CURSORICON_SimulateLoadingFromResourceW param check. VFP6 EXEs now work.

[C. Daniel Mojoli B.]

Alexandre Julliard wrote:

>"C. Daniel Mojoli B." <cdmojoli at idea.com.py> writes:
>
>  
>
>>7) We hit L513 and cause an exception when executing the following
>>expression: bits->idEntries[i]. The problem is that we are looping
>>with our index bound to the garbage number of icon entries! That
>>garbage is almost assured to be too large and we loop past the
>>assigned memory.
>>    
>>
>
>The proper fix is to make sure the code doesn't crash no matter what
>garbage it gets as input. Checking the magic number doesn't guarantee
>that the rest of the file is correct.
>  
>
I agree with the principle, but still advocate the inclusion of the 
magic number patch because it allows to fail-fast. Even when someone 
eventually patches for complete validity and bounds checking, the 
fail-fast magic number test remains useful.

(And yes, Uwe, your test reflected the condition I encountered. Sorry 
for not answering earlier.)