From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:44092) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VHAPI-0002py-A6 for qemu-devel@nongnu.org; Wed, 04 Sep 2013 06:33:58 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VHAP8-0002HV-Fs for qemu-devel@nongnu.org; Wed, 04 Sep 2013 06:33:52 -0400 Date: Wed, 4 Sep 2013 18:18:00 +0800 From: xuanmao_001 References: <2013090411470490667813@163.com> <20130904080040.GA3562@dhcp-200-207.str.redhat.com> <2013090417074860959027@163.com>, <20130904094517.GC3562@dhcp-200-207.str.redhat.com> Mime-Version: 1.0 Message-ID: <2013090418180040633031@163.com> Content-Type: multipart/alternative; boundary="----=_001_NextPart350013220026_=----" Subject: Re: [Qemu-devel] question about qemu disk cache mode Reply-To: xuanmao_001 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Kevin Wolf Cc: qemu-devel , qemu-discuss This is a multi-part message in MIME format. ------=_001_NextPart350013220026_=---- Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 c28sIHRoZSBndWVzdCBkaXNrIHdyaXRlIGNhY2hlIGp1c3Qgb25seSBmb3Igbm9uZSBhbmQgd3Jp dGViYWNrIGNhY2hlIG1vZGU/DQp0aGUgY2FjaGUgYWxsb2NhdGUgb24gaG9zdCBmb3IgZGlzayBm aWxlIGltYWdlIGJ5IHFlbXU/DQoNCg0KDQoNCnh1YW5tYW9fMDAxDQoNCkZyb206IEtldmluIFdv bGYNCkRhdGU6IDIwMTMtMDktMDQgMTc6NDUNClRvOiB4dWFubWFvXzAwMQ0KQ0M6IHFlbXUtZGV2 ZWw7IHFlbXUtZGlzY3Vzcw0KU3ViamVjdDogUmU6IFJlOiBxdWVzdGlvbiBhYm91dCBxZW11IGRp c2sgY2FjaGUgbW9kZQ0KQW0gMDQuMDkuMjAxMyB1bSAxMTowNyBoYXQgeHVhbm1hb18wMDEgZ2Vz Y2hyaWViZW46DQo+IEkgdW5kZXJzdGFuZCB0aGUgcGh5c2ljYWwgZGlzayBjYWNoZSBhbmQgaG9z dCBwYWdlIGNhY2hlLg0KPiAgDQo+IEkgd2FudCB0byB0aGUgZGlmZmVyZW5jZSBiZXR3ZWVuIGd1 ZXN0IGRpc2sgd3JpdGUgY2FjaGUgYW5kIGhvc3QgcGFnZSBjYWNoZQ0KPiB0aGF0IGRlc2NyaWJl ZCB3aXRoICJDYWNoaW5nIG1vZGVzIGluIFFlbXUiIGluDQo+IA0KPiBodHRwczovL2V2ZW50cy5s aW51eGZvdW5kYXRpb24ub3JnL3NsaWRlcy8yMDExL2xpbnV4Y29uLWphcGFuL2xjajIwMTFfaGFq bm9jemkucGRmDQo+IA0KPiAgDQo+IA0KPiAgDQo+IA0KPiBnaXZlIG1lIHNvbWUgbW9yZSBpbmZv cm1hdGlvbiBwbGVhc2UsIHRoYW5rcy4NCg0KSXQgc2ltcGx5IGRlc2NyaWJlcyB3aGV0aGVyIHRo ZSBndWVzdCB3aWxsIHNlZSBhIHZvbGF0aWxlIHdyaXRlIGNhY2hlLg0KVGhpcyBpcyB0aGUgY2Fz ZSBpZiBhbnkgd3JpdGViYWNrIGNhY2hlIGlzIGludm9sdmVkIGluIHRoZSBzdGFjaywgYmUgaXQN CnRoZSBob3N0IGtlcm5lbCBwYWdlIGNhY2hlIG9yIHRoZSBob3N0IGRpc2sgd3JpdGUgY2FjaGUu IE9ubHkgaWYgdGhlDQp3aG9sZSBzdGFjayB1c2VzIHdyaXRldGhyb3VnaCBjYWNoaW5nLCB0aGUg Z3Vlc3Qgd29uJ3Qgc2VlIGEgdm9sYXRpbGUNCndyaXRlIGNhY2hlLg0KDQpLZXZpbg0KDQo+ICAN Cj4g4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB 4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB 4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB 4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB4pSB 4pSB4pSB4pSB4pSBDQo+IHh1YW5tYW9fMDAxDQo+ICANCj4gRnJvbTogS2V2aW4gV29sZg0KPiBE YXRlOiAyMDEzLTA5LTA0IDE2OjAwDQo+IFRvOiB4dWFubWFvXzAwMQ0KPiBDQzogcWVtdS1kZXZl bDsgcWVtdS1kaXNjdXNzDQo+IFN1YmplY3Q6IFJlOiBxdWVzdGlvbiBhYm91dCBxZW11IGRpc2sg Y2FjaGUgbW9kZQ0KPiBBbSAwNC4wOS4yMDEzIHVtIDA1OjQ3IGhhdCB4dWFubWFvXzAwMSBnZXNj aHJpZWJlbjoNCj4gPiBEZWFyIHFlbXVlcnM6DQo+ID4gIA0KPiA+IG15IHFlbXUta3ZtIHZlcnNp b24gaXMgMS4wLjENCj4gPiBJIHdvdWxkIGxpa2UgdG8gZmlndXJlIG91dCB0aGUgcWVtdSBkaXNr IGNhY2hlIG1vZGUuIEkgaGF2ZSB2aXNpdGVkIHRoZQ0KPiA+IHFlbXUtb3B0aW9ucy5oeA0KPiA+ IHRoZXJlIGlzIHR3byBjYWNoZSB0aGF0IEkgZGlkbid0IHVuZGVyc3RhbmQ6IHRoZSBob3N0IGNh Y2hlIHBhZ2UgYW5kIHRoZSBxZW11DQo+ID4gZGlzayB3cml0ZSBjYWNoZS4NCj4gPiAgDQo+ID4g SXMgImhvc3QgcGFnZSBjYWNoZSIgb25seSB1c2VkIGZvciByZWFkLiBhbmQgInFlbXUgZGlzayB3 cml0ZSBjYWNoZSIgdXNlZCBmb3INCj4gPiB3cml0aW5nLg0KPiA+ICANCj4gPiB3aGljaCBjYWNo ZSB0aGUgZGF0YSByZWFjaGVkIGZpcnN0PyBob3N0IHBhZ2UgY2FjaGUgb3IgcWVtdSBkaXNrIHdy aXRlIGNhY2hlPw0KPiAgDQo+IFlvdSdyZSBwcm9iYWJseSBtaXN1bmRlcnN0YW5kaW5nIHRoZSBs YXR0ZXIsIEkgYXNzdW1lIHRoYXQgd2hhdCB5b3UndmUNCj4gcmVhZCBhYm91dCBpcyB0aGUgImRp c2sgY2FjaGUiLCBub3QgYSAicWVtdSBkaXNrIGNhY2hlIi4gVGhpcyBpcyB0aGUNCj4gY2FjaGUg b24geW91ciBoYXJkd2FyZSwgdGhlIHBoeXNpY2FsIGhhcmQgZGlzay4gVGhlIGhvc3QgcGFnZSBj YWNoZSBpcw0KPiB0aGUgY2FjaGluZyB0aGF0IHRoZSBMaW51eCBrZXJuZWwgZG9lcyBmb3IgZXZl cnkgZmlsZSAodW5sZXNzIGl0J3MNCj4gYnlwYXNzZWQgd2l0aCBPX0RJUkVDVCwgd2l0aCBpcyBl eHBvc2VkIGFzIGNhY2hlPW5vbmUvZGlyZWN0c3luYyBpbg0KPiBxZW11KS4gTm9uZSBvZiB0aGlz IGlzIGltcGxlbWVudGVkIGluIG9yIHNwZWNpZmljIHRvIHFlbXUuDQo+ICANCj4gV2hlbiB5b3Ug d3JpdGUgZGF0YSwgaXQgcmVhY2hlcyB0aGUgcGFnZSBjYWNoZSBpbiB0aGUga2VybmVsIGZpcnN0 IChpZg0KPiBpdCBpcyB1c2VkIGF0IGFsbCksIGFuZCB0aGVuIHRoZSBjYWNoZSBvbiB0aGUgaGFy ZCBkaXNrLg0KPiAgDQo+IEhvcGUgdGhpcyBoZWxwcy4NCj4gIA0KPiBLZXZpbg== ------=_001_NextPart350013220026_=---- Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable =EF=BB=BF
so, the guest disk write cache just only for none and writeback cache= =20 mode?
the cache allocate on host for disk file image by qemu?
 
