From owner-freebsd-arm@FreeBSD.ORG Thu Apr 26 12:43:50 2012 Return-Path: Delivered-To: freebsd-arm@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 78D9A106564A for ; Thu, 26 Apr 2012 12:43:50 +0000 (UTC) (envelope-from kp@sigsegv.be) Received: from mercury.codepro.be (mercury.codepro.be [IPv6:2001:4b98:dc0:51:216:3eff:feb7:3147]) by mx1.freebsd.org (Postfix) with ESMTP id 070108FC08 for ; Thu, 26 Apr 2012 12:43:50 +0000 (UTC) Received: from adrastea.jupiter.sigsegv.be (adrastea.jupiter.sigsegv.be [IPv6:2001:6f8:1498:1::3]) by mercury.codepro.be (Postfix) with ESMTP id E6F5BAD; Thu, 26 Apr 2012 14:43:48 +0200 (CEST) Received: from thebe.jupiter.sigsegv.be (thebe.jupiter.sigsegv.be [172.16.1.5]) by adrastea.jupiter.sigsegv.be (Postfix) with ESMTP id 710081FC6; Thu, 26 Apr 2012 14:43:48 +0200 (CEST) Received: by thebe.jupiter.sigsegv.be (Postfix, from userid 1000) id 51809200D1; Thu, 26 Apr 2012 14:43:48 +0200 (CEST) Date: Thu, 26 Apr 2012 14:43:48 +0200 From: Kristof Provost To: Ian Lepore Message-ID: <20120426124347.GL46643@thebe.jupiter.sigsegv.be> References: <201204250920.q3P9KDDc093458@freefall.freebsd.org> <1335364474.66865.36.camel@revolution.hippie.lan> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1335364474.66865.36.camel@revolution.hippie.lan> X-PGP-Fingerprint: E114 D9EA 909E D469 8F57 17A5 7D15 91C6 9EFA F286 User-Agent: Mutt/1.5.21 (2010-09-15) Cc: freebsd-arm@freebsd.org Subject: Re: arm/158950: arm/sheevaplug fails fsx when mmap operations are enabled X-BeenThere: freebsd-arm@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Porting FreeBSD to the StrongARM Processor List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Apr 2012 12:43:50 -0000 On 2012-04-25 08:34:34 (-0600), Ian Lepore wrote: > Whenever I run into IO corruption that comes in 32-byte chunks I > immediately suspect trouble in the busdma cacheline flush code. A > useful first step in narrowing that down is to change the data cache > from write-back to write-through mode and see if the symptoms change. I > think for a sheeva cpu you'd do that in pmap_pte_init_arm10() in > sys/arm/arm/pmap.c. Change the xxx_cache_mode vars to be the same as > they are in the pmap_pte_init_arm9() function just above it. > I've had to disable it in pmap_pte_init_generic() rather than pmap_pte_init_arm10(), but that does seem to prevent the corruption. Regards, Kristof