This is the mail archive of the systemtap@sources.redhat.com 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]

C untapset


Hi -


Here is a scenario which, if we embrace, would have implications on
kprobes reentrancy as well as instrumentation reuse.

The idea is that, for whatever reason, there may be a batch of kprobes
or jprobes-based C code in existence or being written, which someone
may wish to reuse via systemtap.  Here's a limited way to do it:
within this C code, place a marker (or even just documentation) at
interesting control flow points, such as kprobes event handlers.

Then a systemtap user can place an ordinary, possibly kprobe-backed,
probe point into that lower level instrumentation.  The script code
would be able to pull out data from the variables visible to the
target C code.  They could even communicate by setting flags back and
forth.

This is in effect meta-instrumentation, and requires no changes in
current systemtap design.  On the other hand, it would require having
kprobes support full reentrancy (instead of blocking the nested trap).


- FChE

Attachment: pgp00000.pgp
Description: PGP signature


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