From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cuda.sgi.com (cuda1.sgi.com [192.48.157.11]) by oss.sgi.com (8.14.3/8.14.3/SuSE Linux 0.8) with ESMTP id p6644AXv065524 for ; Tue, 5 Jul 2011 23:04:10 -0500 Date: Wed, 6 Jul 2011 14:04:03 +1000 From: Dave Chinner Subject: Re: XFS internal error (memory corruption) Message-ID: <20110706040403.GW1026@dastard> References: <4E12A927.9020102@gmail.com> <20110705130932.GF1026@dastard> <4E1313C1.6020309@gmail.com> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <4E1313C1.6020309@gmail.com> List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 Sender: xfs-bounces@oss.sgi.com Errors-To: xfs-bounces@oss.sgi.com To: =?iso-8859-1?B?VPZy9ms=?= Edwin Cc: xfs-masters@oss.sgi.com, Linux Kernel Mailing List , xfs@oss.sgi.com T24gVHVlLCBKdWwgMDUsIDIwMTEgYXQgMDQ6Mzg6MDlQTSArMDMwMCwgVMO2csO2ayBFZHdpbiB3 cm90ZToKPiBPbiAwNy8wNS8yMDExIDA0OjA5IFBNLCBEYXZlIENoaW5uZXIgd3JvdGU6Cj4gPiBP biBUdWUsIEp1bCAwNSwgMjAxMSBhdCAwOTowMzoxOUFNICswMzAwLCBUw7Zyw7ZrIEVkd2luIHdy b3RlOgo+ID4+IEhpLAo+ID4+Cj4gPj4gWWVzdGVyZGF5IHdoZW4gcnVubmluZyAnc2h1dGRvd24g LVBmaCBub3cnLCBpdCBodW5nIHVzaW5nIDk5JSBDUFUgaW4gc3lzIFsqXQo+ID4+IExvb2tpbmcg YXQgdGhlIGNvbnNvbGUgdGhlcmUgd2FzIGEgbWVzc2FnZSBhYm91dCBYRlMgIkNvcnJ1cHRpb24g b2YgaW4tbWVtb3J5IGRhdGEgZGV0ZWN0ZWQiLCBhbmQgYWJvdXQgWEZTX1dBTlRfQ09SUlVQVEVE X0dPVE8uCj4gPiAKPiA+IFNvIHlvdSBoYWQgYSBidHJlZSBjb3JydXB0aW9uLgo+ID4gCj4gPj4g SGFkIHRvIHNodXRkb3duIHRoZSBtYWNoaW5lIHZpYSBTeXNSUSB1ICsgby4KPiA+Pgo+ID4+IFRv ZGF5IHdoZW4gSSBib290ZWQgSSBnb3QgdGhpcyBtZXNzYWdlOgo+ID4+IFsgICAgOS43ODY0OTRd IFhGUyAobWQxcDIpOiBNb3VudGluZyBGaWxlc3lzdGVtCj4gPj4gWyAgICA5LjkyNzU5MF0gWEZT IChtZDFwMik6IFN0YXJ0aW5nIHJlY292ZXJ5IChsb2dkZXY6IC9kZXYvZGlzay9ieS1pZC9zY3Np LVNBVEFfV0RDX1dENzQwQURGRC0wX1dELVdNQVJGMTAwNzc5Ny1wYXJ0NSkKPiA+PiBbICAgMTAu Mzg1OTQxXSBYRlM6IEludGVybmFsIGVycm9yIFhGU19XQU5UX0NPUlJVUFRFRF9HT1RPIGF0IGxp bmUgMTYzOCBvZiBmaWxlIGZzL3hmcy94ZnNfYWxsb2MuYy4gIENhbGxlciAweGZmZmZmZmZmODEy MmI4MGUKPiA+PiBbICAgMTAuMzg1OTQzXQo+ID4+IFsgICAxMC4zODYwMDddIFBpZDogMTk5MCwg Y29tbTogbW91bnQgTm90IHRhaW50ZWQgMy4wLjAtcmM1ICMxNTUKPiA+PiBbICAgMTAuMzg2MDA5 XSBDYWxsIFRyYWNlOgo+ID4+IFsgICAxMC4zODYwMTRdICBbPGZmZmZmZmZmODEyNTUxY2E+XSB4 