Merge http://gkernel.bkbits.net/misc-2.5

into home.transmeta.com:/home/torvalds/v2.5/linux

drivers/block/deadline-iosched.c

@@ -118,6 +118,8 @@ deadline_find_hash(struct deadline_data *dd, sector_t offset)
 
 	while ((entry = next) != hash_list) {
 		next = entry->next;
+		prefetch(next);
+		
 		drq = list_entry_hash(entry);
 
 		BUG_ON(!drq->hash_valid_count);
@@ -191,6 +193,8 @@ deadline_merge(request_queue_t *q, struct list_head **insert, struct bio *bio)
 	while ((entry = entry->prev) != sort_list) {
 		__rq = list_entry_rq(entry);
 
+		prefetch(entry->prev);
+
 		BUG_ON(__rq->flags & REQ_STARTED);
 
 		if (!(__rq->flags & REQ_CMD))
@@ -298,6 +302,8 @@ static void deadline_move_requests(struct deadline_data *dd, struct request *rq)
 		struct list_head *nxt = rq->queuelist.next;
 		int this_rq_cost;
 
+		prefetch(nxt);
+
 		/*
 		 * take it off the sort and fifo list, move
 		 * to dispatch queue

drivers/block/elevator.c

@@ -272,13 +272,27 @@ void elv_merge_requests(request_queue_t *q, struct request *rq,
 		e->elevator_merge_req_fn(q, rq, next);
 }
 
-/*
- * add_request and next_request are required to be supported, naturally
- */
-void __elv_add_request(request_queue_t *q, struct request *rq,
-			  struct list_head *insert_here)
+void __elv_add_request(request_queue_t *q, struct request *rq, int at_end,
+		       int plug)
 {
-	q->elevator.elevator_add_req_fn(q, rq, insert_here);
+	struct list_head *insert = &q->queue_head;
+
+	if (at_end)
+		insert = insert->prev;
+	if (plug)
+		blk_plug_device(q);
+
+	q->elevator.elevator_add_req_fn(q, rq, insert);
+}
+
+void elv_add_request(request_queue_t *q, struct request *rq, int at_end,
+		     int plug)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(q->queue_lock, flags);
+	__elv_add_request(q, rq, at_end, plug);
+	spin_unlock_irqrestore(q->queue_lock, flags);
 }
 
 static inline struct request *__elv_next_request(request_queue_t *q)
@@ -289,8 +303,14 @@ static inline struct request *__elv_next_request(request_queue_t *q)
 struct request *elv_next_request(request_queue_t *q)
 {
 	struct request *rq;
+	int ret;
 
 	while ((rq = __elv_next_request(q))) {
+		/*
+		 * just mark as started even if we don't start it, a request
+		 * that has been delayed should not be passed by new incoming
+		 * requests
+		 */
 		rq->flags |= REQ_STARTED;
 
 		if (&rq->queuelist == q->last_merge)
@@ -299,20 +319,22 @@ struct request *elv_next_request(request_queue_t *q)
 		if ((rq->flags & REQ_DONTPREP) || !q->prep_rq_fn)
 			break;
 
-		/*
-		 * all ok, break and return it
-		 */
-		if (!q->prep_rq_fn(q, rq))
+		ret = q->prep_rq_fn(q, rq);
+		if (ret == BLKPREP_OK) {
 			break;
-
-		/*
-		 * prep said no-go, kill it
-		 */
+		} else if (ret == BLKPREP_DEFER) {
+			rq = NULL;
+			break;
+		} else if (ret == BLKPREP_KILL) {
 			blkdev_dequeue_request(rq);
-		if (end_that_request_first(rq, 0, rq->nr_sectors))
-			BUG();
-
+			rq->flags |= REQ_QUIET;
+			while (end_that_request_first(rq, 0, rq->nr_sectors))
+				;
 			end_that_request_last(rq);
+		} else {
+			printk("%s: bad return=%d\n", __FUNCTION__, ret);
+			break;
+		}
 	}
 
 	return rq;
@@ -322,6 +344,16 @@ void elv_remove_request(request_queue_t *q, struct request *rq)
 {
 	elevator_t *e = &q->elevator;
 
+	/*
+	 * the main clearing point for q->last_merge is on retrieval of
+	 * request by driver (it calls elv_next_request()), but it _can_
+	 * also happen here if a request is added to the queue but later
+	 * deleted without ever being given to driver (merged with another
+	 * request).
+	 */
+	if (&rq->queuelist == q->last_merge)
+		q->last_merge = NULL;
+
 	if (e->elevator_remove_req_fn)
 		e->elevator_remove_req_fn(q, rq);
 }
@@ -357,6 +389,7 @@ module_init(elevator_global_init);
 
