4.0.2 (zeev) userful "hang" backtrace #1

From: Date: Wed, 23 Aug 2000 04:29:05 +0000
Subject: 4.0.2 (zeev) userful "hang" backtrace #1
Groups: php.dev 
Request: Send a blank email to php-dev+get-30271@lists.php.net to get a copy of this message
backtrace from infinite loop, whereas nothing loops infinitely in my code: Part I: This one is now generated with zeevs 4.0.2 debug build (wihout using .pdb information). You can see that the same functions are called over and over, with modifications in it. The actual trace was 3MB before I killed it, I hope to have included all relevant data. I don´t know if you can debug sth. that´s not a crash... If I replace these binaries (debug+release doesn´t matter) with my 06-31 build, everything works fine (the next one I have is from 07-27, which does not work). Part II: I´ve also attached MUTEK to the release build and then it did hang just like the debug build instead of crashing. It blews up to 9.1MB in RAM (slowly) but then the memory usage does not seems to increase over a long period, than after 20min or so shutdown_memory_manager() is called, then efree is called very much and the memeroy usage gets less and then PHP crashes. Running the release build with MUTEK attached creates an immediate excpetion. I´ve watched the state before that exception and the sams things with same adresses are called over and over. <loop> _efee, 0x0008e6f0 ts_resource_ex, 0x0007d0a0 tsrm_thread_id,0x0007d2f0 tsrm_mutex_lock, 0x0007d330 tsrm_muteck_unlock, 0x0007d350 </loop> the only difference immediately before the crash I´ve noticed, is that the stack trace at _efree() contained: _efree(), 0x0007e6f0, return shutdown_memory_manage(), 0x0007eb30 directly before the crash, the return missed: _efree(), 0x0007e6f0 perhaps that´s related to crashes, I don´t know, I think I´ll have to switch back to 06-30 permanently :( andré

Attachment: [application/octet-stream] trace.zip
« previous php.dev (#30271) next »