ZnNfZXJyb3JfcmVwb3J0KzB4M2EvMHg0MAo+ID4+IFsgICAxMC4zODYwMTddICBbPGZmZmZmZmZm ODEyMmI4MGU+XSA/IHhmc19mcmVlX2V4dGVudCsweGNlLzB4MTIwCj4gPj4gWyAgIDEwLjM4NjAx OV0gIFs8ZmZmZmZmZmY4MTIyN2UwNj5dID8geGZzX2FsbG9jX2xvb2t1cF9lcSsweDE2LzB4MjAK PiA+PiBbICAgMTAuMzg2MDIxXSAgWzxmZmZmZmZmZjgxMjI4ZjRhPl0geGZzX2ZyZWVfYWdfZXh0 ZW50KzB4NmFhLzB4NzgwCj4gPj4gWyAgIDEwLjM4NjAyM10gIFs8ZmZmZmZmZmY4MTIyYjgwZT5d IHhmc19mcmVlX2V4dGVudCsweGNlLzB4MTIwCj4gPj4gWyAgIDEwLjM4NjAyNl0gIFs8ZmZmZmZm ZmY4MTI3YjBmZj5dID8ga21lbV96b25lX2FsbG9jKzB4NWYvMHhlMAo+ID4+IFsgICAxMC4zODYw MjldICBbPGZmZmZmZmZmODEyNjhlOWY+XSB4bG9nX3JlY292ZXJfcHJvY2Vzc19lZmkrMHgxNWYv MHgxYTAKPiA+PiBbICAgMTAuMzg2MDMxXSAgWzxmZmZmZmZmZjgxMjZhYjI2Pl0geGxvZ19yZWNv dmVyX3Byb2Nlc3NfZWZpcy5pc3JhLjQrMHg3Ni8weGMwCj4gPj4gWyAgIDEwLjM4NjAzM10gIFs8 ZmZmZmZmZmY4MTI2ZGU2Mj5dIHhsb2dfcmVjb3Zlcl9maW5pc2grMHgyMi8weGMwCj4gPj4gWyAg IDEwLjM4NjAzNV0gIFs8ZmZmZmZmZmY4MTI2NWFhND5dIHhmc19sb2dfbW91bnRfZmluaXNoKzB4 MjQvMHgzMAo+ID4+IFsgICAxMC4zODYwMzhdICBbPGZmZmZmZmZmODEyNzBhYWI+XSB4ZnNfbW91 bnRmcysweDQ1Yi8weDcyMAo+ID4+IFsgICAxMC4zODYwNDBdICBbPGZmZmZmZmZmODEyODg3NDE+ XSB4ZnNfZnNfZmlsbF9zdXBlcisweDFmMS8weDJlMAo+ID4+IFsgICAxMC4zODYwNDJdICBbPGZm ZmZmZmZmODExNTczYWE+XSBtb3VudF9iZGV2KzB4MWFhLzB4MWYwCj4gPj4gWyAgIDEwLjM4NjA0 NF0gIFs8ZmZmZmZmZmY4MTI4ODU1MD5dID8geGZzX3BhcnNlYXJncysweGI5MC8weGI5MAo+ID4+ IFsgICAxMC4zODYwNDZdICBbPGZmZmZmZmZmODEyODY2YjA+XSB4ZnNfZnNfbW91bnQrMHgxMC8w eDIwCj4gPj4gWyAgIDEwLjM4NjA0OF0gIFs8ZmZmZmZmZmY4MTE1N2MzZT5dIG1vdW50X2ZzKzB4 M2UvMHgxYjAKPiA+PiBbICAgMTAuMzg2MDUxXSAgWzxmZmZmZmZmZjgxMTcxODA3Pl0gdmZzX2tl cm5fbW91bnQrMHg1Ny8weGEwCj4gPj4gWyAgIDEwLjM4NjA1Ml0gIFs8ZmZmZmZmZmY4MTE3MWM0 Zj5dIGRvX2tlcm5fbW91bnQrMHg0Zi8weDEwMAo+ID4+IFsgICAxMC4zODYwNTRdICBbPGZmZmZm ZmZmODExNzMyZGM+XSBkb19tb3VudCsweDE5Yy8weDg0MAo+ID4+IFsgICAxMC4zODYwNTddICBb PGZmZmZmZmZmODExMGZhMTI+XSA/IF9fZ2V0X2ZyZWVfcGFnZXMrMHgxMi8weDUwCj4gPj4gWyAg IDEwLjM4NjA1OV0gIFs8ZmZmZmZmZmY4MTE3MmZjNT5dID8gY29weV9tb3VudF9vcHRpb25zKzB4 MzUvMHgxNzAKPiA+PiBbICAgMTAuMzg2MDYxXSAgWzxmZmZmZmZmZjgxMTczZDBiPl0gc3lzX21v dW50KzB4OGIvMHhlMAo+ID4+IFsgICAxMC4zODYwNjRdICBbPGZmZmZmZmZmODE0YzE5ZmI+XSBz eXN0ZW1fY2FsbF9mYXN0cGF0aCsweDE2LzB4MWIKPiA+PiBbICAgMTAuMzg2MDcxXSBYRlMgKG1k