<div dir="ltr">







<p class="">We're testing Presto (<a href="https://github.com/facebook/presto"><span class="">https://github.com/facebook/presto</span></a>) on Java 8, and ran into a segfault in the VM. Below is the relevant portion from the crash report. You can see the full dump here: <a href="https://gist.github.com/martint/364a12d4f776a6e54911">https://gist.github.com/martint/364a12d4f776a6e54911</a></p>
<pre class="" style="font-family:Consolas,'Liberation Mono',Courier,monospace;font-size:12px;margin-top:0px;margin-bottom:0px;line-height:16.799999237060547px;width:13986px;padding:0px;color:rgb(51,51,51)"><div class="" id="file-hs_err-log-LC91">
Stack: [0x00007fe8a1a6c000,0x00007fe8a1b6d000],  sp=0x00007fe8a1b6abe0,  free space=1018k</div><div class="" id="file-hs_err-log-LC92">Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)</div><div class="" id="file-hs_err-log-LC93">
V  [libjvm.so+0x3fd2f8]  ciSignature::ciSignature(ciKlass*, ciSymbol*, ciMethodType*)+0x178</div><div class="" id="file-hs_err-log-LC94">V  [libjvm.so+0x403ba7]  ciBytecodeStream::get_method(bool&, ciSignature**)+0x4a7</div>
<div class="" id="file-hs_err-log-LC95">V  [libjvm.so+0x332099]  GraphBuilder::invoke(Bytecodes::Code)+0x39</div><div class="" id="file-hs_err-log-LC96">V  [libjvm.so+0x333b3d]  GraphBuilder::iterate_bytecodes_for_block(int)+0x6ad</div>
<div class="" id="file-hs_err-log-LC97">V  [libjvm.so+0x330fa2]  GraphBuilder::iterate_all_blocks(bool)+0xf2</div><div class="" id="file-hs_err-log-LC98">V  [libjvm.so+0x335a97]  GraphBuilder::GraphBuilder(Compilation*, IRScope*)+0x517</div>
<div class="" id="file-hs_err-log-LC99">V  [libjvm.so+0x33e337]  IRScope::IRScope(Compilation*, IRScope*, int, ciMethod*, int, bool)+0x1e7</div><div class="" id="file-hs_err-log-LC100">V  [libjvm.so+0x33e44f]  IR::IR(Compilation*, ciMethod*, int)+0x9f</div>
<div class="" id="file-hs_err-log-LC101">V  [libjvm.so+0x31d74b]  Compilation::build_hir()+0xdb</div><div class="" id="file-hs_err-log-LC102">V  [libjvm.so+0x31db0e]  Compilation::compile_java_method()+0x6e</div><div class="" id="file-hs_err-log-LC103">
V  [libjvm.so+0x31dc3e]  Compilation::compile_method()+0x4e</div><div class="" id="file-hs_err-log-LC104">V  [libjvm.so+0x31df9f]  Compilation::Compilation(AbstractCompiler*, ciEnv*, ciMethod*, int, BufferBlob*)+0x24f</div>
<div class="" id="file-hs_err-log-LC105">V  [libjvm.so+0x31ed39]  Compiler::compile_method(ciEnv*, ciMethod*, int)+0xa9</div><div class="" id="file-hs_err-log-LC106">V  [libjvm.so+0x46982a]  CompileBroker::invoke_compiler_on_method(CompileTask*)+0xc8a</div>
<div class="" id="file-hs_err-log-LC107">V  [libjvm.so+0x46c230]  CompileBroker::compiler_thread_loop()+0x620</div><div class="" id="file-hs_err-log-LC108">V  [libjvm.so+0x9e303f]  JavaThread::thread_main_inner()+0xdf</div>
<div class="" id="file-hs_err-log-LC109">V  [libjvm.so+0x9e3205]  JavaThread::run()+0x1b5</div><div class="" id="file-hs_err-log-LC110">V  [libjvm.so+0x8a00c8]  java_start(Thread*)+0x108</div></pre>
<p class=""><br></p><p class="">Is this a known issue? I couldn't find anything related in the bug database. Otherwise, any suggestions for debugging it would be very much appreciated! </p><div>The problem only seems to happen after Presto has been running for a while and has processed several hundred queries. I’ve tried re-running the queries that were running at the moment it crashed, but it does not reproduce deterministically. Also, we don't see it in Java 7.<br>
</div><div><br></div><p class="">Thanks!</p><p class="">Martin</p><p class=""><br></p><p class=""><br></p></div>