Merge branch 'for-3.14/drivers' of git://git.kernel.dk/linux-block

Pull block IO driver changes from Jens Axboe:

 - bcache update from Kent Overstreet.

 - two bcache fixes from Nicholas Swenson.

 - cciss pci init error fix from Andrew.

 - underflow fix in the parallel IDE pg_write code from Dan Carpenter.
   I'm sure the 1 (or 0) users of that are now happy.

 - two PCI related fixes for sx8 from Jingoo Han.

 - floppy init fix for first block read from Jiri Kosina.

 - pktcdvd error return miss fix from Julia Lawall.

 - removal of IRQF_SHARED from the SEGA Dreamcast CD-ROM code from
   Michael Opdenacker.

 - comment typo fix for the loop driver from Olaf Hering.

 - potential oops fix for null_blk from Raghavendra K T.

 - two fixes from Sam Bradshaw (Micron) for the mtip32xx driver, fixing
   an OOM problem and a problem with handling security locked conditions

* 'for-3.14/drivers' of git://git.kernel.dk/linux-block: (47 commits)
  mg_disk: Spelling s/finised/finished/
  null_blk: Null pointer deference problem in alloc_page_buffers
  mtip32xx: Correctly handle security locked condition
  mtip32xx: Make SGL container per-command to eliminate high order dma allocation
  drivers/block/loop.c: fix comment typo in loop_config_discard
  drivers/block/cciss.c:cciss_init_one(): use proper errnos
  drivers/block/paride/pg.c: underflow bug in pg_write()
  drivers/block/sx8.c: remove unnecessary pci_set_drvdata()
  drivers/block/sx8.c: use module_pci_driver()
  floppy: bail out in open() if drive is not responding to block0 read
  bcache: Fix auxiliary search trees for key size > cacheline size
  bcache: Don't return -EINTR when insert finished
  bcache: Improve bucket_prio() calculation
  bcache: Add bch_bkey_equal_header()
  bcache: update bch_bkey_try_merge
  bcache: Move insert_fixup() to btree_keys_ops
  bcache: Convert sorting to btree_keys
  bcache: Convert debug code to btree_keys
  bcache: Convert btree_iter to struct btree_keys
  bcache: Refactor bset_tree sysfs stats
  ...

block/blk-settings.c

@@ -592,6 +592,10 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
 		ret = -1;
 	}
 
+	t->raid_partial_stripes_expensive =
+		max(t->raid_partial_stripes_expensive,
+		    b->raid_partial_stripes_expensive);
+
 	/* Find lowest common alignment_offset */
 	t->alignment_offset = lcm(t->alignment_offset, alignment)
 		& (max(t->physical_block_size, t->io_min) - 1);

drivers/block/cciss.c

@@ -5004,7 +5004,7 @@ reinit_after_soft_reset:
 
 	i = alloc_cciss_hba(pdev);
 	if (i < 0)
-		return -1;
+		return -ENOMEM;
 
 	h = hba[i];
 	h->pdev = pdev;
@@ -5205,7 +5205,7 @@ clean_no_release_regions:
 	 */
 	pci_set_drvdata(pdev, NULL);
 	free_hba(h);
-	return -1;
+	return -ENODEV;
 }
 
 static void cciss_shutdown(struct pci_dev *pdev)

drivers/block/floppy.c

@@ -3691,9 +3691,12 @@ static int floppy_open(struct block_device *bdev, fmode_t mode)
 	if (!(mode & FMODE_NDELAY)) {
 		if (mode & (FMODE_READ|FMODE_WRITE)) {
 			UDRS->last_checked = 0;
+			clear_bit(FD_OPEN_SHOULD_FAIL_BIT, &UDRS->flags);
 			check_disk_change(bdev);
 			if (test_bit(FD_DISK_CHANGED_BIT, &UDRS->flags))
 				goto out;
+			if (test_bit(FD_OPEN_SHOULD_FAIL_BIT, &UDRS->flags))
+				goto out;
 		}
 		res = -EROFS;
 		if ((mode & FMODE_WRITE) &&
@@ -3746,17 +3749,29 @@ static unsigned int floppy_check_events(struct gendisk *disk,
  * a disk in the drive, and whether that disk is writable.
  */
 
-static void floppy_rb0_complete(struct bio *bio, int err)
+struct rb0_cbdata {
+	int drive;
+	struct completion complete;
+};
+
+static void floppy_rb0_cb(struct bio *bio, int err)
 {
-	complete((struct completion *)bio->bi_private);
+	struct rb0_cbdata *cbdata = (struct rb0_cbdata *)bio->bi_private;
+	int drive = cbdata->drive;
+
+	if (err) {
+		pr_info("floppy: error %d while reading block 0", err);
+		set_bit(FD_OPEN_SHOULD_FAIL_BIT, &UDRS->flags);
+	}
+	complete(&cbdata->complete);
 }
 
-static int __floppy_read_block_0(struct block_device *bdev)
+static int __floppy_read_block_0(struct block_device *bdev, int drive)
 {
 	struct bio bio;
 	struct bio_vec bio_vec;
-	struct completion complete;
 	struct page *page;
+	struct rb0_cbdata cbdata;
 	size_t size;
 
 	page = alloc_page(GFP_NOIO);
@@ -3769,6 +3784,8 @@ static int __floppy_read_block_0(struct block_device *bdev)
 	if (!size)
 		size = 1024;
 
+	cbdata.drive = drive;
+
 	bio_init(&bio);
 	bio.bi_io_vec = &bio_vec;
 	bio_vec.bv_page = page;
@@ -3779,13 +3796,14 @@ static int __floppy_read_block_0(struct block_device *bdev)
 	bio.bi_bdev = bdev;
 	bio.bi_iter.bi_sector = 0;
 	bio.bi_flags = (1 << BIO_QUIET);
-	init_completion(&complete);
-	bio.bi_private = &complete;
-	bio.bi_end_io = floppy_rb0_complete;
+	bio.bi_private = &cbdata;
+	bio.bi_end_io = floppy_rb0_cb;
 
 	submit_bio(READ, &bio);
 	process_fd_request();
-	wait_for_completion(&complete);
+
+	init_completion(&cbdata.complete);
+	wait_for_completion(&cbdata.complete);
 