MXAyKTogRmFpbGVkIHRvIHJlY292ZXIgRUZJcwo+ID4+IFsgICAxMC4zODYwOTddIFhGUyAobWQx cDIpOiBsb2cgbW91bnQgZmluaXNoIGZhaWxlZAo+ID4+IFsgICAxMC40Mjg1NjJdIFhGUyAobWQx cDMpOiBNb3VudGluZyBGaWxlc3lzdGVtCj4gPj4gWyAgIDEwLjYwOTk0OV0gWEZTIChtZDFwMyk6 IEVuZGluZyBjbGVhbiBtb3VudAo+ID4+Cj4gPj4gRldJVyBJIGdvdCBhIG1lc3NhZ2UgYWJvdXQg RUZJcyB5ZXN0ZXJkYXkgdG9vLCBidXQgZXZlcnl0aGluZyBlbHNlIHdvcmtlZDoKPiA+PiBKdWwg IDQgMDk6NDI6NTQgZGViaWFuIGtlcm5lbDogWyAgIDExLjQzOTg2MV0gWEZTIChtZDFwMik6IE1v dW50aW5nIEZpbGVzeXN0ZW0KPiA+PiBKdWwgIDQgMDk6NDI6NTQgZGViaWFuIGtlcm5lbDogWyAg IDExLjU5OTgxNV0gWEZTIChtZDFwMik6IFN0YXJ0aW5nIHJlY292ZXJ5IChsb2dkZXY6IC9kZXYv ZGlzay9ieS1pZC9zY3NpLVNBVEFfV0RDX1dENzQwQURGRC0wX1dELVdNQVJGMTAwNzc5Ny1wYXJ0 NSkKPiA+PiBKdWwgIDQgMDk6NDI6NTQgZGViaWFuIGtlcm5lbDogWyAgIDExLjc4Nzk4MF0gWEZT IChtZDFwMik6IEkvTyBlcnJvciBvY2N1cnJlZDogbWV0YS1kYXRhIGRldiBtZDFwMiBibG9jayAw eDExNzkyNWE4ICAgICAgICgieGZzX3RyYW5zX3JlYWRfYnVmIikgZXJyb3IgNSBidWYgYwo+ID4+ IG91bnQgNDA5Ngo+ID4+IEp1bCAgNCAwOTo0Mjo1NCBkZWJpYW4ga2VybmVsOiBbICAgMTEuNzg4 MDQ0XSBYRlMgKG1kMXAyKTogRmFpbGVkIHRvIHJlY292ZXIgRUZJcwo+ID4+IEp1bCAgNCAwOTo0 Mjo1NCBkZWJpYW4ga2VybmVsOiBbICAgMTEuNzg4MDY1XSBYRlMgKG1kMXAyKTogbG9nIG1vdW50 IGZpbmlzaCBmYWlsZWQKPiA+PiBKdWwgIDQgMDk6NDI6NTQgZGViaWFuIGtlcm5lbDogWyAgIDEx LjgzMTA3N10gWEZTIChtZDFwMyk6IE1vdW50aW5nIEZpbGVzeXN0ZW0KPiA+PiBKdWwgIDQgMDk6 NDI6NTQgZGViaWFuIGtlcm5lbDogWyAgIDEyLjAwOTY0N10gWEZTIChtZDFwMyk6IEVuZGluZyBj bGVhbiBtb3VudAo+ID4gCj4gPiBMb29rcyBsaWtlIHlvdSBtaWdodCBoYXZlIGEgZHlpbmcgZGlz ay4gVGhhdCdzIGEgSU8gZXJyb3Igb24gcmVhZAo+ID4gdGhhdCBoYXMgYmVlbiByZXBvcnRlZCBi YWNrIHRvIFhGUywgYW5kIGl0IHdhcm5lZCB0aGF0IGJhZCB0aGluZ3MKPiA+IGhhcHBlbmVkLiBN YXliZSBYRlMgc2hvdWxkIGhhdmUgc2h1dCBkb3duLCB0aG91Z2guCj4gCj4gQWggSSBoYWQgL2Rl di9zZGQga2lja2VkIG91dCBvZiB0aGUgUkFJRCBhcnJheSwgYW5kIGEgbG90IG9mIFNBVEEgRUgg cmVzZXRzLgo+IFRob3NlIHdlcmUgYXBwYXJlbnRseSBjYXVzZWQgYnkgYSBsb29zZSBTQVRBIGNh YmxlOgo+IGh0dHBzOi8vbGttbC5vcmcvbGttbC8yMDExLzcvNS8yMgo+IFRob3NlIGVycm9ycyB3 ZXJlIGZvciAvZGV2L3NkZCB0aG91Z2gsIG5vdCAvZGV2L3NkaSAodGhlIFhGUyBsb2cgZGlzayku CgpTdGlsbCwgSU8gZXJyb3JzIGluIHRoZSBmaWxlc3lzdGVtIGFyZSBpbXBvcnRhbnQgdG8gcmVw b3J0IHdoZW4KeW91J3ZlIGdvdCBhIGNvcnJ1cHRpb24gb2NjdXJyaW5nLgoKPiBPbmNlIEkgcmUt cGx1Z2dlZCBhbGwgdGhlIFNBVEEgY2FibGVzLCBhbmQgcmUtYWRkZWQgL2Rldi9zZGQgdG8gdGhl