https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-5062 Lustre <p>Cray has recently begun stress testing master clients in preparation for 2.6.0.</p> <p>We hit this bug on several nodes during a stress run last night:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java">2014-05-13T23:00:06.485598-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) page@ffff880221781c00[4 ffff880222618600 3 0 1 (<span class="code-keyword">null</span>) ffff88022328d740 0x0] 2014-05-13T23:00:06.485630-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) vvp-page@ffff880221781ca0(0:0:0) vm@ffffea000e1502a0 200000000000801 4:0 ffff880221781c00 0 lru 2014-05-13T23:00:06.485637-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) lov-page@ffff880221781cf8, raid0 2014-05-13T23:00:06.485672-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) osc-page@ffff880221781d60 0: 1&lt; 0x845fed 1 0 + + &gt; 2&lt; 0 0 4096 0x7 0x8 | (<span class="code-keyword">null</span>) ffff88044ad844f8 ffff88044d7a59e8 &gt; 3&lt; + ffff88044e055080 1 16 0 &gt; 4&lt; 0 0 8 35586048 - | - - - + &gt; 5&lt; - - - + | 0 - | 0 - -&gt; 2014-05-13T23:00:06.485681-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) end page@ffff880221781c00 2014-05-13T23:00:06.485689-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:824:osc_req_attr_set()) dump uncover page! 2014-05-13T23:00:06.485700-05:00 c1-0c0s2n1 Pid: 14298, comm: nsystst 2014-05-13T23:00:06.485710-05:00 c1-0c0s2n1 Call Trace: 2014-05-13T23:00:06.485724-05:00 c1-0c0s2n1 [&lt;ffffffff81005eb9&gt;] try_stack_unwind+0x169/0x1b0 2014-05-13T23:00:06.485746-05:00 c1-0c0s2n1 [&lt;ffffffff81004919&gt;] dump_trace+0x89/0x450 2014-05-13T23:00:06.485757-05:00 c1-0c0s2n1 [&lt;ffffffffa023b8d7&gt;] libcfs_debug_dumpstack+0x57/0x80 [libcfs] 2014-05-13T23:00:06.485784-05:00 c1-0c0s2n1 [&lt;ffffffffa092f348&gt;] osc_req_attr_set+0x6b8/0x740 [osc] 2014-05-13T23:00:06.485796-05:00 c1-0c0s2n1 [&lt;ffffffffa038dbf1&gt;] cl_req_attr_set+0xd1/0x220 [obdclass] 2014-05-13T23:00:06.485807-05:00 c1-0c0s2n1 [&lt;ffffffffa091a6da&gt;] osc_build_rpc+0x4fa/0x15d0 [osc] 2014-05-13T23:00:06.485817-05:00 c1-0c0s2n1 [&lt;ffffffffa093c199&gt;] osc_io_unplug0+0x1419/0x2020 [osc] 2014-05-13T23:00:06.485837-05:00 c1-0c0s2n1 [&lt;ffffffffa093ddb1&gt;] osc_io_unplug+0x11/0x20 [osc] 2014-05-13T23:00:06.485844-05:00 c1-0c0s2n1 [&lt;ffffffffa093dfa1&gt;] osc_queue_sync_pages+0x1e1/0x380 [osc] 2014-05-13T23:00:06.485852-05:00 c1-0c0s2n1 [&lt;ffffffffa092f90d&gt;] osc_io_submit+0x2dd/0x4e0 [osc] 2014-05-13T23:00:06.485871-05:00 c1-0c0s2n1 [&lt;ffffffffa038d40e&gt;] cl_io_submit_rw+0x6e/0x170 [obdclass] 2014-05-13T23:00:06.485881-05:00 c1-0c0s2n1 [&lt;ffffffffa076123e&gt;] lov_io_submit+0x2ee/0x4d0 [lov] 2014-05-13T23:00:06.485894-05:00 c1-0c0s2n1 [&lt;ffffffffa038d40e&gt;] cl_io_submit_rw+0x6e/0x170 [obdclass] 2014-05-13T23:00:06.485900-05:00 c1-0c0s2n1 [&lt;ffffffffa038f3fa&gt;] cl_io_read_page+0x19a/0x1b0 [obdclass] 2014-05-13T23:00:06.485907-05:00 c1-0c0s2n1 [&lt;ffffffffa081bf7c&gt;] ll_readpage+0x16c/0x1b0 [lustre] 2014-05-13T23:00:06.485920-05:00 c1-0c0s2n1 [&lt;ffffffff810fdc68&gt;] generic_file_aio_read+0x268/0x740 2014-05-13T23:00:06.485932-05:00 c1-0c0s2n1 [&lt;ffffffffa084e0b0&gt;] vvp_io_read_start+0x2a0/0x430 [lustre] 2014-05-13T23:00:06.485938-05:00 c1-0c0s2n1 [&lt;ffffffffa038d6b2&gt;] cl_io_start+0x72/0x140 [obdclass] 2014-05-13T23:00:06.485947-05:00 c1-0c0s2n1 [&lt;ffffffffa0391234&gt;] cl_io_loop+0xb4/0x1b0 [obdclass] 2014-05-13T23:00:06.485959-05:00 c1-0c0s2n1 [&lt;ffffffffa07e9ccc&gt;] ll_file_io_generic+0x58c/0x8a0 [lustre] 2014-05-13T23:00:06.485967-05:00 c1-0c0s2n1 [&lt;ffffffffa07ea49c&gt;] ll_file_aio_read+0x22c/0x290 [lustre] 2014-05-13T23:00:06.485994-05:00 c1-0c0s2n1 [&lt;ffffffff8115517b&gt;] do_sync_readv_writev+0xdb/0x120 2014-05-13T23:00:06.486000-05:00 c1-0c0s2n1 [&lt;ffffffff81156084&gt;] do_readv_writev+0xd4/0x1e0 2014-05-13T23:00:06.486010-05:00 c1-0c0s2n1 [&lt;ffffffff811563c5&gt;] vfs_readv+0x45/0x60 2014-05-13T23:00:06.486016-05:00 c1-0c0s2n1 [&lt;ffffffff81156505&gt;] sys_readv+0x55/0xc0 2014-05-13T23:00:06.486023-05:00 c1-0c0s2n1 [&lt;ffffffff8142286b&gt;] system_call_fastpath+0x16/0x1b 2014-05-13T23:00:06.486031-05:00 c1-0c0s2n1 [&lt;000000002006d41b&gt;] 0x2006d41b 2014-05-13T23:00:06.510785-05:00 c1-0c0s2n1 LustreError: 14298:0:(osc_io.c:826:osc_req_attr_set()) LBUG </pre> </div></div> <p>I'll make the dump+logs available shortly.