size art

[Dmitry Timoshkov]

"Dmitry Timoshkov" <dmitry at baikal.ru> wrote:

> Changelog:
>     Relax a bit PE consistence checks.
>     Return BINARY_DOS type if extended header was not recognized.

--- cvs/hq/wine/memory/virtual.c Sat Aug 17 18:42:29 2002
+++ wine/memory/virtual.c Thu Aug 22 15:00:03 2002
@@ -645,7 +645,7 @@
         DWORD size;
 
         /* a few sanity checks */
-        size = sec->VirtualAddress + ROUND_SIZE( sec->VirtualAddress, sec->Misc.VirtualSize );
+        size = sec->VirtualAddress + ROUND_SIZE( 0, sec->Misc.VirtualSize );
         if (sec->VirtualAddress > total_size || size > total_size || size < sec->VirtualAddress)
         {
             ERR_(module)( "Section %.8s too large (%lx+%lx/%lx)\n",

Apparently the above bits of the patch are wrong. Please ignore them.
Perhaps (size > total_size) check should be completely dropped and
PE image mapper should be prepared to handle sec->Misc.VirtualSize == 0
taking sec->SizeOfRawData into account instead?

-- 
Dmitry.