This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
[Bug kprobes/2062] Return probes does not scale well on SMP box
- From: "jkenisto at us dot ibm dot com" <sourceware-bugzilla at sourceware dot org>
- To: systemtap at sources dot redhat dot com
- Date: 20 Apr 2006 00:22:38 -0000
- Subject: [Bug kprobes/2062] Return probes does not scale well on SMP box
- References: <20051216010933.2062.anil.s.keshavamurthy@intel.com>
- Reply-to: sourceware-bugzilla at sourceware dot org
------- Additional Comments From jkenisto at us dot ibm dot com 2006-04-20 00:22 -------
(In reply to comment #10)
> If it were a matter only of accelerating *allocation* and *deallocation* of the
> kretprobe_instances, sure, per-cpu free_instances pools would be handy. But the
> code also needs to deal with *lookup*, which needs to search cross-cpus thus a
> shared used_instances.
Insertion, lookup, and removal of a kretprobe_instance is always done by the
task associated with the kretprobe_instance. So if each task had its own list
of active kretprobe_instances, these activities wouldn't require locking.
Even given the current arrangement (global hash table, hashed by task), we could
lock only the current task's hash bucket, as proposed in Comment #2, and have
fairly low contention.
>
> Perhaps the single global kretprobe_lock could be replaced with per-kretprobe
locks?
Yeah, Comment #2 proposes that.
--
http://sourceware.org/bugzilla/show_bug.cgi?id=2062
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.