0000 Date: Tue, 05 Feb 2008 08:33:00 -0000 Message-ID: <20080205083302.22347.qmail@sourceware.org> From: "pmachata at redhat dot com" To: frysk-bugzilla@sourceware.org In-Reply-To: <20080204174048.5724.swagiaal@redhat.com> References: <20080204174048.5724.swagiaal@redhat.com> Reply-To: sourceware-bugzilla@sourceware.org Subject: [Bug general/5724] ftrace needs a -all option... X-Bugzilla-Reason: AssignedTo Mailing-List: contact frysk-bugzilla-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Post: List-Help: , Sender: frysk-bugzilla-owner@sourceware.org X-SW-Source: 2008-q1/txt/msg00276.txt.bz2 List-Id: ------- Additional Comments From pmachata at redhat dot com 2008-02-05 08:33 ------- I believe you almost never need to trace every symbol in the application. If the -all defaulted to syscalls and -plt=@MAIN, i.e. what ltrace does, that I would probably find useful. Maybe even -plt=, which is ALL library calls except for the calls in dynamic linker. But with the way ftrace works (i.e. over ptrace), that's prohibitively slow as each function call imposes two context switches (to debugger on SIGTRAP, and back on PTRACE_CONT). So all in all, I just don't think that -all would be very useful. -- What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WONTFIX http://sourceware.org/bugzilla/show_bug.cgi?id=5724 ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.