Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:52428
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain nebm.ist.utl.pt from 193.136.128.22 cause and error)
Content-Type: text/plain; charset=utf-8; format=flowed; delsp=yes
To: internals@lists.php.net
References: <BANLkTinNZmE9NszP4nqeGpx3ox0eBWx6RA@mail.gmail.com>
Date: Tue, 17 May 2011 15:21:59 +0100
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Organization: =?utf-8?Q?N=C3=BAcleo_de_Eng=2E_Biom=C3=A9di?=
 =?utf-8?Q?ca_do_I=2ES=2ET=2E?=
Message-ID: <op.vvmp6xm9idpuyk@clk-0081.clk-domain>
In-Reply-To: <BANLkTinNZmE9NszP4nqeGpx3ox0eBWx6RA@mail.gmail.com>
User-Agent: Opera Mail/11.10 (Win32)
Subject: Re: [PHP-DEV] Need karma for committing test patches
From: glopes@nebm.ist.utl.pt ("Gustavo Lopes")

Em Tue, 17 May 2011 14:40:53 +0100, Alexey Shein <confik@gmail.com>  
escreveu:

> 2) fix for http://bugs.php.net/48203: there's a segfault when
> CURLOPT_STDERR file pointer is closed before calling curl_exec, i.e.
> like this:
>
> $fp = fopen(dirname(__FILE__) . '/bug48203.tmp', 'w');
>
> $ch = curl_init();
>
> curl_setopt($ch, CURLOPT_VERBOSE, 1);
> curl_setopt($ch, CURLOPT_STDERR, $fp);
> curl_setopt($ch, CURLOPT_URL, getenv("PHP_CURL_HTTP_REMOTE_SERVER"));
>
> fclose($fp); // <-- premature close of $fp caused a crash!
>
> curl_exec($ch); // segfault
>
> All tests run ok on php5.3 php5.4 and trunk.
> Could somebody review this patch since I'm new to developing php and
> could make some silly mistakes?
>

A few remarks:

* Isn't this supposed to be fixed? Was there something that triggered a  
regression?
* If this strategy is used (checking whether the resource associated with  
the stored zval is valid), how about curl_multi_exec?
* I think a better strategy would be to just dup the file descriptor  
gotten after the cast in curl_setopt, store it (instead of storing the  
zval) and close it on curl handle destruction. This way we wouldn't have  
to worry about zval refcounts or whether the file descriptor obtained is  
still valid. Could you see if this other strategy works? (otherwise I can  
do it later this week)

-- 
Gustavo Lopes