ide barrier support, #2

@@ -240,11 +240,40 @@
 	INIT_LIST_HEAD(&q->plug_list);
 
 	blk_queue_activity_fn(q, NULL, NULL);
+
+	q->ordered = QUEUE_ORDERED_NONE;
 }
 
 EXPORT_SYMBOL(blk_queue_make_request);
 
 /**
+ * blk_queue_ordered - does this queue support ordered writes
+ * @q:     the request queue
+ * @flag:  see below
+ *
+ * Description:
+ *   For journalled file systems, doing ordered writes on a commit
+ *   block instead of explicitly doing wait_on_buffer (which is bad
+ *   for performance) can be a big win. Block drivers supporting this
+ *   feature should call this function and indicate so.
+ *
+ *   SCSI drivers usually need to support ordered tags, while others
+ *   may have to do a complete drive cache flush if they are using write
+ *   back caching (or not and lying about it)
+ *
+ *   With this in mind, the values are
+ *             QUEUE_ORDERED_NONE:	the default, doesn't support barrier
+ *             QUEUE_ORDERED_TAG:	supports ordered tags
+ *             QUEUE_ORDERED_FLUSH:	supports barrier through cache flush
+ **/
+void blk_queue_ordered(request_queue_t *q, int flag)
+{
+	q->ordered = flag;
+}
+
+EXPORT_SYMBOL(blk_queue_ordered);
+
+/**
  * blk_queue_bounce_limit - set bounce buffer limit for queue
  * @q:  the request queue for the device
  * @dma_addr:   bus address limit

@@ -1820,6 +1849,8 @@
 
 	if (unlikely(!q))
 		return;
+
+	WARN_ON(!req->ref_count);
 	if (unlikely(--req->ref_count))
 		return;
 

@@ -1986,7 +2017,7 @@
 static int __make_request(request_queue_t *q, struct bio *bio)
 {
 	struct request *req, *freereq = NULL;
-	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra;
+	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra, err;
 	sector_t sector;
 
 	sector = bio->bi_sector;

@@ -2004,7 +2035,11 @@
 
 	spin_lock_prefetch(q->queue_lock);
 
-	barrier = test_bit(BIO_RW_BARRIER, &bio->bi_rw);
+	barrier = bio_barrier(bio);
+	if (barrier && (q->ordered == QUEUE_ORDERED_NONE)) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
 
 	ra = bio->bi_rw & (1 << BIO_RW_AHEAD);
 

@@ -2086,6 +2121,7 @@
 			/*
 			 * READA bit set
 			 */
+			err = -EWOULDBLOCK;
 			if (ra)
 				goto end_io;
 	

@@ -2141,7 +2177,7 @@
 	return 0;
 
 end_io:
-	bio_endio(bio, nr_sectors << 9, -EWOULDBLOCK);
+	bio_endio(bio, nr_sectors << 9, err);
 	return 0;
 }
 

@@ -1688,6 +1688,12 @@
 	if (drive->id->cfs_enable_2 & 0x3000)
 		write_cache(drive, (id->cfs_enable_2 & 0x3000));
 
+	drive->wcache = drive->id->cfs_enable_1 & 0x20;
+	if (drive->wcache) {
+		printk("%s: write cache enabled\n", drive->name);
+		blk_queue_ordered(drive->queue, QUEUE_ORDERED_FLUSH);
+	}
+
 #ifdef CONFIG_BLK_DEV_IDE_TCQ_DEFAULT
 	if (drive->using_dma)
 		HWIF(drive)->ide_dma_queued_on(drive);

@@ -1755,7 +1761,7 @@
 		drive->wcache = 0;
 		/* Cache enabled ? */
 		if (drive->id->csfo & 1)
-		drive->wcache = 1;
+			drive->wcache = 1;
 		/* Cache command set available ? */
 		if (drive->id->cfs_enable_1 & (1<<5))
 			drive->wcache = 1;

279 lines not on this page · raw message

@@ -879,7 +879,7 @@
 		conf = mddev_to_conf(mddev);
 		bio = r1_bio->master_bio;
 		switch(r1_bio->cmd) {
-		case SPECIAL:
+		case WRITESYNC:
 			sync_request_write(mddev, r1_bio);
 			break;
 		case READ:

