From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37621) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YdySL-0006YJ-K5 for qemu-devel@nongnu.org; Fri, 03 Apr 2015 06:04:07 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YdySI-0001Dz-4B for qemu-devel@nongnu.org; Fri, 03 Apr 2015 06:04:05 -0400 Received: from smtp.sangfor.com.cn ([58.251.49.30]:47807 helo=mail.sangfor.com.cn) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YdySG-00016j-Bo for qemu-devel@nongnu.org; Fri, 03 Apr 2015 06:04:02 -0400 Date: Fri, 3 Apr 2015 18:04:08 +0800 From: leijian Mime-Version: 1.0 Message-ID: <201504031804076871637@sangfor.com.cn> Content-Type: multipart/alternative; boundary="----=_001_NextPart028844402688_=----" Subject: [Qemu-devel] [Snapshot Bug?]Qcow2 meta data corruption Reply-To: leijian List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel , stefanha , kwolf This is a multi-part message in MIME format. ------=_001_NextPart028844402688_=---- Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: base64 SGkgYWxsLA0KDQpUaGVyZSB3YXMgYSBwcm9ibGVtIGFib3V0IHFjb3cyIGltYWdlIGZpbGUgaGFw cGVuZWQgaW4gbXkgc2VydmFsIHZtcyBhbmQgSSBjb3VsZCBub3QgZmlndXJlIGl0IG91dCwNCnNv IGhhdmUgdG8gYXNrIGZvciBzb21lIGhlbHAuDQoNCkhlcmUgaXMgdGhlIHRoaW5nOg0KDQpBdCBm aXJzdCwgSSBmb3VuZCB0aGVyZSB3ZXJlIHNvbWUgZGF0YSBjb3JydXB0aW9uIGluIGEgdm0sIHNv IEkgZGlkIHFlbXUtaW1nIGNoZWNrIHRvIGFsbCBteSB2bXMuDQoNCnBhcnRzIG9mIGNoZWNrIHJl cG9ydDoNCg0KMy1MZWFrZWQgY2x1c3RlciAyOTI2MjI5IHJlZmNvdW50PTEgcmVmZXJlbmNlPTAN CjQtTGVha2VkIGNsdXN0ZXIgMzAyMTE4MSByZWZjb3VudD0xIHJlZmVyZW5jZT0wDQo1LUxlYWtl ZCBjbHVzdGVyIDMwMjExODIgcmVmY291bnQ9MSByZWZlcmVuY2U9MA0KNi1MZWFrZWQgY2x1c3Rl ciAzMDIxMTgzIHJlZmNvdW50PTEgcmVmZXJlbmNlPTANCjctTGVha2VkIGNsdXN0ZXIgMzAyMTE4 NCByZWZjb3VudD0xIHJlZmVyZW5jZT0wDQo4LUVSUk9SIGNsdXN0ZXIgMzEwMjU0NyByZWZjb3Vu dD0zIHJlZmVyZW5jZT00DQo5LUVSUk9SIGNsdXN0ZXIgMzExMTUzNiByZWZjb3VudD0zIHJlZmVy ZW5jZT00DQoxMC1FUlJPUiBjbHVzdGVyIDMxMTMzNjkgcmVmY291bnQ9MyByZWZlcmVuY2U9NA0K MTEtRVJST1IgY2x1c3RlciAzMjM1NTkwIHJlZmNvdW50PTEwIHJlZmVyZW5jZT0xMQ0KMTItRVJS T1IgY2x1c3RlciAzMjM1NTkxIHJlZmNvdW50PTEwIHJlZmVyZW5jZT0xMQ0KDQo0MjMtV2Fybmlu ZzogY2x1c3RlciBvZmZzZXQ9MHhjMDAwYzAwMDIwMDAwIGlzIGFmdGVyIHRoZSBlbmQgb2YgdGhl IGltYWdlIGZpbGUsIGNhbid0IHByb3Blcmx5IGNoZWNrIHJlZmNvdW50cy4NCjQyNC1XYXJuaW5n OiBjbHVzdGVyIG9mZnNldD0weGMwMDBjMDAwYzAwMDAgaXMgYWZ0ZXIgdGhlIGVuZCBvZiB0aGUg aW1hZ2UgZmlsZSwgY2FuJ3QgcHJvcGVybHkgY2hlY2sgcmVmY291bnRzLg0KNDI1LVdhcm5pbmc6 IGNsdXN0ZXIgb2Zmc2V0PTB4YzAwMDEwMDBjMDAwMCBpcyBhZnRlciB0aGUgZW5kIG9mIHRoZSBp