Date: Thu, 7 Jan 2010 01:19:01 +0000 (UTC) From: Attilio Rao <attilio@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r201709 - head/sys/kern Message-ID: <201001070119.o071J1V9098305@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: attilio Date: Thu Jan 7 01:19:01 2010 New Revision: 201709 URL: http://svn.freebsd.org/changeset/base/201709 Log: Tweak comments. Modified: head/sys/kern/kern_lock.c Modified: head/sys/kern/kern_lock.c ============================================================================== --- head/sys/kern/kern_lock.c Thu Jan 7 01:10:49 2010 (r201708) +++ head/sys/kern/kern_lock.c Thu Jan 7 01:19:01 2010 (r201709) @@ -277,6 +277,11 @@ wakeupshlk(struct lock *lk, const char * * avoid a starvation for the threads sleeping on the shared * queue by giving them precedence and cleaning up the * exclusive waiters bit anyway. + * Please note that lk_exslpfail count may be lying about + * the real number of waiters with the LK_SLEEPFAIL flag on + * because they may be used in conjuction with interruptible + * sleeps so lk_exslpfail is consider as a 'upper limit' + * bound, considering the edge cases. */ realexslp = sleepq_sleepcnt(&lk->lock_object, SQ_EXCLUSIVE_QUEUE); @@ -943,6 +948,12 @@ __lockmgr_args(struct lock *lk, u_int fl * empty avoid a starvation for the threads sleeping * on the shared queue by giving them precedence * and cleaning up the exclusive waiters bit anyway. + * Please note that lk_exslpfail count may be lying + * about the real number of waiters with the + * LK_SLEEPFAIL flag on because they may be used in + * conjuction with interruptible sleeps so + * lk_exslpfail is consider as a 'upper limit' bound, + * considering the edge cases. */ MPASS((x & LK_EXCLUSIVE_SPINNERS) == 0); realexslp = sleepq_sleepcnt(&lk->lock_object, @@ -1046,6 +1057,13 @@ __lockmgr_args(struct lock *lk, u_int fl * threads sleeping on the shared queue by * giving them precedence and cleaning up the * exclusive waiters bit anyway. + * Please note that lk_exslpfail count may be + * lying about the real number of waiters with + * the LK_SLEEPFAIL flag on because they may + * be used in conjuction with interruptible + * sleeps so lk_exslpfail is consider as a + * 'upper limit' bound, considering the edge + * cases. */ if (v & LK_EXCLUSIVE_WAITERS) { queue = SQ_EXCLUSIVE_QUEUE;
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201001070119.o071J1V9098305>