<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
<!DOCTYPE bugzilla SYSTEM "https://bugs.webkit.org/page.cgi?id=bugzilla.dtd">

<bugzilla version="5.0.4.1"
          urlbase="https://bugs.webkit.org/"
          
          maintainer="admin@webkit.org"
>

    <bug>
          <bug_id>265469</bug_id>
          
          <creation_ts>2023-11-28 11:30:04 -0800</creation_ts>
          <short_desc>Race condition in ARM64 disassembler initialization</short_desc>
          <delta_ts>2023-11-30 10:51:59 -0800</delta_ts>
          <reporter_accessible>1</reporter_accessible>
          <cclist_accessible>1</cclist_accessible>
          <classification_id>1</classification_id>
          <classification>Unclassified</classification>
          <product>WebKit</product>
          <component>JavaScriptCore</component>
          <version>WebKit Nightly Build</version>
          <rep_platform>Unspecified</rep_platform>
          <op_sys>Unspecified</op_sys>
          <bug_status>RESOLVED</bug_status>
          <resolution>FIXED</resolution>
          
          
          <bug_file_loc></bug_file_loc>
          <status_whiteboard></status_whiteboard>
          <keywords>InRadar</keywords>
          <priority>P2</priority>
          <bug_severity>Normal</bug_severity>
          <target_milestone>---</target_milestone>
          
          
          <everconfirmed>1</everconfirmed>
          <reporter name="David Degazio">d_degazio</reporter>
          <assigned_to name="David Degazio">d_degazio</assigned_to>
          <cc>webkit-bug-importer</cc>
          

      

      

      

          <comment_sort_order>oldest_to_newest</comment_sort_order>  
          <long_desc isprivate="0" >
    <commentid>1995365</commentid>
    <comment_count>0</comment_count>
    <who name="David Degazio">d_degazio</who>
    <bug_when>2023-11-28 11:30:04 -0800</bug_when>
    <thetext>rdar://118890976

It&apos;s rare, but possible for two threads to initialize the ARM64 disassembler in JSC at the same time. If this happens, we run into problems in the following code:

    if (!opcodeTable[opcodeGroupNumber])
        opcodeTable[opcodeGroupNumber] = newOpcodeGroup;
    else
        lastGroups[opcodeGroupNumber]-&gt;setNext(newOpcodeGroup);
    lastGroups[opcodeGroupNumber] = newOpcodeGroup;

This code builds an array of linked lists of opcode groups. We can get an interleaving where:
  - Thread 1 hits the if statement, and sees the opcode table entry is null.
  - Thread 1 sets the opcode table entry to a new, non-null group.
  - Thread 2 hits the if statement, and sees the opcode table entry is now non-null.
  - Thread 2 continues to setNext on an element of lastGroups, but it&apos;s still null at this point! So we crash.

To fix this, we should just protect this initialization function with a lock.</thetext>
  </long_desc><long_desc isprivate="0" >
    <commentid>1995387</commentid>
    <comment_count>1</comment_count>
    <who name="David Degazio">d_degazio</who>
    <bug_when>2023-11-28 13:56:33 -0800</bug_when>
    <thetext>Pull request: https://github.com/WebKit/WebKit/pull/21015</thetext>
  </long_desc><long_desc isprivate="0" >
    <commentid>1995935</commentid>
    <comment_count>2</comment_count>
    <who name="EWS">ews-feeder</who>
    <bug_when>2023-11-30 10:51:57 -0800</bug_when>
    <thetext>Committed 271350@main (48684d06eb78): &lt;https://commits.webkit.org/271350@main&gt;

Reviewed commits have been landed. Closing PR #21015 and removing active labels.</thetext>
  </long_desc>
      
      

    </bug>

</bugzilla>