New patches for 2.2.18pre24 raw IO (fix for bounce buffer copy)

Hi,

I have pushed another set of raw IO patches out, this time to fix a bug with bounce buffer copying when running on highmem boxes. It is likely to affect any bounce buffer copies using non-page-aligned accesses if both highmem and normal pages are involved in the kiobuf.

The specific new patch added in this patchset is attached below. The full set has been uploaded as

	kiobuf-2.2.18pre24-B.tar.gz

at

ftp.*.kernel.org:/pub/linux/kernel/people/sct/raw-io/
and ftp.uk.linux.org:/pub/linux/sct/fs/raw-io/

This one really should kill all known bugs, dead. Please stress it out and let me know if anybody encounters any further problems. A merge of all of the pending raw IO fixes into 2.4 should be happening soon once the current VM changes for marking pages dirty are working.

Cheers,
Stephen

@@ -211,10 +211,10 @@
 		unsigned long kin, kout;
 		int pagelen = length;
 		
+		if ((pagelen+offset) > PAGE_SIZE)
+			pagelen = PAGE_SIZE - offset;
+			
 		if (bounce_page) {
-			if ((pagelen+offset) > PAGE_SIZE)
-				pagelen = PAGE_SIZE - offset;
-		
 			if (direction == COPY_TO_BOUNCE) {
 				kin  = kmap(page, KM_READ);
 				kout = kmap(bounce_page, KM_WRITE);

This fix is missing:

@@ -455,7 +455,7 @@
 	unsigned long		ptr, end;
 	int			err;
 	struct mm_struct *	mm;
-	struct vm_area_struct *	vma = 0;
+	struct vm_area_struct *	vma;
 	unsigned long		page;
 	struct page *		map;
 	int			doublepage = 0;

@@ -478,6 +478,7 @@
 		return err;
 
  repeat:
+	vma = NULL;
 	down(&mm->mmap_sem);
 
 	err = -EFAULT;

Andrea
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Hi,