Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:57576
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain zend.com designates 212.199.177.89 as permitted sender)
Message-ID: <4F26BA53.7040004@zend.com>
Date: Mon, 30 Jan 2012 19:42:11 +0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:9.0) Gecko/20111222 Thunderbird/9.0
MIME-Version: 1.0
To: yoram bar haim <yoram.b@zend.com>, <mike@php.net>
CC: PHP Internals <internals@lists.php.net>
References: <201201301729.12063.yoram.b@zend.com>
In-Reply-To: <201201301729.12063.yoram.b@zend.com>
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] Re: One more critical problem in 5.4
From: dmitry@zend.com (Dmitry Stogov)

Hi Mike,

I confirm the bug. In case of empty output HTTP headers are sent to 
stdout or stderr instead of FastCGI stream.

To reproduce I configured nginx to use PHP FastCGI server on UNIX 
socket, then launched (sapi/cgi/php-cgi -b /tmp/fcgi-php5) and performed 
several request to empty PHP file. The HTTP headers were printed in 
console running PHP.

[dmitry@tpl2 CGI-DEBUG]$ sapi/cgi/php-cgi -b /tmp/fcgi-php5
X-Powered-By: PHP/5.4.0RC7-dev
Content-type: text/html

X-Powered-By: PHP/5.4.0RC7-dev
Content-type: text/html

I afraid Yoram's patch is too radical and may have side effects.
Mike, could you please think about a safer way of fixing.

Thanks. Dmitry.


On 01/30/2012 07:29 PM, yoram bar haim wrote:
> I Debugged the issue described here by lior.
> the problem is :
> in php_request_shutdown() we call
>       sapi_send_headers() after
>       php_output_deactivate().
>
> at this point,
> in main/output.c,
> OG(flags)&  PHP_OUTPUT_ACTIVATED is false so
>        php_output_write_unbuffered() calls
>                php_output_stderr() instead of
>                sapi_module handler.
>
> the following patch solves the issue but It might be a wrong solution (because
> other thins are dependent on this order) so I'm not suggesting it as fix.
>
> --- main/main.c.orig    2012-01-30 17:25:44.000000000 +0200
> +++ main/main.c 2012-01-30 17:09:05.000000000 +0200
> @@ -1738,12 +1738,13 @@
>                  } else {
>                          php_output_end_all(TSRMLS_C);
>                  }
> +               sapi_send_headers(TSRMLS_C);
>                  php_output_deactivate(TSRMLS_C);
>          } zend_end_try();
>
>          /* 4. Send the set HTTP headers (note: This must be done AFTER
> php_output_discard_all() / php_output_end_all() !!) */
>          zend_try {
> -               sapi_send_headers(TSRMLS_C);
> +/*             sapi_send_headers(TSRMLS_C); */
>          } zend_end_try();
>
>          /* 5. Reset max_execution_time (no longer executing php code after
> response sent) */
>