@@ -989,7 +989,7 @@
 
 	r1_bio->mddev = mddev;
 	r1_bio->sector = sector_nr;
-	r1_bio->cmd = SPECIAL;
+	r1_bio->cmd = WRITESYNC;
 	r1_bio->read_disk = disk;
 
 	bio = r1_bio->master_bio;

@@ -2665,6 +2665,9 @@
 	if (rw == READ && buffer_dirty(bh))
 		buffer_error();
 
+	if (buffer_ordered(bh) && (rw == WRITE))
+		rw = WRITEBARRIER;
+
 	/* Only clear out a write error when rewriting */
 	if (test_set_buffer_req(bh) && rw == WRITE)
 		clear_buffer_write_io_error(bh);

@@ -32,6 +32,7 @@
 		set_buffer_uptodate(bh);
 	else
 		clear_buffer_uptodate(bh);
+	clear_buffer_ordered(bh);
 	unlock_buffer(bh);
 }
 

@@ -474,6 +475,7 @@
 				clear_buffer_dirty(bh);
 				set_buffer_uptodate(bh);
 				bh->b_end_io = journal_end_buffer_io_sync;
+				set_buffer_ordered(bh);
 				submit_bh(WRITE, bh);
 			}
 			cond_resched();

@@ -193,6 +193,8 @@
 	__REQ_PM_SUSPEND,	/* suspend request */
 	__REQ_PM_RESUME,	/* resume request */
 	__REQ_PM_SHUTDOWN,	/* shutdown request */
+	__REQ_BAR_PREFLUSH,	/* barrier pre-flush done */
+	__REQ_BAR_POSTFLUSH,	/* barrier post-flush */
 	__REQ_NR_BITS,		/* stops here */
 };
 

@@ -218,6 +220,8 @@
 #define REQ_PM_SUSPEND	(1 << __REQ_PM_SUSPEND)
 #define REQ_PM_RESUME	(1 << __REQ_PM_RESUME)
 #define REQ_PM_SHUTDOWN	(1 << __REQ_PM_SHUTDOWN)
+#define REQ_BAR_PREFLUSH	(1 << __REQ_BAR_PREFLUSH)
+#define REQ_BAR_POSTFLUSH	(1 << __REQ_BAR_POSTFLUSH)
 
 /*
  * State information carried for REQ_PM_SUSPEND and REQ_PM_RESUME

@@ -344,6 +348,8 @@
 	unsigned long		seg_boundary_mask;
 	unsigned int		dma_alignment;
 
+	unsigned int		ordered;
+
 	struct blk_queue_tag	*queue_tags;
 
 	atomic_t		refcnt;

@@ -368,6 +374,13 @@
 #define QUEUE_FLAG_WRITEFULL	4	/* read queue has been filled */
 #define QUEUE_FLAG_DEAD		5	/* queue being torn down */
 
+/*
+ * write barrier support
+ */
+#define QUEUE_ORDERED_NONE	0	/* no support */
+#define QUEUE_ORDERED_TAG	1	/* supported by tags */
+#define QUEUE_ORDERED_FLUSH	2	/* supported by cache flush */
+
 #define blk_queue_plugged(q)	!list_empty(&(q)->plug_list)
 #define blk_queue_tagged(q)	test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
 #define blk_fs_request(rq)	((rq)->flags & REQ_CMD)

@@ -379,6 +392,10 @@
 #define blk_pm_request(rq)	\
 	((rq)->flags & (REQ_PM_SUSPEND | REQ_PM_RESUME))
 
+#define blk_barrier_rq(rq)	((rq)->flags & REQ_HARDBARRIER)
+#define blk_barrier_preflush(rq)	((rq)->flags & REQ_BAR_PREFLUSH)
+#define blk_barrier_postflush(rq)	((rq)->flags & REQ_BAR_POSTFLUSH)
+
 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
 
 #define rq_data_dir(rq)		((rq)->flags & 1)

@@ -561,6 +578,7 @@
 extern void blk_queue_merge_bvec(request_queue_t *, merge_bvec_fn *);
 extern void blk_queue_dma_alignment(request_queue_t *, int);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
+extern void blk_queue_ordered(request_queue_t *, int);
 
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);