xuanmao_001
 
From: Kevin Wolf<= /DIV>
Date: 2013-09-04 17:45
Subject: Re: Re: question about qemu disk cache=20 mode
Am 04.09.2013 um 11:07 hat xuanmao_001 = geschrieben:
> I understand the physical disk cac= he and host page cache.
>  
> I want to the difference between&n= bsp;guest disk write cache and host page&nbs= p;cache
> that described with "Caching modes = ;in Qemu" in
> 
> https://events.linuxfoundation.org/slides/2011/linuxcon-jap= an/lcj2011_hajnoczi.pdf
> 
>  
> 
>  
> 
> give me some more information plea= se, thanks.
 
It simply describes whether the guest w= ill see a volatile write cache.
This is the case if any writeback = cache is involved in the stack, be it
the host kernel page cache or the = host disk write cache. Only if the
whole stack uses writethrough caching, the&n= bsp;guest won't see a volatile
write cache.
 
Kevin
 
>  
> =E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2=94=81=E2= =94=81
> xuanmao_001
>  
> From: Kevin Wolf
> Date: 2013-09-04 16:00
> To: xuanmao_001
> CC: qemu-devel; qemu-discuss
> Subject: Re: question about qemu d= isk cache mode
> Am 04.09.2013 um 05:47 hat xuanmao= _001 geschrieben:
> > Dear qemuers:
> >  
> > my qemu-kvm version is 1.0.1<= /DIV>
> > I would like to figure o= ut the qemu disk cache mode. I have&nbs= p;visited the
> > qemu-options.hx
> > there is two cache that = I didn't understand: the host cache page&nbs= p;and the qemu
> > disk write cache.
> >  
> > Is "host page cache" only&nbs= p;used for read. and "qemu disk write c= ache" used for
> > writing.
> >  
> > which cache the data reached&= nbsp;first? host page cache or qemu disk&nbs= p;write cache?
>  
> You're probably misunderstanding the la= tter, I assume that what you've
> read about is the "disk cache",&nb= sp;not a "qemu disk cache". This is the=
> cache on your hardware, the physic= al hard disk. The host page cache is
> the caching that the Linux kernel&= nbsp;does for every file (unless it's
> bypassed with O_DIRECT, with is ex= posed as cache=3Dnone/directsync in
> qemu). None of this is implemented=  in or specific to qemu.
>  
> When you write data, it reaches&nb= sp;the page cache in the kernel first (= if
> it is used at all), and then&= nbsp;the cache on the hard disk.
>  
> Hope this helps.
>  
> Kevin
------=_001_NextPart350013220026_=------