bWFnZSBmaWxlLCBjYW4ndCBwcm9wZXJseSBjaGVjayByZWZjb3VudHMuDQo0MjYtV2FybmluZzog Y2x1c3RlciBvZmZzZXQ9MHhjMDAwYzAwMGMwMDAwIGlzIGFmdGVyIHRoZSBlbmQgb2YgdGhlIGlt YWdlIGZpbGUsIGNhbid0IHByb3Blcmx5IGNoZWNrIHJlZmNvdW50cy4NCjQyNy1XYXJuaW5nOiBj bHVzdGVyIG9mZnNldD0weGMwMDBjMDAwYzAwMDAgaXMgYWZ0ZXIgdGhlIGVuZCBvZiB0aGUgaW1h Z2UgZmlsZSwgY2FuJ3QgcHJvcGVybHkgY2hlY2sgcmVmY291bnRzLg0KNDI4LVdhcm5pbmc6IGNs dXN0ZXIgb2Zmc2V0PTB4YzAwMGMwMDBjMDAwMCBpcyBhZnRlciB0aGUgZW5kIG9mIHRoZSBpbWFn ZSBmaWxlLCBjYW4ndCBwcm9wZXJseSBjaGVjayByZWZjb3VudHMuDQo0MjktV2FybmluZzogY2x1 c3RlciBvZmZzZXQ9MHhjMDAwYzAwMGMwMDAwIGlzIGFmdGVyIHRoZSBlbmQgb2YgdGhlIGltYWdl IGZpbGUsIGNhbid0IHByb3Blcmx5IGNoZWNrIHJlZmNvdW50cy4NCjQzMC1XYXJuaW5nOiBjbHVz dGVyIG9mZnNldD0weGMwMDBjMDAwMTAwMDAgaXMgYWZ0ZXIgdGhlIGVuZCBvZiB0aGUgaW1hZ2Ug ZmlsZSwgY2FuJ3QgcHJvcGVybHkgY2hlY2sgcmVmY291bnRzLg0KDQoNCkFmdGVyIGEgZnV0aGVy IGxvb2sgaW4sIEkgZm91bmQgdHdvIGwyIGVudHJpZXMgcG9pbnQgdG8gdGhlIHNhbWUgY2x1c3Rl ciwgYW5kIHRoYXQgd2FzIGZvdW5kIGluIHNlcnZhbCBxY293MiBpbWFnZSBmaWxlcyBvZiBkaWZm ZXJlbnQgdm1zLg0KTGlrZSB0aGlzOg0KDQp0YWJsZSBlbnRyeSBjb25mbGljdCAod2l0aCBvdXIg cWNvdzIgY2hlY2sgdG9vbCk6IA0KYSB0YWJsZSBvZmZzZXQgOiAweDAwMDAwMDkzZjcwODAwMDAg bGV2ZWwgOiAyLCBsMSB0YWJsZSBlbnRyeSAxMDAsIGwyIHRhYmxlIGVudHJ5IDcNCmIgdGFibGUg b2Zmc2V0IDogMHgwMDAwMDA5M2Y3MDgwMDAwIGxldmVsIDogMiwgbDEgdGFibGUgZW50cnkgNSwg bDIgdGFibGUgZW50cnkgNw0KdGFibGUgZW50cnkgY29uZmxpY3QgOiANCmEgdGFibGUgb2Zmc2V0 IDogMHgwMDAwMDAwMGEwMWUwMDAwIGxldmVsIDogMiwgbDEgdGFibGUgZW50cnkgMTAwLCBsMiB0 YWJsZSBlbnRyeSAxOQ0KYiB0YWJsZSBvZmZzZXQgOiAweDAwMDAwMDAwYTAxZTAwMDAgbGV2ZWwg OiAyLCBsMSB0YWJsZSBlbnRyeSA1LCBsMiB0YWJsZSBlbnRyeSAxOQ0KdGFibGUgZW50cnkgY29u ZmxpY3QgOiANCmEgdGFibGUgb2Zmc2V0IDogMHgwMDAwMDAwMGEwMWQwMDAwIGxldmVsIDogMiwg bDEgdGFibGUgZW50cnkgMTAwLCBsMiB0YWJsZSBlbnRyeSAxOA0KYiB0YWJsZSBvZmZzZXQgOiAw eDAwMDAwMDAwYTAxZDAwMDAgbGV2ZWwgOiAyLCBsMSB0YWJsZSBlbnRyeSA1LCBsMiB0YWJsZSBl bnRyeSAxOA0KdGFibGUgZW50cnkgY29uZmxpY3QgOiANCmEgdGFibGUgb2Zmc2V0IDogMHgwMDAw MDAwMGEwMWMwMDAwIGxldmVsIDogMiwgbDEgdGFibGUgZW50cnkgMTAwLCBsMiB0YWJsZSBlbnRy eSAxNw0KYiB0YWJsZSBvZmZzZXQgOiAweDAwMDAwMDAwYTAxYzAwMDAgbGV2ZWwgOiAyLCBsMSB0 YWJsZSBlbnRyeSA1LCBsMiB0YWJsZSBlbnRyeSAxNw0KdGFibGUgZW50cnkgY29uZmxpY3QgOiAN CmEgdGFibGUgb2Zmc2V0IDogMHgwMDAwMDAwMGEwMWIwMDAwIGxldmVsIDogMiwgbDEgdGFibGUg ZW50cnkgMTAwLCBsMiB0YWJsZSBlbnRyeSAxNg0KYiB0YWJsZSBvZmZzZXQgOiAweDAwMDAwMDAw YTAxYjAwMDAgbGV2ZWwgOiAyLCBsMSB0YWJsZSBlbnRyeSA1LCBsMiB0YWJsZSBlbnRyeSAxNg0K DQpJIHRoaW5rIHRoZSBwcm9ibGVtIGlzIHJlbGF0ZSB0byB0aGUgc25hcHNob3QgY3JlYXRlLCBk