@@ -26,6 +26,7 @@
 	BH_Delay,	/* Buffer is not yet allocated on disk */
 	BH_Boundary,	/* Block is followed by a discontiguity */
 	BH_Write_EIO,	/* I/O error on write */
+	BH_Ordered,	/* ordered write */
 
 	BH_PrivateStart,/* not a state bit, but the first bit available
 			 * for private allocation by other entities

@@ -117,7 +118,8 @@
 BUFFER_FNS(Async_Write, async_write)
 BUFFER_FNS(Delay, delay)
 BUFFER_FNS(Boundary, boundary)
-BUFFER_FNS(Write_EIO,write_io_error)
+BUFFER_FNS(Write_EIO, write_io_error)
+BUFFER_FNS(Ordered, ordered)
 
 #define bh_offset(bh)		((unsigned long)(bh)->b_data & ~PAGE_MASK)
 #define touch_buffer(bh)	mark_page_accessed(bh->b_page)

@@ -81,7 +81,8 @@
 #define READ 0
 #define WRITE 1
 #define READA 2		/* read-ahead  - don't block if no resources */
-#define SPECIAL 4	/* For non-blockdevice requests in request queue */
+#define WRITESYNC 4
+#define WRITEBARRIER	5	/* 1st bit, write, 3rd barrier */
 
 #define SEL_IN		1
 #define SEL_OUT		2

@@ -728,6 +728,7 @@
 	unsigned ata_flash	: 1;	/* 1=present, 0=default */
 	unsigned blocked        : 1;	/* 1=powermanagment told us not to do anything, so sleep nicely */
 	unsigned vdma		: 1;	/* 1=doing PIO over DMA 0=doing normal DMA */
+	unsigned doing_barrier	: 1;	/* state, 1=currently doing flush */
 	unsigned addressing;		/*      : 3;
 					 *  0=28-bit
 					 *  1=48-bit

@@ -773,6 +774,7 @@
 	int		forced_lun;	/* if hdxlun was given at boot */
 	int		lun;		/* logical unit */
 	int		crc_count;	/* crc counter to reduce drive speed */
+	char		special_buf[8];	/* private command buffer */
 	struct list_head list;
 	struct device	gendev;
 	struct semaphore gendev_rel_sem;	/* to deal with device release() */

@@ -240,11 +240,40 @@
 	INIT_LIST_HEAD(&q->plug_list);
 
 	blk_queue_activity_fn(q, NULL, NULL);
+
+	q->ordered = QUEUE_ORDERED_NONE;
 }
 
 EXPORT_SYMBOL(blk_queue_make_request);
 
 /**
+ * blk_queue_ordered - does this queue support ordered writes
+ * @q:     the request queue
+ * @flag:  see below
+ *
+ * Description:
+ *   For journalled file systems, doing ordered writes on a commit
+ *   block instead of explicitly doing wait_on_buffer (which is bad
+ *   for performance) can be a big win. Block drivers supporting this
+ *   feature should call this function and indicate so.
+ *
+ *   SCSI drivers usually need to support ordered tags, while others
+ *   may have to do a complete drive cache flush if they are using write
+ *   back caching (or not and lying about it)
+ *
+ *   With this in mind, the values are
+ *             QUEUE_ORDERED_NONE:	the default, doesn't support barrier
+ *             QUEUE_ORDERED_TAG:	supports ordered tags
+ *             QUEUE_ORDERED_FLUSH:	supports barrier through cache flush
+ **/
+void blk_queue_ordered(request_queue_t *q, int flag)
+{
+	q->ordered = flag;
+}
+
+EXPORT_SYMBOL(blk_queue_ordered);
+
+/**
  * blk_queue_bounce_limit - set bounce buffer limit for queue
  * @q:  the request queue for the device
  * @dma_addr:   bus address limit

@@ -1820,6 +1849,8 @@
 
 	if (unlikely(!q))
 		return;
+
+	WARN_ON(!req->ref_count);
 	if (unlikely(--req->ref_count))
 		return;
 

@@ -1986,7 +2017,7 @@
 static int __make_request(request_queue_t *q, struct bio *bio)
 {
 	struct request *req, *freereq = NULL;
-	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra;
+	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra, err;
 	sector_t sector;
 
 	sector = bio->bi_sector;

@@ -2004,7 +2035,11 @@
 
 	spin_lock_prefetch(q->queue_lock);
 