 EXPORT_SYMBOL(elevator_noop);
 
+EXPORT_SYMBOL(elv_add_request);
 EXPORT_SYMBOL(__elv_add_request);
 EXPORT_SYMBOL(elv_next_request);
 EXPORT_SYMBOL(elv_remove_request);

drivers/block/umem.c

@@ -548,12 +548,7 @@ static void process_page(unsigned long data)
 
 		return_bio = bio->bi_next;
 		bio->bi_next = NULL;
-		/* should use bio_endio(), however already cleared
-		 * BIO_UPTODATE. so set bio->bi_size = 0 manually to indicate
-		 * completely done
-		 */
-		bio->bi_size = 0;
-		bio->bi_end_io(bio, bytes, 0);
+		bio_endio(bio, bio->bi_size, 0);
 	}
 }
 

drivers/ide/ide-disk.c

@@ -1610,56 +1610,6 @@ static void idedisk_add_settings(ide_drive_t *drive)
 #endif
 }
 
-static int idedisk_suspend(struct device *dev, u32 state, u32 level)
-{
-	ide_drive_t *drive = dev->driver_data;
-
-	printk("Suspending device %p\n", dev->driver_data);
-
-	/* I hope that every freeze operation from the upper levels have
-	 * already been done...
-	 */
-
-	if (level != SUSPEND_SAVE_STATE)
-		return 0;
-	BUG_ON(in_interrupt());
-
-	printk("Waiting for commands to finish\n");
-
-	/* wait until all commands are finished */
-	/* FIXME: waiting for spinlocks should be done instead. */
-	if (!(HWGROUP(drive)))
-		printk("No hwgroup?\n");
-	while (HWGROUP(drive)->handler)
-		yield();
-
-	/* set the drive to standby */
-	printk(KERN_INFO "suspending: %s ", drive->name);
-	if (drive->driver) {
-		if (drive->driver->standby)
-			drive->driver->standby(drive);
-	}
-	drive->blocked = 1;
-
-	while (HWGROUP(drive)->handler)
-		yield();
-
-	return 0;
-}
-
-static int idedisk_resume(struct device *dev, u32 level)
-{
-	ide_drive_t *drive = dev->driver_data;
-
-	if (level != RESUME_RESTORE_STATE)
-		return 0;
-	if (!drive->blocked)
-		panic("ide: Resume but not suspended?\n");
-
-	drive->blocked = 0;
-	return 0;
-}
-
 
 /* This is just a hook for the overall driver tree.
  */

drivers/ide/ide-floppy.c

@@ -1238,6 +1238,21 @@ static void idefloppy_create_rw_cmd (idefloppy_floppy_t *floppy, idefloppy_pc_t
 	set_bit(PC_DMA_RECOMMENDED, &pc->flags);
 }
 
+static int
+idefloppy_blockpc_cmd(idefloppy_floppy_t *floppy, idefloppy_pc_t *pc, struct request *rq)
+{
+	/*
+	 * just support eject for now, it would not be hard to make the
+	 * REQ_BLOCK_PC support fully-featured
+	 */
+	if (rq->cmd[0] != IDEFLOPPY_START_STOP_CMD)
+		return 1;
+
+	idefloppy_init_pc(pc);
+	memcpy(pc->c, rq->cmd, sizeof(pc->c));
+	return 0;
+}
+
 /*
  *	idefloppy_do_request is our request handling function.	
  */
@@ -1280,6 +1295,12 @@ static ide_startstop_t idefloppy_do_request (ide_drive_t *drive, struct request
 		idefloppy_create_rw_cmd(floppy, pc, rq, block);
 	} else if (rq->flags & REQ_SPECIAL) {
 		pc = (idefloppy_pc_t *) rq->buffer;
+	} else if (rq->flags & REQ_BLOCK_PC) {
+		pc = idefloppy_next_pc_storage(drive);
+		if (idefloppy_blockpc_cmd(floppy, pc, rq)) {
+			idefloppy_do_end_request(drive, 0, 0);
+			return ide_stopped;
+		}
 	} else {
 		blk_dump_rq_flags(rq,
 			"ide-floppy: unsupported command in queue");

drivers/ide/ide.c

@@ -878,13 +878,12 @@ ide_startstop_t start_request (ide_drive_t *drive, struct request *rq)
 {
 	ide_startstop_t startstop;
 	unsigned long block;
-	ide_hwif_t *hwif = HWIF(drive);
 
 	BUG_ON(!(rq->flags & REQ_STARTED));
 