ZWxldGUuIEJ1dCBJIGNhbnQgcmVwcm9kdWNlIGl0IC4NCkNhbiBBbnlvbmUgZ2l2ZSBhIGhpbnQg YWJvdXQgaG93IHRoaXMgaGFwcGVuPw0KDQpRZW11IHZlcnNpb24gMi4wLjEsIEkgZG93bmxvYWQg dGhlIHNvdXJjZSBjb2RlIGFuZCBtYWtlIGluc3RhbGwgaXQuDQoNClFlbXUgcGFyYW1ldGVyczoN Ci91c3IvYmluL2t2bSAtY2hhcmRldiBzb2NrZXQsaWQ9cW1wLHBhdGg9L3Zhci9ydW4vcWVtdS1z ZXJ2ZXIvNTg1NTg5OTYzOTgzOC5xbXAsc2VydmVyLG5vd2FpdCAtbW9uIGNoYXJkZXY9cW1wLG1v ZGU9Y29udHJvbCAtdm5jIDowLHdlYnNvY2tldCx0bz0yMDAgLWVuYWJsZS1rdm0gLXBpZGZpbGUg L3Zhci9ydW4vcWVtdS1zZXJ2ZXIvNTg1NTg5OTYzOTgzOC5waWQgLWRhZW1vbml6ZSAtbmFtZSB5 Zk1haWxTdnItMjAwLjIwMC4wLjE0IC1zbXAgc29ja2V0cz0xLGNvcmVzPTQgLWNwdSBjb3JlMmR1 byxodl9zcGlubG9ja3M9MHhmZmZmLGh2X3JlbGF4ZWQsaHZfdGltZSxodl92YXBpYywrc3NlNC4x LCtzc2U0LjIsK3gyYXBpYywrZXJtcywrc21lcCwrZnNnc2Jhc2UsK2YxNmMsK2RjYSwrcGNpZCwr cGRjbSwreHRwciwraHQsK3NzLCthY3BpLCtkcyAtbm9kZWZhdWx0cyAtdmdhIGNpcnJ1cyAtayBl bi11cyAtYm9vdCBtZW51PW9uLHNwbGFzaC10aW1lPTgwMDAgLW0gODE5MiAtdXNiIC1kcml2ZSBp Zj1ub25lLGlkPWRyaXZlLWlkZTAsbWVkaWE9Y2Ryb20sYWlvPW5hdGl2ZSAtZGV2aWNlIGlkZS1j ZCxidXM9aWRlLjAsdW5pdD0wLGRyaXZlPWRyaXZlLWlkZTAsaWQ9aWRlMCAtZHJpdmUgZmlsZT0v c2YvZGF0YS8zNmI4MmE3MjBkM2EyNzgwMDFiYTkwNGU4MGMyMGMxM2VfZWNmNGJiYmYzZTk0L2lt YWdlcy9ob3N0LWVjZjRiYmJmM2U5NC83ODRmM2YwODUzMmEveWZNYWlsU3ZyLTIwMC4yMDAuMC4x NC52bS92bS1kaXNrLTEucWNvdzIsaWY9bm9uZSxpZD1kcml2ZS12aXJ0aW8xLGNhY2hlPW5vbmUs YWlvPW5hdGl2ZSAtZGV2aWNlIHZpcnRpby1ibGstcGNpLGRyaXZlPWRyaXZlLXZpcnRpbzEsaWQ9 dmlydGlvMSxidXM9cGNpLjAsYWRkcj0weGIgLWRyaXZlIGZpbGU9L3NmL2RhdGEvMzZiODJhNzIw ZDNhMjc4MDAxYmE5MDRlODBjMjBjMTNlX2VjZjRiYmJmM2U5NC9pbWFnZXMvaG9zdC1lY2Y0YmJi ZjNlOTQvNzg0ZjNmMDg1MzJhL3lmTWFpbFN2ci0yMDAuMjAwLjAuMTQudm0vdm0tZGlzay0yLnFj b3cyLGlmPW5vbmUsaWQ9ZHJpdmUtdmlydGlvMixjYWNoZT1ub25lLGFpbz1uYXRpdmUgLWRldmlj ZSB2aXJ0aW8tYmxrLXBjaSxkcml2ZT1kcml2ZS12aXJ0aW8yLGlkPXZpcnRpbzIsYnVzPXBjaS4w LGFkZHI9MHhjLGJvb3RpbmRleD0xMDEgLW5ldGRldiB0eXBlPXRhcCxpZD1uZXQwLGlmbmFtZT01 ODU1ODk5NjM5ODM4MDAsc2NyaXB0PS9zZi9ldGMva3ZtL3Z0cC1icmlkZ2Usdmhvc3Q9b24sdmhv c3Rmb3JjZT1vbiAtZGV2aWNlIHZpcnRpby1uZXQtcGNpLHJvbWZpbGU9LG1hYz1GRTpGQzpGRTpG MDpBQjpCQSxuZXRkZXY9bmV0MCxidXM9cGNpLjAsYWRkcj0weDEyLGlkPW5ldDAgLXJ0YyBkcmlm dGZpeD1zbGV3LGNsb2NrPXJ0LGJhc2U9bG9jYWx0aW1lIC1nbG9iYWwga3ZtLXBpdC5sb3N0X3Rp Y2tfcG9saWN5PWRpc2NhcmQgLWdsb2JhbCBQSUlYNF9QTS5kaXNhYmxlX3MzPTEgLWdsb2JhbCBQ SUlYNF9QTS5kaXNhYmxlX3M0PTENCg0KVGhhbmtzDQpTYW5nZm9yIFZULg0KDQoNCg0KDQoNCmxl aWppYW4= ------=_001_NextPart028844402688_=---- Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: quoted-printable
Hi all,
 