-	barrier = test_bit(BIO_RW_BARRIER, &bio->bi_rw);
+	barrier = bio_barrier(bio);
+	if (barrier && (q->ordered == QUEUE_ORDERED_NONE)) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
 
 	ra = bio->bi_rw & (1 << BIO_RW_AHEAD);
 

@@ -2086,6 +2121,7 @@
 			/*
 			 * READA bit set
 			 */
+			err = -EWOULDBLOCK;
 			if (ra)
 				goto end_io;
 	

@@ -2141,7 +2177,7 @@
 	return 0;
 
 end_io:
-	bio_endio(bio, nr_sectors << 9, -EWOULDBLOCK);
+	bio_endio(bio, nr_sectors << 9, err);
 	return 0;
 }
 

@@ -1371,6 +1371,7 @@
 static int write_cache (ide_drive_t *drive, int arg)
 {
 	ide_task_t args;
+	int err;
 
 	if (!(drive->id->cfs_enable_2 & 0x3000))
 		return 1;

@@ -1380,7 +1381,10 @@
 			SETFEATURES_EN_WCACHE : SETFEATURES_DIS_WCACHE;
 	args.tfRegister[IDE_COMMAND_OFFSET]	= WIN_SETFEATURES;
 	args.command_type			= ide_cmd_type_parser(&args);
-	(void) ide_raw_taskfile(drive, &args, NULL);
+
+	err = ide_raw_taskfile(drive, &args, NULL);
+	if (err)
+		return err;
 
 	drive->wcache = arg;
 	return 0;

@@ -1688,6 +1692,12 @@
 	if (drive->id->cfs_enable_2 & 0x3000)
 		write_cache(drive, (id->cfs_enable_2 & 0x3000));
 
+	drive->wcache = drive->id->cfs_enable_1 & 0x20;
+	if (drive->wcache) {
+		printk("%s: write cache enabled\n", drive->name);
+		blk_queue_ordered(drive->queue, QUEUE_ORDERED_FLUSH);
+	}
+
 #ifdef CONFIG_BLK_DEV_IDE_TCQ_DEFAULT
 	if (drive->using_dma)
 		HWIF(drive)->ide_dma_queued_on(drive);

@@ -1755,7 +1765,7 @@
 		drive->wcache = 0;
 		/* Cache enabled ? */
 		if (drive->id->csfo & 1)
-		drive->wcache = 1;
+			drive->wcache = 1;
 		/* Cache command set available ? */
 		if (drive->id->cfs_enable_1 & (1<<5))
 			drive->wcache = 1;

284 lines not on this page · raw message

@@ -879,7 +879,7 @@
 		conf = mddev_to_conf(mddev);
 		bio = r1_bio->master_bio;
 		switch(r1_bio->cmd) {
-		case SPECIAL:
+		case WRITESYNC:
 			sync_request_write(mddev, r1_bio);
 			break;
 		case READ:

@@ -989,7 +989,7 @@
 
 	r1_bio->mddev = mddev;
 	r1_bio->sector = sector_nr;
-	r1_bio->cmd = SPECIAL;
+	r1_bio->cmd = WRITESYNC;
 	r1_bio->read_disk = disk;
 
 	bio = r1_bio->master_bio;

@@ -2665,6 +2665,9 @@
 	if (rw == READ && buffer_dirty(bh))
 		buffer_error();
 
+	if (buffer_ordered(bh) && (rw == WRITE))
+		rw = WRITEBARRIER;
+
 	/* Only clear out a write error when rewriting */
 	if (test_set_buffer_req(bh) && rw == WRITE)
 		clear_buffer_write_io_error(bh);

@@ -613,7 +613,9 @@
 	{
 		struct buffer_head *bh = jh2bh(descriptor);
 		set_buffer_uptodate(bh);
+		set_buffer_ordered(bh);
 		sync_dirty_buffer(bh);
+		clear_buffer_ordered(bh);
 		if (unlikely(!buffer_uptodate(bh)))
 			err = -EIO;
 		put_bh(bh);		/* One for getblk() */

