[PATCH] deprecate memclear_highpage_flush deprecation warnings

[PATCH] deprecate memclear_highpage_flush deprecation warnings

[Christoph Hellwig]

Recent builds get tons of warnings about memclear_highpage_flush
beeing deprecated.  Turns out it's replaced by zero_user_page
which takes an additional argument.

Now folks, deprecated is for actual functionality going away, there
is no need to mark the old name deprecated for such a trivial
paramter change and rename.  This stuff should go to Linus in one
patch that doesn't create utterly useless warnings and keeps around
stale interfaces.

Here's a patch to kill memclear_highpage_flush and convert the reaming
user to make the build a littler more silent, it's more than noisy
enough due to all the useless addition of __deprecated or __must_check
to widely used functionality and gcc stupid false positives.


Signed-off-by: Christoph Hellwig <hch@lst.de>

Index: linux-2.6/fs/nfs/read.c
===================================================================
--- linux-2.6.orig/fs/nfs/read.c	2007-05-15 14:19:59.000000000 +0200
+++ linux-2.6/fs/nfs/read.c	2007-05-15 14:21:24.000000000 +0200
@@ -79,7 +79,7 @@ void nfs_readdata_release(void *data)
 static
 int nfs_return_empty_page(struct page *page)
 {
-	memclear_highpage_flush(page, 0, PAGE_CACHE_SIZE);
+	zero_user_page(page, 0, PAGE_CACHE_SIZE, KM_USER0);
 	SetPageUptodate(page);
 	unlock_page(page);
 	return 0;
@@ -103,10 +103,10 @@ static void nfs_readpage_truncate_uninit
 	pglen = PAGE_CACHE_SIZE - base;
 	for (;;) {
 		if (remainder <= pglen) {
-			memclear_highpage_flush(*pages, base, remainder);
+			zero_user_page(*pages, base, remainder, KM_USER0);
 			break;
 		}
-		memclear_highpage_flush(*pages, base, pglen);
+		zero_user_page(*pages, base, pglen, KM_USER0);
 		pages++;
 		remainder -= pglen;
 		pglen = PAGE_CACHE_SIZE;
@@ -130,7 +130,7 @@ static int nfs_readpage_async(struct nfs
 		return PTR_ERR(new);
 	}
 	if (len < PAGE_CACHE_SIZE)
-		memclear_highpage_flush(page, len, PAGE_CACHE_SIZE - len);
+		zero_user_page(page, len, PAGE_CACHE_SIZE - len, KM_USER0);
 
 	nfs_list_add_request(new, &one_request);
 	if (NFS_SERVER(inode)->rsize < PAGE_CACHE_SIZE)
@@ -532,7 +532,7 @@ readpage_async_filler(void *data, struct
 			return PTR_ERR(new);
 	}
 	if (len < PAGE_CACHE_SIZE)
-		memclear_highpage_flush(page, len, PAGE_CACHE_SIZE - len);
+		zero_user_page(page, len, PAGE_CACHE_SIZE - len, KM_USER0);
 	nfs_pageio_add_request(desc->pgio, new);
 	return 0;
 }
Index: linux-2.6/fs/nfs/write.c
===================================================================
--- linux-2.6.orig/fs/nfs/write.c	2007-05-15 14:19:59.000000000 +0200
+++ linux-2.6/fs/nfs/write.c	2007-05-15 14:21:21.000000000 +0200
@@ -168,7 +168,7 @@ static void nfs_mark_uptodate(struct pag
 	if (count != nfs_page_length(page))
 		return;
 	if (count != PAGE_CACHE_SIZE)
-		memclear_highpage_flush(page, count, PAGE_CACHE_SIZE - count);
+		zero_user_page(page, count, PAGE_CACHE_SIZE - count, KM_USER0);
 	SetPageUptodate(page);
 }
 
