Re: Re: APC-3.0.10 issue

From: Date: Tue, 23 May 2006 03:53:30 +0000
Subject: Re: Re: APC-3.0.10 issue
References: 1 2 3  Groups: php.pecl.dev 
Request: Send a blank email to pecl-dev+get-3796@lists.php.net to get a copy of this message
It'll be helpful to see some more symbols in this frame.... Obviously *something* is not initialized properly. Give these gdb commands a try:
(gdb) p execute_data.opline->op1.u.constant [Ordinary IS_STRING .var = "loaded" .len = 6] (gdb) p execute_data.opline->result.u.var $6 = 1 (gdb) p op_array->T $9 = 340 Well crap, that's unfortunate becuase it all looks (mostly) normal...
The op_array is saying it needs 340 temp vars allocated and EX(Ts) has a nice little address somewhere on the stack right about where we'd expect it to be... The op itself, which happens to be the first op in the op_array, is storing its result into temp var #1 (which is well within the 340 limit). What's a little odd is that it's going into #1... not #0 which is what I'd expect from the first op. This doesn't feel like the cause, but it's odd, and odd is all I've got atm.... It's also curious that resolution of this constant *fails* and it's in the fallback string treatment that the actual crash occours. Based on your application logic are you expecting this constant "loaded" to be undefined at the time of its use? Can you share the relevant few lines of code? It would be at the beginning of a function without parameters (at least, none in the delaration -- func_get_args() not withstanding), probably an initial statement like: $foo = loaded; or if (loaded) { or something involving that constant... Include a few subsequent lines of context if you don't mind... -Sara

« previous php.pecl.dev (#3796) next »