@@ -193,6 +193,8 @@
 	__REQ_PM_SUSPEND,	/* suspend request */
 	__REQ_PM_RESUME,	/* resume request */
 	__REQ_PM_SHUTDOWN,	/* shutdown request */
+	__REQ_BAR_PREFLUSH,	/* barrier pre-flush done */
+	__REQ_BAR_POSTFLUSH,	/* barrier post-flush */
 	__REQ_NR_BITS,		/* stops here */
 };
 

@@ -218,6 +220,8 @@
 #define REQ_PM_SUSPEND	(1 << __REQ_PM_SUSPEND)
 #define REQ_PM_RESUME	(1 << __REQ_PM_RESUME)
 #define REQ_PM_SHUTDOWN	(1 << __REQ_PM_SHUTDOWN)
+#define REQ_BAR_PREFLUSH	(1 << __REQ_BAR_PREFLUSH)
+#define REQ_BAR_POSTFLUSH	(1 << __REQ_BAR_POSTFLUSH)
 
 /*
  * State information carried for REQ_PM_SUSPEND and REQ_PM_RESUME

@@ -344,6 +348,8 @@
 	unsigned long		seg_boundary_mask;
 	unsigned int		dma_alignment;
 
+	unsigned int		ordered;
+
 	struct blk_queue_tag	*queue_tags;
 
 	atomic_t		refcnt;

@@ -368,6 +374,13 @@
 #define QUEUE_FLAG_WRITEFULL	4	/* read queue has been filled */
 #define QUEUE_FLAG_DEAD		5	/* queue being torn down */
 
+/*
+ * write barrier support
+ */
+#define QUEUE_ORDERED_NONE	0	/* no support */
+#define QUEUE_ORDERED_TAG	1	/* supported by tags */
+#define QUEUE_ORDERED_FLUSH	2	/* supported by cache flush */
+
 #define blk_queue_plugged(q)	!list_empty(&(q)->plug_list)
 #define blk_queue_tagged(q)	test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
 #define blk_fs_request(rq)	((rq)->flags & REQ_CMD)

@@ -379,6 +392,10 @@
 #define blk_pm_request(rq)	\
 	((rq)->flags & (REQ_PM_SUSPEND | REQ_PM_RESUME))
 
+#define blk_barrier_rq(rq)	((rq)->flags & REQ_HARDBARRIER)
+#define blk_barrier_preflush(rq)	((rq)->flags & REQ_BAR_PREFLUSH)
+#define blk_barrier_postflush(rq)	((rq)->flags & REQ_BAR_POSTFLUSH)
+
 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
 
 #define rq_data_dir(rq)		((rq)->flags & 1)

@@ -561,6 +578,7 @@
 extern void blk_queue_merge_bvec(request_queue_t *, merge_bvec_fn *);
 extern void blk_queue_dma_alignment(request_queue_t *, int);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
+extern void blk_queue_ordered(request_queue_t *, int);
 
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);

@@ -26,6 +26,7 @@
 	BH_Delay,	/* Buffer is not yet allocated on disk */
 	BH_Boundary,	/* Block is followed by a discontiguity */
 	BH_Write_EIO,	/* I/O error on write */
+	BH_Ordered,	/* ordered write */
 
 	BH_PrivateStart,/* not a state bit, but the first bit available
 			 * for private allocation by other entities

@@ -117,7 +118,8 @@
 BUFFER_FNS(Async_Write, async_write)
 BUFFER_FNS(Delay, delay)
 BUFFER_FNS(Boundary, boundary)
-BUFFER_FNS(Write_EIO,write_io_error)
+BUFFER_FNS(Write_EIO, write_io_error)
+BUFFER_FNS(Ordered, ordered)
 
 #define bh_offset(bh)		((unsigned long)(bh)->b_data & ~PAGE_MASK)
 #define touch_buffer(bh)	mark_page_accessed(bh->b_page)

@@ -81,7 +81,8 @@
 #define READ 0
 #define WRITE 1
 #define READA 2		/* read-ahead  - don't block if no resources */
-#define SPECIAL 4	/* For non-blockdevice requests in request queue */
+#define WRITESYNC 4
+#define WRITEBARRIER	5	/* 1st bit, write, 3rd barrier */
 
