[Ocfs2-devel] Re: [PATCH]2.6 friendly access to bh devices

* [Ocfs2-devel] Re: [PATCH]2.6 friendly access to bh devices
  2004-02-17 16:53 [Ocfs2-devel] " Rusty Lynch
@ 2004-02-17 21:27 ` Mark Fasheh
  0 siblings, 0 replies; 3+ messages in thread
From: Mark Fasheh @ 2004-02-17 21:27 UTC (permalink / raw)
  To: ocfs2-devel

Cool, this one's been committed too. I added an extra set of parens around
the _b in _hashfn.
	--Mark

On Tue, Feb 17, 2004 at 02:53:11PM -0800, Rusty Lynch wrote:
> The new buffer_head struct references a block_device which points
> to a specific device instead of directly pointing to a device.
> 
> This patch adds a new BH_GET_DEVICE(bh) macro which will expand
> correctly across 2.4 and 2.6 builds.
> 
>     --rusty
> 
> Index: src/hash.c
> ===================================================================
> --- src/hash.c	(revision 31)
> +++ src/hash.c	(working copy)
> @@ -685,7 +685,7 @@
>  		(((block)<<(ocfs_bh_hash_shift - 6)) ^ ((block) >> 13) ^ \
>  		((block) << (ocfs_bh_hash_shift - 12))))
>  #define ocfs_bh_sem_hash_fn(_b)   \
> -	(_hashfn((unsigned int)((_b)->b_dev), (_b)->b_blocknr) & ocfs_bh_hash_shift)
> +	(_hashfn((unsigned int)BH_GET_DEVICE(_b), (_b)->b_blocknr) & ocfs_bh_hash_shift)
>  
>  int ocfs_bh_sem_hash_init()
>  {
> @@ -748,7 +748,7 @@
>  		}
>  		sem = list_entry (iter, ocfs_bh_sem, s_list);
>  		if (sem->s_blocknr == bh->b_blocknr &&
> -		    sem->s_dev == bh->b_dev) {
> +		    sem->s_dev == BH_GET_DEVICE(bh)) {
>  			if (atomic_read(&sem->s_refcnt)==0) {
>  				if (sem->s_bh) {
>  					LOG_ERROR_STR("refcount was zero but s_bh not NULL!");
> @@ -800,7 +800,7 @@
>  			newsem->s_bh = NULL;
>  			atomic_set(&newsem->s_refcnt, 0);
>  			newsem->s_blocknr = bh->b_blocknr;
> -			newsem->s_dev = bh->b_dev;
> +			newsem->s_dev = BH_GET_DEVICE(bh);
>  			init_MUTEX (&newsem->s_sem);
>  			init_waitqueue_head(&newsem->s_wait);
>  			newsem->s_pid = 0;
> Index: src/inc/io.h
> ===================================================================
> --- src/inc/io.h	(revision 31)
> +++ src/inc/io.h	(working copy)
> @@ -166,8 +166,12 @@
>  }
>  #endif
>  
> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)
> +#define BH_GET_DEVICE(bh) ((bh->b_bdev)->bd_dev)
> +#else
> +#define BH_GET_DEVICE(bh) (bh->b_dev)
> +#endif
>  
> -
>  static inline void OCFS_BH_PUT_DATA(struct buffer_head *bh)
>  {
>  	kunmap((bh)->b_page);
--
Mark Fasheh
Software Developer, Oracle Corp
mark.fasheh@oracle.com

^ permalink raw reply	[flat|nested] 3+ messages in thread