IFJBSUQgYXJyYXksIGFsbCBzZWVtZWQgd2VsbDoKPiAKPiBKdWwgIDQgMDk6NTk6MjIgZGViaWFu IGtlcm5lbDogWyAgIDExLjYxMzA3M10gWEZTIChtZDFwMik6IE1vdW50aW5nIEZpbGVzeXN0ZW0K PiBKdWwgIDQgMDk6NTk6MjIgZGViaWFuIGtlcm5lbDogWyAgIDExLjc4MjA1MV0gWEZTIChtZDFw Mik6IFN0YXJ0aW5nIHJlY292ZXJ5IChsb2dkZXY6IC9kZXYvZGlzay9ieS1pZC9zY3NpLVNBVEFf V0RDX1dENzQwQURGRC0wX1dELVdNQVJGMTAwNzc5Ny1wYXJ0NSkKPiBKdWwgIDQgMDk6NTk6MjIg ZGViaWFuIGtlcm5lbDogWyAgIDEyLjAyMDc2Nl0gWEZTIChtZDFwMik6IEVuZGluZyByZWNvdmVy eSAobG9nZGV2OiAvZGV2L2Rpc2svYnktaWQvc2NzaS1TQVRBX1dEQ19XRDc0MEFERkQtMF9XRC1X TUFSRjEwMDc3OTctcGFydDUpCj4gSnVsICA0IDA5OjU5OjIyIGRlYmlhbiBrZXJuZWw6IFsgICAx Mi4wMzEwMzhdIFhGUyAobWQxcDMpOiBNb3VudGluZyBGaWxlc3lzdGVtCj4gSnVsICA0IDA5OjU5 OjIyIGRlYmlhbiBrZXJuZWw6IFsgICAxMi4xOTYyMzhdIFhGUyAobWQxcDMpOiBFbmRpbmcgY2xl YW4gbW91bnQKPiAKPiBCdXQgd2hlbiBJIHNodXRkb3duIHRoZSBzeXN0ZW0gbGF0ZXIgdGhhdCBk YXkgSSBnb3QgdGhhdCBsb2cgY29ycnVwdGlvbiBlcnJvci4KPiBDb3VsZCBpdCBiZSB0aGF0IHRo ZSBsb2cgIC8gc29tZSBkYXRhIG9uIHRoZSBkaXNrIHdhcyBzdGlsbCBjb3JydXB0ZWQgKGRlc3Bp dGUgdGhlIHN1Y2Nlc3NmdWwgbW91bnQsIGFuZCBSQUlEIHJlc3luYyksIHdoaWNoIGNhdXNlZCB0 aGUKPiBsb2cgZXJyb3IgbGF0ZXI/CgpNb3N0IGxpa2VseSB0aGUgZnJlZSBzcGFjZSBidHJlZSB3 YXMgY29ycnVwdGVkIGJ5IHdoYXRldmVyIFJBSUQKcHJvYmxlbSB5b3UgaGFkLiBpLmUuIGl0IGRp ZG4ndCByZWNvdmVyIGNsZWFubHkuCgo+IFRoZSBsb2cgaXMgbm90IG9uIHRoZSBSQUlEIGFycmF5 IHRob3VnaCwgaXQgaXMgb24gYSBzZXBhcmF0ZSBkaXNrLAo+IGFuZCB0aGF0IGRpc2sgaXMgaW5k ZWVkIG9sZGVyIHRoYW4gYWxsIHRoZSBvdGhlciBkaXNrcy4KCkxvc2UgdGhlIGxvZywgY29ycnVw dCB5b3VyIGZpbGVzeXN0ZW0uIGkuZS4gdGhlIGV4dGVybmFsIGxvZyBuZWVkcwp0byBoYXZlIOOB n2hlIHNhbWUgcmVkdW5kYW5jeSBhcyB0aGUgcmVzdCBvZiB0aGUgZmlsZXN5c3RlbS4uLgoKPiBT dGlsbAo+IFNNQVJUIGRvZXNuJ3Qgc2hvdyBhbnkgZXJyb3JzLCBhbmQgSSd2ZSBydW4gYSBTTUFS VCBzaG9ydAo+IHNlbGYtdGVzdCwgYW5kIGNvbnZleWFuY2Ugc2VsZi10ZXN0LCBhbmQgdGhvc2Ug ZGlkbid0IGxvZyBhbnkKPiBlcnJvcnMgZWl0aGVyLgoKU01BUlQgaXNuJ3QgdmVyeSBzbWFydCBh Ym91dCByZXBvcnRpbmcgZXJyb3JzLiBBbmQgaXQgY2FuJ3QgcmVwb3J0CmVycm9ycyBjYXVzZWQg YnkgYmFkIGNhYmxlcy4uLi4KCj4gPiBPYnZpb3VzbHkgLSB5b3UndmUgZ290IGNvcnJ1cHRlZCBm cmVlIHNwYWNlIGJ0cmVlcyB0aGFua3MgdG8gdGhlIElPCj4gPiBlcnJvciBkdXJpbmcgcmVjb3Zl cnkgYW5kIHRoZSBsYXRlciBvcGVyYXRpb25zIHRoYXQgd2VyZSBkb25lIG9uIGl0Lgo+ID4gTm93 