From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cornelia Huck Subject: Re: [PATCH v3] KVM: remove buggy vcpu id check on vcpu creation Date: Thu, 21 Apr 2016 14:26:19 +0200 Message-ID: <20160421142619.2ba2c296.cornelia.huck@de.ibm.com> References: <146116689259.20666.15860134511726195550.stgit@bahia.huguette.org> <20160420182909.GB4044@potion> <20160421132958.0e9292d5@bahia.huguette.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Radim =?UTF-8?B?S3LEjW3DocWZ?= , Paolo Bonzini , james.hogan@imgtec.com, mingo@redhat.com, linux-mips@linux-mips.org, kvm@vger.kernel.org, linux-kernel@vger.kernel.org, qemu-ppc@nongnu.org, Paul Mackerras , David Gibson , David Hildenbrand To: Greg Kurz Return-path: Received: from e06smtp16.uk.ibm.com ([195.75.94.112]:36302 "EHLO e06smtp16.uk.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751672AbcDUM03 convert rfc822-to-8bit (ORCPT ); Thu, 21 Apr 2016 08:26:29 -0400 Received: from localhost by e06smtp16.uk.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 21 Apr 2016 13:26:27 +0100 In-Reply-To: <20160421132958.0e9292d5@bahia.huguette.org> Sender: kvm-owner@vger.kernel.org List-ID: On Thu, 21 Apr 2016 13:29:58 +0200 Greg Kurz wrote: > On Wed, 20 Apr 2016 20:29:09 +0200 > Radim Kr=C4=8Dm=C3=A1=C5=99 wrote: >=20 > > 2016-04-20 17:44+0200, Greg Kurz: > > > Commit 338c7dbadd26 ("KVM: Improve create VCPU parameter (CVE-201= 3-4587)") > > > introduced a check to prevent potential kernel memory corruption = in case > > > the vcpu id is too great. > > >=20 > > > Unfortunately this check assumes vcpu ids grow in sequence with a= common > > > difference of 1, which is wrong: archs are free to use vcpu id as= they fit. > > > For example, QEMU originated vcpu ids for PowerPC cpus running in= boot3s_hv > > > mode, can grow with a common difference of 2, 4 or 8: if KVM_MAX_= VCPUS is > > > 1024, guests may be limited down to 128 vcpus on POWER8. > > >=20 > > > This means the check does not belong here and should be moved to = some arch > > > specific function: kvm_arch_vcpu_create() looks like a good candi= date. > > >=20 > > > ARM and s390 already have such a check. > > >=20 > > > I could not spot any path in the PowerPC or common KVM code where= a vcpu > > > id is used as described in the above commit: I believe PowerPC ca= n live > > > without this check. =20 > >=20 > > The only problematic path I see is kvm_get_vcpu_by_id(), which retu= rns > > NULL for any id above KVM_MAX_VCPUS. >=20 > Oops my bad, I started to work on a 4.4 tree and I missed this check = brought > by commit c896939f7cff (KVM: use heuristic for fast VCPU lookup by id= ). >=20 > But again, I believe the check is wrong there also: the changelog jus= t mentions > this is a fastpath for the usual case where "VCPU ids match the array= index"... > why does the patch add a NULL return path if id >=3D KVM_MAX_VCPUS ? Probably because noone considered power :) >=20 > > kvm_vm_ioctl_create_vcpu() uses kvm_get_vcpu_by_id() to check for > > duplicate ids, so PowerPC could end up with many VCPUs of the same = id. > > I'm not sure what could fail, but code doesn't expect this situatio= n. > > Patching kvm_get_vcpu_by_id() is easy, though. > >=20 >=20 > Something like this ? >=20 > if (id < 0) > return NULL; > if (id < KVM_MAX_VCPUS) > vcpu =3D kvm_get_vcpu(kvm, id); >=20 > In the same patch ? >=20 > > Second issue is that Documentation/virtual/kvm/api.txt says > > 4.7 KVM_CREATE_VCPU > > [...] > > This API adds a vcpu to a virtual machine. The vcpu id is a smal= l > > integer in the range [0, max_vcpus). > >=20 >=20 > Yeah and I find the meaning of max_vcpus is unclear. >=20 > Here it is considered as a limit for vcpu id, but if you look at the = code, > KVM_MAX_VCPUS is also used as a limit for the number of vcpus: >=20 > virt/kvm/kvm_main.c: if (atomic_read(&kvm->online_vcpus) =3D=3D KV= M_MAX_VCPUS) { >=20 > > so we'd remove those two lines and change the API too. The change = would > > be somewhat backward compatible, but doesn't PowerPC use high vcpu_= id > > just because KVM is lacking an API to set DT ID? >=20 > This is related to a limitation when running in book3s_hv mode with c= pus > that support SMT (multiple HW threads): all HW threads within a core > cannot be running in different guests at the same time.=20 >=20 > We solve this by using a vcpu numbering scheme as follows: >=20 > vcpu_id[N] =3D (N * thread_per_core_guest) / threads_per_core_host + = N % threads_per_core_guest >=20 > where N means "the Nth vcpu presented to the guest". This allows to h= ave groups of vcpus > that can be scheduled to run on the same real core. >=20 > So, in the "worst" case where we want to run a guest with 1 thread/co= re and the host > has 8 threads/core, we will need the vcpu_id limit to be 8*KVM_MAX_VC= PUS. >=20 > > x86 (APIC ID) is affected by this and ARM (MP ID) probably too. > >=20 >=20 > x86 is limited to KVM_MAX_VCPUS (=3D=3D 255) vcpus: it won't be affec= ted if we also > patch kvm_get_vcpu_by_id() like suggested above. >=20 > Depending on the platform, ARM can be limited to VGIC_V3_MAX_CPUS (=3D= =3D 255) or > VGIC_V8_MAX_CPUS (=3D=3D 8). I guess it won't be affected either. =46or s390, it's either 64 (no esca) or 248 (esca). >=20 > > (Maybe it is time to decouple VCPU ID used in KVM interfaces from > > architecture dependent CPU ID that the guest uses ... >=20 > Maybe... I did not get that far. It seems that the various architectures are more different than I thought... wasn't aware of the complicated situation on power, for example.