</p> <p>I note this bug was seen by Oleg during patch review for the readahead changes in <a href="https://jira.whamcloud.com/browse/LU-3321" title="2.x single thread/process throughput degraded from 1.8" class="issue-link" data-issue-key="LU-3321"><del>LU-3321</del></a>, see his comments here:</p> <p><a href="http://review.whamcloud.com/#/c/8523/" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#/c/8523/</a></p> <p>I also note a previous instance of this assertion was fixed in <a href="https://jira.whamcloud.com/browse/LU-1650" title="crash of lustre clients in osc_req_attr_set() routine" class="issue-link" data-issue-key="LU-1650"><del>LU-1650</del></a>:<br/> <a href="https://jira.hpdd.intel.com/browse/LU-1650" class="external-link" target="_blank" rel="nofollow noopener">https://jira.hpdd.intel.com/browse/LU-1650</a></p> SLES11SP3 clients, master (v2_5_58_0-105-gb7f1952), Cray stress testing. LU-5062

LBUG: osc_req_attr_set

Bug Critical Resolved Fixed Jinshan Xiong Patrick Farrell osc readahead Wed, 14 May 2014 20:10:59 +0000 Tue, 20 Jan 2015 21:58:51 +0000 Tue, 24 Jun 2014 17:13:32 +0000 Lustre 2.6.0 Lustre 2.6.0 Lustre 2.5.4 0 7 <p>Patrick</p> <p>How up to date is your master? Are any patches applied?</p> <p>Peter</p> <p>Peter -</p> <p>Current as of mid-day yesterday, no patches. We're doing this stress testing to try to help improve release quality, so we're doing it unmodified.</p> <p>Most recent commit is:</p> <p>commit b7f1952882e9e133b43acb22d010767c664539b9<br/> Author: Mikhail Pershin &lt;mike.pershin@intel.com&gt;<br/> Date: Thu May 8 22:29:24 2014 +0400</p> <p> <a href="https://jira.whamcloud.com/browse/LU-4916" title="mount failure when adding failover node to mkfs.lustre" class="issue-link" data-issue-key="LU-4916"><del>LU-4916</del></a> lwp: don't add connection for missed LWP</p> <p> The LWP isn't needed for some MDT-MDT connections,<br/> e.g. for MDT0-MDT1, but lustre_lwp_add_conn() is called<br/> for any 'add mdc' marker and cause error.</p> <p> Patch add LWP connection only where it is needed.<br/> &#8212;</p> <p>Looking at commits since then, the only possibly related commit is one for <a href="https://jira.whamcloud.com/browse/LU-4977" title="Deadlock in balance_dirty_pages()" class="issue-link" data-issue-key="LU-4977"><del>LU-4977</del></a>, but that's not really down this code path.</p> <p>Dump is up at:<br/> ftp.whamcloud.com:/uploads/<a href="https://jira.whamcloud.com/browse/LU-5062" title="LBUG: osc_req_attr_set" class="issue-link" data-issue-key="LU-5062"><del>LU-5062</del></a>/LU_5062_140514.tar.gz</p> <p>Jinshan,<br/> Can you please comment on this one?<br/> Thank you!</p> <p>Do you have patch <a href="http://review.whamcloud.com/10220" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/10220</a> applied on your tree? Probably this is a known problem.</p> <p>Jinshan,</p> <p>From the git history, it looks like we had that commit... But it does look like it would fix the problem, so it's possible I had older source than I thought I did. I will test again and if we continue to see this problem, I'll let you know.</p> <p>Patrick, yes, please try it. Thank you very much for helping us to improve our release quality.</p> <p>OK, same software environment (SLES11SP3 client, Cray stress testing), current master - 2.5.59 tag.