 #ifdef DEBUG
 	printk("%s: start_request: current=0x%08lx\n",
-		hwif->name, (unsigned long) rq);
+		HWIF(drive)->name, (unsigned long) rq);
 #endif
 
 	/* bail early if we've exceeded max_failures */
@@ -910,7 +909,7 @@ ide_startstop_t start_request (ide_drive_t *drive, struct request *rq)
 		block = 1;  /* redirect MBR access to EZ-Drive partn table */
 
 #if (DISK_RECOVERY_TIME > 0)
-	while ((read_timer() - hwif->last_time) < DISK_RECOVERY_TIME);
+	while ((read_timer() - HWIF(drive)->last_time) < DISK_RECOVERY_TIME);
 #endif
 
 	SELECT_DRIVE(drive);
@@ -1128,9 +1127,15 @@ void ide_do_request (ide_hwgroup_t *hwgroup, int masked_irq)
 			break;
 		}
 
+		/*
+		 * we know that the queue isn't empty, but this can happen
+		 * if the q->prep_rq_fn() decides to kill a request
+		 */
 		rq = elv_next_request(&drive->queue);
-		if (!rq)
+		if (!rq) {
+			hwgroup->busy = !!ata_pending_commands(drive);
 			break;
+		}
 
 		if (!rq->bio && ata_pending_commands(drive))
 			break;
@@ -1515,10 +1520,8 @@ int ide_do_drive_cmd (ide_drive_t *drive, struct request *rq, ide_action_t actio
 {
 	unsigned long flags;
 	ide_hwgroup_t *hwgroup = HWGROUP(drive);
-	unsigned int major = HWIF(drive)->major;
-	request_queue_t *q = &drive->queue;
-	struct list_head *queue_head = &q->queue_head;
 	DECLARE_COMPLETION(wait);
+	int insert_end = 1, err;
 
 #ifdef CONFIG_BLK_DEV_PDC4030
 	if (HWIF(drive)->chipset == ide_pdc4030 && rq->buffer != NULL)
@@ -1540,29 +1543,35 @@ int ide_do_drive_cmd (ide_drive_t *drive, struct request *rq, ide_action_t actio
 	}
 
 	rq->rq_disk = drive->disk;
-	if (action == ide_wait)
+
+	/*
+	 * we need to hold an extra reference to request for safe inspection
+	 * after completion
+	 */
+	if (action == ide_wait) {
+		rq->ref_count++;
 		rq->waiting = &wait;
+	}
+
 	spin_lock_irqsave(&ide_lock, flags);
-	if (blk_queue_empty(q) || action == ide_preempt) {
-		if (action == ide_preempt)
+	if (action == ide_preempt) {
 		hwgroup->rq = NULL;
-	} else {
-		if (action == ide_wait || action == ide_end) {
-			queue_head = queue_head->prev;
-		} else
-			queue_head = queue_head->next;
+		insert_end = 0;
 	}
-	q->elevator.elevator_add_req_fn(q, rq, queue_head);
+	__elv_add_request(&drive->queue, rq, insert_end, 0);
 	ide_do_request(hwgroup, 0);
 	spin_unlock_irqrestore(&ide_lock, flags);
+
+	err = 0;
 	if (action == ide_wait) {
-		/* wait for it to be serviced */
 		wait_for_completion(&wait);
-		/* return -EIO if errors */
-		return rq->errors ? -EIO : 0;
+		if (rq->errors)
+			err = -EIO;
+
+		blk_put_request(rq);
 	}
-	return 0;
 
+	return err;
 }
 
 EXPORT_SYMBOL(ide_do_drive_cmd);
@@ -3369,7 +3378,7 @@ int ide_register_driver(ide_driver_t *driver)
 		list_del_init(&drive->list);
 		ata_attach(drive);
 	}
-	driver->gen_driver.name = driver->name;
+	driver->gen_driver.name = (char *) driver->name;
 	driver->gen_driver.bus = &ide_bus_type;
 	driver->gen_driver.remove = ide_drive_remove;
 	return driver_register(&driver->gen_driver);

drivers/md/linear.c

@@ -52,19 +52,21 @@ static inline dev_info_t *which_dev(mddev_t *mddev, sector_t sector)
  *	@bio: the buffer head that's been built up so far
  *	@biovec: the request that could be merged to it.
  *
- *	Return 1 if the merge is not permitted (because the
- *	result would cross a device boundary), 0 otherwise.
+ *	Return amount of bytes we can take at this offset
  */
 static int linear_mergeable_bvec(request_queue_t *q, struct bio *bio, struct bio_vec *biovec)
 {
 	mddev_t *mddev = q->queuedata;
-	dev_info_t *dev0, *dev1;
+	dev_info_t *dev0;
+	int maxsectors, bio_sectors = (bio->bi_size + biovec->bv_len) >> 9;
 
 	dev0 = which_dev(mddev, bio->bi_sector);
-	dev1 = which_dev(mddev, bio->bi_sector +
-			        ((bio->bi_size + biovec->bv_len - 1) >> 9));
+	maxsectors = (dev0->size << 1) - (bio->bi_sector - (dev0->offset<<1));
 
