better SMP support for kprobes

[Frank Ch. Eigler]

Hi -


richardj_moore wrote:

> I think this discussion has arisen before. We don't block even though we
> appear to use copy_from_user this is because we intercept faults and call
> back to the probe handler to do with them what they will.  [...]

Has this control flow seen much "in the wild", or is it largely
theoretical?  Even if there is some experience with it, it would
be better if an alternate routine were available that was
- explicitly safe to call even from interrupt context (perhaps
  passing an explicit reference to the user process in question)
- failed via a return code rather than a nested interrupt

- FChE
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/systemtap/attachments/20050203/7864bcd4/attachment.sig>