Index: linux-2.6/fs/xfs/linux-2.6/xfs_lrw.c
===================================================================
--- linux-2.6.orig/fs/xfs/linux-2.6/xfs_lrw.c	2007-05-15 14:19:59.000000000 +0200
+++ linux-2.6/fs/xfs/linux-2.6/xfs_lrw.c	2007-05-15 14:21:27.000000000 +0200
@@ -159,7 +159,7 @@ xfs_iozero(
 		if (status)
 			goto unlock;
 
-		memclear_highpage_flush(page, offset, bytes);
+		zero_user_page(page, offset, bytes, KM_USER0);
 
 		status = mapping->a_ops->commit_write(NULL, page, offset,
 							offset + bytes);
Index: linux-2.6/include/linux/highmem.h
===================================================================
--- linux-2.6.orig/include/linux/highmem.h	2007-05-15 14:19:59.000000000 +0200
+++ linux-2.6/include/linux/highmem.h	2007-05-15 14:21:12.000000000 +0200
@@ -110,12 +110,6 @@ static inline void clear_highpage(struct
 		kunmap_atomic(kaddr, (km_type));		\
 	} while (0)
 
-static inline void __deprecated memclear_highpage_flush(struct page *page,
-			unsigned int offset, unsigned int size)
-{
-	zero_user_page(page, offset, size, KM_USER0);
-}
-
 #ifndef __HAVE_ARCH_COPY_USER_HIGHPAGE
 
 static inline void copy_user_highpage(struct page *to, struct page *from,

Re: [PATCH] deprecate memclear_highpage_flush deprecation warnings

[Andrew Morton]

On Tue, 15 May 2007 14:24:40 +0200 Christoph Hellwig <hch@lst.de> wrote:

> Recent builds get tons of warnings about memclear_highpage_flush
> beeing deprecated.  Turns out it's replaced by zero_user_page
> which takes an additional argument.
> 
> Now folks, deprecated is for actual functionality going away, there
> is no need to mark the old name deprecated for such a trivial
> paramter change and rename.

Yes, it's rather trivial, but memclear_highpage_flush() *is* going away.

>  This stuff should go to Linus in one
> patch that doesn't create utterly useless warnings and keeps around
> stale interfaces.
> 
> Here's a patch to kill memclear_highpage_flush and convert the reaming
> user to make the build a littler more silent, it's more than noisy
> enough due to all the useless addition of __deprecated or __must_check
> to widely used functionality and gcc stupid false positives.

Patches have been prepared which convert all in-kernel users.  I thought
I'd dtrt and feed them through maintainers, but that takes time.  The
laggards are on cc ;)

Re: [PATCH] deprecate memclear_highpage_flush deprecation warnings

[Christoph Hellwig]

On Tue, May 15, 2007 at 09:00:32AM -0700, Andrew Morton wrote:
> On Tue, 15 May 2007 14:24:40 +0200 Christoph Hellwig <hch@lst.de> wrote:
> 
> > Recent builds get tons of warnings about memclear_highpage_flush
> > beeing deprecated.  Turns out it's replaced by zero_user_page
> > which takes an additional argument.
> > 
> > Now folks, deprecated is for actual functionality going away, there
> > is no need to mark the old name deprecated for such a trivial
> > paramter change and rename.
> 
> Yes, it's rather trivial, but memclear_highpage_flush() *is* going away.

Well, it's renamed to zero_user_page and gained an argument.  We do
changed like that all the time and don't put silly deprecation warnings
in.

> > Here's a patch to kill memclear_highpage_flush and convert the reaming
> > user to make the build a littler more silent, it's more than noisy
> > enough due to all the useless addition of __deprecated or __must_check
> > to widely used functionality and gcc stupid false positives.
> 
> Patches have been prepared which convert all in-kernel users.  I thought
> I'd dtrt and feed them through maintainers, but that takes time.  The
> laggards are on cc ;)

No, there is absolutely no point in feeding this through maintainers.
For something as trivial as a rename and adding an argument doing one
patch is enough and the right thing.  We can't feed any trivial global
search & replace through maintainer and generate warnings in the
meantime.