> 21 jun 2015 kl. 20:09 skrev Richard <replies-lists-e7x6-php@xxxxxxxxxxxxxxxxxxxxx>: > > > > ------------ Original Message ------------ >> Date: Sunday, June 21, 2015 12:22:28 AM +0200 >> From: Anatol Belski <anatol.php@xxxxxxxxxx> >> >> Hi Frank, >> >>> -----Original Message----- >>> From: Frank Arensmeier [mailto:farensmeier@xxxxxxxxx] >>> Sent: Sunday, June 21, 2015 12:05 AM >>> >>> Hi List! >>> >>> I am working on a CLI script (PHP 5.5.25) that parses and imports >>> large amounts of data into a database. Usually, the script runs >>> for 5-15 minutes. For the most part, the script is working just >>> fine. But for some reason, it seems to hit some limit and PHP >>> crashes with a segmentation fault. I managed to hook GDB to >>> the PHP process and got a backtrace. Unfortunately, I am not so >>> much wiser than before. >>> >>> Can someone be so kind and point me into the right direction and >>> give me a hint why PHP crashes? The backtrace: >>> >>> Program received signal EXC_BAD_ACCESS, Could not access memory. >>> Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000018 >>> 0x000000010e8f416f in zend_signal_handler_unblock () >>> (gdb) bt >>> # 0 0x000000010e8f416f in zend_signal_handler_unblock () >>> # 1 0x000000010e8b47d5 in _zend_mm_alloc_int () >>> # 2 0x000000010e884cd4 in vspprintf () >>> # 3 0x000000010f5e112f in xdebug_error_cb () >>> # 4 0x000000010e8d5875 in zend_error () >>> # 5 0x000000010e8f4121 in zend_signal_handler () >>> # 6 0x000000010e8f3fae in zend_signal_handler_defer () >>> # 7 <signal handler called> >>> # 8 0x00007fff855fce9a in write$NOCANCEL () >>> # 9 0x00007fff8fae8970 in _swrite () >>> # 10 0x00007fff8fae15e9 in __sflush () >>> [.] >>> >>> I have already tried to give PHP more memory (512MB) but that had >>> no impact on the segmentation fault. Any ideas? >>> >> You can try to disable xdebug and see what happens. >> >> Regards >> >> Anatol > > Are you trying to read in a whole data set, process it and then spit > it out or are you reading it incrementally (by line or defined > chunk)? Based on what you said, I suspect the former -- the whole > data set. In general, reading/writing incrementally (if possible) > will keep you from running into resource (e.g., memory) limits and > is a more robust long-term approach as increases in the input data > set size are less likely to cause unexpected processing changes > (e.g., resource limit issues) over time. Hi Richard! Thanks for the heads up. You are absolutely right that it is always a good idea to household with resources wherever possible. In this case, Xdebug was the root cause of the segfault. It turned out that similar bugs have been reported already. Hopefully there will be an update for Xdebug soon. /frank > > > > > > -- > PHP General Mailing List (http://www.php.net/) > To unsubscribe, visit: http://www.php.net/unsub.php > -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
