From mboxrd@z Thu Jan 1 00:00:00 1970 From: Felipe Balbi Subject: Re: [RFC/NOT FOR MERGING] HACK: add global/private timers for A9 Date: Wed, 3 Jun 2015 16:04:55 -0500 Message-ID: <20150603210455.GD6042@saruman.tx.rr.com> References: <1433363565-17725-1-git-send-email-balbi@ti.com> <2466648.WjMxjIpAUG@wuerfel> Reply-To: Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="ILuaRSyQpoVaJ1HG" Return-path: Received: from bear.ext.ti.com ([192.94.94.41]:48798 "EHLO bear.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751753AbbFCVHo (ORCPT ); Wed, 3 Jun 2015 17:07:44 -0400 Content-Disposition: inline In-Reply-To: <2466648.WjMxjIpAUG@wuerfel> Sender: linux-omap-owner@vger.kernel.org List-Id: linux-omap@vger.kernel.org To: Arnd Bergmann Cc: linux-arm-kernel@lists.infradead.org, Felipe Balbi , Tony Lindgren , Russell King , Linux OMAP Mailing List --ILuaRSyQpoVaJ1HG Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Wed, Jun 03, 2015 at 10:55:27PM +0200, Arnd Bergmann wrote: > On Wednesday 03 June 2015 15:32:45 Felipe Balbi wrote: > >=20 > > Hi Tony and Russell, > >=20 > > AM43xx, even though it's a single processor A9, it still has TWD and gl= obal > > timer. I was doing some profiling with RT v4.0 and latency is 3.5x lowe= r just > > by switching from gptimer to twd/global. > >=20 > > The only problem is that currently, is_smp() check prevents me from usi= ng twd > > with AM43xx (that's why it's commented below, for testing purposes). > >=20 > > In the hopes that we can start a, hopefully, small thread around the su= bject, > > I'm sending this HACK which I used to get TWD and global timer enabled = so I > > could measure latencies with cyclictest. > >=20 > > Is it so that TWD shouldn't be available on UP integrations of ARM's Co= rtex-A > > processors ? > >=20 > >=20 >=20 > I wondered about this recently when looking at something unrelated > and noticed that the check had been introduced as part of > 904464b91eca8 ("ARM: 7655/1: smp_twd: make twd_local_timer_of_register() > no-op for nosmp"). >=20 > I suspect this was just the wrong fix at the time, and that the > real culprit is either alloc_percpu() or request_percpu_irq() > getting called too early on a machine without SMP support. >=20 > Possibly the problem is already resolved independently, if you > didn't run into it. no, no splats, nothing at all. See [1] [1] http://hastebin.com/helekubutu --=20 balbi --ILuaRSyQpoVaJ1HG Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJVb2v3AAoJEIaOsuA1yqREQdcQALRHC/6lQbZtb7xnjcU39MiC HP/kQOgTRQzP9JKUnnykRkzD3kJN8/ucc9QbBjenyDS4CZmbEAq8rxq7WxRNzl/d Tq3SPaLERQRc9NQlgXnHP9VNFKPREpF8p60n9nfag003xpZZ6eHmKLqRfT4+uQdw huU5zh8XMa02MyXxDeL2Hbmy8H0XPM6aqIqaMpQAbHwP6uIkCF+HY4L6D4dev1IV Hjo5/FmNs+bon0B8WyAY9icYeebYF7lgmqxSL2+Hp14RuAs4w6cDZqKDz5uCQQgx TPQVkyUps+P4ciKXF270/rancwe00EvfM0xbJApZ5vOtk9qLxcKFD1ZI9UDGDyCb +kO7E9aWNjuiZQQ22yVCKzZ4glYWZrCmNTM5Pqn8uDcxtlI2/FdZk8u7BrtrAiUL ztkg6jsR+BrxW/qidu/iajBeRfCSfylcD2aLRe/ulkQQmhNebKYdQllWERyAGGll IAg3Fi7PojhufXxlOSFDlL2tf4d5Py/W5jLSauZddkOnwDepGkscfKW8JSoDnaSV kRk/V3li68kRQDf52Z3FWAvw8UpHOHssWLK+KZUvjXTVKhGiCfb7qpcDNH740tY0 XHsr1x/Tg/zAG9Zi2BVtoqrHJzILWg7D1K4EElebgSh/eQIRWX3NbdSuNlZWwGWk FDEJHNSSoFhfFjS/1loU =ee6V -----END PGP SIGNATURE----- --ILuaRSyQpoVaJ1HG--