This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
[Bug uprobes/10458] uaddr() returns one past current instruction for uprobes
- From: "mjw at redhat dot com" <sourceware-bugzilla at sourceware dot org>
- To: systemtap at sources dot redhat dot com
- Date: 31 Jul 2009 18:56:44 -0000
- Subject: [Bug uprobes/10458] uaddr() returns one past current instruction for uprobes
- References: <20090729111010.10458.mjw@redhat.com>
- Reply-to: sourceware-bugzilla at sourceware dot org
------- Additional Comments From mjw at redhat dot com 2009-07-31 18:56 -------
(In reply to comment #11)
> Looks good, but keep in mind that the (corrected) IP when you're in a kretprobe
> or uretprobe handler is just the address of the return-probe trampoline.
This was mostly to make sure that we are always at the actual probe address and
not accidentally one past (possibly in the middle of the actual instruction, and
in the worse case just one past the end of the function).
The next layer/level deals with the actual symbol/unwind resolving. But we might
indeed always have to special case the [ku]retprobe case by checking CONTEXT->pi
being set anyway. If so, then we might indeed never need the actual address in
the first place. Thanks for the reminder. I'll think about it.
--
http://sourceware.org/bugzilla/show_bug.cgi?id=10458
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.