[Bug 16443] Regression in ntdll breaks IE6 setup

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=16443


Anastasius Focht <focht at gmx.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |focht at gmx.net




--- Comment #3 from Anastasius Focht <focht at gmx.net>  2008-12-10 05:48:42 ---
Hello,

seems the commit really needs to be reverted to work around such crap.

Yes, I admit it was wrong place to fix.
It was meant to work around Winedevice which treats drivers the same way as
standard win32 subsystem PE when it comes to loading and resolving imports.
The difference is that the entry point should not be called automagically by
loader in this case (native subsystem).
This is done by Winedevice - the driver init routine is explicitly resolved 
and called, passing proper parameters.

Apart from that technical aspects this is also a bug with M$ IE6 package.
Broken stuff as usual... I wonder how such stuff gets shipped, passing QA.

The 'corpol.dll' binary is actually invalid from technical standpoint, the
optional header subsystem attribute should never be "native".
If you look at dll imports you see standard win32 subsystem imports which is
invalid for "true" native subsystem binaries.
This is in fact a standard win32 subsystem binary, despite the incorrect
subsystem value.

It looks like the binary was later corrected by Services Packs/Windows updates
(Subsystem: 0x0003  -> WINDOWS_CUI).

Regards


-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.