This is the mail archive of the ecos-bugs@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug 19973] New: sparclite temporary breakpoints can remain set


http://bugs.ecos.sourceware.org/show_bug.cgi?id=19973

           Summary: sparclite temporary breakpoints can remain set
           Product: eCos
           Version: unknown
          Platform: Other
        OS/Version: Other
            Status: RESOLVED
          Severity: normal
          Priority: low
         Component: Other
        AssignedTo: ebachalo at schuilenburg dot org
        ReportedBy: nobody at cygnus dot com
                CC: jifl at ecoscentric dot com


Using GDBtk and the eCos twothreads example program for
testing sparclite eval board with cygmon, I was testing
switching thread contexts.

What I did was step one thread, then switch to another, then
step that thread, then switch to another, then maybe "next" on
a thread, switch, maybe "finish" on a thread, etc.etc. ad
nauseum.

Eventually I saw that GDB was stopping at the wrong places,
and instead of getting "Breakpoint" messages, I got messages
of the form "Program received Trace/breakpoint trap" on the
console.

If I then "continue"d, then it kept hitting these traps, even though
I had never set a breakpoint, and info break was reporting
no breakpoints present.

It looks to me like GDB/Cygmon is forgetting about the
temporary breakpoints it sets, after a point.

How-To-Repeat:
I couldn't reproduce this in any accurate way, just by
switching threads and stepping them, switching and stepping
them, ad nauseam, until you notice that the wrong things
are happening.

If you then just use "continue" from that point you notice that
the breakpoints keep on getting hit, even though none are
set, and they don't clear themselves.

It is easily reproducable, its just long-winded to find out
exactly what steps are required to do it :-). Try it and see
- you may need a mixture of "step", "next", and "finish".


---------------------------------------------------------------------------

Originator:
Jonathan Larmour

Organization:
cygnus

Audit-Trail:
Responsible-Changed-From-To: alexs->ebachalo 
Responsible-Changed-By: alexs 
Responsible-Changed-When: Tue Apr 27 06:12:45 PDT 1999 
Responsible-Changed-Why:  
Hi Eric 

This is another problem thrown up by eCos QA. 

There is a problem with the QA procedure in that it assumes you can 
switch between thread contexts and single step the thread whose context you 
switched to. You should only be able to step the current thread. 

However, this PR reports a problem with temporary breakpoints being 
seemingly left active. It could be a GDB problem or a cygmon problem. 
Which one we are not sure. I will forward Mark Salter a copy of this 
so that he is aware of this problem. 

Cheers 
-- Alex

Unformatted:
Originator:  

page: www.cygnus.com/product/ecc-pr.html

Send_PR_form: Sent_from_www.cygnus.com

------- Additional Comments From alexs at ecoscentric dot com  2003-24-04 17:20 BST -------
Tools issue



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]