[Bug runtime/2637] skipped probes in FC6

[jkenisto at us dot ibm dot com]

------- Additional Comments From jkenisto at us dot ibm dot com  2006-05-31 21:58 -------
I believe that we have concluded that this is NOT a kprobes bug, and in fact
reflects a fix in kprobes's fault handling.  See Anil's analysis in Comment #10.
Kprobes handlers aren't supposed to sleep.  When called from a handler, a
function that accesses a non-resident user page must fail.  That's what's
happening now.  What was happening before was that handler slept (BAD, BAD)
while the non-resident page was brought it.  So the memory access succeeded, but
at the risk of hanging the system.

The real bug, as I see it, is that the script now terminates when the memory
access fails.

I'm not sure whether this is a runtime bug or a translator bug.  I'm reassigning
this as a runtime bug because most of the proposed solutions involve at least
some modification of the runtime.  See
http://sources.redhat.com/ml/systemtap/2006-q2/msg00493.html

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|systemtap at sources dot    |hunt at redhat dot com
                   |redhat dot com              |
             Status|NEW                         |ASSIGNED
          Component|kprobes                     |runtime


http://sourceware.org/bugzilla/show_bug.cgi?id=2637

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.