IGxvZyByZWNvdmVyeSBjYW4ndCBjb21wbGV0ZSB3aXRob3V0IGhpdHRpbmcgdGhvc2UgY29ycnVw dGlvbnMuCj4gCj4gSXMgdGhlIGNvcnJ1cHRpb24gaW4gdGhlIGxvZywgb3IgaW4gdGhlIEZTIGl0 c2VsZj8KPiBUaGUgRlMgaXRzZWxmIGNvdWxkJ3ZlIGJlZW4gZGFtYWdlZCB3aGVuIGtpY2tlZCBv dXQgb2YgdGhlIFJBSUQuCgpUaGUgRlMgaXRzZWxmLgoKPiBBbHRob3VnaCBpcyB0aGVyZSBzdXBw b3NlZCB0byBiZSBhIHBlcmZvcm1hbmNlIGJlbmVmaXQgZnJvbSBoYXZpbmcKPiBhIHNlcGFyYXRl IGxvZyBkaXNrIHdpdGggWEZTPwoKVGhlcmUgdXNlZCB0byBiZS4gTm93IGV2ZXJ5b25lIGp1c3Qg dXNlcyBkZWxheWVkIGxvZ2dpbmcsIHdoaWNoIGlzCmZhciBmYXN0ZXIgYW5kIG1vcmUgc2NhbGFi bGUgdGhhdCBldmVuIHVzaW5nIGFuIGV4dGVybmFsIGxvZy4KCj4gSUlSQyBpdCBoYXMgYSBkaXNh ZHZhbnRhZ2UgdGhhdCB5b3UgY2FuJ3QgdXNlIGJhcnJpZXJzIHByb3Blcmx5LgoKVGhhdCBtb3N0 bHkgd29ya3Mgbm93IChyZWNlbnQga2VybmVscyksIGJ1dCB5b3UgdGFrZSBhIGhpdCBpbgpqb3Vy bmFsIElPIHdhaXRpbmcgc3luY2hyb25vdXNseSBmb3IgdGhlIGRhdGEgZGV2aWNlIGNhY2hlcyB0 byBiZQpmbHVzaGVkIGJlZm9yZSB3cml0aW5nIHRvIHRoZSBsb2cgZGV2aWNlLgoKPiBCdXQgaWYg SSdkIG1vdmUgdGhlIGxvZyB0byBiZSBzdG9yZWQgd2l0aCB0aGUgRlMgKG9uIHRoZSBSQUlEMTAK PiBhcnJheSksIHdpbGwgWEZTIGJlIGFibGUgdG8gdXNlIGJhcnJpZXJzIHRoZXJlLCBvciBSQUlE IHN0aWxsCj4gcHJldmVudHMgYmFycmllcnMgZnJvbSB3b3JraW5nPwoKUkFJRCBkb2VzIG5vdCBw cmV2ZW50IGJhcnJpZXJzIGZyb20gd29ya2luZyBhbnkgbW9yZS4gVGhleSBhcmUganVzdApyZWFs bHkgc2xvdyBvbiBzb2Z0d2FyZSBSQUlENS82Li4uLgoKQ2hlZXJzLAoKRGF2ZS4KLS0gCkRhdmUg Q2hpbm5lcgpkYXZpZEBmcm9tb3JiaXQuY29tCgpfX19fX19fX19fX19fX19fX19fX19fX19fX19f X19fX19fX19fX19fX19fX19fXwp4ZnMgbWFpbGluZyBsaXN0Cnhmc0Bvc3Muc2dpLmNvbQpodHRw Oi8vb3NzLnNnaS5jb20vbWFpbG1hbi9saXN0aW5mby94ZnMK From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750932Ab1GFEEK (ORCPT ); Wed, 6 Jul 2011 00:04:10 -0400 Received: from ipmail06.adl2.internode.on.net ([150.101.137.129]:7749 "EHLO ipmail06.adl2.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750735Ab1GFEEI (ORCPT ); Wed, 6 Jul 2011 00:04:08 -0400 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: AjsFAJPbE055LCkB/2dsb2JhbABTg31Fo0t4iHqyTpEFDoEdg3+BDASifg Date: Wed, 6 Jul 2011 14:04:03 +1000 From: Dave Chinner To: =?iso-8859-1?B?VPZy9ms=?