-	return dev0 != dev1;
+	if (bio_sectors <= maxsectors)
+		return biovec->bv_len;
+
+	return (maxsectors << 9) - bio->bi_size;
 }
 
 static int linear_run (mddev_t *mddev)

drivers/md/raid0.c

@@ -168,8 +168,7 @@ static int create_strip_zones (mddev_t *mddev)
  *	@bio: the buffer head that's been built up so far
  *	@biovec: the request that could be merged to it.
  *
- *	Return 1 if the merge is not permitted (because the
- *	result would cross a chunk boundary), 0 otherwise.
+ *	Return amount of bytes we can accept at this offset
  */
 static int raid0_mergeable_bvec(request_queue_t *q, struct bio *bio, struct bio_vec *biovec)
 {
@@ -182,7 +181,7 @@ static int raid0_mergeable_bvec(request_queue_t *q, struct bio *bio, struct bio_
 	block = bio->bi_sector >> 1;
 	bio_sz = (bio->bi_size + biovec->bv_len) >> 10;
 
-	return chunk_size < ((block & (chunk_size - 1)) + bio_sz);
+	return (chunk_size - ((block & (chunk_size - 1)) + bio_sz)) << 10;
 }
 
 static int raid0_run (mddev_t *mddev)

drivers/scsi/scsi_lib.c

@@ -240,7 +240,7 @@ void scsi_queue_next_request(request_queue_t * q, Scsi_Cmnd * SCpnt)
 		SCpnt->request->special = (void *) SCpnt;
 		if(blk_rq_tagged(SCpnt->request))
 			blk_queue_end_tag(q, SCpnt->request);
-		_elv_add_request(q, SCpnt->request, 0, 0);
+		__elv_add_request(q, SCpnt->request, 0, 0);
 	}
 
 	/*
@@ -951,7 +951,7 @@ void scsi_request_fn(request_queue_t * q)
 				SCpnt->request->flags |= REQ_SPECIAL;
 				if(blk_rq_tagged(SCpnt->request))
 					blk_queue_end_tag(q, SCpnt->request);
-				_elv_add_request(q, SCpnt->request, 0, 0);
+				__elv_add_request(q, SCpnt->request, 0, 0);
 				break;
 			}
 

drivers/scsi/sr_ioctl.c

@@ -160,13 +160,11 @@ int sr_do_ioctl(Scsi_CD *cd, struct cdrom_generic_command *cgc)
 			if (!cgc->quiet)
 				printk(KERN_ERR "%s: CDROM (ioctl) reports ILLEGAL "
 				       "REQUEST.\n", cd->cdi.name);
+			err = -EIO;
 			if (SRpnt->sr_sense_buffer[12] == 0x20 &&
-			    SRpnt->sr_sense_buffer[13] == 0x00) {
+			    SRpnt->sr_sense_buffer[13] == 0x00)
 				/* sense: Invalid command operation code */
 				err = -EDRIVE_CANT_DO_THIS;
-			} else {
-				err = -EINVAL;
-			}
 #ifdef DEBUG
 			print_command(cgc->cmd);
 			print_req_sense("sr", SRpnt);

fs/bio.c

@@ -122,6 +122,7 @@ inline void bio_init(struct bio *bio)
 	bio->bi_max_vecs = 0;
 	bio->bi_end_io = NULL;
 	atomic_set(&bio->bi_cnt, 1);
+	bio->bi_private = NULL;
 }
 
 /**
@@ -354,7 +355,7 @@ int bio_get_nr_vecs(struct block_device *bdev)
 	request_queue_t *q = bdev_get_queue(bdev);
 	int nr_pages;
 
-	nr_pages = q->max_sectors >> (PAGE_SHIFT - 9);
+	nr_pages = ((q->max_sectors << 9) + PAGE_SIZE - 1) >> PAGE_SHIFT;
 	if (nr_pages > q->max_phys_segments)
 		nr_pages = q->max_phys_segments;
 	if (nr_pages > q->max_hw_segments)
@@ -385,13 +386,13 @@ int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 	 * cloned bio must not modify vec list
 	 */
 	if (unlikely(bio_flagged(bio, BIO_CLONED)))
-		return 1;
+		return 0;
 
 	if (bio->bi_vcnt >= bio->bi_max_vecs)
-		return 1;
+		return 0;
 
 	if (((bio->bi_size + len) >> 9) > q->max_sectors)
-		return 1;
+		return 0;
 
