From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:53410) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bEtKO-0003bi-AX for qemu-devel@nongnu.org; Mon, 20 Jun 2016 03:09:01 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bEtKJ-0004a0-9l for qemu-devel@nongnu.org; Mon, 20 Jun 2016 03:08:59 -0400 Message-ID: <1466406504.24271.136.camel@kernel.crashing.org> From: Benjamin Herrenschmidt Date: Mon, 20 Jun 2016 17:08:24 +1000 In-Reply-To: <1466375718.24271.132.camel@kernel.crashing.org> References: <1465795496-15071-1-git-send-email-clg@kaod.org> <1465795496-15071-2-git-send-email-clg@kaod.org> <20160616010702.GI28087@voom.fritz.box> <20160617022731.GA19581@voom.fritz.box> <57639095.5010305@kaod.org> <576392B1.6030204@kaod.org> <5763A258.2010408@redhat.com> <5763D3EF.6060305@kaod.org> <5763D8D1.70701@redhat.com> <3258f4f3-6f5d-bb1f-9cac-2d19cac7ab53@kaod.org> <1466292910.24271.95.camel@kernel.crashing.org> <1ecfdda6-49a9-a737-80fd-a229c5a1fcf1@kaod.org> <3B48DA40-D79B-416A-84E7-4743F60C2657@suse.de> <1466374521.24271.130.camel@kernel.crashing.org> <1466375718.24271.132.camel@kernel.crashing.org> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [Qemu-ppc] [PATCH 01/10] ppc: Fix rfi/rfid/hrfi/... emulation List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?ISO-8859-1?Q?C=E9dric?= Le Goater , Alexander Graf Cc: Thomas Huth , David Gibson , qemu-ppc@nongnu.org, qemu-devel@nongnu.org On Mon, 2016-06-20 at 08:35 +1000, Benjamin Herrenschmidt wrote: >=C2=A0, > Ok so Darwin uses some of the HSPRG etc... that we don't emulate on > the 970 so I suspect it never worked. I've added them but something > else breaks, I'll dig, I have the kernel source so it shouldn't be > too hard > ;-) Well, it ended up being harder than I thought. In fact I'm not there yet ! So the Darwin VM layer dies early because HID5 isn't set properly, thus dcbz is doing 128 bytes clear instead of 32, clobbering things. It looks like Apple's kernel doesn't set the right default in HID5, so we have to either do it in openbios or in qemu. Alex, what did you find out back in the day ? Or you never bothered running a 64-bit MacOS under PR KVM ? That fixed, it dies elsewhere in something related to page faults, still digging. Cheers, Ben.