</p> <p>Still hit it, and with the same test:</p> <p>2014-05-15T17:18:35.970514-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) page@ffff880b2372cc00<span class="error">&#91;4 ffff880b2040eb10 3 0 1 (null) ffff881037c8ac00 0x0&#93;</span><br/> 2014-05-15T17:18:35.997237-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) vvp-page@ffff880b2372cca0(0:0:0) vm@ffffea0038eaadb8 600000000000801 4:0 ffff880b2372cc00 0 lru<br/> 2014-05-15T17:18:36.022514-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) lov-page@ffff880b2372ccf8, raid0<br/> 2014-05-15T17:18:36.054138-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) osc-page@ffff880b2372cd60 0: 1&lt; 0x845fed 1 0 + + &gt; 2&lt; 0 0 4096 0x7 0x8 | (null) ffff881837716578 ffff880200914ea8 &gt; 3&lt; + ffff8810379f7800 1 16 0 &gt; 4&lt; 0 0 8 43520000 - | - - - + &gt; 5&lt; - - - + | 0 - | 0 - -&gt;<br/> 2014-05-15T17:18:36.079394-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) end page@ffff880b2372cc00<br/> 2014-05-15T17:18:36.104691-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:824:osc_req_attr_set()) dump uncover page!<br/> 2014-05-15T17:18:36.104732-05:00 c0-0c2s0n3 Pid: 28354, comm: nsystst<br/> 2014-05-15T17:18:36.104746-05:00 c0-0c2s0n3 Call Trace:<br/> 2014-05-15T17:18:36.104756-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff81005eb9&gt;&#93;</span> try_stack_unwind+0x169/0x1b0<br/> 2014-05-15T17:18:36.129940-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff81004939&gt;&#93;</span> dump_trace+0x89/0x450<br/> 2014-05-15T17:18:36.129981-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa02218c7&gt;&#93;</span> libcfs_debug_dumpstack+0x57/0x80 <span class="error">&#91;libcfs&#93;</span><br/> 2014-05-15T17:18:36.129993-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0918348&gt;&#93;</span> osc_req_attr_set+0x6b8/0x740 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.155180-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0374be1&gt;&#93;</span> cl_req_attr_set+0xd1/0x220 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.155220-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa09036da&gt;&#93;</span> osc_build_rpc+0x4fa/0x15d0 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.180416-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0925199&gt;&#93;</span> osc_io_unplug0+0x1419/0x2020 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.180456-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0926db1&gt;&#93;</span> osc_io_unplug+0x11/0x20 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.180467-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0926fa1&gt;&#93;</span> osc_queue_sync_pages+0x1e1/0x380 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.205619-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa091890d&gt;&#93;</span> osc_io_submit+0x2dd/0x4e0 <span class="error">&#91;osc&#93;</span><br/> 2014-05-15T17:18:36.205659-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa03743fe&gt;&#93;</span> cl_io_submit_rw+0x6e/0x170 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.230909-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa074a23e&gt;&#93;</span> lov_io_submit+0x2ee/0x4d0 <span class="error">&#91;lov&#93;</span><br/> 2014-05-15T17:18:36.230948-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa03743fe&gt;&#93;</span> cl_io_submit_rw+0x6e/0x170 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.256143-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa03763ea&gt;&#93;</span> cl_io_read_page+0x19a/0x1b0 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.256185-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0804f4c&gt;&#93;</span> ll_readpage+0x16c/0x1b0 <span class="error">&#91;lustre&#93;</span><br/> 2014-05-15T17:18:36.256195-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff810ffec8&gt;&#93;</span> generic_file_aio_read+0x268/0x740<br/> 2014-05-15T17:18:36.281326-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0837030&gt;&#93;</span> vvp_io_read_start+0x2a0/0x430 <span class="error">&#91;lustre&#93;</span><br/> 2014-05-15T17:18:36.281367-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa03746a2&gt;&#93;</span> cl_io_start+0x72/0x140 