From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Date: Sun, 26 Jul 2009 17:53:22 +0000 Subject: Re: [GIT PULL] Add KVM support for Book3s_64 (PPC64) hosts v3 Message-Id: <4A6C9812.4060306@redhat.com> List-Id: References: <1248453028-49627-1-git-send-email-agraf@suse.de> In-Reply-To: <1248453028-49627-1-git-send-email-agraf@suse.de> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: kvm-ppc@vger.kernel.org On 07/26/2009 08:25 PM, Alexander Graf wrote: >>> >>>>> Perhaps we can get away with disabling 32-on-64 for the >>>>> configurations that break to start with. >>>> >>>> 32-on-64 only breaks for big endian which nobody except for this >>>> target implements. Since my PPC enabling patches for qemu change >>>> the type to uint64_t always, nobody is broken right now. >>>> >>> >>> That assumes you implement set_bit_u64(), no? >> >> No, little endian works as-is :-). > > > Let me rephrase: It shouldn't if I understand little endian correctly :). > > I haven't actually verified that 32 on 64 doesn't break. But then > again that's a rather uncommon configuration for x86 anyways, but > really common on ppc. I meant disable 32-on-64 ppc. 32-on-64 x86 is supported and works (though it is rare). Given that you say 32-on-64 ppc is common, it could only be a short term fix until we implemlement u64 bitops. Alternatively we can mangle the bitmap when copying it to userspace. -- Do not meddle in the internals of kernels, for they are subtle and quick to panic.