 	__free_page(page);
 
@@ -3827,7 +3845,7 @@ static int floppy_revalidate(struct gendisk *disk)
 			UDRS->generation++;
 		if (drive_no_geom(drive)) {
 			/* auto-sensing */
-			res = __floppy_read_block_0(opened_bdev[drive]);
+			res = __floppy_read_block_0(opened_bdev[drive], drive);
 		} else {
 			if (cf)
 				poll_drive(false, FD_RAW_NEED_DISK);

drivers/block/loop.c

@@ -799,7 +799,7 @@ static void loop_config_discard(struct loop_device *lo)
 
 	/*
 	 * We use punch hole to reclaim the free space used by the
-	 * image a.k.a. discard. However we do support discard if
+	 * image a.k.a. discard. However we do not support discard if
 	 * encryption is enabled, because it may give an attacker
 	 * useful information.
 	 */

drivers/block/mg_disk.c

@@ -915,7 +915,7 @@ static int mg_probe(struct platform_device *plat_dev)
 
 	/* disk reset */
 	if (prv_data->dev_attr == MG_STORAGE_DEV) {
-		/* If POR seq. not yet finised, wait */
+		/* If POR seq. not yet finished, wait */
 		err = mg_wait_rstout(host->rstout, MG_TMAX_RSTOUT);
 		if (err)
 			goto probe_err_3b;

drivers/block/mtip32xx/mtip32xx.c

@@ -41,10 +41,31 @@
 #include "mtip32xx.h"
 
 #define HW_CMD_SLOT_SZ		(MTIP_MAX_COMMAND_SLOTS * 32)
-#define HW_CMD_TBL_SZ		(AHCI_CMD_TBL_HDR_SZ + (MTIP_MAX_SG * 16))
-#define HW_CMD_TBL_AR_SZ	(HW_CMD_TBL_SZ * MTIP_MAX_COMMAND_SLOTS)
-#define HW_PORT_PRIV_DMA_SZ \
-		(HW_CMD_SLOT_SZ + HW_CMD_TBL_AR_SZ + AHCI_RX_FIS_SZ)
+
+/* DMA region containing RX Fis, Identify, RLE10, and SMART buffers */
+#define AHCI_RX_FIS_SZ          0x100
+#define AHCI_RX_FIS_OFFSET      0x0
+#define AHCI_IDFY_SZ            ATA_SECT_SIZE
+#define AHCI_IDFY_OFFSET        0x400
+#define AHCI_SECTBUF_SZ         ATA_SECT_SIZE
+#define AHCI_SECTBUF_OFFSET     0x800
+#define AHCI_SMARTBUF_SZ        ATA_SECT_SIZE
+#define AHCI_SMARTBUF_OFFSET    0xC00
+/* 0x100 + 0x200 + 0x200 + 0x200 is smaller than 4k but we pad it out */
+#define BLOCK_DMA_ALLOC_SZ      4096
+
+/* DMA region containing command table (should be 8192 bytes) */
+#define AHCI_CMD_SLOT_SZ        sizeof(struct mtip_cmd_hdr)
+#define AHCI_CMD_TBL_SZ         (MTIP_MAX_COMMAND_SLOTS * AHCI_CMD_SLOT_SZ)
+#define AHCI_CMD_TBL_OFFSET     0x0
+
+/* DMA region per command (contains header and SGL) */
+#define AHCI_CMD_TBL_HDR_SZ     0x80
+#define AHCI_CMD_TBL_HDR_OFFSET 0x0
+#define AHCI_CMD_TBL_SGL_SZ     (MTIP_MAX_SG * sizeof(struct mtip_cmd_sg))
+#define AHCI_CMD_TBL_SGL_OFFSET AHCI_CMD_TBL_HDR_SZ
+#define CMD_DMA_ALLOC_SZ        (AHCI_CMD_TBL_SGL_SZ + AHCI_CMD_TBL_HDR_SZ)
+
 
 #define HOST_CAP_NZDMA		(1 << 19)
 #define HOST_HSORG		0xFC
@@ -899,8 +920,9 @@ static void mtip_handle_tfe(struct driver_data *dd)
 			fail_reason = "thermal shutdown";
 		}
 		if (buf[288] == 0xBF) {
+			set_bit(MTIP_DDF_SEC_LOCK_BIT, &dd->dd_flag);
 			dev_info(&dd->pdev->dev,
-				"Drive indicates rebuild has failed.\n");
+				"Drive indicates rebuild has failed. Secure erase required.\n");
 			fail_all_ncq_cmds = 1;
 			fail_reason = "rebuild failed";
 		}
@@ -1566,6 +1588,12 @@ static int mtip_get_identify(struct mtip_port *port, void __user *user_buffer)
 	}
 #endif
 
+	/* Check security locked state */
+	if (port->identify[128] & 0x4)
+		set_bit(MTIP_DDF_SEC_LOCK_BIT, &port->dd->dd_flag);
+	else
+		clear_bit(MTIP_DDF_SEC_LOCK_BIT, &port->dd->dd_flag);
+
 #ifdef MTIP_TRIM /* Disabling TRIM support temporarily */
 	/* Demux ID.DRAT & ID.RZAT to determine trim support */
 	if (port->identify[69] & (1 << 14) && port->identify[69] & (1 << 5))
@@ -1887,6 +1915,10 @@ static void mtip_dump_identify(struct mtip_port *port)
 	strlcpy(cbuf, (char *)(port->identify+27), 41);
 	dev_info(&port->dd->pdev->dev, "Model: %s\n", cbuf);
 
+	dev_info(&port->dd->pdev->dev, "Security: %04x %s\n",
+		port->identify[128],
+		port->identify[128] & 0x4 ? "(LOCKED)" : "");
+
 	if (mtip_hw_get_capacity(port->dd, &sectors))
 		dev_info(&port->dd->pdev->dev,
 			"Capacity: %llu sectors (%llu MB)\n",
@@ -3312,6 +3344,118 @@ st_out:
 	return 0;
 }
 