= Edwin Cc: xfs-masters@oss.sgi.com, xfs@oss.sgi.com, Linux Kernel Mailing List Subject: Re: XFS internal error (memory corruption) Message-ID: <20110706040403.GW1026@dastard> References: <4E12A927.9020102@gmail.com> <20110705130932.GF1026@dastard> <4E1313C1.6020309@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <4E1313C1.6020309@gmail.com> User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Jul 05, 2011 at 04:38:09PM +0300, Török Edwin wrote: > On 07/05/2011 04:09 PM, Dave Chinner wrote: > > On Tue, Jul 05, 2011 at 09:03:19AM +0300, Török Edwin wrote: > >> Hi, > >> > >> Yesterday when running 'shutdown -Pfh now', it hung using 99% CPU in sys [*] > >> Looking at the console there was a message about XFS "Corruption of in-memory data detected", and about XFS_WANT_CORRUPTED_GOTO. > > > > So you had a btree corruption. > > > >> Had to shutdown the machine via SysRQ u + o. > >> > >> Today when I booted I got this message: > >> [ 9.786494] XFS (md1p2): Mounting Filesystem > >> [ 9.927590] XFS (md1p2): Starting recovery (logdev: /dev/disk/by-id/scsi-SATA_WDC_WD740ADFD-0_WD-WMARF1007797-part5) > >> [ 10.385941] XFS: Internal error XFS_WANT_CORRUPTED_GOTO at line 1638 of file fs/xfs/xfs_alloc.c. Caller 0xffffffff8122b80e > >> [ 10.385943] > >> [ 10.386007] Pid: 1990, comm: mount Not tainted 3.0.0-rc5 #155 > >> [ 10.386009] Call Trace: > >> [ 10.386014] [] xfs_error_report+0x3a/0x40 > >> [ 10.386017] [] ? xfs_free_extent+0xce/0x120 > >> [ 10.386019] [] ? xfs_alloc_lookup_eq+0x16/0x20 > >> [ 10.386021] [] xfs_free_ag_extent+0x6aa/0x780 > >> [ 10.386023] [] xfs_free_extent+0xce/0x120 > >> [ 10.386026] [] ? kmem_zone_alloc+0x5f/0xe0 > >> [ 10.386029] [] xlog_recover_process_efi+0x15f/0x1a0 > >> [ 10.386031] [] xlog_recover_process_efis.isra.4+0x76/0xc0 > >> [ 10.386033] [] xlog_recover_finish+0x22/0xc0 > >> [ 10.386035] [] xfs_log_mount_finish+0x24/0x30 > >> [ 10.386038] [] xfs_mountfs+0x45b/0x720 > >> [ 10.386040] [] xfs_fs_fill_super+0x1f1/0x2e0 > >> [ 10.386042] [] mount_bdev+0x1aa/0x1f0 > >> [ 10.386044] [] ? xfs_parseargs+0xb90/0xb90 > >> [ 10.386046] [] xfs_fs_mount+0x10/0x20 > >> [ 10.386048] [] mount_fs+0x3e/0x1b0 > >> [ 10.386051] [] vfs_kern_mount+0x57/0xa0 > >> [ 10.386052] [] do_kern_mount+0x4f/0x100 > >> [ 10.386054] [] do_mount+0x19c/0x840 > >> [ 10.386057] [] ? __get_free_pages+0x12/0x50 > >> [ 10.386059] [] ? copy_mount_options+0x35/0x170 > >> [ 10.386061] [] sys_mount+0x8b/0xe0 > >> [ 10.386064] [] system_call_fastpath+0x16/0x1b > >> [ 10.386071] XFS (md1p2): Failed to recover EFIs > >> [ 10.386097] XFS (md1p2): log mount finish failed > >> [ 10.428562] XFS (md1p3): Mounting Filesystem > >> [ 10.609949] XFS (md1p3): Ending clean mount > >> > >> FWIW I got a message about EFIs yesterday too, but everything else worked: > >> Jul 4 09:42:54 debian kernel: [ 11.439861] XFS (md1p2): Mounting Filesystem > >> Jul 4 09:42:54 debian kernel: [ 11.599815] XFS (md1p2): Starting recovery (logdev: /dev/disk/by-id/scsi-SATA_WDC_WD740ADFD-0_WD-WMARF1007797-part5) > >> Jul 4 09:42:54 debian kernel: [ 11.787980] XFS (md1p2): I/O error occurred: meta-data dev md1p2 block 0x117925a8 ("xfs_trans_read_buf") error 5 buf c > >> ount 4096 > >> Jul 4 09:42:54 debian kernel: [ 11.788044] XFS (md1p2): Failed to recover EFIs > >> Jul 4 09:42:54 debian kernel: [ 11.788065] XFS (md1p2): log mount finish failed > >> Jul 4 09:42:54 debian kernel: [ 11.831077] XFS (md1p3): Mounting Filesystem > >> Jul 4 09:42:54 debian kernel: [ 12.009647] XFS (md1p3): Ending clean mount > > > > Looks like you might have a dying disk. That's a IO error on read > > that has been reported back to XFS, and it warned that bad things > > happened. Maybe XFS should have shut down, though. > > Ah I had /dev/sdd kicked out of the RAID array, and a lot of SATA EH resets. > Those were apparently caused by a loose SATA cable: > https://lkml.org/lkml/2011/7/5/22 > Those errors were for /dev/sdd though, not /dev/sdi (the XFS log disk). Still, IO errors in the filesystem are important to report when you've got a corruption occurring. > Once I re-plugged all the SATA cables, and re-added /dev/sdd to the RAID array, all seemed well: > > Jul 4 09:59:22 debian kernel: [ 11.613073] XFS (md1p2): Mounting Filesystem > Jul 4 09:59:22 debian kernel: [ 11.782051] XFS (md1p2): Starting recovery (logdev: /dev/disk/by-id/scsi-SATA_WDC_WD740ADFD-0_WD-WMARF1007797-part5) > Jul 4 09:59:22 debian kernel: [ 12.020766] XFS (md1p2): Ending recovery (logdev: /dev/disk/by-id/scsi-SATA_WDC_WD740ADFD-0_WD-WMARF1007797-part5) > Jul 4 09:59:22 debian kernel: [ 12.031038] XFS (md1p3): Mounting Filesystem > Jul 4 09:59:22 debian kernel: [ 12.196238] XFS (md1p3): Ending clean mount > > But when I shutdown the system later that day I got that log corruption error. > Could it be that the log / some data on the disk was still corrupted (despite the successful mount, and RAID resync), which caused the > log error later? Most likely the free space btree was corrupted by whatever RAID problem you had. i.e. it didn't recover cleanly. > The log is not on the RAID array though, it is on a separate disk, > and that disk is indeed older than all the other disks. Lose the log, corrupt your filesystem. i.e. the external log needs to have たhe same redundancy as the rest of the filesystem... > Still > SMART doesn't show any errors, and I've run a SMART short > self-test, and conveyance self-test, and those didn't log any > errors either. SMART isn't very smart about reporting errors. And it can't report errors caused by bad cables.... > > Obviously - you've got corrupted free space btrees thanks to the IO > > error during recovery and the later operations that were done on it. > > Now log recovery can't complete without hitting those corruptions. > > Is the corruption in the log, or in the FS itself? > The FS itself could've been damaged when kicked out of the RAID. The FS itself. > Although is there supposed to be a performance benefit from having > a separate log disk with XFS? There used to be. Now everyone just uses delayed logging, which is far faster and more scalable that even using an external log. > IIRC it has a disadvantage that you can't use barriers properly. That mostly works now (recent kernels), but you take a hit in journal IO waiting synchronously for the data device caches to be flushed before writing to the log device. > But if I'd move the log to be stored with the FS (on the RAID10 > array), will XFS be able to use barriers there, or RAID still > prevents barriers from working? RAID does not prevent barriers from working any more. They are just really slow on software RAID5/6.... Cheers, Dave. -- Dave Chinner david@fromorbit.com