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.306578-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa0378224&gt;&#93;</span> cl_io_loop+0xb4/0x1b0 <span class="error">&#91;obdclass&#93;</span><br/> 2014-05-15T17:18:36.306618-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa07d2ccc&gt;&#93;</span> ll_file_io_generic+0x58c/0x8a0 <span class="error">&#91;lustre&#93;</span><br/> 2014-05-15T17:18:36.306630-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa07d349c&gt;&#93;</span> ll_file_aio_read+0x22c/0x290 <span class="error">&#91;lustre&#93;</span><br/> 2014-05-15T17:18:36.331852-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffffa07d3e45&gt;&#93;</span> ll_file_read+0x1e5/0x270 <span class="error">&#91;lustre&#93;</span><br/> 2014-05-15T17:18:36.331893-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff811585f8&gt;&#93;</span> vfs_read+0xc8/0x180<br/> 2014-05-15T17:18:36.331904-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff811587b5&gt;&#93;</span> sys_read+0x55/0x90<br/> 2014-05-15T17:18:36.357172-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;ffffffff81427a2b&gt;&#93;</span> system_call_fastpath+0x16/0x1b<br/> 2014-05-15T17:18:36.357219-05:00 c0-0c2s0n3 <span class="error">&#91;&lt;000000002001d680&gt;&#93;</span> 0x2001d680<br/> 2014-05-15T17:18:36.357233-05:00 c0-0c2s0n3 LustreError: 28354:0:(osc_io.c:826:osc_req_attr_set()) LBUG</p> <p>I will look at this issue.</p> <p>Hi Patrick, can you please upload one more crash dump for me? It looks like the previous upload has problem and I can't extract lustre log from there. Please make sure the ko files match the running code.</p> <p>Ugh, sorry for the incorrect KO files... There's a small issue in our automated dump gathering.</p> <p>I did this one by hand and checked that the KOs are right. This is the dump for the report I just made a few hours ago:</p> <p>ftp.whamcloud.com<br/> uploads/<a href="https://jira.whamcloud.com/browse/LU-5062" title="LBUG: osc_req_attr_set" class="issue-link" data-issue-key="LU-5062"><del>LU-5062</del></a>/LU_5062_140515.tar.gz</p> <p><del>The timer is saying 9 minutes to complete the upload.</del><br/> Upload complete.</p> <p>Thank you for taking a look.</p> <p>Hi Patrick,</p> <p>I still can't extract lustre log from the crash dump file. This is what I got when I was trying to extract it.</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>crash&gt; extend /home/jay/crashdump/lustre.so /home/jay/crashdump/lustre.so: shared object loaded crash&gt; lustre -l x lustre_walk_cpus(0, 5, 1) cmd: p (*cfs_trace_data[0])[0].tcd.tcd_cur_pages p (*cfs_trace_data[0])[0].tcd.tcd_pages.next lustre: gdb request failed: "p (*cfs_trace_data[0])[0].tcd.tcd_cur_pages" crash&gt; p cfs_trace_data cfs_trace_data = $7 = {0x1400000004, 0x554e4700000003, 0x5b9a0443e17b2baa, 0xcd0dd360156499dc, 0x5abd76e8, 0x0, 0xffffffffa022c9e0, 0xffffffffa024f201} </pre> </div></div> <p>cfs_trace_data should be an array of pointers. However, the memory cfs_trace_data pointing to seems to be mangled. Can you run crash on your local side to see if you can get the log? Otherwise, please check if the dump gathering script is correct.</p> <p>Jinshan,</p> <p>Hmm. The problem might be SLES11SP3... Xyratex modified lustre.so so it could get logs out. I'm not familiar with what they changed. (We also had to update our version of Crash to something more recent than we wer eusing.) I'll attach that file as well.</p> <p>In any case, I was able to extract the logs. I'll note there did appear to be a few corrupt messages (I got some error messages when using lctl to process the binary logs), so I'll include the binary logs as well as the processed ones. Attaching in a moment.</p> <p>log is the raw binary log, log.sort is the processed log.</p> <p>It turned out that this is a race when a file with empty layout turned into raid0 layout.</p> <p>Jinshan,<br/> Are you planning to make a patch for this?</p> <p>Cray hit this again today <span class="error">&#91;Update: Looking internally, we&#39;ve hit it several times recently in testing.