+/*
+ * DMA region teardown
+ *
+ * @dd Pointer to driver_data structure
+ *
+ * return value
+ *      None
+ */
+static void mtip_dma_free(struct driver_data *dd)
+{
+	int i;
+	struct mtip_port *port = dd->port;
+
+	if (port->block1)
+		dmam_free_coherent(&dd->pdev->dev, BLOCK_DMA_ALLOC_SZ,
+					port->block1, port->block1_dma);
+
+	if (port->command_list) {
+		dmam_free_coherent(&dd->pdev->dev, AHCI_CMD_TBL_SZ,
+				port->command_list, port->command_list_dma);
+	}
+
+	for (i = 0; i < MTIP_MAX_COMMAND_SLOTS; i++) {
+		if (port->commands[i].command)
+			dmam_free_coherent(&dd->pdev->dev, CMD_DMA_ALLOC_SZ,
+				port->commands[i].command,
+				port->commands[i].command_dma);
+	}
+}
+
+/*
+ * DMA region setup
+ *
+ * @dd Pointer to driver_data structure
+ *
+ * return value
+ *      -ENOMEM Not enough free DMA region space to initialize driver
+ */
+static int mtip_dma_alloc(struct driver_data *dd)
+{
+	struct mtip_port *port = dd->port;
+	int i, rv = 0;
+	u32 host_cap_64 = readl(dd->mmio + HOST_CAP) & HOST_CAP_64;
+
+	/* Allocate dma memory for RX Fis, Identify, and Sector Bufffer */
+	port->block1 =
+		dmam_alloc_coherent(&dd->pdev->dev, BLOCK_DMA_ALLOC_SZ,
+					&port->block1_dma, GFP_KERNEL);
+	if (!port->block1)
+		return -ENOMEM;
+	memset(port->block1, 0, BLOCK_DMA_ALLOC_SZ);
+
+	/* Allocate dma memory for command list */
+	port->command_list =
+		dmam_alloc_coherent(&dd->pdev->dev, AHCI_CMD_TBL_SZ,
+					&port->command_list_dma, GFP_KERNEL);
+	if (!port->command_list) {
+		dmam_free_coherent(&dd->pdev->dev, BLOCK_DMA_ALLOC_SZ,
+					port->block1, port->block1_dma);
+		port->block1 = NULL;
+		port->block1_dma = 0;
+		return -ENOMEM;
+	}
+	memset(port->command_list, 0, AHCI_CMD_TBL_SZ);
+
+	/* Setup all pointers into first DMA region */
+	port->rxfis         = port->block1 + AHCI_RX_FIS_OFFSET;
+	port->rxfis_dma     = port->block1_dma + AHCI_RX_FIS_OFFSET;
+	port->identify      = port->block1 + AHCI_IDFY_OFFSET;
+	port->identify_dma  = port->block1_dma + AHCI_IDFY_OFFSET;
+	port->log_buf       = port->block1 + AHCI_SECTBUF_OFFSET;
+	port->log_buf_dma   = port->block1_dma + AHCI_SECTBUF_OFFSET;
+	port->smart_buf     = port->block1 + AHCI_SMARTBUF_OFFSET;
+	port->smart_buf_dma = port->block1_dma + AHCI_SMARTBUF_OFFSET;
+
+	/* Setup per command SGL DMA region */
+
+	/* Point the command headers at the command tables */
+	for (i = 0; i < MTIP_MAX_COMMAND_SLOTS; i++) {
+		port->commands[i].command =
+			dmam_alloc_coherent(&dd->pdev->dev, CMD_DMA_ALLOC_SZ,
+				&port->commands[i].command_dma, GFP_KERNEL);
+		if (!port->commands[i].command) {
+			rv = -ENOMEM;
+			mtip_dma_free(dd);
+			return rv;
+		}
+		memset(port->commands[i].command, 0, CMD_DMA_ALLOC_SZ);
+
+		port->commands[i].command_header = port->command_list +
+					(sizeof(struct mtip_cmd_hdr) * i);
+		port->commands[i].command_header_dma =
+					dd->port->command_list_dma +
+					(sizeof(struct mtip_cmd_hdr) * i);
+
+		if (host_cap_64)
+			port->commands[i].command_header->ctbau =
+				__force_bit2int cpu_to_le32(
+				(port->commands[i].command_dma >> 16) >> 16);
+
+		port->commands[i].command_header->ctba =
+				__force_bit2int cpu_to_le32(
+				port->commands[i].command_dma & 0xFFFFFFFF);
+
+		sg_init_table(port->commands[i].sg, MTIP_MAX_SG);
+
+		/* Mark command as currently inactive */
+		atomic_set(&dd->port->commands[i].active, 0);
+	}
+	return 0;
+}
+
 /*
  * Called once for each card.
  *
@@ -3370,83 +3514,10 @@ static int mtip_hw_init(struct driver_data *dd)
 	dd->port->mmio	= dd->mmio + PORT_OFFSET;
 	dd->port->dd	= dd;
 
-	/* Allocate memory for the command list. */
-	dd->port->command_list =
-		dmam_alloc_coherent(&dd->pdev->dev,
-			HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 4),
-			&dd->port->command_list_dma,
-			GFP_KERNEL);
-	if (!dd->port->command_list) {
-		dev_err(&dd->pdev->dev,
-			"Memory allocation: command list\n");
-		rv = -ENOMEM;
+	/* DMA allocations */
+	rv = mtip_dma_alloc(dd);
+	if (rv < 0)
 		goto out1;
-	}
-
-	/* Clear the memory we have allocated. */
-	memset(dd->port->command_list,
-		0,
-		HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 4));
-
-	/* Setup the addresse of the RX FIS. */
-	dd->port->rxfis	    = dd->port->command_list + HW_CMD_SLOT_SZ;
-	dd->port->rxfis_dma = dd->port->command_list_dma + HW_CMD_SLOT_SZ;
-
-	/* Setup the address of the command tables. */
-	dd->port->command_table	  = dd->port->rxfis + AHCI_RX_FIS_SZ;
-	dd->port->command_tbl_dma = dd->port->rxfis_dma + AHCI_RX_FIS_SZ;
-
-	/* Setup the address of the identify data. */
-	dd->port->identify     = dd->port->command_table +
-					HW_CMD_TBL_AR_SZ;
-	dd->port->identify_dma = dd->port->command_tbl_dma +
-					HW_CMD_TBL_AR_SZ;
-
-	/* Setup the address of the sector buffer - for some non-ncq cmds */
-	dd->port->sector_buffer	= (void *) dd->port->identify + ATA_SECT_SIZE;
-	dd->port->sector_buffer_dma = dd->port->identify_dma + ATA_SECT_SIZE;
-
-	/* Setup the address of the log buf - for read log command */
-	dd->port->log_buf = (void *)dd->port->sector_buffer  + ATA_SECT_SIZE;
-	dd->port->log_buf_dma = dd->port->sector_buffer_dma + ATA_SECT_SIZE;
-
-	/* Setup the address of the smart buf - for smart read data command */
-	dd->port->smart_buf = (void *)dd->port->log_buf  + ATA_SECT_SIZE;
-	dd->port->smart_buf_dma = dd->port->log_buf_dma + ATA_SECT_SIZE;
-
-
-	/* Point the command headers at the command tables. */
-	for (i = 0; i < num_command_slots; i++) {
-		dd->port->commands[i].command_header =
-					dd->port->command_list +
-					(sizeof(struct mtip_cmd_hdr) * i);
-		dd->port->commands[i].command_header_dma =
-					dd->port->command_list_dma +
-					(sizeof(struct mtip_cmd_hdr) * i);
-
-		dd->port->commands[i].command =
-			dd->port->command_table + (HW_CMD_TBL_SZ * i);
-		dd->port->commands[i].command_dma =
-			dd->port->command_tbl_dma + (HW_CMD_TBL_SZ * i);
-
-		if (readl(dd->mmio + HOST_CAP) & HOST_CAP_64)
-			dd->port->commands[i].command_header->ctbau =
-			__force_bit2int cpu_to_le32(
-			(dd->port->commands[i].command_dma >> 16) >> 16);
-		dd->port->commands[i].command_header->ctba =
-			__force_bit2int cpu_to_le32(
-			dd->port->commands[i].command_dma & 0xFFFFFFFF);
-
-		/*
-		 * If this is not done, a bug is reported by the stock
-		 * FC11 i386. Due to the fact that it has lots of kernel
-		 * debugging enabled.
-		 */
-		sg_init_table(dd->port->commands[i].sg, MTIP_MAX_SG);
-
-		/* Mark all commands as currently inactive.*/
-		atomic_set(&dd->port->commands[i].active, 0);
-	}
 
