Re: [Dprobes] Questions about dprobes

[Richard J M. <ric...@uk...>]

Will, I seem to remember that we integrated this for X86 about May 2003.
I'll check on the status of other platforms. I wrote the white papers
before this change. However, the latest presentation does mention that it's
no longer an issue, but you won't have seen these unless you were at my
talk in Reykjavik University earlier this year. I guess I ought to make the
latest presentation available more generally.

And of course --stop-cpus is no longer relevant and is therefore retired.
We should remove all mention of it.

- -
Richard J Moore
IBM Advanced Linux Response Team - Linux Technology Centre
http://oss.software.ibm.com/developerworks/opensource/linux
Tel: (+44) (0)1962-817072


                                                                           
             Will Cohen                                                    
             <wcohen@redhat.                                               
             com>                                                       To 
                                     Richard J Moore/UK/IBM@IBMGB          
             19/08/2004                                                 cc 
             19:07                   dp...@ww... 
                                     ,                                     
                                     dpr...@ww...f.i 
                                     bm.com, S Vamsikrishna                
                                     <vam...@in...>            
                                                                       bcc 
                                                                           
                                                                   Subject 
                                     Re: [Dprobes] Questions about dprobes 
                                                                           
                                                                           




Richard J Moore wrote:
>
>
>
> The SMP misses problem has not been an issue for a long time. We
> implemented code to execute the original instruction out-of-context so
that
> the INT3 was never removed during single-stepping.

That is great. How long has that been in there? None of the papers or
documentation I read mentioned that. It might be good to point out that
is no longer an issue in papers and documentation.

-Will Cohen

>
> - -
> Richard J Moore
> IBM Advanced Linux Response Team - Linux Technology Centre
> http://oss.software.ibm.com/developerworks/opensource/linux
> Tel: (+44) (0)1962-817072