Jani Taskinen schrieb:
What about fixing it also in PHP_5_2 ?? Or doesn't this exist there? ;)
Sounds quite critical to be in there..

Good question.

A precondition for the crash to happen is that ext/mysqli allocates a too small result buffer. This could happen due to bug which broke BC in 5.3.0. That bug has been fixed in the commit immerdiately followed after this commit: http://news.php.net/php.cvs/60379

I was wrong that the crash can be reproduced with the test 005.phpt and PHP 5.0.x-5.2.x. I checked again and found that I had messed up my PHP sources causing a similar but unrelated crash.

I also checked a couple of SQL statements that caused headaches during the mysqlnd development (e.g. SELECT UNION) and the MySQL bugs database for a bug which may lead to a too small result buffer in order to break ext/mysqli: without success.

Anyway, I can patch PHP 5.2 as well, just in case someone finds a way to make a current MySQL server report faulty meta data (length must be reported as 0 but the actual data returned needs to be longer than 256 bytes).


Ulf Wendel, MySQL
Sun Microsystems GmbH,   Sonnenallee 1,   D-85551 Kirchheim-Heimstetten
Geschaeftsfuehrer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Haering     Muenchen: HRB161028

PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php

Reply via email to