 	/*
 	 * we might loose a segment or two here, but rather that than
@@ -404,7 +405,7 @@ int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 
 	if (fail_segments) {
 		if (retried_segments)
-			return 1;
+			return 0;
 
 		bio->bi_flags &= ~(1 << BIO_SEG_VALID);
 		retried_segments = 1;
@@ -425,18 +426,24 @@ int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
 	 * depending on offset), it can specify a merge_bvec_fn in the
 	 * queue to get further control
 	 */
-	if (q->merge_bvec_fn && q->merge_bvec_fn(q, bio, bvec)) {
+	if (q->merge_bvec_fn) {
+		/*
+		 * merge_bvec_fn() returns number of bytes it can accept
+		 * at this offset
+		 */
+		if (q->merge_bvec_fn(q, bio, bvec) < len) {
 			bvec->bv_page = NULL;
 			bvec->bv_len = 0;
 			bvec->bv_offset = 0;
-		return 1;
+			return 0;
+		}
 	}
 
 	bio->bi_vcnt++;
 	bio->bi_phys_segments++;
 	bio->bi_hw_segments++;
 	bio->bi_size += len;
-	return 0;
+	return len;
 }
 
 /**
@@ -446,14 +453,15 @@ int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
  * @error:	error, if any
  *
  * Description:
- *   bio_endio() will end I/O @bytes_done number of bytes. This may be just
- *   a partial part of the bio, or it may be the whole bio. bio_endio() is
- *   the preferred way to end I/O on a bio, it takes care of decrementing
+ *   bio_endio() will end I/O on @bytes_done number of bytes. This may be
+ *   just a partial part of the bio, or it may be the whole bio. bio_endio()
+ *   is the preferred way to end I/O on a bio, it takes care of decrementing
  *   bi_size and clearing BIO_UPTODATE on error. @error is 0 on success, and
  *   and one of the established -Exxxx (-EIO, for instance) error values in
- *   case something went wrong.
+ *   case something went wrong. Noone should call bi_end_io() directly on
+ *   a bio unless they own it and thus know that it has an end_io function.
  **/
-int bio_endio(struct bio *bio, unsigned int bytes_done, int error)
+void bio_endio(struct bio *bio, unsigned int bytes_done, int error)
 {
 	if (error)
 		clear_bit(BIO_UPTODATE, &bio->bi_flags);
@@ -465,7 +473,9 @@ int bio_endio(struct bio *bio, unsigned int bytes_done, int error)
 	}
 
 	bio->bi_size -= bytes_done;
-	return bio->bi_end_io(bio, bytes_done, error);
+
+	if (bio->bi_end_io)
+		bio->bi_end_io(bio, bytes_done, error);
 }
 
 static void __init biovec_init_pools(void)
@@ -537,7 +547,7 @@ static int __init init_bio(void)
 	return 0;
 }
 
-module_init(init_bio);
+subsys_initcall(init_bio);
 
 EXPORT_SYMBOL(bio_alloc);
 EXPORT_SYMBOL(bio_put);

fs/direct-io.c

@@ -417,12 +417,12 @@ dio_bio_add_page(struct dio *dio, struct page *page,
 
 	/* Take a ref against the page each time it is placed into a BIO */
 	page_cache_get(page);
-	if (bio_add_page(dio->bio, page, bv_len, bv_offset)) {
+	if (bio_add_page(dio->bio, page, bv_len, bv_offset) < bv_len) {
 		dio_bio_submit(dio);
 		ret = dio_new_bio(dio, blkno);
 		if (ret == 0) {
 			ret = bio_add_page(dio->bio, page, bv_len, bv_offset);
-			BUG_ON(ret != 0);
+			BUG_ON(ret < bv_len);
 		} else {
 			/* The page didn't make it into a BIO */
 			page_cache_release(page);

fs/mpage.c

@@ -176,6 +176,7 @@ do_mpage_readpage(struct bio *bio, struct page *page, unsigned nr_pages,
 	unsigned first_hole = blocks_per_page;
 	struct block_device *bdev = NULL;
 	struct buffer_head bh;
+	int length;
 
 	if (page_has_buffers(page))
 		goto confused;
@@ -233,7 +234,8 @@ do_mpage_readpage(struct bio *bio, struct page *page, unsigned nr_pages,
 			goto confused;
 	}
 