 	/* Setup the pointers to the extended s_active and CI registers. */
 	for (i = 0; i < dd->slot_groups; i++) {
@@ -3594,12 +3665,8 @@ out3:
 
 out2:
 	mtip_deinit_port(dd->port);
+	mtip_dma_free(dd);
 
-	/* Free the command/command header memory. */
-	dmam_free_coherent(&dd->pdev->dev,
-				HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 4),
-				dd->port->command_list,
-				dd->port->command_list_dma);
 out1:
 	/* Free the memory allocated for the for structure. */
 	kfree(dd->port);
@@ -3622,7 +3689,8 @@ static int mtip_hw_exit(struct driver_data *dd)
 	 * saves its state.
 	 */
 	if (!dd->sr) {
-		if (!test_bit(MTIP_DDF_REBUILD_FAILED_BIT, &dd->dd_flag))
+		if (!test_bit(MTIP_PF_REBUILD_BIT, &dd->port->flags) &&
+		    !test_bit(MTIP_DDF_SEC_LOCK_BIT, &dd->dd_flag))
 			if (mtip_standby_immediate(dd->port))
 				dev_warn(&dd->pdev->dev,
 					"STANDBY IMMEDIATE failed\n");
@@ -3641,11 +3709,9 @@ static int mtip_hw_exit(struct driver_data *dd)
 	irq_set_affinity_hint(dd->pdev->irq, NULL);
 	devm_free_irq(&dd->pdev->dev, dd->pdev->irq, dd);
 
-	/* Free the command/command header memory. */
-	dmam_free_coherent(&dd->pdev->dev,
-			HW_PORT_PRIV_DMA_SZ + (ATA_SECT_SIZE * 4),
-			dd->port->command_list,
-			dd->port->command_list_dma);
+	/* Free dma regions */
+	mtip_dma_free(dd);
+
 	/* Free the memory allocated for the for structure. */
 	kfree(dd->port);
 	dd->port = NULL;

drivers/block/mtip32xx/mtip32xx.h

@@ -69,7 +69,7 @@
  * Maximum number of scatter gather entries
  * a single command may have.
  */
-#define MTIP_MAX_SG		128
+#define MTIP_MAX_SG		504
 
 /*
  * Maximum number of slot groups (Command Issue & s_active registers)
@@ -92,7 +92,7 @@
 
 /* Driver name and version strings */
 #define MTIP_DRV_NAME		"mtip32xx"
-#define MTIP_DRV_VERSION	"1.2.6os3"
+#define MTIP_DRV_VERSION	"1.3.0"
 
 /* Maximum number of minor device numbers per device. */
 #define MTIP_MAX_MINORS		16
@@ -391,15 +391,13 @@ struct mtip_port {
 	 */
 	dma_addr_t rxfis_dma;
 	/*
-	 * Pointer to the beginning of the command table memory as used
-	 * by the driver.
+	 * Pointer to the DMA region for RX Fis, Identify, RLE10, and SMART
 	 */
-	void *command_table;
+	void *block1;
 	/*
-	 * Pointer to the beginning of the command table memory as used
-	 * by the DMA.
+	 * DMA address of region for RX Fis, Identify, RLE10, and SMART
 	 */
-	dma_addr_t command_tbl_dma;
+	dma_addr_t block1_dma;
 	/*
 	 * Pointer to the beginning of the identify data memory as used
 	 * by the driver.

drivers/block/null_blk.c

@@ -616,6 +616,11 @@ static int __init null_init(void)
 		irqmode = NULL_IRQ_NONE;
 	}
 #endif
+	if (bs > PAGE_SIZE) {
+		pr_warn("null_blk: invalid block size\n");
+		pr_warn("null_blk: defaults block size to %lu\n", PAGE_SIZE);
+		bs = PAGE_SIZE;
+	}
 
 	if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
 		if (submit_queues < nr_online_nodes) {

drivers/block/paride/pg.c

@@ -581,7 +581,7 @@ static ssize_t pg_write(struct file *filp, const char __user *buf, size_t count,
 
 	if (hdr.magic != PG_MAGIC)
 		return -EINVAL;
-	if (hdr.dlen > PG_MAX_DATA)
+	if (hdr.dlen < 0 || hdr.dlen > PG_MAX_DATA)
 		return -EINVAL;
 	if ((count - hs) > PG_MAX_DATA)
 		return -EINVAL;

drivers/block/pktcdvd.c

@@ -706,7 +706,9 @@ static int pkt_generic_packet(struct pktcdvd_device *pd, struct packet_command *
 			     WRITE : READ, __GFP_WAIT);
 
 	if (cgc->buflen) {
-		if (blk_rq_map_kern(q, rq, cgc->buffer, cgc->buflen, __GFP_WAIT))
+		ret = blk_rq_map_kern(q, rq, cgc->buffer, cgc->buflen,
+				      __GFP_WAIT);
+		if (ret)
 			goto out;
 	}
 

drivers/block/sx8.c

@@ -1744,20 +1744,6 @@ static void carm_remove_one (struct pci_dev *pdev)
 	kfree(host);
 	pci_release_regions(pdev);
 	pci_disable_device(pdev);
-	pci_set_drvdata(pdev, NULL);
 }
 
