From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nikunj A Dadhania Subject: Re: [RFC PATCH v1 3/5] KVM: Add paravirt kvm_flush_tlb_others Date: Wed, 02 May 2012 14:21:14 +0530 Message-ID: <87obq7ge4d.fsf@linux.vnet.ibm.com> References: <20120427161727.27082.43096.stgit@abhimanyu> <20120427162401.27082.59387.stgit@abhimanyu> <4F9D32B4.8040002@redhat.com> <1335865176.13683.120.camel@twins> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: mingo@elte.hu, jeremy@goop.org, mtosatti@redhat.com, kvm@vger.kernel.org, x86@kernel.org, vatsa@linux.vnet.ibm.com, linux-kernel@vger.kernel.org, hpa@zytor.com To: Peter Zijlstra , Avi Kivity Return-path: In-Reply-To: <1335865176.13683.120.camel@twins> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On Tue, 01 May 2012 11:39:36 +0200, Peter Zijlstra wrote: > On Sun, 2012-04-29 at 15:23 +0300, Avi Kivity wrote: > > On 04/27/2012 07:24 PM, Nikunj A. Dadhania wrote: > > > flush_tlb_others_ipi depends on lot of statics in tlb.c. Replica= ted > > > the flush_tlb_others_ipi as kvm_flush_tlb_others to further adapt= to > > > paravirtualization. > > > > > > Use the vcpu state information inside the kvm_flush_tlb_others to > > > avoid sending ipi to pre-empted vcpus. > > > > > > * Do not send ipi's to offline vcpus and set flush_on_enter flag > >=20 > > get_user_pages_fast() depends on the IPI to hold off page table tea= rdown > > while they are locklessly walked with interrupts disabled. If a vc= pu > > were to be preempted while in this critical section, another vcpu > > tearing down page tables would go ahead and destroy them. when the > > preempted vcpu resumes it then touches the freed pages. > >=20 > > We could try to teach kvm and get_user_pages_fast() about this, but= this > > is intrusive. Another option is to replace the cpu_relax() loop wi= th > > something that sleeps and is then woken up by the TLB IPI handler i= f needed. >=20 > I think something like >=20 > select HAVE_RCU_TABLE_FREE if PARAVIRT >=20 > or somesuch is just about all it takes. > [root@krm1 linux]# grep HAVE_RCU_TABLE .config CONFIG_HAVE_RCU_TABLE_FREE=3Dy [root@krm1 linux]# make -j32 -s mm/memory.c: In function =E2=80=98tlb_remove_table_one=E2=80=99: mm/memory.c:315: error: implicit declaration of function =E2=80=98__tlb= _remove_table=E2=80=99 I suppose we need to have __tlb_remove_table. Trying to understand what needs to be done there. Regards Nikunj