From: Operating system: Windows 7 PHP version: 5.3.6 Package: Reproducible crash Bug Type: Bug Bug description:PHP crash when using closures + extract(EXTR_REFS)
Description: ------------ See test script. PHP 5.3.5 is not affected. Test script: --------------- // Initially $Object is not a reference and contains a "pointer" // to an stdClass object. $Object =new stdClass; /**/ echo 'New: '; debug_zval_dump( $Object); // $Object becomes a reference to the "pointer" to an stdClass. $Object =&$Object; /**/ echo 'Self-reference: '; debug_zval_dump( $Object); // Now we import $Object into closure by value. In theory, // $Object, that is inside closure, should not be a reference, but rather // should be a variable, that "points" to stdClass (i.e. an equivalent of // $ImportedObject in expression "$ImportedObject =$Object"). $closure =function() use( $Object) { // Once you manipulate $Object, you get PHP crashed. $Object->x =10; //debug_zval_dump( $Object); }; // By calling extract() we make $Object to reference a new stdClass instance. extract( array( 'Object' =>new stdClass), EXTR_REFS); echo 'After extract: '; debug_zval_dump( $Object); // now we execute closure and get PHP crashed $closure(); Expected result: ---------------- PHP should not crash. Actual result: -------------- PHP crashes. If you put die() right before $closure(), then you get following output: line 1: New: object(stdClass)#1 (0) refcount(2){ line 2: } line 3: Self-reference: object(stdClass)#1 (0) refcount(1){ line 4: } line 5: After extract: object(stdClass)#3 (0) refcount(2){ line 6: } Some questions regarding that output: 1) why there is refcount(2) in the first line? Isn't the object referenced only once and by $Object variable? I would expect to see refcount(1) here. As you can see on line 3, refcount seems to become correct after self-referencing is being made. 2) why line 5 says object(stdClass)#3, while there were only two (and not 3) stdClass objects allocated? If you comment out closure's definition, then you get object(stdClass)#2 (an expected output). Does closure clone $Object when you say "use($Object)"? Shouldn't the stdClass object be simply "referenced" by the "use($Object)"? -- Edit bug report at http://bugs.php.net/bug.php?id=54860&edit=1 -- Try a snapshot (PHP 5.2): http://bugs.php.net/fix.php?id=54860&r=trysnapshot52 Try a snapshot (PHP 5.3): http://bugs.php.net/fix.php?id=54860&r=trysnapshot53 Try a snapshot (trunk): http://bugs.php.net/fix.php?id=54860&r=trysnapshottrunk Fixed in SVN: http://bugs.php.net/fix.php?id=54860&r=fixed Fixed in SVN and need be documented: http://bugs.php.net/fix.php?id=54860&r=needdocs Fixed in release: http://bugs.php.net/fix.php?id=54860&r=alreadyfixed Need backtrace: http://bugs.php.net/fix.php?id=54860&r=needtrace Need Reproduce Script: http://bugs.php.net/fix.php?id=54860&r=needscript Try newer version: http://bugs.php.net/fix.php?id=54860&r=oldversion Not developer issue: http://bugs.php.net/fix.php?id=54860&r=support Expected behavior: http://bugs.php.net/fix.php?id=54860&r=notwrong Not enough info: http://bugs.php.net/fix.php?id=54860&r=notenoughinfo Submitted twice: http://bugs.php.net/fix.php?id=54860&r=submittedtwice register_globals: http://bugs.php.net/fix.php?id=54860&r=globals PHP 4 support discontinued: http://bugs.php.net/fix.php?id=54860&r=php4 Daylight Savings: http://bugs.php.net/fix.php?id=54860&r=dst IIS Stability: http://bugs.php.net/fix.php?id=54860&r=isapi Install GNU Sed: http://bugs.php.net/fix.php?id=54860&r=gnused Floating point limitations: http://bugs.php.net/fix.php?id=54860&r=float No Zend Extensions: http://bugs.php.net/fix.php?id=54860&r=nozend MySQL Configuration Error: http://bugs.php.net/fix.php?id=54860&r=mysqlcfg