-static int __init carm_init(void)
-{
-	return pci_register_driver(&carm_driver);
-}
-
-static void __exit carm_exit(void)
-{
-	pci_unregister_driver(&carm_driver);
-}
-
-module_init(carm_init);
-module_exit(carm_exit);
-
-
+module_pci_driver(carm_driver);

drivers/cdrom/gdrom.c

@@ -561,11 +561,11 @@ static int gdrom_set_interrupt_handlers(void)
 	int err;
 
 	err = request_irq(HW_EVENT_GDROM_CMD, gdrom_command_interrupt,
-		IRQF_DISABLED, "gdrom_command", &gd);
+		0, "gdrom_command", &gd);
 	if (err)
 		return err;
 	err = request_irq(HW_EVENT_GDROM_DMA, gdrom_dma_interrupt,
-		IRQF_DISABLED, "gdrom_dma", &gd);
+		0, "gdrom_dma", &gd);
 	if (err)
 		free_irq(HW_EVENT_GDROM_CMD, &gd);
 	return err;

drivers/md/bcache/Makefile

@@ -1,7 +1,8 @@
 
 obj-$(CONFIG_BCACHE)	+= bcache.o
 
-bcache-y		:= alloc.o btree.o bset.o io.o journal.o writeback.o\
-	movinggc.o request.o super.o sysfs.o debug.o util.o trace.o stats.o closure.o
+bcache-y		:= alloc.o bset.o btree.o closure.o debug.o extents.o\
+	io.o journal.o movinggc.o request.o stats.o super.o sysfs.o trace.o\
+	util.o writeback.o
 
 CFLAGS_request.o	+= -Iblock

drivers/md/bcache/alloc.c

@@ -132,10 +132,16 @@ bool bch_bucket_add_unused(struct cache *ca, struct bucket *b)
 {
 	BUG_ON(GC_MARK(b) || GC_SECTORS_USED(b));
 
-	if (fifo_used(&ca->free) > ca->watermark[WATERMARK_MOVINGGC] &&
-	    CACHE_REPLACEMENT(&ca->sb) == CACHE_REPLACEMENT_FIFO)
-		return false;
+	if (CACHE_REPLACEMENT(&ca->sb) == CACHE_REPLACEMENT_FIFO) {
+		unsigned i;
 
+		for (i = 0; i < RESERVE_NONE; i++)
+			if (!fifo_full(&ca->free[i]))
+				goto add;
+
+		return false;
+	}
+add:
 	b->prio = 0;
 
 	if (can_inc_bucket_gen(b) &&
@@ -162,8 +168,21 @@ static void invalidate_one_bucket(struct cache *ca, struct bucket *b)
 	fifo_push(&ca->free_inc, b - ca->buckets);
 }
 
-#define bucket_prio(b)				\
-	(((unsigned) (b->prio - ca->set->min_prio)) * GC_SECTORS_USED(b))
+/*
+ * Determines what order we're going to reuse buckets, smallest bucket_prio()
+ * first: we also take into account the number of sectors of live data in that
+ * bucket, and in order for that multiply to make sense we have to scale bucket
+ *
+ * Thus, we scale the bucket priorities so that the bucket with the smallest
+ * prio is worth 1/8th of what INITIAL_PRIO is worth.
+ */
+
+#define bucket_prio(b)							\
+({									\
+	unsigned min_prio = (INITIAL_PRIO - ca->set->min_prio) / 8;	\
+									\
+	(b->prio - ca->set->min_prio + min_prio) * GC_SECTORS_USED(b);	\
+})
 
 #define bucket_max_cmp(l, r)	(bucket_prio(l) < bucket_prio(r))
 #define bucket_min_cmp(l, r)	(bucket_prio(l) > bucket_prio(r))
@@ -304,6 +323,21 @@ do {									\
 	__set_current_state(TASK_RUNNING);				\
 } while (0)
 
+static int bch_allocator_push(struct cache *ca, long bucket)
+{
+	unsigned i;
+
+	/* Prios/gens are actually the most important reserve */
+	if (fifo_push(&ca->free[RESERVE_PRIO], bucket))
+		return true;
+
+	for (i = 0; i < RESERVE_NR; i++)
+		if (fifo_push(&ca->free[i], bucket))
+			return true;
+
+	return false;
+}
+
 static int bch_allocator_thread(void *arg)
 {
 	struct cache *ca = arg;
@@ -336,9 +370,7 @@ static int bch_allocator_thread(void *arg)
 				mutex_lock(&ca->set->bucket_lock);
 			}
 
-			allocator_wait(ca, !fifo_full(&ca->free));
-
-			fifo_push(&ca->free, bucket);
+			allocator_wait(ca, bch_allocator_push(ca, bucket));
 			wake_up(&ca->set->bucket_wait);
 		}
 
@@ -365,34 +397,29 @@ static int bch_allocator_thread(void *arg)
 	}
 }
 
-long bch_bucket_alloc(struct cache *ca, unsigned watermark, bool wait)
+long bch_bucket_alloc(struct cache *ca, unsigned reserve, bool wait)
 {
 	DEFINE_WAIT(w);
 	struct bucket *b;
 	long r;
 
 	/* fastpath */
-	if (fifo_used(&ca->free) > ca->watermark[watermark]) {
-		fifo_pop(&ca->free, r);
+	if (fifo_pop(&ca->free[RESERVE_NONE], r) ||
+	    fifo_pop(&ca->free[reserve], r))
 		goto out;
-	}
 
 	if (!wait)
 		return -1;
 
-	while (1) {
-		if (fifo_used(&ca->free) > ca->watermark[watermark]) {
-			fifo_pop(&ca->free, r);
-			break;
-		}
-
+	do {
 		prepare_to_wait(&ca->set->bucket_wait, &w,
 				TASK_UNINTERRUPTIBLE);
 
 		mutex_unlock(&ca->set->bucket_lock);
 		schedule();
 		mutex_lock(&ca->set->bucket_lock);
-	}
+	} while (!fifo_pop(&ca->free[RESERVE_NONE], r) &&
+		 !fifo_pop(&ca->free[reserve], r));
 
 	finish_wait(&ca->set->bucket_wait, &w);
 out:
@@ -401,12 +428,14 @@ out:
 	if (expensive_debug_checks(ca->set)) {
 		size_t iter;
 		long i;
+		unsigned j;
 
 		for (iter = 0; iter < prio_buckets(ca) * 2; iter++)
 			BUG_ON(ca->prio_buckets[iter] == (uint64_t) r);
 