There was a problem about qcow2 image&n= bsp;file happened in my serval vms and = I could not figure it out,
so have to ask for some help.
 
Here is the thing:
 
At first, I found there were some = data corruption in a vm, so I did = qemu-img check to all my vms.
 
parts of check report:
 
3-Leaked cluster 2926229 refcount=3D1 reference= =3D0
4-Leaked cluster 3021181 refcount=3D1 reference= =3D0
5-Leaked cluster 3021182 refcount=3D1 reference= =3D0
6-Leaked cluster 3021183 refcount=3D1 reference= =3D0
7-Leaked cluster 3021184 refcount=3D1 reference= =3D0
8-ERROR cluster 3102547 refcount=3D3 reference=3D= 4
9-ERROR cluster 3111536 refcount=3D3 reference=3D= 4
10-ERROR cluster 3113369 refcount=3D3 reference= =3D4
11-ERROR cluster 3235590 refcount=3D10 reference= =3D11
12-ERROR cluster 3235591 refcount=3D10 reference= =3D11
 
423-Warning: cluster offset=3D0xc000c00020000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
424-Warning: cluster offset=3D0xc000c000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
425-Warning: cluster offset=3D0xc0001000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
426-Warning: cluster offset=3D0xc000c000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
427-Warning: cluster offset=3D0xc000c000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
428-Warning: cluster offset=3D0xc000c000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
429-Warning: cluster offset=3D0xc000c000c0000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
430-Warning: cluster offset=3D0xc000c00010000 is = after the end of the image file, can't&= nbsp;properly check refcounts.
 