 #define SEL_IN		1
 #define SEL_OUT		2

@@ -728,6 +728,7 @@
 	unsigned ata_flash	: 1;	/* 1=present, 0=default */
 	unsigned blocked        : 1;	/* 1=powermanagment told us not to do anything, so sleep nicely */
 	unsigned vdma		: 1;	/* 1=doing PIO over DMA 0=doing normal DMA */
+	unsigned doing_barrier	: 1;	/* state, 1=currently doing flush */
 	unsigned addressing;		/*      : 3;
 					 *  0=28-bit
 					 *  1=48-bit

@@ -773,6 +774,7 @@
 	int		forced_lun;	/* if hdxlun was given at boot */
 	int		lun;		/* logical unit */
 	int		crc_count;	/* crc counter to reduce drive speed */
+	char		special_buf[8];	/* private command buffer */
 	struct list_head list;
 	struct device	gendev;
 	struct semaphore gendev_rel_sem;	/* to deal with device release() */

@@ -240,11 +240,40 @@
 	INIT_LIST_HEAD(&q->plug_list);
 
 	blk_queue_activity_fn(q, NULL, NULL);
+
+	q->ordered = QUEUE_ORDERED_NONE;
 }
 
 EXPORT_SYMBOL(blk_queue_make_request);
 
 /**
+ * blk_queue_ordered - does this queue support ordered writes
+ * @q:     the request queue
+ * @flag:  see below
+ *
+ * Description:
+ *   For journalled file systems, doing ordered writes on a commit
+ *   block instead of explicitly doing wait_on_buffer (which is bad
+ *   for performance) can be a big win. Block drivers supporting this
+ *   feature should call this function and indicate so.
+ *
+ *   SCSI drivers usually need to support ordered tags, while others
+ *   may have to do a complete drive cache flush if they are using write
+ *   back caching (or not and lying about it)
+ *
+ *   With this in mind, the values are
+ *             QUEUE_ORDERED_NONE:	the default, doesn't support barrier
+ *             QUEUE_ORDERED_TAG:	supports ordered tags
+ *             QUEUE_ORDERED_FLUSH:	supports barrier through cache flush
+ **/
+void blk_queue_ordered(request_queue_t *q, int flag)
+{
+	q->ordered = flag;
+}
+
+EXPORT_SYMBOL(blk_queue_ordered);
+
+/**
  * blk_queue_bounce_limit - set bounce buffer limit for queue
  * @q:  the request queue for the device
  * @dma_addr:   bus address limit

@@ -1820,6 +1849,8 @@
 
 	if (unlikely(!q))
 		return;
+
+	WARN_ON(!req->ref_count);
 	if (unlikely(--req->ref_count))
 		return;
 

@@ -1986,7 +2017,7 @@
 static int __make_request(request_queue_t *q, struct bio *bio)
 {
 	struct request *req, *freereq = NULL;
-	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra;
+	int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, ra, err;
 	sector_t sector;
 
 	sector = bio->bi_sector;

@@ -2004,7 +2035,11 @@
 
 	spin_lock_prefetch(q->queue_lock);
 
-	barrier = test_bit(BIO_RW_BARRIER, &bio->bi_rw);
+	barrier = bio_barrier(bio);
+	if (barrier && (q->ordered == QUEUE_ORDERED_NONE)) {
+		err = -EOPNOTSUPP;
+		goto end_io;
+	}
 
 	ra = bio->bi_rw & (1 << BIO_RW_AHEAD);
 

@@ -2086,6 +2121,7 @@
 			/*
 			 * READA bit set
 			 */
+			err = -EWOULDBLOCK;
 			if (ra)
 				goto end_io;
 	

@@ -2141,7 +2177,7 @@
 	return 0;
 
 end_io:
-	bio_endio(bio, nr_sectors << 9, -EWOULDBLOCK);
+	bio_endio(bio, nr_sectors << 9, err);
 	return 0;
 }
 

@@ -1371,6 +1371,7 @@
 static int write_cache (ide_drive_t *drive, int arg)
 {
 	ide_task_t args;
+	int err;
 
 	if (!(drive->id->cfs_enable_2 & 0x3000))
 		return 1;

@@ -1380,7 +1381,10 @@
 			SETFEATURES_EN_WCACHE : SETFEATURES_DIS_WCACHE;
 	args.tfRegister[IDE_COMMAND_OFFSET]	= WIN_SETFEATURES;
 	args.command_type			= ide_cmd_type_parser(&args);
-	(void) ide_raw_taskfile(drive, &args, NULL);
+
+	err = ide_raw_taskfile(drive, &args, NULL);
+	if (err)
+		return err;
 
 	drive->wcache = arg;
 	return 0;

