Fw: [ltc-interlock] [RFC] draft NFS trace hooks (fwd)

[Tony Reix]

Le mercredi 23 août 2006 à 11:36 -0400, Frank Ch. Eigler a écrit :
> Tony Reix <tony.reix@bull.net> writes:
> 
> > [...] I'm not sure I perfectly understood the explanations, but it
> > seems to me that "dynamic" probes do require gcc to provide
> > debugging information that prevent it to run full optimizations.
> 
> No, we don't interfere with gcc optimization, and in exchange may
> receive less-than-perfect debugging information.  Systemtap can
> usually make do with that, or tell the user otherwise during
> script translation.

Ooops Sorry, I misunderstood a sentence in your paper. The one talking
about "Since markers don't rely on debugging information, ...".

However your analysis of performance impact of SystemTap seems really
promising. I guess you need more figures to convince "only performance
in mind" people ...

If I remember well, Mainframe-class computers have such features and are
able to trace all the system at will. AIX does, I think.

Probably a future step will be to provide the FFDC feature (First
Failure Data Capture = first time a failure occurs, all data needed to
understand the problems is already available) ?

Regards,

Tony