[rfc][icedtea-web] (C++) More precise check of browser table and plugin table callback structures in NP_Initialize

[Adam Domurad]

Hi all. I have tested this and there does not seem to be any regressions - considering the recent problem reported with 1.2, can anyone give it an OK for 1.2 ?

----- Original Message -----
From: "Adam Domurad" <adomurad at redhat.com>
To: distro-pkg-dev at openjdk.java.net
Sent: Monday, June 18, 2012 12:12:09 PM
Subject: [rfc][icedtea-web] (C++) More precise check of browser table and	plugin table callback structures in NP_Initialize

Hey all. This is motivated by the recent version of Google Chrome
erroring and not running the plugin, giving 'Invalid plugin table.' I
was speaking with Deepak on ways to solve this issue and came up with
this solution.

The previous code checked the size of NPPluginFunc's (as well as
NPNetscapeFunc's) against the size reported by the browser. This code
merely requires the browser to pass a copy of these structures big
enough to support the functions we actually use. 

The previous code could in theory write past the end of an object as
well, as it only warned if NPPluginFunc was not big enough. 

The code now always errors if functionality we need is not provided.

Changelog:
2012-06-18  Adam Domurad  <adomurad at redhat.com>

	Allow passing of plugin tables and browser tables in NP_Initialize that
	are not the expected length but still large enough for our purposes.
	* plugin/icedteanp/IcedTeaNPPlugin.cc
	(initialize_browser_functions): New function to check size of passed
	browser function table, and initialize 'browser_functions' global
	variable.
	(initialize_plugin_table): New function to check size of passed
	plugin function table, and initialize proper plugin callbacks.
	(NP_Initialize): Make use of initialization helper functions, get
	rid of old size tests and error if the helper functions fail.