 
After a futher look in, I found tw= o l2 entries point to the same cluster,=  and that was found in serval qcow2&nbs= p;image files of different vms.
Like this:
 
table entry conflict (with our qcow2 check=20 tool): 
a table offset : 0x00000093f7080000 level&nb= sp;: 2, l1 table entry 100, l2 table&nb= sp;entry 7
b table offset : 0x00000093f7080000 level&nb= sp;: 2, l1 table entry 5, l2 table = ;entry 7
table entry conflict : 
a table offset : 0x00000000a01e0000 level&nb= sp;: 2, l1 table entry 100, l2 table&nb= sp;entry 19
b table offset : 0x00000000a01e0000 level&nb= sp;: 2, l1 table entry 5, l2 table = ;entry 19
table entry conflict : 
a table offset : 0x00000000a01d0000 level&nb= sp;: 2, l1 table entry 100, l2 table&nb= sp;entry 18
b table offset : 0x00000000a01d0000 level&nb= sp;: 2, l1 table entry 5, l2 table = ;entry 18
table entry conflict : 
a table offset : 0x00000000a01c0000 level&nb= sp;: 2, l1 table entry 100, l2 table&nb= sp;entry 17
b table offset : 0x00000000a01c0000 level&nb= sp;: 2, l1 table entry 5, l2 table = ;entry 17
table entry conflict : 
a table offset : 0x00000000a01b0000 level&nb= sp;: 2, l1 table entry 100, l2 table&nb= sp;entry 16
b table offset : 0x00000000a01b0000 level&nb= sp;: 2, l1 table entry 5, l2 table = ;entry 16
 
I think the problem is relate to t= he snapshot create, delete. But I cant = reproduce it .
Can Anyone give a hint about how t= his happen?
 
Qemu version 2.0.1, I download the sour= ce code and make install it.
 
Qemu parameters:
/usr/bin/kvm -chardev socket,id=3Dqmp,path=3D/var/run/qemu-= server/5855899639838.qmp,server,nowait -mon chardev=3Dqmp,mode= =3Dcontrol -vnc :0,websocket,to=3D200 -enable-kvm -pid= file /var/run/qemu-server/5855899639838.pid -daemonize -nam= e yfMailSvr-200.200.0.14 -smp sockets=3D1,cores=3D4 -c= pu core2duo,hv_spinlocks=3D0xffff,hv_relaxed,hv_time,hv_vapic,+sse4.1= ,+sse4.2,+x2apic,+erms,+smep,+fsgsbase,+f16c,+dca,+pcid,+pdcm,+xtpr,+ht,+s= s,+acpi,+ds -nodefaults -vga cirrus -k en-us = ;-boot menu=3Don,splash-time=3D8000 -m 8192 -usb = -drive if=3Dnone,id=3Ddrive-ide0,media=3Dcdrom,aio=3Dnative -dev= ice ide-cd,bus=3Dide.0,unit=3D0,drive=3Ddrive-ide0,id=3Dide0 -dr= ive file=3D/sf/data/36b82a720d3a278001ba904e80c20c13e_ecf4bbbf3e94/im= ages/host-ecf4bbbf3e94/784f3f08532a/yfMailSvr-200.200.0.14.vm/vm-disk-1.qc= ow2,if=3Dnone,id=3Ddrive-virtio1,cache=3Dnone,aio=3Dnative -device&nb= sp;virtio-blk-pci,drive=3Ddrive-virtio1,id=3Dvirtio1,bus=3Dpci.0,addr=3D0x= b -drive file=3D/sf/data/36b82a720d3a278001ba904e80c20c13e_ecf4b= bbf3e94/images/host-ecf4bbbf3e94/784f3f08532a/yfMailSvr-200.200.0.14.vm/vm= -disk-2.qcow2,if=3Dnone,id=3Ddrive-virtio2,cache=3Dnone,aio=3Dnative = -device virtio-blk-pci,drive=3Ddrive-virtio2,id=3Dvirtio2,bus=3Dpci.0= ,addr=3D0xc,bootindex=3D101 -netdev type=3Dtap,id=3Dnet0,ifname= =3D585589963983800,script=3D/sf/etc/kvm/vtp-bridge,vhost=3Don,vhostforce= =3Don -device virtio-net-pci,romfile=3D,mac=3DFE:FC:FE:F0:AB:BA,= netdev=3Dnet0,bus=3Dpci.0,addr=3D0x12,id=3Dnet0 -rtc driftfix=3D= slew,clock=3Drt,base=3Dlocaltime -global kvm-pit.lost_tick_polic= y=3Ddiscard -global PIIX4_PM.disable_s3=3D1 -global PI= IX4_PM.disable_s4=3D1
 
Thanks
Sangfor VT.
 
 
leijian
------=_001_NextPart028844402688_=------