-	if (bio_add_page(bio, page, first_hole << blkbits, 0)) {
+	length = first_hole << blkbits;
+	if (bio_add_page(bio, page, length, 0) < length) {
 		bio = mpage_bio_submit(READ, bio);
 		goto alloc_new;
 	}
@@ -334,6 +336,7 @@ mpage_writepage(struct bio *bio, struct page *page, get_block_t get_block,
 	int boundary = 0;
 	sector_t boundary_block = 0;
 	struct block_device *boundary_bdev = NULL;
+	int length;
 
 	if (page_has_buffers(page)) {
 		struct buffer_head *head = page_buffers(page);
@@ -467,7 +470,8 @@ mpage_writepage(struct bio *bio, struct page *page, get_block_t get_block,
 			try_to_free_buffers(page);
 	}
 
-	if (bio_add_page(bio, page, first_unmapped << blkbits, 0)) {
+	length = first_unmapped << blkbits;
+	if (bio_add_page(bio, page, length, 0) < length) {
 		bio = mpage_bio_submit(WRITE, bio);
 		goto alloc_new;
 	}

fs/pipe.c

@@ -109,7 +109,7 @@ pipe_read(struct file *filp, char *buf, size_t count, loff_t *ppos)
 			break;
 		}
 		if (do_wakeup) {
-			wake_up_interruptible(PIPE_WAIT(*inode));
+			wake_up_interruptible_sync(PIPE_WAIT(*inode));
  			kill_fasync(PIPE_FASYNC_WRITERS(*inode), SIGIO, POLL_OUT);
 		}
 		pipe_wait(inode);
@@ -117,7 +117,7 @@ pipe_read(struct file *filp, char *buf, size_t count, loff_t *ppos)
 	up(PIPE_SEM(*inode));
 	/* Signal writers asynchronously that there is more room.  */
 	if (do_wakeup) {
-		wake_up_interruptible_sync(PIPE_WAIT(*inode));
+		wake_up_interruptible(PIPE_WAIT(*inode));
 		kill_fasync(PIPE_FASYNC_WRITERS(*inode), SIGIO, POLL_OUT);
 	}
 	if (ret > 0)

fs/xfs/pagebuf/page_buf.c

@@ -1448,7 +1448,7 @@ pagebuf_iorequest(			/* start real I/O		*/
 		if (nbytes > size)
 			nbytes = size;
 
-		if (bio_add_page(bio, pb->pb_pages[map_i], nbytes, offset))
+		if (bio_add_page(bio, pb->pb_pages[map_i], nbytes, offset) < nbytes)
 			break;
 
 		offset = 0;

include/asm-i386/ide.h

@@ -70,6 +70,7 @@ static __inline__ void ide_init_default_hwifs(void)
 	int index;
 
 	for(index = 0; index < MAX_HWIFS; index++) {
+		memset(&hw, 0, sizeof hw);
 		ide_init_hwif_ports(&hw, ide_default_io_base(index), 0, NULL);
 		hw.irq = ide_default_irq(ide_default_io_base(index));
 		ide_register_hw(&hw, NULL);

include/linux/bio.h

@@ -101,6 +101,7 @@ struct bio {
 #define BIO_EOF		2	/* out-out-bounds error */
 #define BIO_SEG_VALID	3	/* nr_hw_seg valid */
 #define BIO_CLONED	4	/* doesn't own data */
+#define BIO_BOUNCED	5	/* bio is a bounce bio */
 #define bio_flagged(bio, flag)	((bio)->bi_flags & (1 << (flag)))
 
 /*
@@ -201,7 +202,7 @@ struct bio {
 extern struct bio *bio_alloc(int, int);
 extern void bio_put(struct bio *);
 
-extern int bio_endio(struct bio *, unsigned int, int);
+extern void bio_endio(struct bio *, unsigned int, int);
 struct request_queue;
 extern inline int bio_phys_segments(struct request_queue *, struct bio *);
 extern inline int bio_hw_segments(struct request_queue *, struct bio *);

include/linux/blk.h

@@ -39,33 +39,20 @@ void initrd_init(void);
  */
 
 extern int end_that_request_first(struct request *, int, int);
+extern int end_that_request_chunk(struct request *, int, int);
 extern void end_that_request_last(struct request *);
 struct request *elv_next_request(request_queue_t *q);
 
 static inline void blkdev_dequeue_request(struct request *req)
 {
-	list_del(&req->queuelist);
+	BUG_ON(list_empty(&req->queuelist));
+
+	list_del_init(&req->queuelist);
 