-		fifo_for_each(i, &ca->free, iter)
-			BUG_ON(i == r);
+		for (j = 0; j < RESERVE_NR; j++)
+			fifo_for_each(i, &ca->free[j], iter)
+				BUG_ON(i == r);
 		fifo_for_each(i, &ca->free_inc, iter)
 			BUG_ON(i == r);
 		fifo_for_each(i, &ca->unused, iter)
@@ -419,7 +448,7 @@ out:
 
 	SET_GC_SECTORS_USED(b, ca->sb.bucket_size);
 
-	if (watermark <= WATERMARK_METADATA) {
+	if (reserve <= RESERVE_PRIO) {
 		SET_GC_MARK(b, GC_MARK_METADATA);
 		SET_GC_MOVE(b, 0);
 		b->prio = BTREE_PRIO;
@@ -445,7 +474,7 @@ void bch_bucket_free(struct cache_set *c, struct bkey *k)
 	}
 }
 
-int __bch_bucket_alloc_set(struct cache_set *c, unsigned watermark,
+int __bch_bucket_alloc_set(struct cache_set *c, unsigned reserve,
 			   struct bkey *k, int n, bool wait)
 {
 	int i;
@@ -459,7 +488,7 @@ int __bch_bucket_alloc_set(struct cache_set *c, unsigned watermark,
 
 	for (i = 0; i < n; i++) {
 		struct cache *ca = c->cache_by_alloc[i];
-		long b = bch_bucket_alloc(ca, watermark, wait);
+		long b = bch_bucket_alloc(ca, reserve, wait);
 
 		if (b == -1)
 			goto err;
@@ -478,12 +507,12 @@ err:
 	return -1;
 }
 
-int bch_bucket_alloc_set(struct cache_set *c, unsigned watermark,
+int bch_bucket_alloc_set(struct cache_set *c, unsigned reserve,
 			 struct bkey *k, int n, bool wait)
 {
 	int ret;
 	mutex_lock(&c->bucket_lock);
-	ret = __bch_bucket_alloc_set(c, watermark, k, n, wait);
+	ret = __bch_bucket_alloc_set(c, reserve, k, n, wait);
 	mutex_unlock(&c->bucket_lock);
 	return ret;
 }
@@ -573,8 +602,8 @@ bool bch_alloc_sectors(struct cache_set *c, struct bkey *k, unsigned sectors,
 
 	while (!(b = pick_data_bucket(c, k, write_point, &alloc.key))) {
 		unsigned watermark = write_prio
-			? WATERMARK_MOVINGGC
-			: WATERMARK_NONE;
+			? RESERVE_MOVINGGC
+			: RESERVE_NONE;
 
 		spin_unlock(&c->data_bucket_lock);
 
@@ -689,7 +718,7 @@ int bch_cache_allocator_init(struct cache *ca)
 	 * Then 8 for btree allocations
 	 * Then half for the moving garbage collector
 	 */
-
+#if 0
 	ca->watermark[WATERMARK_PRIO] = 0;
 
 	ca->watermark[WATERMARK_METADATA] = prio_buckets(ca);
@@ -699,6 +728,6 @@ int bch_cache_allocator_init(struct cache *ca)
 
 	ca->watermark[WATERMARK_NONE] = ca->free.size / 2 +
 		ca->watermark[WATERMARK_MOVINGGC];
-
+#endif
 	return 0;
 }

drivers/md/bcache/bcache.h

@@ -187,6 +187,7 @@
 #include <linux/types.h>
 #include <linux/workqueue.h>
 
+#include "bset.h"
 #include "util.h"
 #include "closure.h"
 
@@ -309,7 +310,8 @@ struct cached_dev {
 	struct cache_sb		sb;
 	struct bio		sb_bio;
 	struct bio_vec		sb_bv[1];
-	struct closure_with_waitlist sb_write;
+	struct closure		sb_write;
+	struct semaphore	sb_write_mutex;
 
 	/* Refcount on the cache set. Always nonzero when we're caching. */
 	atomic_t		count;
@@ -382,12 +384,12 @@ struct cached_dev {
 	unsigned		writeback_rate_p_term_inverse;
 };
 
-enum alloc_watermarks {
-	WATERMARK_PRIO,
-	WATERMARK_METADATA,
-	WATERMARK_MOVINGGC,
-	WATERMARK_NONE,
-	WATERMARK_MAX
+enum alloc_reserve {
+	RESERVE_BTREE,
+	RESERVE_PRIO,
+	RESERVE_MOVINGGC,
+	RESERVE_NONE,
+	RESERVE_NR,
 };
 
 struct cache {
@@ -399,8 +401,6 @@ struct cache {
 	struct kobject		kobj;
 	struct block_device	*bdev;
 
-	unsigned		watermark[WATERMARK_MAX];
-
 	struct task_struct	*alloc_thread;
 
 	struct closure		prio;
@@ -429,7 +429,7 @@ struct cache {
 	 * because all the data they contained was overwritten), so we only
 	 * need to discard them before they can be moved to the free list.
 	 */
-	DECLARE_FIFO(long, free);
+	DECLARE_FIFO(long, free)[RESERVE_NR];
 	DECLARE_FIFO(long, free_inc);
 	DECLARE_FIFO(long, unused);
 
@@ -514,7 +514,8 @@ struct cache_set {
 	uint64_t		cached_dev_sectors;
 	struct closure		caching;
 
-	struct closure_with_waitlist sb_write;
+	struct closure		sb_write;
+	struct semaphore	sb_write_mutex;
 
 	mempool_t		*search;
 	mempool_t		*bio_meta;
@@ -629,13 +630,15 @@ struct cache_set {
 
 #ifdef CONFIG_BCACHE_DEBUG
 	struct btree		*verify_data;
+	struct bset		*verify_ondisk;
 	struct mutex		verify_lock;
 #endif
 
 	unsigned		nr_uuids;
 	struct uuid_entry	*uuids;
 	BKEY_PADDED(uuid_bucket);
-	struct closure_with_waitlist uuid_write;
+	struct closure		uuid_write;
+	struct semaphore	uuid_write_mutex;
 
