Edit report at https://bugs.php.net/bug.php?id=60978&edit=1
ID: 60978 Patch added by: larue...@php.net Reported by: the...@php.net Summary: exit code incorrect Status: Assigned Type: Bug Package: Scripting Engine problem Operating System: Windows PHP Version: 5.4.0RC7 Assigned To: derick Block user comment: N Private report: N New Comment: The following patch has been added/updated: Patch Name: wrong_exit_code.patch Revision: 1328522147 URL: https://bugs.php.net/patch-display.php?bug=60978&patch=wrong_exit_code.patch&revision=1328522147 Previous Comments: ------------------------------------------------------------------------ [2012-02-05 13:36:33] the...@php.net Checking SVN history revealed r322922 as the cause - see http://svn.php.net/viewvc?view=revision&revision=322922. Reverting it fixes this bug. The commit mentions a relation to bug #60218 but I'm not sure why it "Reinstated correct return values". Before this fix the code in zend_execute_API.c read: zend_execute(new_op_array TSRMLS_CC); [...] retval = SUCCESS; After the fix it read: zend_try { zend_execute(new_op_array TSRMLS_CC); } zend_end_try(); [...] retval = SUCCESS; Finally, r322922 changed it to: retval = SUCCESS; zend_try { zend_execute(new_op_array TSRMLS_CC); } zend_catch { retval = FAILURE; } zend_end_try(); The zend_catch "block" is executed whenever SETJMP returns non-zero, so basically when LONGJMP is called, which is the case for zend_bailout(), which again is used by exit() and die(). To my eyes, this *changed* the return value instead of reinstating it. @derick, maybe you can shed some light on this commit? ------------------------------------------------------------------------ [2012-02-05 10:54:30] the...@php.net Test suite: <?php if (!isset($argv[1])) { exit("*** Usage: php exit.php /path/to/php/binary\n"); } $php= realpath($argv[1]); if (!is_executable($php)) { exit("*** PHP Binary '$argv[1]' is not executable\n"); } $tests= array( "echo '';" => 0, "exit('test');" => 0, "exit(2);" => 2, "fatal();" => 255, "\$->" => 254, "throw new Exception('test');" => 255 ); foreach ($tests as $source => $expected) { $cmd= '"'.$php.'" -r "'.$source.'" 2>&1'; $output= array(); exec($cmd, $output, $actual); if ($actual === $expected) { printf("%-30s: [OK]\n", $source); } else { printf("%-30s: [FAIL, expect %d, have %d (%s)]\n", $source, $expected, $actual, implode(' ', $output)); } } ?> ------------------------------------------------------------------------ [2012-02-05 10:01:41] the...@php.net ...oh, and: * Uncaught exception exit, `php -r 'throw new Exception("test");' ; echo $?` = 255 ------------------------------------------------------------------------ [2012-02-05 09:59:54] the...@php.net Your patch breaks the exitcode 254 for parse errors. We have the following cases and expected exit codes: * Clean exit, `php -r 'echo "";' ; echo $?` = 0 * Clean explicit exit with message, `php -r 'exit("test");' ; echo $?` = 0 * Clean explicit exit with exitcode, `php -r 'exit(2);' ; echo $?` = 2 * Exit after a fatal error, `php -r '$fatal->error();' ; echo $?` = 255 * Exit after compile error, `php -r '$->' ; echo $?` = 254 ------------------------------------------------------------------------ [2012-02-04 20:43:08] php-dev at zerocue dot com Alright, in 5.3.10@php_cli:1023 the call to zend_eval_string_ex() never returned after a zend_bailout(), if it had this bug would have been present there because zend_bailout always returns FAILURE even in non-failure cases. In 5.4 and trunk, the zend_eval_string_ex() does return and hits the exit_status=254 line. The proper fix would be to have zend_bailout not return FAILURE unless it did fail but this is a much larger change. In this case the patch simply ignores the return state of the zend_eval_string_ex() and always returns the EG(exit_status) value. ------------------------------------------------------------------------ The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at https://bugs.php.net/bug.php?id=60978 -- Edit this bug report at https://bugs.php.net/bug.php?id=60978&edit=1