[patch 6/15] larger b_size, and misc fixlets

[patch 6/15] larger b_size, and misc fixlets

[Andrew Morton]

Miscellany.

- make the printk in buffer_io_error() sector_t-aware.

- Some buffer.c cleanups from AntonA: remove a couple of !uptodate
  checks, and set a new buffer's b_blocknr to -1 in a more sensible
  place.

- Make buffer_head.b_size a 32-bit quantity.  Needed for 64k pagesize
  on ia64.  Does not increase sizeof(struct buffer_head).

=====================================

--- 2.5.16/fs/buffer.c~sector_t-printing	Sun May 19 11:49:47 2002
+++ 2.5.16-akpm/fs/buffer.c	Sun May 19 12:02:57 2002
@@ -179,8 +179,8 @@ __clear_page_buffers(struct page *page)
 
 static void buffer_io_error(struct buffer_head *bh)
 {
-	printk(KERN_ERR "Buffer I/O error on device %s, logical block %ld\n",
-			bdevname(bh->b_bdev), bh->b_blocknr);
+	printk(KERN_ERR "Buffer I/O error on device %s, logical block %Ld\n",
+			bdevname(bh->b_bdev), (u64)bh->b_blocknr);
 }
 
 /*
@@ -189,12 +189,12 @@ static void buffer_io_error(struct buffe
  */
 void end_buffer_io_sync(struct buffer_head *bh, int uptodate)
 {
-	if (!uptodate)
-		buffer_io_error(bh);
-	if (uptodate)
+	if (uptodate) {
 		set_buffer_uptodate(bh);
-	else
+	} else {
+		buffer_io_error(bh);
 		clear_buffer_uptodate(bh);
+	}
 	unlock_buffer(bh);
 	put_bh(bh);
 }
@@ -519,14 +519,12 @@ static void end_buffer_async_read(struct
 
 	BUG_ON(!buffer_async_read(bh));
 
-	if (!uptodate)
-		buffer_io_error(bh);
-
 	page = bh->b_page;
 	if (uptodate) {
 		set_buffer_uptodate(bh);
 	} else {
 		clear_buffer_uptodate(bh);
+		buffer_io_error(bh);
 		SetPageError(page);
 	}
 
@@ -579,13 +577,11 @@ static void end_buffer_async_write(struc
 
 	BUG_ON(!buffer_async_write(bh));
 
-	if (!uptodate)
-		buffer_io_error(bh);
-
 	page = bh->b_page;
 	if (uptodate) {
 		set_buffer_uptodate(bh);
 	} else {
+		buffer_io_error(bh);
 		clear_buffer_uptodate(bh);
 		SetPageError(page);
 	}
@@ -907,6 +903,7 @@ try_again:
 
 		bh->b_bdev = NULL;
 		bh->b_this_page = head;
+		bh->b_blocknr = -1;
 		head = bh;
 
 		bh->b_state = 0;
@@ -2442,7 +2439,6 @@ static void init_buffer_head(void *data,
 		struct buffer_head * bh = (struct buffer_head *)data;
 
 		memset(bh, 0, sizeof(*bh));
-		bh->b_blocknr = -1;
 		INIT_LIST_HEAD(&bh->b_assoc_buffers);
 	}
 }
--- 2.5.16/include/linux/buffer_head.h~sector_t-printing	Sun May 19 11:49:47 2002
+++ 2.5.16-akpm/include/linux/buffer_head.h	Sun May 19 12:02:57 2002
@@ -44,7 +44,7 @@ struct buffer_head {
 	struct page *b_page;		/* the page this bh is mapped to */
 
 	sector_t b_blocknr;		/* block number */
-	unsigned short b_size;		/* block size */
+	u32 b_size;			/* block size */
 	char *b_data;			/* pointer to data block */
 
 	struct block_device *b_bdev;


-

Re: [patch 6/15] larger b_size, and misc fixlets

[Andreas Dilger]

On May 19, 2002  12:39 -0700, Andrew Morton wrote:
> - make the printk in buffer_io_error() sector_t-aware.
> =====================================
> --- 2.5.16/fs/buffer.c~sector_t-printing	Sun May 19 11:49:47 2002
> +++ 2.5.16-akpm/fs/buffer.c	Sun May 19 12:02:57 2002
> @@ -179,8 +179,8 @@ __clear_page_buffers(struct page *page)
>  
>  static void buffer_io_error(struct buffer_head *bh)
>  {
> -	printk(KERN_ERR "Buffer I/O error on device %s, logical block %ld\n",
> -			bdevname(bh->b_bdev), bh->b_blocknr);
> +	printk(KERN_ERR "Buffer I/O error on device %s, logical block %Ld\n",
> +			bdevname(bh->b_bdev), (u64)bh->b_blocknr);
>  }