&#93;</span>.</p> <p>Also: It looks like <a href="https://jira.whamcloud.com/browse/LU-3729" title="uncovered page in osc_req_attr_set" class="issue-link" data-issue-key="LU-3729"><del>LU-3729</del></a> may be a duplicate of this? Though there have been several osc_req_attr_set cases fixed since that bug was opened, I believe.</p> <p>I'll start to work on this.</p> <p>Hi Patrick,</p> <p>Can you try this patch: <a href="http://review.whamcloud.com/10760?" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/10760?</a></p> <p>Jinshan - While testing the patch, I hit the assertion from <a href="https://jira.whamcloud.com/browse/LU-4900" title="cl_use_try()) ASSERTION( result != -38 ) failed" class="issue-link" data-issue-key="LU-4900"><del>LU-4900</del></a>, same stack trace as there as well:<br/> 2014-06-19T22:02:47.307062-05:00 c0-0c2s6n0 LustreError: 13331:0:(cl_lock.c:1122:cl_use_try()) ASSERTION( result != -38 ) failed:<br/> 2014-06-19T22:02:47.307096-05:00 c0-0c2s6n0 LustreError: 13331:0:(cl_lock.c:1122:cl_use_try()) LBUG</p> <p>It's probably just random, since that bug's known to exist, but I wanted to let you know in case there's a possible relationship. It's worth noting I don't think we've ever hit <a href="https://jira.whamcloud.com/browse/LU-4900" title="cl_use_try()) ASSERTION( result != -38 ) failed" class="issue-link" data-issue-key="LU-4900"><del>LU-4900</del></a> before. (Still is likely to be just chance.)</p> <p>You used patch set 1, is that right?</p> <p>Yes, I did, because I had it built and the system set up before I saw your response in Gerrit, and then I didn't realize the other code affected more than that sanity testing. Oops.</p> <p>Having looked at <a href="https://jira.whamcloud.com/browse/LU-4900" title="cl_use_try()) ASSERTION( result != -38 ) failed" class="issue-link" data-issue-key="LU-4900"><del>LU-4900</del></a>, I understand now why you did what you did in patch set 2. I see now that's called in the macro in cl_use_try.</p> <p>So it looks like your change to vvp_io_read_lock is getting this kind of cl_lock in to the cache, and then in to cl_use_try, where it asserts.<br/> And since cl_use_try is only called for cached locks, we don't hit that assertion all the time...</p> <p>I'll do another run later with patch set 2. Is the data from running only patch set 1 helpful in proving the vvp_io_read_lock part of the patch, or should I just ignore it? <img class="emoticon" src="https://jira.whamcloud.com/images/icons/emoticons/wink.png" height="16" width="16" align="absmiddle" alt="" border="0"/></p> <p>Sorry about that.</p> <p>Just out of curiosity: I'm still a little puzzled. If I'm reading cl_use_try correctly, as long as there's at least one slice with a valid clo_use, rc shouldn't be ENOSYS, so no assertion. Is the error case here a cl_lock with only a single slice, then?</p> <blockquote> <p>Just out of curiosity: I'm still a little puzzled. If I'm reading cl_use_try correctly, as long as there's at least one slice with a valid clo_use, rc shouldn't be ENOSYS, so no assertion. Is the error case here a cl_lock with only a single slice, then?</p></blockquote> <p>For cl_lock of an empty file, it will have two slices: ccc and lov. However, neither of them has -&gt;clo_use defined this is why cl_use_try() will see ENOSYS error.</p> <p>I don't need the data from patch set 1, thanks.</p> <p>Landed for 2.6</p> <p>Oleg Drokin (oleg.drokin@intel.com) merged in patch <a href="http://review.whamcloud.com/12139/" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/12139/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-5062" title="LBUG: osc_req_attr_set" class="issue-link" data-issue-key="LU-5062"><del>LU-5062</del></a> llite: Solve a race to access lli_has_smd in read case<br/> Project: fs/lustre-release<br/> Branch: b2_5<br/> Current Patch Set: <br/> Commit: 078dee1448c2995257923dc712943eab2c78e48e</p> Related LU-5685 Development Rank 1|hzwmi7: Rank (Obsolete) 13981 Severity