2012/3/19 Gleb Smirnoff <gleb...@freebsd.org>: > Davide, > > On Fri, Mar 16, 2012 at 08:32:11PM +0000, Davide Italiano wrote: > D> Author: davide > D> Date: Fri Mar 16 20:32:11 2012 > D> New Revision: 233045 > D> URL: http://svn.freebsd.org/changeset/base/233045 > D> > D> Log: > D> Add rudimentary profiling of the hash table used in the in the umtx code > to > D> hold active lock queues. > D> > D> Reviewed by: attilio > D> Approved by: davidxu, gnn (mentor) > D> MFC after: 3 weeks > D> > D> Modified: > D> head/sys/conf/NOTES > D> head/sys/conf/options > D> head/sys/kern/kern_umtx.c > > ... > > D> static void > D> umtxq_sysinit(void *arg __unused) > D> { > D> @@ -232,8 +265,15 @@ umtxq_sysinit(void *arg __unused) > D> TAILQ_INIT(&umtxq_chains[i][j].uc_pi_list); > D> umtxq_chains[i][j].uc_busy = 0; > D> umtxq_chains[i][j].uc_waiters = 0; > D> + #ifdef UMTX_PROFILING > D> + umtxq_chains[i][j].length = 0; > D> + umtxq_chains[i][j].max_length = 0; > D> + #endif > D> } > D> } > D> + #ifdef UMTX_PROFILING > D> + umtx_init_profiling(); > D> + #endif > D> mtx_init(&umtx_lock, "umtx lock", NULL, MTX_SPIN); > D> EVENTHANDLER_REGISTER(process_exec, umtx_exec_hook, NULL, > D> EVENTHANDLER_PRI_ANY); > D> @@ -384,6 +424,14 @@ umtxq_insert_queue(struct umtx_q *uq, in > D> > D> TAILQ_INSERT_TAIL(&uh->head, uq, uq_link); > D> uh->length++; > D> + #ifdef UMTX_PROFILING > D> + uc->length++; > D> + if (uc->length > uc->max_length) { > D> + uc->max_length = uc->length; > D> + if (uc->max_length > max_length) > D> + max_length = uc->max_length; > D> + } > D> + #endif > D> uq->uq_flags |= UQF_UMTXQ; > D> uq->uq_cur_queue = uh; > D> return; > D> @@ -401,6 +449,9 @@ umtxq_remove_queue(struct umtx_q *uq, in > D> uh = uq->uq_cur_queue; > D> TAILQ_REMOVE(&uh->head, uq, uq_link); > D> uh->length--; > D> + #ifdef UMTX_PROFILING > D> + uc->length--; > D> + #endif > D> uq->uq_flags &= ~UQF_UMTXQ; > D> if (TAILQ_EMPTY(&uh->head)) { > D> KASSERT(uh->length == 0, > > These indented ifdefs look like a major violation of style used throughout > the FreeBSD kernel code. Can you please keep with common style? > > -- > Totus tuus, Glebius.
Heh, sorry, also Juli Mallet noticed this, I'm writing a fix for this and after I'll have approval from my mentor I'll commit. Davide _______________________________________________ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
