Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753381AbcL3Joe (ORCPT ); Fri, 30 Dec 2016 04:44:34 -0500 Received: from mail-wj0-f179.google.com ([209.85.210.179]:33068 "EHLO mail-wj0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752520AbcL3Jo3 (ORCPT ); Fri, 30 Dec 2016 04:44:29 -0500 Subject: Re: [BUG] ARM64: amlogic: gxbb: unhandled level 2 translation fault (11) To: Heinrich Schuchardt , Carlo Caione , Kevin Hilman References: <20161224140028.4205-1-xypron.glpk@gmx.de> <478f1429-03f2-a944-8d47-3757a4d44a40@baylibre.com> <130a3716-a643-d14f-fd7c-9ad3bb67f3a7@gmx.de> Cc: linux-amlogic@lists.infradead.org, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org From: Neil Armstrong Organization: Baylibre Message-ID: Date: Fri, 30 Dec 2016 10:44:26 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.3.0 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5099 Lines: 119 On 12/30/2016 09:51 AM, Neil Armstrong wrote: > On 12/29/2016 10:18 PM, Heinrich Schuchardt wrote: >> On 12/29/2016 10:07 AM, Neil Armstrong wrote: >>> On 12/24/2016 03:00 PM, Heinrich Schuchardt wrote: >>>> When trying to run sddm on an Hardkernel Odroid C2 I invariably run into the >>>> translation fault below. >>>> >>>> The following mail thread relates this kind of problem to TLB (translation >>>> lookaside buffer) broadcasting. >>>> >>>> https://lkml.org/lkml/2014/4/15/207 >>>> >>>> [ 3163.014263] sddm[1851]: unhandled level 2 translation fault (11) at 0x00000160, esr 0x82000006 >>>> [ 3163.017287] pgd = ffff80007bf86000 >>>> [ 3163.020589] [00000160] *pgd=000000007a8a3003 >>>> [ 3163.024733] , *pud=000000007be9c003 >>>> [ 3163.028095] , *pmd=0000000000000000 >>>> >>>> >>>> [ 3163.033026] CPU: 1 PID: 1851 Comm: sddm Not tainted 4.9.0-next-20161212-r022-arm64 #1 >>>> [ 3163.040831] Hardware name: Hardkernel ODROID-C2 (DT) >>>> [ 3163.045698] task: ffff80007bc6d780 task.stack: ffff80007c524000 >>>> [ 3163.051563] PC is at 0x160 >>>> [ 3163.054231] LR is at 0xffff9a9fbc98 >>>> [ 3163.057686] pc : [<0000000000000160>] lr : [<0000ffff9a9fbc98>] pstate: 40000000 >>>> [ 3163.065022] sp : 0000ffffd7180130 >>>> [ 3163.068281] x29: 0000ffffd7180130 x28: 0000ffffd7180288 >>>> [ 3163.073538] x27: 0000ffff9aa94000 x26: 0000000000000001 >>>> [ 3163.078798] x25: 0000000000000000 x24: 0000ffffd7180410 >>>> [ 3163.084060] x23: 000000000e0c2190 x22: 000000000e0ca5c0 >>>> [ 3163.089322] x21: 0000ffff9ac35000 x20: 0000000000454fa9 >>>> [ 3163.094583] x19: 0000000000454fa8 x18: 000000000e0b5938 >>>> [ 3163.099843] x17: 0000ffff9a3f2988 x16: 0000ffff9ac36aa0 >>>> [ 3163.105105] x15: 0000000000000000 x14: 0000000000000000 >>>> [ 3163.110367] x13: 6d00640064007300 x12: 0800000005000000 >>>> [ 3163.115627] x11: 0000040000000000 x10: 0000a00000000000 >>>> [ 3163.120889] x9 : 00003fffffffffff x8 : 0000000000000000 >>>> [ 3163.126150] x7 : 000000000e0cb520 x6 : 0000000000454fc0 >>>> [ 3163.131412] x5 : 0000ffffd717ffd8 x4 : 000000000e0cb510 >>>> [ 