#40933 [Opn->Fbk]: Streams blocking under concurrent access

ID: 40933 Updated by: jani@php.net Reported By: loony at loonybin dot org -Status: Open +Status: Feedback Bug Type: Streams related Operating System: Linux 2.6.20 - FC6 PHP Version: 5.2.1 New Comment: Does that script work in any PHP version? I also get some E_NOTICE for that ob_flush(); call.. Previous Comments: ------------------------------------------------------------------------ [2007-07-13 02:27:36] loony at loonybin dot org Tried php5.2-200707122230 and the behavior does not seem to have changed. ------------------------------------------------------------------------ [2007-07-12 21:16:39] jani@php.net Please try using this CVS snapshot: http://snaps.php.net/php5.2-latest.tar.gz For Windows (zip): http://snaps.php.net/win32/php5.2-win32-latest.zip For Windows (installer): http://snaps.php.net/win32/php5.2-win32-installer-latest.msi ------------------------------------------------------------------------ [2007-03-28 04:32:29] loony at loonybin dot org Description: ------------ Tail for php. Open the file, output the current content... then wait for more - if no more content arrives, then return number of bytes read after a while... Running 1 execution of this code works fine. Executing it from two browsers at the same time blocks. Happens on Fedora Core 6 default 5.1.6 and also after upgrading to 5.2.1 Reproduce code: --------------- http://ns1.loonybin.org/php/tail.tar.gz - 45 lines Expected result: ---------------- See content of file printed immediately. If any content is appended to the file within 15 seconds from request start, that content should be printed as well. This works with 1 execution of the code. Should yield same result if executed multiple times. Actual result: -------------- 1 run of the code works. Multiple executions however block on semop / epoll_wait. semop(4358158, 0x15470c, 1) = 0 <2.194313> epoll_wait(16, {{EPOLLIN, {u32=2156798632, u64=13801426109869204136}}}, 2, -1) = 1 <3.738842> As soon as the first execution of the code has ended, the second will run as expected and finish successfully (as long as max_execution_time is not reached). ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/?id=40933&edit=1