This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap 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]

Re: [RFC] Systemtap translator support for hardware breakpoints on


> The translator need not veto such a script request.  It should be able
> to fulfill it by rounding up/aligning down, then filtering the
> resulting callbacks.

... and that is exactly what I described in the next paragraph.
There is no way to "filter" for .read, and the only way for .write
is to track changed values rather than writes per se.  So this means
something different (i.e. .change) than what .write means in the
sense of the actual hw_breakpoint layer (and the hardware feature).
Hence it is IMHO wrong to claim to satisfy any .write probe request
by implementing .change instead.  It could just warn, but I don't
see why it shouldn't give a hard error that says "try .change instead".


Thanks,
Roland


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