svn commit: r187156 - head/tools/sched

[John Baldwin]

On Tuesday 13 January 2009 11:44:18 am John Baldwin wrote:
> Author: jhb
> Date: Tue Jan 13 16:44:18 2009
> New Revision: 187156
> URL: http://svn.freebsd.org/changeset/base/187156
> 
> Log:
>   Add support for two new event source types:
>   - Callwheels traced via KTR_CALLOUT.  Each CPU is assigned a callwheel
>     source.  The events on this source are the execution of individual callout
>     routines.  Each routine shows up as a green rectangle while it is executed
>     and the event details include the function pointer and argument.
>   - Locks traced via KTR_LOCK.  Currently, each lock name is assigned an event
>     source (since the existing KTR_LOCK traces only include lock names and
>     not pointers).  This does mean that if multiple locks of the same name are
>     manipulated, the source line for that name may be confusing.  However, for
>     many cases this can be useful.  Locks are blue when they are held and
>     purple when contested.  The contention support is a bit weak due to
>     limitations in the rw_rlock() and mtx_lock_spin() logging messages
>     currently.  I also have not added support for contention on lockmgr,
>     sx, or rmlocks yet.  What is there now can be profitably used to examine
>     activity on Giant however.
>   - Expand the width of the event source names column a bit to allow for some
>     of the longer names of these new source types.

For some examples, check out www.freebsd.org/~jhb/sysctl_before.png and
sysctl_after.png.  Those are screenshots of 'sysctl -a' showing lock activity.
In particular, in 'after' I removed (via #if 0) the debug.hashstat sysctls
(from sys/kern/vfs_cache.c).  In the 'before' graph one of those two sysctls
(the raw dump of the hash table depths) took 7 milliseconds to run on my test
machine (2 x 3.4 GHz Intel dual-core) holding Giant the entire time which caused
softclock to block on Giant for a couple of milliseconds.  (These traces were
performed on 7-stable.)

Once I get more caught up I plan on sticking those sysctl's under some sort of
#ifdef that isn't on by default.

-- 
John Baldwin