3163.136680] x3 : 0000000000000004 x2 : f2f9022b551b3900 >>>> [ 3163.141935] x1 : 0000000000000160 x0 : 000000000e0ca5c0 >>>> >>>> Best regards >>>> >>>> Heinrich Schuchardt >>> >>> Hi Heinrich, >>> >>> I personally never had this issue even while loading huge applications loke LibreOffice and Gnome environment. >>> >>> I will have a look and try to reproduce this issue, can you provide us your configuration and user-space complete use case ? >>> >>> Neil >>> >> >> Hello Neil, >> >> the kernel is build with >> https://github.com/xypron/kernel-odroid-c2/tree/f8d565ff755e92fd585f5ae10123ce20abe03968 >> >> Especially look at the patch directory and config/config-next-20161212. >> >> The userland is Debian Stretch with this package: >> https://packages.debian.org/stretch/sddm >> >> The link >> https://www.spinics.net/lists/arm-kernel/msg550204.html >> that you mention in a separate mail just links to this very thread due >> to linux-arm-kernel@lists.infradead.org being in copy. >> >> Best regards >> >> Heinrich Schuchardt >> > > Hi, > > Thanks for the details, but why do you use the next-20161212 tag ? does it work with 4.10-rc1 or previous next tags ? > > Neil > Hi Heinrich, I'm able to reproduce the bug using SDMM from Ubuntu running a 4.10-rc1 patched with memory zones : [ 17.988446] sddm-greeter[2366]: unhandled level 2 translation fault (11) at 0x00000000, esr 0x92000006 [ 17.988451] pgd = ffff80003c3ee000 [ 17.988472] [00000000] *pgd=00000000398bd003 [ 17.988474] , *pud=00000000398bf003 [ 17.990477] , *pmd=0000000000000000 [ 17.990485] CPU: 3 PID: 2366 Comm: sddm-greeter Not tainted 4.10.0-rc1-00004-gd3f812382-dirty #488 [ 17.990487] Hardware name: Amlogic Meson GXBB P200 Development Board (DT) [ 17.990489] task: ffff80003c160000 task.stack: ffff80003c314000 [ 17.990493] PC is at 0xffffb2999994 [ 17.990495] LR is at 0xffffb299f774 [ 17.990497] pc : [<0000ffffb2999994>] lr : [<0000ffffb299f774>] pstate: 20000000 [ 17.990503] sp : 0000ffffd2d1d5b0 [ 17.990504] x29: 0000ffffd2d1d5b0 x28: 0000ffffad10e010 [ 17.990508] x27: 0000000009c376d0 x26: 0000000000000001 [ 17.990511] x25: 0000ffffb2b67000 x24: 0000ffffad10e010 [ 17.990514] x23: 0000000000000000 x22: 0000000000000000 [ 17.990517] x21: 0000000009c376d0 x20: 0000ffffb2fcdcb8 [ 17.990520] x19: 0000ffffb2b2b000 x18: 0000000009c78450 [ 17.990523] x17: 0000ffffb2b68068 x16: 0000ffffb2022158 [ 17.990526] x15: 0000ffffb2fcd000 x14: 0000000000000000 [ 17.990529] x13: aaaaaaaaaaaaaaab x12: 0000000000010000 [ 17.990532] x11: 0000000000000008 x10: 0000000009c3b980 [ 17.990535] x9 : 0000000009c38c40 x8 : 0000ffffd2d1d4e0 [ 17.990538] x7 : 0000000000000000 x6 : 0000000009c3b998 [ 17.990541] x5 : 0000000009c3b980 x4 : 0000800000000000 [ 17.990544] x3 : 00000000fffffff0 x2 : 0000ffffad10e010 [ 17.990547] x1 : 0000000009c379d8 x0 : 0000000000000000 Looking about other occurrence of such error, it seems it may be a issue from sdmm instead. I'll continue searching, Neil