From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mario Smarduch Subject: stage2_unmap_vm - softlockup watchdog Date: Tue, 12 Jan 2016 17:20:12 -0800 Message-ID: <5695A64C.2010701@samsung.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from localhost (localhost [127.0.0.1]) by mm01.cs.columbia.edu (Postfix) with ESMTP id 46FB749B2B for ; Tue, 12 Jan 2016 20:16:12 -0500 (EST) Received: from mm01.cs.columbia.edu ([127.0.0.1]) by localhost (mm01.cs.columbia.edu [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9MJPMZWe+hVj for ; Tue, 12 Jan 2016 20:16:10 -0500 (EST) Received: from usmailout2.samsung.com (mailout2.w2.samsung.com [211.189.100.12]) by mm01.cs.columbia.edu (Postfix) with ESMTPS id 8C63D49B15 for ; Tue, 12 Jan 2016 20:16:10 -0500 (EST) Received: from uscpsbgm2.samsung.com (u115.gpu85.samsung.co.kr [203.254.195.115]) by mailout2.w2.samsung.com (Oracle Communications Messaging Server 7.0.5.31.0 64bit (built May 5 2014)) with ESMTP id <0O0V0035HADQD520@mailout2.w2.samsung.com> for kvmarm@lists.cs.columbia.edu; Tue, 12 Jan 2016 20:20:14 -0500 (EST) List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: kvmarm-bounces@lists.cs.columbia.edu Sender: kvmarm-bounces@lists.cs.columbia.edu To: "kvmarm@lists.cs.columbia.edu" , Christoffer Dall , Marc Zyngier Cc: "linux-arm-kernel@lists.infradead.org" List-Id: kvmarm@lists.cs.columbia.edu While running tests on foundation and fast models. stage2_unmap_vm() triggers the softlockup watchdog (on a reboot). I have not tried to reproduce on hardware or work with the thresholds (watchdog_thresh()) these could be more aggressive and usually are in more critical environments. I think that calling 'touch_softlockup_watchdog()' someplace while walking second stage tables should prevent this. On hardware huge VMs may be susceptible to this. This is only likely to happen when main memory is fragmented and 2nd stage tables are backed by pages. With huge pages only everything appears to work fine. - Mario From mboxrd@z Thu Jan 1 00:00:00 1970 From: m.smarduch@samsung.com (Mario Smarduch) Date: Tue, 12 Jan 2016 17:20:12 -0800 Subject: stage2_unmap_vm - softlockup watchdog Message-ID: <5695A64C.2010701@samsung.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org While running tests on foundation and fast models. stage2_unmap_vm() triggers the softlockup watchdog (on a reboot). I have not tried to reproduce on hardware or work with the thresholds (watchdog_thresh()) these could be more aggressive and usually are in more critical environments. I think that calling 'touch_softlockup_watchdog()' someplace while walking second stage tables should prevent this. On hardware huge VMs may be susceptible to this. This is only likely to happen when main memory is fragmented and 2nd stage tables are backed by pages. With huge pages only everything appears to work fine. - Mario