@@ -1688,6 +1692,12 @@
 	if (drive->id->cfs_enable_2 & 0x3000)
 		write_cache(drive, (id->cfs_enable_2 & 0x3000));
 
+	drive->wcache = drive->id->cfs_enable_1 & 0x20;
+	if (drive->wcache) {
+		printk("%s: write cache enabled\n", drive->name);
+		blk_queue_ordered(drive->queue, QUEUE_ORDERED_FLUSH);
+	}
+
 #ifdef CONFIG_BLK_DEV_IDE_TCQ_DEFAULT
 	if (drive->using_dma)
 		HWIF(drive)->ide_dma_queued_on(drive);

@@ -1755,7 +1765,7 @@
 		drive->wcache = 0;
 		/* Cache enabled ? */
 		if (drive->id->csfo & 1)
-		drive->wcache = 1;
+			drive->wcache = 1;
 		/* Cache command set available ? */
 		if (drive->id->cfs_enable_1 & (1<<5))
 			drive->wcache = 1;

295 lines not on this page · raw message

@@ -879,7 +879,7 @@
 		conf = mddev_to_conf(mddev);
 		bio = r1_bio->master_bio;
 		switch(r1_bio->cmd) {
-		case SPECIAL:
+		case WRITESYNC:
 			sync_request_write(mddev, r1_bio);
 			break;
 		case READ:

@@ -989,7 +989,7 @@
 
 	r1_bio->mddev = mddev;
 	r1_bio->sector = sector_nr;
-	r1_bio->cmd = SPECIAL;
+	r1_bio->cmd = WRITESYNC;
 	r1_bio->read_disk = disk;
 
 	bio = r1_bio->master_bio;

@@ -2665,6 +2665,9 @@
 	if (rw == READ && buffer_dirty(bh))
 		buffer_error();
 
+	if (buffer_ordered(bh) && (rw == WRITE))
+		rw = WRITESYNC;
+
 	/* Only clear out a write error when rewriting */
 	if (test_set_buffer_req(bh) && rw == WRITE)
 		clear_buffer_write_io_error(bh);

@@ -613,7 +613,9 @@
 	{
 		struct buffer_head *bh = jh2bh(descriptor);
 		set_buffer_uptodate(bh);
+		set_buffer_ordered(bh);
 		sync_dirty_buffer(bh);
+		clear_buffer_ordered(bh);
 		if (unlikely(!buffer_uptodate(bh)))
 			err = -EIO;
 		put_bh(bh);		/* One for getblk() */

@@ -1403,12 +1403,12 @@
 
 submit_io:
 	if (likely(bio->bi_size)) {
-		if (pb->pb_flags & PBF_READ) {
-			submit_bio(READ, bio);
-		} else {
-			submit_bio(WRITE, bio);
-		}
-
+		int cmd = WRITE;
+		if (pb->pb_flags & PBF_READ)
+			cmd = READ;
+		else if (pb->pb_flags & PBF_FLUSH)
+			cmd = WRITESYNC;
+		submit_bio(cmd, bio);
 		if (size)
 			goto next_chunk;
 	} else {

@@ -193,6 +193,8 @@
 	__REQ_PM_SUSPEND,	/* suspend request */
 	__REQ_PM_RESUME,	/* resume request */
 	__REQ_PM_SHUTDOWN,	/* shutdown request */
+	__REQ_BAR_PREFLUSH,	/* barrier pre-flush done */
+	__REQ_BAR_POSTFLUSH,	/* barrier post-flush */
 	__REQ_NR_BITS,		/* stops here */
 };
 

@@ -218,6 +220,8 @@
 #define REQ_PM_SUSPEND	(1 << __REQ_PM_SUSPEND)
 #define REQ_PM_RESUME	(1 << __REQ_PM_RESUME)
 #define REQ_PM_SHUTDOWN	(1 << __REQ_PM_SHUTDOWN)
+#define REQ_BAR_PREFLUSH	(1 << __REQ_BAR_PREFLUSH)
+#define REQ_BAR_POSTFLUSH	(1 << __REQ_BAR_POSTFLUSH)
 
