[hbirr] 13173: Fixed the range which must be zeroed if the write starts after the end of the file.

Fixed the range which must be zeroed if the write starts after the end
of the file.
Modified: branches/ext2/reactos/drivers/fs/ext2/src/write.c
  _____  

Modified: branches/ext2/reactos/drivers/fs/ext2/src/write.c
--- branches/ext2/reactos/drivers/fs/ext2/src/write.c	2005-01-20
23:21:14 UTC (rev 13172)
+++ branches/ext2/reactos/drivers/fs/ext2/src/write.c	2005-01-20
23:24:51 UTC (rev 13173)
@@ -544,7 +544,7 @@

 
 			LogicalBlockSize = EXT2_MIN_BLOCK_SIZE <<
PtrVCB->LogBlockSize;
 		
-			if ( ByteOffset.QuadPart + WriteLength >=
PtrReqdFCB->CommonFCBHeader.FileSize.QuadPart )
+			if ( ByteOffset.QuadPart + WriteLength >
PtrReqdFCB->CommonFCBHeader.FileSize.QuadPart )
 			{
 				if( PagingIo )
 				{
@@ -621,7 +621,7 @@
 
 							//	Start
offset is the New File size
 
StartOffsetForZeroing.QuadPart = 
-
PtrReqdFCB->CommonFCBHeader.FileSize.QuadPart + WriteLength;
+
ByteOffset.QuadPart + WriteLength;
 							//	End
offset is the New Allocation size
 
EndOffsetForZeroing.QuadPart =
PtrReqdFCB->CommonFCBHeader.AllocationSize.QuadPart;
 						}

[Gunnar Dalsnes]

IRC RELAY:
Filip said that he had tested this driver on Windows and run some 
Hardware Compatibility Tests, and it worked fine. But he also said "some 
of the code paths aren't excercised on windows", so some bugs might be 
"hiding". Still i think its strange that it should take different paths 
on Ros than on Windows. It could also be that Windows is more forgiving 
to prog. errors in some cases.

Gunnar

hbirr(a)svn.reactos.com wrote:
>   Fixed the range which must be zeroed if the write
> starts after the end of the file.
>  
>  Modified: branches/ext2/reactos/drivers/fs/ext2/src/write.c