breakpoints/2165: Stepping over longjmp presumably broken for glibc

[eggert at gnu dot org]

>Number:         2165
>Category:       breakpoints
>Synopsis:       Stepping over longjmp presumably broken for glibc
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Thu Aug 31 19:38:01 UTC 2006
>Closed-Date:
>Last-Modified:
>Originator:     Paul Eggert
>Release:        CVS
>Organization:
>Environment:
glibc
>Description:
This problem report follows up on the thread
"Stepping over longjmp presumably broken for glibc"
<http://sources.redhat.com/ml/gdb/2005-12/msg00177.html>
that was on the GDB mailing list late last year.

I was wondering whether Daniel Jacobowitz's suggestion to
single-step through longjmp needed any help from glibc to
become practical, or whether GDB can do it on its own
without any changes from glibc.  If the former, what sort of
help would be useful?  For example, does GDB need help to
find out where the longjmp machine code begins and ends?
>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: