From mboxrd@z Thu Jan 1 00:00:00 1970 From: Xiao Guangrong Subject: Re: [RFC PATCH V1 0/6] mm: add a new option MREMAP_DUP to mmrep syscall Date: Mon, 06 Jan 2014 15:41:52 +0800 Message-ID: <52CA5E40.6040603@linux.vnet.ibm.com> References: <1368093011-4867-1-git-send-email-wenchaolinux@gmail.com> <20130509141329.GC11497@suse.de> <518C5B5E.4010706@gmail.com> <52AFE828.3010500@linux.vnet.ibm.com> <20131230202342.GA7973@amt.cnet> <943AC3BD-C4EB-4B6C-BE34-AB921938AAF0@linux.vnet.ibm.com> <20131231185328.GA22414@amt.cnet> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Cc: Stefan Hajnoczi , wenchao , Mel Gorman , linux-mm@kvack.org, Andrew Morton , hughd@google.com, walken@google.com, Alexander Viro , kirill.shutemov@linux.intel.com, Anthony Liguori , KVM To: Marcelo Tosatti Return-path: In-Reply-To: <20131231185328.GA22414@amt.cnet> Sender: owner-linux-mm@kvack.org List-Id: kvm.vger.kernel.org On 01/01/2014 02:53 AM, Marcelo Tosatti wrote: > On Tue, Dec 31, 2013 at 08:06:51PM +0800, Xiao Guangrong wrote: >> >> On Dec 31, 2013, at 4:23 AM, Marcelo Tosatti wro= te: >> >>> On Tue, Dec 17, 2013 at 01:59:04PM +0800, Xiao Guangrong wrote: >>>> >>>> CCed KVM guys. >>>> >>>> On 05/10/2013 01:11 PM, Stefan Hajnoczi wrote: >>>>> On Fri, May 10, 2013 at 4:28 AM, wenchao w= rote: >>>>>> =E4=BA=8E 2013-5-9 22:13, Mel Gorman =E5=86=99=E9=81=93: >>>>>> >>>>>>> On Thu, May 09, 2013 at 05:50:05PM +0800, wenchaolinux@gmail.com = wrote: >>>>>>>> >>>>>>>> From: Wenchao Xia >>>>>>>> >>>>>>>> This serial try to enable mremap syscall to cow some private me= mory >>>>>>>> region, >>>>>>>> just like what fork() did. As a result, user space application w= ould got >>>>>>>> a >>>>>>>> mirror of those region, and it can be used as a snapshot for fur= ther >>>>>>>> processing. >>>>>>>> >>>>>>> >>>>>>> What not just fork()? Even if the application was threaded it sho= uld be >>>>>>> managable to handle fork just for processing the private memory r= egion >>>>>>> in question. I'm having trouble figuring out what sort of applica= tion >>>>>>> would require an interface like this. >>>>>>> >>>>>> It have some troubles: parent - child communication, sometimes >>>>>> page copy. >>>>>> I'd like to snapshot qemu guest's RAM, currently solution is: >>>>>> 1) fork() >>>>>> 2) pipe guest RAM data from child to parent. >>>>>> 3) parent write down the contents. >>>>>> >>>>>> To avoid complex communication for data control, and file content >>>>>> protecting, So let parent instead of child handling the data with >>>>>> a pipe, but this brings additional copy(). I think an explicit API >>>>>> cow mapping an memory region inside one process, could avoid it, >>>>>> and faster and cow less pages, also make user space code nicer. >>>>> >>>>> A new Linux-specific API is not portable and not available on exist= ing >>>>> hosts. Since QEMU supports non-Linux host operating systems the >>>>> fork() approach is preferable. >>>>> >>>>> If you're worried about the memory copy - which should be benchmark= ed >>>>> - then vmsplice(2) can be used in the child process and splice(2) c= an >>>>> be used in the parent. It probably doesn't help though since QEMU >>>>> scans RAM pages to find all-zero pages before sending them over the >>>>> socket, and at that point the memory copy might not make much >>>>> difference. >>>>> >>>>> Perhaps other applications can use this new flag better, but for QE= MU >>>>> I think fork()'s portability is more important than the convenience= of >>>>> accessing the CoW pages in the same process. >>>> >>>> Yup, I agree with you that the new syscall sometimes is not a good s= olution. >>>> >>>> Currently, we're working on live-update[1] that will be enabled on Q= emu firstly, >>>> this feature let the guest run on the new Qemu binary smoothly witho= ut >>>> restart, it's good for us to do security-update. >>>> >>>> In this case, we need to move the guest memory on old qemu instance = to the >>>> new one, fork() can not help because we need to exec() a new instanc= e, after >>>> that all memory mapping will be destroyed. >>>> >>>> We tried to enable SPLICE_F_MOVE[2] for vmsplice() to move the memor= y without >>>> memory-copy but the performance isn't so good as we expected: it's d= ue to >>>> some limitations: the page-size, lock, message-size limitation on pi= pe, etc. >>>> Of course, we will continue to improve this, but wenchao's patch see= ms a new >>>> direction for us. >>>> >>>> To coordinate with your fork() approach, maybe we can introduce a ne= w flag >>>> for VMA, something like: VM_KEEP_ONEXEC, to tell exec() to do not de= stroy >>>> this VMA. How about this or you guy have new idea? Really appreciate= for your >>>> suggestion. >>>> >>>> [1] http://marc.info/?l=3Dqemu-devel&m=3D138597598700844&w=3D2 >>>> [2] https://lkml.org/lkml/2013/10/25/285 >>> >>> Hi, >>> >> >> Hi Marcelo, >> >> >>> What is the purpose of snapshotting guest RAM here, in the context of >>> local migration? >> >> RAM-shapshotting and local-migration are on the different ways. >> Why i asked for your guy=E2=80=99s suggestion here is beacuse i thou= ght >> they need do a same thing that moves memory from one process >> to another in a efficient way. Your idea? :) >=20 > Another possibility is to use memory that is not anonymous for guest > RAM, such as hugetlbfs or tmpfs.=20 >=20 > IIRC ksm and thp have limitations wrt tmpfs. Yes, KSM and THP are what we're concerning about. >=20 > Still curious about RAM snapshotting. Wen Chao, could you please tell it more? -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org