Not that I'm a 64-bit system user/developer, but it is my understanding
that u64 == long on a 64-bit platform, so your cast to u64 does not
actually change the type of b_blocknr as far as printk is concerned.
You would need to cast it to unsigned long long instead.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re: [patch 6/15] larger b_size, and misc fixlets

[Andrew Morton]

Andreas Dilger wrote:
> 
> On May 19, 2002  12:39 -0700, Andrew Morton wrote:
> > - make the printk in buffer_io_error() sector_t-aware.
> > =====================================
> > --- 2.5.16/fs/buffer.c~sector_t-printing      Sun May 19 11:49:47 2002
> > +++ 2.5.16-akpm/fs/buffer.c   Sun May 19 12:02:57 2002
> > @@ -179,8 +179,8 @@ __clear_page_buffers(struct page *page)
> >
> >  static void buffer_io_error(struct buffer_head *bh)
> >  {
> > -     printk(KERN_ERR "Buffer I/O error on device %s, logical block %ld\n",
> > -                     bdevname(bh->b_bdev), bh->b_blocknr);
> > +     printk(KERN_ERR "Buffer I/O error on device %s, logical block %Ld\n",
> > +                     bdevname(bh->b_bdev), (u64)bh->b_blocknr);
> >  }
> 
> Not that I'm a 64-bit system user/developer, but it is my understanding
> that u64 == long on a 64-bit platform, so your cast to u64 does not
> actually change the type of b_blocknr as far as printk is concerned.
> You would need to cast it to unsigned long long instead.
> 

Yes, I suppose so.  That more closely matches what "%L" does.

-

Re: [patch 6/15] larger b_size, and misc fixlets

[Anton Altaparmakov]

On Sun, 19 May 2002, Andrew Morton wrote:
> Andreas Dilger wrote:
> > On May 19, 2002  12:39 -0700, Andrew Morton wrote:
> > > -     printk(KERN_ERR "Buffer I/O error on device %s, logical block %ld\n",
> > > -                     bdevname(bh->b_bdev), bh->b_blocknr);
> > > +     printk(KERN_ERR "Buffer I/O error on device %s, logical block %Ld\n",
> > > +                     bdevname(bh->b_bdev), (u64)bh->b_blocknr);
> > 
> > Not that I'm a 64-bit system user/developer, but it is my understanding
> > that u64 == long on a 64-bit platform, so your cast to u64 does not
> > actually change the type of b_blocknr as far as printk is concerned.
> > You would need to cast it to unsigned long long instead.
> 
> Yes, I suppose so.  That more closely matches what "%L" does.

/me can't help it: Didn't I say earlier on that one has to use (unsigned)
long long and not u64? (-; But noone would listen...

But to be fully correct, if you want unsigned long long you really ought
to write %Lu not %Ld... (-8

Cheers,
	Anton
-- 
Anton Altaparmakov <aia21 at cantab.net> (replace at with @)
Linux NTFS maintainer / IRC: #ntfs on irc.openprojects.net
WWW: http://linux-ntfs.sf.net/ & http://www-stu.christs.cam.ac.uk/~aia21/

Re: [patch 6/15] larger b_size, and misc fixlets

[Peter Chubb]

>>>>> "Anton" == Anton Altaparmakov <aia21@cantab.net> writes:

>> > Not that I'm a 64-bit system user/developer, but it is my
>> understanding > that u64 == long on a 64-bit platform, so your cast
>> to u64 does not > actually change the type of b_blocknr as far as
>> printk is concerned.  > You would need to cast it to unsigned long
>> long instead.
>> 
>> Yes, I suppose so.  That more closely matches what "%L" does.

Anton> /me can't help it: Didn't I say earlier on that one has to use
Anton> (unsigned) long long and not u64? (-; But noone would listen...

The current C standard guarantees that long long is *at least* 64
bits.

So you're right.  Can we introduce a new pair of types, ull_t and ll_t
to reduce typing?  (unsigned long long) tends to make lines overflow
the 80-char boundary bit much, but (ull_t) is a lot shorter.

Peter C