gdb/1933: When calling a function from gdb, unexpected SIGTRAPs are received

Peter Wainwright prw@ceiriog1.demon.co.uk
Tue May 3 13:08:00 GMT 2005


The following reply was made to PR gdb/1933; it has been noted by GNATS.

From: Peter Wainwright <prw@ceiriog1.demon.co.uk>
To: urs.hunkeler@epfl.ch
Cc: gdb-gnats@sources.redhat.com, nobody@sources.redhat.com,
        gdb-prs@sources.redhat.com
Subject: Re: gdb/1933: When calling a function from gdb, unexpected
	SIGTRAPs are received
Date: Mon, 02 May 2005 18:55:53 +0100

 That's surprising, Urs.  I am now working on a vanilla 2.6.11.5
 (kernel.org) kernel and my problem has disappeared.  I managed
 to get a test case debugging a very simple "Hello World"
 type application, and managed to replicate the problem consistently.
 The problem was the second consecutive invocation of a function
 after a single step.  After the single step, "print $eflags"
 showed that the trap flag (0x100) was still set.  For some reason
 this did not affect the first invocation of the function
 (you call this a "call dummy"?), however, the second would
 always break after every instruction, as you would expect since
 the trap flag was not cleared properly.  I suspect the kernel
 implementation of ptrace (arch/i386/ptrace.c) or something similar.
 -- 
 Peter Wainwright <prw@ceiriog1.demon.co.uk>



More information about the Gdb-prs mailing list