 	if (req->q)
 		elv_remove_request(req->q, req);
 }
 
-#define _elv_add_request_core(q, rq, where, plug)			\
-	do {								\
-		if ((plug))						\
-			blk_plug_device((q));				\
-		(q)->elevator.elevator_add_req_fn((q), (rq), (where));	\
-	} while (0)
-
-#define _elv_add_request(q, rq, back, p) do {				      \
-	if ((back))							      \
-		_elv_add_request_core((q), (rq), (q)->queue_head.prev, (p));  \
-	else								      \
-		_elv_add_request_core((q), (rq), &(q)->queue_head, (p));      \
-} while (0)
-
-#define elv_add_request(q, rq, back) _elv_add_request((q), (rq), (back), 1)
-
 #if defined(MAJOR_NR) || defined(IDE_DRIVER)
 #if (MAJOR_NR != SCSI_TAPE_MAJOR) && (MAJOR_NR != OSST_MAJOR)
 #if !defined(IDE_DRIVER)

include/linux/blkdev.h

@@ -26,6 +26,8 @@ struct request {
 	struct list_head queuelist; /* looking for ->queue? you must _not_
 				     * access it directly, use
 				     * blkdev_dequeue_request! */
+	int ref_count;
+
 	void *elevator_private;
 
 	unsigned char cmd[16];
@@ -215,6 +217,7 @@ struct request_queue
 	unsigned int		max_segment_size;
 
 	unsigned long		seg_boundary_mask;
+	unsigned int		dma_alignment;
 
 	wait_queue_head_t	queue_wait;
 
@@ -254,6 +257,13 @@ struct request_queue
  */
 #define blk_queue_headactive(q, head_active)
 
+/*
+ * q->prep_rq_fn return values
+ */
+#define BLKPREP_OK		0	/* serve it */
+#define BLKPREP_KILL		1	/* fatal error, kill */
+#define BLKPREP_DEFER		2	/* leave on queue */
+
 extern unsigned long blk_max_low_pfn, blk_max_pfn;
 
 /*
@@ -268,7 +278,7 @@ extern unsigned long blk_max_low_pfn, blk_max_pfn;
 #define BLK_BOUNCE_ISA		(ISA_DMA_THRESHOLD)
 
 extern int init_emergency_isa_pool(void);
-void blk_queue_bounce(request_queue_t *q, struct bio **bio);
+inline void blk_queue_bounce(request_queue_t *q, struct bio **bio);
 
 #define rq_for_each_bio(bio, rq)	\
 	if ((rq->bio))			\
@@ -339,6 +349,7 @@ extern void blk_queue_segment_boundary(request_queue_t *, unsigned long);
 extern void blk_queue_assign_lock(request_queue_t *, spinlock_t *);
 extern void blk_queue_prep_rq(request_queue_t *, prep_rq_fn *pfn);
 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 int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
@@ -385,6 +396,21 @@ static inline int bdev_hardsect_size(struct block_device *bdev)
 	return queue_hardsect_size(bdev_get_queue(bdev));
 }
 
+static inline int queue_dma_alignment(request_queue_t *q)
+{
+	int retval = 511;
+
+	if (q && q->dma_alignment)
+		retval = q->dma_alignment;
+
+	return retval;
+}
+
+static inline int bdev_dma_aligment(struct block_device *bdev)
+{
+	return queue_dma_alignment(bdev_get_queue(bdev));
+}
+
 #define blk_finished_io(nsects)	do { } while (0)
 #define blk_started_io(nsects)	do { } while (0)
 

include/linux/elevator.h

@@ -40,8 +40,8 @@ struct elevator_s
 /*
  * block elevator interface
  */
-extern void __elv_add_request(request_queue_t *, struct request *,
-			      struct list_head *);
+extern void elv_add_request(request_queue_t *, struct request *, int, int);
+extern void __elv_add_request(request_queue_t *, struct request *, int, int);
 extern int elv_merge(request_queue_t *, struct list_head **, struct bio *);
 extern void elv_merge_requests(request_queue_t *, struct request *,
 			       struct request *);
@@ -50,6 +50,9 @@ extern void elv_remove_request(request_queue_t *, struct request *);
 extern int elv_queue_empty(request_queue_t *);
 extern inline struct list_head *elv_get_sort_head(request_queue_t *, struct request *);
 
+#define __elv_add_request_pos(q, rq, pos)	\
+	(q)->elevator.elevator_add_req_fn((q), (rq), (pos))
+
 /*
  * noop I/O scheduler. always merges, always inserts new request at tail
  */

mm/highmem.c

@@ -366,34 +366,13 @@ static int bounce_end_io_read_isa(struct bio *bio, unsigned int bytes_done, int
 	return 0;
 }
 
