Experiences with cvs runtime

[Frank Ch. Eigler]

Hi -

hunt wrote:

> [...]
> > - many runtime (and one relayfs) files use smp_processor_id() without
> > disabling preemption
> 
> [...]
> We should probably disable preemption during all probes. 

Yes, quite possibly, but I don't understand all the implications of
putting a preempt_disable/enable() into strategic places within the
kprobes layer or into the generated code.


> Unless the script compiler can guarantee that scripts won't call
> something that tries to reschedule during a probe.  [...]

The translator would certainly not emit any code directly that intends
to reschedule.  Probes are supposed to never block.


- 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/20050427/237d8bd9/attachment.sig>