https://bugs.dragonflybsd.org/https://bugs.dragonflybsd.org/favicon.ico?16293952082008-11-12T06:53:03ZDragonFlyBSD bugtrackerDragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=55232008-11-12T06:53:03Zdillon
<ul></ul><p>:The lighttpd process got stuck in the select state unkillably. I decided to <br />:reboot and got a panic.<br />:<br />:syncing disks... 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87<br />:giving up on 1 buffers<br />:Debugger("busy buffer problem")<br />:Stopped at Debugger+0x34: movb $0,in_Debugger.3949<br />:db> trace<br />:Debugger(c05382b5,c053829c,1,0,0) at Debugger+0x34<br />:boot(0,caf18d34,c04d2fdd,caf18cf0,6) at boot+0x1e2<br />:sys_reboot(caf18cf0,6,55d93,0,c9708c58) at sys_reboot+0x23<br />:syscall2(caf18d40) at syscall2+0x1e9<br />:Xint0x80_syscall() at Xint0x80_syscall+0x36<br />:db><br />:<br />:Needless to say, lighttpd serves from a hammer partition and also runs from <br />:another. Kernel and vmcore are also available:<br />:<br />:http://160.114.134.7/~szg/crash081111/ </p>
<pre><code>Excellent core dump, I have downloaded it and tracked the problem down<br /> to a deadlock in the HAMMER code. I am working on a patch now.</code></pre>
<pre><code>-Matt<br /> Matthew Dillon <br /> &lt;<a class="email" href="mailto:dillon@backplane.com">dillon@backplane.com</a>&gt;</code></pre> DragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=55262008-11-12T13:25:03Zdillon
<ul></ul><p>This may take another day. It turns out to be a deep vnode lock being<br /> acquired from inside a B-tree cursor. The cursor is holding a B-tree<br /> node lock and I can't easily release it.</p>
<pre><code>-Matt</code></pre> DragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=55302008-11-17T00:23:03Zbastyaelvtars
<ul></ul><p>Oops, I misconfigured my mail client, this was actually me.<br />@Matt: thanks for looking into it and do not hurry since I couldn't <br />reproduce it since then. :-)</p>
<p>"DFBSD" <<a class="email" href="mailto:bastyaelvtars@gmail.com">bastyaelvtars@gmail.com</a>> wrote in message <br />news:4919d02a$0$882$415eb37d@crater_reader.dragonflybsd.org...</p>
<blockquote>
<p>The lighttpd process got stuck in the select state unkillably. I decided <br />to reboot and got a panic.</p>
<p>syncing disks... 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 87 <br />87<br />giving up on 1 buffers<br />Debugger("busy buffer problem")<br />Stopped at Debugger+0x34: movb $0,in_Debugger.3949<br />db> trace<br />Debugger(c05382b5,c053829c,1,0,0) at Debugger+0x34<br />boot(0,caf18d34,c04d2fdd,caf18cf0,6) at boot+0x1e2<br />sys_reboot(caf18cf0,6,55d93,0,c9708c58) at sys_reboot+0x23<br />syscall2(caf18d40) at syscall2+0x1e9<br />Xint0x80_syscall() at Xint0x80_syscall+0x36<br />db></p>
<p>Needless to say, lighttpd serves from a hammer partition and also runs <br />from another. Kernel and vmcore are also available:</p>
<p><a class="external" href="http://160.114.134.7/~szg/crash081111/">http://160.114.134.7/~szg/crash081111/</a></p>
</blockquote> DragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=55742008-12-01T21:02:00Zaoiko
<ul></ul><p>Any progress with this bug?</p>
<p>Aggelos</p> DragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=55862008-12-01T23:52:01Zdillon
<ul></ul><p>:Matthew Dillon wrote:<br />:> This may take another day. It turns out to be a deep vnode lock being<br />:> acquired from inside a B-tree cursor. The cursor is holding a B-tree<br />:> node lock and I can't easily release it.<br />:> <br />:> -Matt<br />:<br />:Any progress with this bug?<br />:<br />:Aggelos</p>
<pre><code>This should be fixed now.</code></pre>
<pre><code>-Matt<br /> Matthew Dillon <br /> &lt;<a class="email" href="mailto:dillon@backplane.com">dillon@backplane.com</a>&gt;</code></pre> DragonFlyBSD - Bug #1162: Possible hammer problemhttps://bugs.dragonflybsd.org/issues/1162?journal_id=56032008-12-04T01:53:01Zbastyaelvtars
<ul></ul><p>"Matthew Dillon" <<a class="email" href="mailto:dillon@apollo.backplane.com">dillon@apollo.backplane.com</a>> wrote in message <br />news:<a class="email" href="mailto:200812011549.mB1Fnthj051582@apollo.backplane.com">200812011549.mB1Fnthj051582@apollo.backplane.com</a>...</p>
<blockquote>
<p>:Matthew Dillon wrote:<br />:> This may take another day. It turns out to be a deep vnode lock <br />being<br />:> acquired from inside a B-tree cursor. The cursor is holding a <br />B-tree<br />:> node lock and I can't easily release it.<br />:><br />:> -Matt<br />:<br />:Any progress with this bug?<br />:<br />:Aggelos</p>
<p>This should be fixed now.</p>
</blockquote>
<p>Also in 2.0 or whatever it is called now? :-P</p>