 	/*
 	 * A btree node on disk could have too many bsets for an iterator to fit
@@ -643,13 +646,7 @@ struct cache_set {
 	 */
 	mempool_t		*fill_iter;
 
-	/*
-	 * btree_sort() is a merge sort and requires temporary space - single
-	 * element mempool
-	 */
-	struct mutex		sort_lock;
-	struct bset		*sort;
-	unsigned		sort_crit_factor;
+	struct bset_sort_state	sort;
 
 	/* List of buckets we're currently writing data to */
 	struct list_head	data_buckets;
@@ -665,7 +662,6 @@ struct cache_set {
 	unsigned		congested_read_threshold_us;
 	unsigned		congested_write_threshold_us;
 
-	struct time_stats	sort_time;
 	struct time_stats	btree_gc_time;
 	struct time_stats	btree_split_time;
 	struct time_stats	btree_read_time;
@@ -683,9 +679,9 @@ struct cache_set {
 	unsigned		error_decay;
 
 	unsigned short		journal_delay_ms;
+	bool			expensive_debug_checks;
 	unsigned		verify:1;
 	unsigned		key_merging_disabled:1;
-	unsigned		expensive_debug_checks:1;
 	unsigned		gc_always_rewrite:1;
 	unsigned		shrinker_disabled:1;
 	unsigned		copy_gc_enabled:1;
@@ -707,13 +703,8 @@ struct bbio {
 	struct bio		bio;
 };
 
-static inline unsigned local_clock_us(void)
-{
-	return local_clock() >> 10;
-}
-
 #define BTREE_PRIO		USHRT_MAX
-#define INITIAL_PRIO		32768
+#define INITIAL_PRIO		32768U
 
 #define btree_bytes(c)		((c)->btree_pages * PAGE_SIZE)
 #define btree_blocks(b)							\
@@ -726,21 +717,6 @@ static inline unsigned local_clock_us(void)
 #define bucket_bytes(c)		((c)->sb.bucket_size << 9)
 #define block_bytes(c)		((c)->sb.block_size << 9)
 
-#define __set_bytes(i, k)	(sizeof(*(i)) + (k) * sizeof(uint64_t))
-#define set_bytes(i)		__set_bytes(i, i->keys)
-
-#define __set_blocks(i, k, c)	DIV_ROUND_UP(__set_bytes(i, k), block_bytes(c))
-#define set_blocks(i, c)	__set_blocks(i, (i)->keys, c)
-
-#define node(i, j)		((struct bkey *) ((i)->d + (j)))
-#define end(i)			node(i, (i)->keys)
-
-#define index(i, b)							\
-	((size_t) (((void *) i - (void *) (b)->sets[0].data) /		\
-		   block_bytes(b->c)))
-
-#define btree_data_space(b)	(PAGE_SIZE << (b)->page_order)
-
 #define prios_per_bucket(c)				\
 	((bucket_bytes(c) - sizeof(struct prio_set)) /	\
 	 sizeof(struct bucket_disk))
@@ -783,20 +759,34 @@ static inline struct bucket *PTR_BUCKET(struct cache_set *c,
 	return PTR_CACHE(c, k, ptr)->buckets + PTR_BUCKET_NR(c, k, ptr);
 }
 
-/* Btree key macros */
-
-static inline void bkey_init(struct bkey *k)
+static inline uint8_t gen_after(uint8_t a, uint8_t b)
 {
-	*k = ZERO_KEY;
+	uint8_t r = a - b;
+	return r > 128U ? 0 : r;
 }
 
+static inline uint8_t ptr_stale(struct cache_set *c, const struct bkey *k,
+				unsigned i)
+{
+	return gen_after(PTR_BUCKET(c, k, i)->gen, PTR_GEN(k, i));
+}
+
+static inline bool ptr_available(struct cache_set *c, const struct bkey *k,
+				 unsigned i)
+{
+	return (PTR_DEV(k, i) < MAX_CACHES_PER_SET) && PTR_CACHE(c, k, i);
+}
+
+/* Btree key macros */
+
 /*
  * This is used for various on disk data structures - cache_sb, prio_set, bset,
  * jset: The checksum is _always_ the first 8 bytes of these structs
  */
 #define csum_set(i)							\
 	bch_crc64(((void *) (i)) + sizeof(uint64_t),			\
-	      ((void *) end(i)) - (((void *) (i)) + sizeof(uint64_t)))
+		  ((void *) bset_bkey_last(i)) -			\
+		  (((void *) (i)) + sizeof(uint64_t)))
 
 /* Error handling macros */
 

drivers/md/bcache/bset.h

@@ -1,7 +1,11 @@
 #ifndef _BCACHE_BSET_H
 #define _BCACHE_BSET_H
 
-#include <linux/slab.h>
+#include <linux/bcache.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+
+#include "util.h" /* for time_stats */
 
 /*
  * BKEYS:
@@ -142,20 +146,13 @@
  * first key in that range of bytes again.
  */
 
-/* Btree key comparison/iteration */
+struct btree_keys;
+struct btree_iter;
+struct btree_iter_set;
+struct bkey_float;
 
 #define MAX_BSETS		4U
 
-struct btree_iter {
-	size_t size, used;
-#ifdef CONFIG_BCACHE_DEBUG
-	struct btree *b;
-#endif
-	struct btree_iter_set {
-		struct bkey *k, *end;
-	} data[MAX_BSETS];
-};
-
 struct bset_tree {
 	/*
 	 * We construct a binary tree in an array as if the array
@@ -165,14 +162,14 @@ struct bset_tree {
 	 */
 
 	/* size of the binary tree and prev array */
-	unsigned	size;
+	unsigned		size;
 
 	/* function of size - precalculated for to_inorder() */
-	unsigned	extra;
+	unsigned		extra;
 
 	/* copy of the last key in the set */
-	struct bkey	end;
-	struct bkey_float *tree;
+	struct bkey		end;
+	struct bkey_float	*tree;
 
 	/*
 	 * The nodes in the bset tree point to specific keys - this
@@ -182,12 +179,219 @@ struct bset_tree {
 	 * to keep bkey_float to 4 bytes and prev isn't used in the fast
 	 * path.
 	 */
-	uint8_t		*prev;
+	uint8_t			*prev;
 
 	/* The actual btree node, with pointers to each sorted set */
-	struct bset	*data;
+	struct bset		*data;
 };
 
