From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ryan Harper Subject: Re: [Qemu-devel] [PATCH] Fix segfault with ram_size > 4095M without kvm Date: Thu, 4 Mar 2010 15:34:34 -0600 Message-ID: <20100304213434.GA17350@us.ibm.com> References: <20100223151314.GN17350@us.ibm.com> <4B840A17.9050809@aurel32.net> <20100304212724.GH5860@hall.aurel32.net> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Ryan Harper , qemu-devel@nongnu.org, kvm@vger.kernel.org, Anthony Liguori To: Aurelien Jarno Return-path: Received: from e5.ny.us.ibm.com ([32.97.182.145]:46883 "EHLO e5.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752824Ab0CDVel (ORCPT ); Thu, 4 Mar 2010 16:34:41 -0500 Received: from d01relay03.pok.ibm.com (d01relay03.pok.ibm.com [9.56.227.235]) by e5.ny.us.ibm.com (8.14.3/8.13.1) with ESMTP id o24LKlkM025148 for ; Thu, 4 Mar 2010 16:20:47 -0500 Received: from d01av03.pok.ibm.com (d01av03.pok.ibm.com [9.56.224.217]) by d01relay03.pok.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id o24LYemg152992 for ; Thu, 4 Mar 2010 16:34:40 -0500 Received: from d01av03.pok.ibm.com (loopback [127.0.0.1]) by d01av03.pok.ibm.com (8.14.3/8.13.1/NCO v10.0 AVout) with ESMTP id o24LYdKN018055 for ; Thu, 4 Mar 2010 18:34:40 -0300 Content-Disposition: inline In-Reply-To: <20100304212724.GH5860@hall.aurel32.net> Sender: kvm-owner@vger.kernel.org List-ID: * Aurelien Jarno [2010-03-04 15:27]: > On Tue, Feb 23, 2010 at 06:02:15PM +0100, Aurelien Jarno wrote: > > Ryan Harper a =E9crit : > > > Currently, x86_64-softmmu qemu segfaults when trying to use > 409= 5M memsize. > > > This patch adds a simple check and error message (much like the 2= 047 limit on > > > 32-bit hosts) on ram_size in the control path after we determine = we're > > > not using kvm > > >=20 > > > Upstream qemu-kvm is affected if using the -no-kvm option; this p= atch address > > > the segfault there as well. > >=20 > > It looks like workarounding the real bug. At some point both > > i386-softmmu (via PAE) and x86_64-softmmu were able to support > 4G= B of > > memory. I remember adding the support long time ago, and testing it= with > > 32GB of emulated RAM. >=20 > I have looked into that, and actually one patch to get full support f= or > > 4GB of memory was not merged: Thanks for looking into this. >=20 > diff --git a/exec.c b/exec.c > index 8389c54..b0bb058 100644 > --- a/exec.c > +++ b/exec.c > @@ -166,7 +166,7 @@ typedef struct PhysPageDesc { > */ > #define L1_BITS (TARGET_VIRT_ADDR_SPACE_BITS - L2_BITS - TARGET_PAGE= _BITS) > #else > -#define L1_BITS (32 - L2_BITS - TARGET_PAGE_BITS) > +#define L1_BITS (TARGET_PHYS_ADDR_SPACE_BITS - L2_BITS - TARGET_PAGE= _BITS) > #endif >=20 > #define L1_SIZE (1 << L1_BITS) >=20 > While this patch is acceptable for qemu i386, it creates a big L1 tab= le > for x86_64 or other 64-bit architectures, resulting in huge memory=20 > overhead. >=20 > The recent multilevel tables patches from Richard Henderson should fi= x=20 > the problem for HEAD (I haven't found time to look at them in details= ). >=20 > As this is not something we really want to backport, your patch makes > sense in stable-0.12. Anthony, do you want me to resend and rebase against 0.12-stable? --=20 Ryan Harper Software Engineer; Linux Technology Center IBM Corp., Austin, Tx ryanh@us.ibm.com