This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: prologue detection
- From: Jim Keniston <jkenisto at us dot ibm dot com>
- To: "Frank Ch. Eigler" <fche at elastic dot org>
- Cc: SystemTAP <systemtap at sources dot redhat dot com>
- Date: 07 Apr 2006 14:34:44 -0700
- Subject: Re: prologue detection
- Organization:
- References: <20060407142133.GB5854@elastic.org>
On Fri, 2006-04-07 at 07:21, Frank Ch. Eigler wrote:
> Hi -
>
> While trying to run some measurements on the static instrumentation
> code, I came across an amusing and troubling bug in the translator's
> prologue-detection heuristic. For some classes of functions (in this
> case, prologue-less ones), it would get it totally wrong, and
> designate an instruction within a subsequent block. I've seen one
> case where the probe was placed *beyond* the requested function.
>
> I'm taking a little detour to try to simplify & fix this part up. In
> the mean time, you may want to double-check the placement of probes in
> your scripts, especially when on short funtions. Please send me
> anything suspicious that you may have recently seen.
>
> - FChE
As you may remember, bugzilla #2453 documents one such bug that in turn
revealed a powerpc kprobes bug. There was supposed to be a separate bug
opened for the prologue-detection error, but I don't know whether that
happened.
Jim