 /*
  * State information carried for REQ_PM_SUSPEND and REQ_PM_RESUME

@@ -344,6 +348,8 @@
 	unsigned long		seg_boundary_mask;
 	unsigned int		dma_alignment;
 
+	unsigned int		ordered;
+
 	struct blk_queue_tag	*queue_tags;
 
 	atomic_t		refcnt;

@@ -368,6 +374,13 @@
 #define QUEUE_FLAG_WRITEFULL	4	/* read queue has been filled */
 #define QUEUE_FLAG_DEAD		5	/* queue being torn down */
 
+/*
+ * write barrier support
+ */
+#define QUEUE_ORDERED_NONE	0	/* no support */
+#define QUEUE_ORDERED_TAG	1	/* supported by tags */
+#define QUEUE_ORDERED_FLUSH	2	/* supported by cache flush */
+
 #define blk_queue_plugged(q)	!list_empty(&(q)->plug_list)
 #define blk_queue_tagged(q)	test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
 #define blk_fs_request(rq)	((rq)->flags & REQ_CMD)

@@ -379,6 +392,10 @@
 #define blk_pm_request(rq)	\
 	((rq)->flags & (REQ_PM_SUSPEND | REQ_PM_RESUME))
 
+#define blk_barrier_rq(rq)	((rq)->flags & REQ_HARDBARRIER)
+#define blk_barrier_preflush(rq)	((rq)->flags & REQ_BAR_PREFLUSH)
+#define blk_barrier_postflush(rq)	((rq)->flags & REQ_BAR_POSTFLUSH)
+
 #define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
 
 #define rq_data_dir(rq)		((rq)->flags & 1)

@@ -561,6 +578,7 @@
 extern void blk_queue_merge_bvec(request_queue_t *, merge_bvec_fn *);
 extern void blk_queue_dma_alignment(request_queue_t *, int);
 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
+extern void blk_queue_ordered(request_queue_t *, int);
 
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);

@@ -26,6 +26,7 @@
 	BH_Delay,	/* Buffer is not yet allocated on disk */
 	BH_Boundary,	/* Block is followed by a discontiguity */
 	BH_Write_EIO,	/* I/O error on write */
+	BH_Ordered,	/* ordered write */
 
 	BH_PrivateStart,/* not a state bit, but the first bit available
 			 * for private allocation by other entities

@@ -117,7 +118,8 @@
 BUFFER_FNS(Async_Write, async_write)
 BUFFER_FNS(Delay, delay)
 BUFFER_FNS(Boundary, boundary)
-BUFFER_FNS(Write_EIO,write_io_error)
+BUFFER_FNS(Write_EIO, write_io_error)
+BUFFER_FNS(Ordered, ordered)
 
 #define bh_offset(bh)		((unsigned long)(bh)->b_data & ~PAGE_MASK)
 #define touch_buffer(bh)	mark_page_accessed(bh->b_page)

@@ -81,7 +81,7 @@
 #define READ 0
 #define WRITE 1
 #define READA 2		/* read-ahead  - don't block if no resources */
-#define SPECIAL 4	/* For non-blockdevice requests in request queue */
+#define WRITESYNC	((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
 
 #define SEL_IN		1
 #define SEL_OUT		2

@@ -728,6 +728,8 @@
 	unsigned ata_flash	: 1;	/* 1=present, 0=default */
 	unsigned blocked        : 1;	/* 1=powermanagment told us not to do anything, so sleep nicely */
 	unsigned vdma		: 1;	/* 1=doing PIO over DMA 0=doing normal DMA */
+	unsigned doing_barrier	: 1;	/* state, 1=currently doing flush */
+	unsigned last_rq_flush	: 1;	/* last rq was a flush */
 	unsigned addressing;		/*      : 3;
 					 *  0=28-bit
 					 *  1=48-bit

@@ -773,6 +775,7 @@
 	int		forced_lun;	/* if hdxlun was given at boot */
 	int		lun;		/* logical unit */
 	int		crc_count;	/* crc counter to reduce drive speed */
+	char		special_buf[8];	/* private command buffer */
 	struct list_head list;
 	struct device	gendev;
 	struct semaphore gendev_rel_sem;	/* to deal with device release() */