[PATCH] Initial md/raid5 support for 2.5 (with bio)

With this patch raid5 works.  There is still some more
work to though.

- uses bio instead of buffer_head
- stripe cache is now a fixed size.
   If read requests are smaller, we read the whole block anyway
   If write reqeusts are smaller, we pre-read.
- stripe_head is now variable sized with an array of structures at
  the end.  We allocate extra space depending on how many devices
  are in the array.
  stripe_head has it's very own slab cache.
- store and use bdev for each device in array

by-passing the cache for reads is currently disabled.  I need to
think through the implications (and implementation) of allowing
large bion that are larger than the stripe cache to go directly
to the device (if it isn't failed of-course).

drivers/md/xor.c

@@ -26,31 +26,30 @@
 static struct xor_block_template *active_template;
 
 void
-xor_block(unsigned int count, struct buffer_head **bh_ptr)
+xor_block(unsigned int count, unsigned int bytes, void **ptr)
 {
 	unsigned long *p0, *p1, *p2, *p3, *p4;
-	unsigned long bytes = bh_ptr[0]->b_size;
 
-	p0 = (unsigned long *) bh_ptr[0]->b_data;
-	p1 = (unsigned long *) bh_ptr[1]->b_data;
+	p0 = (unsigned long *) ptr[0];
+	p1 = (unsigned long *) ptr[1];
 	if (count == 2) {
 		active_template->do_2(bytes, p0, p1);
 		return;
 	}
 
-	p2 = (unsigned long *) bh_ptr[2]->b_data;
+	p2 = (unsigned long *) ptr[2];
 	if (count == 3) {
 		active_template->do_3(bytes, p0, p1, p2);
 		return;
 	}
 
-	p3 = (unsigned long *) bh_ptr[3]->b_data;
+	p3 = (unsigned long *) ptr[3];
 	if (count == 4) {
 		active_template->do_4(bytes, p0, p1, p2, p3);
 		return;
 	}
 
-	p4 = (unsigned long *) bh_ptr[4]->b_data;
+	p4 = (unsigned long *) ptr[4];
 	active_template->do_5(bytes, p0, p1, p2, p3, p4);
 }
 

include/linux/raid/raid5.h

@@ -7,21 +7,21 @@
 /*
  *
  * Each stripe contains one buffer per disc.  Each buffer can be in
- * one of a number of states determined by bh_state.  Changes between
+ * one of a number of states stored in "flags".  Changes between
  * these states happen *almost* exclusively under a per-stripe
- * spinlock.  Some very specific changes can happen in b_end_io, and
+ * spinlock.  Some very specific changes can happen in bi_end_io, and
  * these are not protected by the spin lock.
  *
- * The bh_state bits that are used to represent these states are:
- *   BH_Uptodate, BH_Lock
+ * The flag bits that are used to represent these states are:
+ *   R5_UPTODATE and R5_LOCKED
  *
- * State Empty == !Uptodate, !Lock
+ * State Empty == !UPTODATE, !LOCK
  *        We have no data, and there is no active request
- * State Want == !Uptodate, Lock
+ * State Want == !UPTODATE, LOCK
  *        A read request is being submitted for this block
- * State Dirty == Uptodate, Lock
+ * State Dirty == UPTODATE, LOCK
  *        Some new data is in this buffer, and it is being written out
- * State Clean == Uptodate, !Lock
+ * State Clean == UPTODATE, !LOCK
  *        We have valid data which is the same as on disc
  *
  * The possible state transitions are:
@@ -124,24 +124,29 @@
  * plus raid5d if it is handling it, plus one for each active request
  * on a cached buffer.
  */
+
 struct stripe_head {
 	struct stripe_head	*hash_next, **hash_pprev; /* hash pointers */
 	struct list_head	lru;			/* inactive_list or handle_list */
 	struct raid5_private_data	*raid_conf;
-	struct buffer_head	*bh_cache[MD_SB_DISKS];	/* buffered copy */
-	struct buffer_head	*bh_read[MD_SB_DISKS];	/* read request buffers of the MD device */
-	struct buffer_head	*bh_write[MD_SB_DISKS];	/* write request buffers of the MD device */
-	struct buffer_head	*bh_written[MD_SB_DISKS]; /* write request buffers of the MD device that have been scheduled for write */
-	struct page		*bh_page[MD_SB_DISKS];	/* saved bh_cache[n]->b_page when reading around the cache */
-	unsigned long		sector;			/* sector of this row */
-	int			size;			/* buffers size */
+	sector_t		sector;			/* sector of this row */
 	int			pd_idx;			/* parity disk index */
 	unsigned long		state;			/* state flags */
 	atomic_t		count;			/* nr of active thread/requests */
 	spinlock_t		lock;
-	int			sync_redone;
+	struct r5dev {
+		struct bio	req;
+		struct bio_vec	vec;
+		struct page	*page;
+		struct bio	*toread, *towrite, *written;
+		sector_t	sector;			/* sector of this page */
+		unsigned long	flags;
+	} dev[1]; /* allocated with extra space depending of RAID geometry */
 };
-
+/* Flags */
+#define	R5_UPTODATE	0	/* page contains current data */
+#define	R5_LOCKED	1	/* IO has been submitted on "req" */
+#define	R5_OVERWRITE	2	/* towrite covers whole page */
 
 /*
  * Write method
@@ -187,6 +192,7 @@ struct stripe_head {
 
 struct disk_info {
 	kdev_t	dev;
+	struct block_device *bdev;
 	int	operational;
 	int	number;
 	int	raid_disk;
@@ -201,7 +207,6 @@ struct raid5_private_data {
 	mdk_thread_t		*thread, *resync_thread;
 	struct disk_info	disks[MD_SB_DISKS];
 	struct disk_info	*spare;
-	int			buffer_size;
 	int			chunk_size, level, algorithm;
 	int			raid_disks, working_disks, failed_disks;
 	int			resync_parity;
@@ -210,6 +215,9 @@ struct raid5_private_data {
 	struct list_head	handle_list; /* stripes needing handling */
 	struct list_head	delayed_list; /* stripes that have plugged requests */
 	atomic_t		preread_active_stripes; /* stripes with scheduled io */
+
+	char			cache_name[20];
+	kmem_cache_t		*slab_cache; /* for allocating stripes */
 	/*
 	 * Free stripes pool
 	 */

include/linux/raid/xor.h

@@ -5,7 +5,7 @@
 
 #define MAX_XOR_BLOCKS 5
 
-extern void xor_block(unsigned int count, struct buffer_head **bh_ptr);
+extern void xor_block(unsigned int count, unsigned int bytes, void **ptr);
 
 struct xor_block_template {
         struct xor_block_template *next;