-void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
+void __blk_queue_bounce(request_queue_t *q, struct bio **bio_orig, int bio_gfp,
+			mempool_t *pool)
 {
 	struct page *page;
 	struct bio *bio = NULL;
-	int i, rw = bio_data_dir(*bio_orig), bio_gfp;
+	int i, rw = bio_data_dir(*bio_orig);
 	struct bio_vec *to, *from;
-	mempool_t *pool;
-	unsigned long pfn = q->bounce_pfn;
-	int gfp = q->bounce_gfp;
-
-	BUG_ON((*bio_orig)->bi_idx);
-
-	/*
-	 * for non-isa bounce case, just check if the bounce pfn is equal
-	 * to or bigger than the highest pfn in the system -- in that case,
-	 * don't waste time iterating over bio segments
-	 */
-	if (!(gfp & GFP_DMA)) {
-		if (pfn >= blk_max_pfn)
-			return;
-
-		bio_gfp = GFP_NOHIGHIO;
-		pool = page_pool;
-	} else {
-		BUG_ON(!isa_page_pool);
-		bio_gfp = GFP_NOIO;
-		pool = isa_page_pool;
-	}
 
 	bio_for_each_segment(from, *bio_orig, i) {
 		page = from->bv_page;
@@ -401,7 +380,7 @@ void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
 		/*
 		 * is destination page below bounce pfn?
 		 */
-		if ((page - page_zone(page)->zone_mem_map) + (page_zone(page)->zone_start_pfn) < pfn)
+		if ((page - page_zone(page)->zone_mem_map) + (page_zone(page)->zone_start_pfn) < q->bounce_pfn)
 			continue;
 
 		/*
@@ -412,11 +391,11 @@ void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
 
 		to = bio->bi_io_vec + i;
 
-		to->bv_page = mempool_alloc(pool, gfp);
+		to->bv_page = mempool_alloc(pool, q->bounce_gfp);
 		to->bv_len = from->bv_len;
 		to->bv_offset = from->bv_offset;
 
-		if (rw & WRITE) {
+		if (rw == WRITE) {
 			char *vto, *vfrom;
 
 			vto = page_address(to->bv_page) + to->bv_offset;
@@ -437,15 +416,16 @@ void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
 	 * pages
 	 */
 	bio_for_each_segment(from, *bio_orig, i) {
-		to = &bio->bi_io_vec[i];
+		to = bio_iovec_idx(bio, i);
 		if (!to->bv_page) {
 			to->bv_page = from->bv_page;
 			to->bv_len = from->bv_len;
-			to->bv_offset = to->bv_offset;
+			to->bv_offset = from->bv_offset;
 		}
 	}
 
 	bio->bi_bdev = (*bio_orig)->bi_bdev;
+	bio->bi_flags |= (1 << BIO_BOUNCED);
 	bio->bi_sector = (*bio_orig)->bi_sector;
 	bio->bi_rw = (*bio_orig)->bi_rw;
 
@@ -454,14 +434,12 @@ void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
 	bio->bi_size = (*bio_orig)->bi_size;
 
 	if (pool == page_pool) {
-		if (rw & WRITE)
 		bio->bi_end_io = bounce_end_io_write;
-		else
+		if (rw == READ)
 			bio->bi_end_io = bounce_end_io_read;
 	} else {
-		if (rw & WRITE)
 		bio->bi_end_io = bounce_end_io_write_isa;
-		else
+		if (rw == READ)
 			bio->bi_end_io = bounce_end_io_read_isa;
 	}
 
@@ -469,6 +447,37 @@ void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
 	*bio_orig = bio;
 }
 
+inline void blk_queue_bounce(request_queue_t *q, struct bio **bio_orig)
+{
+	mempool_t *pool;
+	int bio_gfp;
+
+	BUG_ON((*bio_orig)->bi_idx);
+
+	/*
+	 * for non-isa bounce case, just check if the bounce pfn is equal
+	 * to or bigger than the highest pfn in the system -- in that case,
+	 * don't waste time iterating over bio segments
+	 */
+	if (!(q->bounce_gfp & GFP_DMA)) {
+		if (q->bounce_pfn >= blk_max_pfn)
+			return;
+
+		bio_gfp = GFP_NOHIGHIO;
+		pool = page_pool;
+	} else {
+		BUG_ON(!isa_page_pool);
+
+		bio_gfp = GFP_NOIO;
+		pool = isa_page_pool;
+	}
+
+	/*
+	 * slow path
+	 */
+	__blk_queue_bounce(q, bio_orig, bio_gfp, pool);
+}
+
 #if defined(CONFIG_DEBUG_HIGHMEM) && defined(CONFIG_HIGHMEM)
 void check_highmem_ptes(void)
 {