From owner-freebsd-hackers@freebsd.org Wed Oct 2 16:39:57 2019 Return-Path: Delivered-To: freebsd-hackers@mailman.nyi.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mailman.nyi.freebsd.org (Postfix) with ESMTP id CA8801336FF for ; Wed, 2 Oct 2019 16:39:57 +0000 (UTC) (envelope-from kostikbel@gmail.com) Received: from kib.kiev.ua (kib.kiev.ua [IPv6:2001:470:d5e7:1::1]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) server-signature RSA-PSS (4096 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 46k2185cQRz4g0k for ; Wed, 2 Oct 2019 16:39:56 +0000 (UTC) (envelope-from kostikbel@gmail.com) Received: from tom.home (kib@localhost [127.0.0.1]) by kib.kiev.ua (8.15.2/8.15.2) with ESMTPS id x92Gdkc7073201 (version=TLSv1.3 cipher=TLS_AES_256_GCM_SHA384 bits=256 verify=NO); Wed, 2 Oct 2019 19:39:50 +0300 (EEST) (envelope-from kostikbel@gmail.com) DKIM-Filter: OpenDKIM Filter v2.10.3 kib.kiev.ua x92Gdkc7073201 Received: (from kostik@localhost) by tom.home (8.15.2/8.15.2/Submit) id x92GdkIS073200; Wed, 2 Oct 2019 19:39:46 +0300 (EEST) (envelope-from kostikbel@gmail.com) X-Authentication-Warning: tom.home: kostik set sender to kostikbel@gmail.com using -f Date: Wed, 2 Oct 2019 19:39:46 +0300 From: Konstantin Belousov To: Bruce Evans Cc: Sebastian Huber , FreeBSD Subject: Re: Why is tc_get_timecount() called two times in tc_init()? Message-ID: <20191002163946.GE44691@kib.kiev.ua> References: <0e27fb3e-0f60-68e1-dbba-f17c3d91c332@embedded-brains.de> <20191002140040.GA44691@kib.kiev.ua> <20191003013314.O2151@besplex.bde.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20191003013314.O2151@besplex.bde.org> User-Agent: Mutt/1.12.2 (2019-09-21) X-Spam-Status: No, score=-1.0 required=5.0 tests=ALL_TRUSTED,BAYES_00, DKIM_ADSP_CUSTOM_MED,FORGED_GMAIL_RCVD,FREEMAIL_FROM, NML_ADSP_CUSTOM_MED autolearn=no autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on tom.home X-Rspamd-Queue-Id: 46k2185cQRz4g0k X-Spamd-Bar: / Authentication-Results: mx1.freebsd.org; dkim=none; dmarc=fail reason="No valid SPF, No valid DKIM" header.from=gmail.com (policy=none); spf=softfail (mx1.freebsd.org: 2001:470:d5e7:1::1 is neither permitted nor denied by domain of kostikbel@gmail.com) smtp.mailfrom=kostikbel@gmail.com X-Spamd-Result: default: False [-0.99 / 15.00]; ARC_NA(0.00)[]; NEURAL_HAM_MEDIUM(-0.99)[-0.994,0]; DMARC_POLICY_SOFTFAIL(0.10)[gmail.com : No valid SPF, No valid DKIM,none]; FROM_HAS_DN(0.00)[]; RCPT_COUNT_THREE(0.00)[3]; FREEMAIL_FROM(0.00)[gmail.com]; IP_SCORE(0.00)[ip: (-2.68), ipnet: 2001:470::/32(-4.53), asn: 6939(-3.34), country: US(-0.05)]; MIME_GOOD(-0.10)[text/plain]; HAS_XAW(0.00)[]; NEURAL_HAM_LONG(-1.00)[-1.000,0]; R_SPF_SOFTFAIL(0.00)[~all:c]; IP_SCORE_FREEMAIL(0.00)[]; TO_MATCH_ENVRCPT_SOME(0.00)[]; TO_DN_ALL(0.00)[]; SUBJECT_ENDS_QUESTION(1.00)[]; FREEMAIL_TO(0.00)[optusnet.com.au]; FROM_EQ_ENVFROM(0.00)[]; R_DKIM_NA(0.00)[]; FREEMAIL_ENVFROM(0.00)[gmail.com]; ASN(0.00)[asn:6939, ipnet:2001:470::/32, country:US]; MIME_TRACE(0.00)[0:+]; RCVD_TLS_ALL(0.00)[]; RCVD_COUNT_TWO(0.00)[2] X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Technical Discussions relating to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 02 Oct 2019 16:39:57 -0000 On Thu, Oct 03, 2019 at 02:25:46AM +1000, Bruce Evans wrote: > On Wed, 2 Oct 2019, Konstantin Belousov wrote: > > > On Tue, Oct 01, 2019 at 01:11:18PM +0200, Sebastian Huber wrote: > >> Hello, > >> > >> since this commit > >> > >> https://github.com/freebsd/freebsd/commit/307f787e5a7f > >> > > It is not very useful to pass github hashes around. > > > > I think that the addition of the second tc_get_timecount() was done > > earlier, in r95530, and there it has semi-useful comment > > + /* Warm up new timecounter. */ > > + (void)newtc->tc_get_timecount(newtc); > > + (void)newtc->tc_get_timecount(newtc); > > > > The commit message is not helpful at all. > > The comment was correct when I added it in r48887. Then it was only > attached to the first tc_get_timecounter() call which is necessary to > start up or sync the hardware timecounter in some cases, e.g., for the > i8254. After the warmup, r48887 called the preexisting function > tc_switch_timecount() switch the software state. This function was > unused (ifdefed out) but required only minor modifications. > > r95530 removes tc_switch_timecount() and replaces it by a second call > to tc_get_timecounter() and an assignment of the new timecounter pointer > to the active timecounter pointer, and removes the blank line that > separates the warmup from the application. The second call and the > assignment are all that is left of the function after moving its > initialization. > > The second warmup looks like nonsense in both versions, and it is > unclear how activation by plain assignment can work in the second > version (the first version was in FreeBSD-3 or 4 and was locked by > splclock()). The assignment is sloppy about memory ordering. The > second warmup may have helped by accidentally forcing the assignment > to be after the first warmup. (Some timecounters that need at least > 1 warmup, e.g., the i854, use atomic ops that accidentally give > sufficient ordering.) Later work on ordering may have reduced the > sloppiness. > > > I do not see a timecounter which would need two get_timecount() calls > > to start working properly now, but I can imagine that at time it was. > > I think it never helped much. For the TSC, the 2 calls are ordered only > relatively each other on a single CPU. They are not ordered relative to > memory. For the i8254, 1 call is enough. The ACPI timer does hardware > accesses so it is in between. So the conclusion is that the second call can be removed, am I right ? > > > I added Bruce to Cc: to may be get more context and explanation. > > > >> tc_get_timecount() is called two times in tc_init(). > >> > >> /* > >> * Initialize a new timecounter and possibly use it. > >> */ > >> void > >> tc_init(struct timecounter *tc) > >> { > >> [...] > >> if (tc->tc_quality == timecounter->tc_quality && > >> tc->tc_frequency < timecounter->tc_frequency) > >> return; > >> (void)tc->tc_get_timecount(tc); > >> (void)tc->tc_get_timecount(tc); > >> timecounter = tc; > >> } > >> > >> What is the reason for this procedure? > > Bruce