+struct btree_keys_ops {
+	bool		(*sort_cmp)(struct btree_iter_set,
+				    struct btree_iter_set);
+	struct bkey	*(*sort_fixup)(struct btree_iter *, struct bkey *);
+	bool		(*insert_fixup)(struct btree_keys *, struct bkey *,
+					struct btree_iter *, struct bkey *);
+	bool		(*key_invalid)(struct btree_keys *,
+				       const struct bkey *);
+	bool		(*key_bad)(struct btree_keys *, const struct bkey *);
+	bool		(*key_merge)(struct btree_keys *,
+				     struct bkey *, struct bkey *);
+	void		(*key_to_text)(char *, size_t, const struct bkey *);
+	void		(*key_dump)(struct btree_keys *, const struct bkey *);
+
+	/*
+	 * Only used for deciding whether to use START_KEY(k) or just the key
+	 * itself in a couple places
+	 */
+	bool		is_extents;
+};
+
+struct btree_keys {
+	const struct btree_keys_ops	*ops;
+	uint8_t			page_order;
+	uint8_t			nsets;
+	unsigned		last_set_unwritten:1;
+	bool			*expensive_debug_checks;
+
+	/*
+	 * Sets of sorted keys - the real btree node - plus a binary search tree
+	 *
+	 * set[0] is special; set[0]->tree, set[0]->prev and set[0]->data point
+	 * to the memory we have allocated for this btree node. Additionally,
+	 * set[0]->data points to the entire btree node as it exists on disk.
+	 */
+	struct bset_tree	set[MAX_BSETS];
+};
+
+static inline struct bset_tree *bset_tree_last(struct btree_keys *b)
+{
+	return b->set + b->nsets;
+}
+
+static inline bool bset_written(struct btree_keys *b, struct bset_tree *t)
+{
+	return t <= b->set + b->nsets - b->last_set_unwritten;
+}
+
+static inline bool bkey_written(struct btree_keys *b, struct bkey *k)
+{
+	return !b->last_set_unwritten || k < b->set[b->nsets].data->start;
+}
+
+static inline unsigned bset_byte_offset(struct btree_keys *b, struct bset *i)
+{
+	return ((size_t) i) - ((size_t) b->set->data);
+}
+
+static inline unsigned bset_sector_offset(struct btree_keys *b, struct bset *i)
+{
+	return bset_byte_offset(b, i) >> 9;
+}
+
+#define __set_bytes(i, k)	(sizeof(*(i)) + (k) * sizeof(uint64_t))
+#define set_bytes(i)		__set_bytes(i, i->keys)
+
+#define __set_blocks(i, k, block_bytes)				\
+	DIV_ROUND_UP(__set_bytes(i, k), block_bytes)
+#define set_blocks(i, block_bytes)				\
+	__set_blocks(i, (i)->keys, block_bytes)
+
+static inline size_t bch_btree_keys_u64s_remaining(struct btree_keys *b)
+{
+	struct bset_tree *t = bset_tree_last(b);
+
+	BUG_ON((PAGE_SIZE << b->page_order) <
+	       (bset_byte_offset(b, t->data) + set_bytes(t->data)));
+
+	if (!b->last_set_unwritten)
+		return 0;
+
+	return ((PAGE_SIZE << b->page_order) -
+		(bset_byte_offset(b, t->data) + set_bytes(t->data))) /
+		sizeof(u64);
+}
+
+static inline struct bset *bset_next_set(struct btree_keys *b,
+					 unsigned block_bytes)
+{
+	struct bset *i = bset_tree_last(b)->data;
+
+	return ((void *) i) + roundup(set_bytes(i), block_bytes);
+}
+
+void bch_btree_keys_free(struct btree_keys *);
+int bch_btree_keys_alloc(struct btree_keys *, unsigned, gfp_t);
+void bch_btree_keys_init(struct btree_keys *, const struct btree_keys_ops *,
+			 bool *);
+
+void bch_bset_init_next(struct btree_keys *, struct bset *, uint64_t);
+void bch_bset_build_written_tree(struct btree_keys *);
+void bch_bset_fix_invalidated_key(struct btree_keys *, struct bkey *);
+bool bch_bkey_try_merge(struct btree_keys *, struct bkey *, struct bkey *);
+void bch_bset_insert(struct btree_keys *, struct bkey *, struct bkey *);
+unsigned bch_btree_insert_key(struct btree_keys *, struct bkey *,
+			      struct bkey *);
+
+enum {
+	BTREE_INSERT_STATUS_NO_INSERT = 0,
+	BTREE_INSERT_STATUS_INSERT,
+	BTREE_INSERT_STATUS_BACK_MERGE,
+	BTREE_INSERT_STATUS_OVERWROTE,
+	BTREE_INSERT_STATUS_FRONT_MERGE,
+};
+
+/* Btree key iteration */
+
+struct btree_iter {
+	size_t size, used;
+#ifdef CONFIG_BCACHE_DEBUG
+	struct btree_keys *b;
+#endif
+	struct btree_iter_set {
+		struct bkey *k, *end;
+	} data[MAX_BSETS];
+};
+
+typedef bool (*ptr_filter_fn)(struct btree_keys *, const struct bkey *);
+
+struct bkey *bch_btree_iter_next(struct btree_iter *);
+struct bkey *bch_btree_iter_next_filter(struct btree_iter *,
+					struct btree_keys *, ptr_filter_fn);
+
+void bch_btree_iter_push(struct btree_iter *, struct bkey *, struct bkey *);
+struct bkey *bch_btree_iter_init(struct btree_keys *, struct btree_iter *,
+				 struct bkey *);
+
+struct bkey *__bch_bset_search(struct btree_keys *, struct bset_tree *,
+			       const struct bkey *);
+
+/*
+ * Returns the first key that is strictly greater than search
+ */
+static inline struct bkey *bch_bset_search(struct btree_keys *b,
+					   struct bset_tree *t,
+					   const struct bkey *search)
+{
+	return search ? __bch_bset_search(b, t, search) : t->data->start;
+}
+
+#define for_each_key_filter(b, k, iter, filter)				\
+	for (bch_btree_iter_init((b), (iter), NULL);			\
+	     ((k) = bch_btree_iter_next_filter((iter), (b), filter));)
+
+#define for_each_key(b, k, iter)					\
+	for (bch_btree_iter_init((b), (iter), NULL);			\
+	     ((k) = bch_btree_iter_next(iter));)
+
+/* Sorting */
+
+struct bset_sort_state {
+	mempool_t		*pool;
+
+	unsigned		page_order;
+	unsigned		crit_factor;
+
+	struct time_stats	time;
+};
+
+void bch_bset_sort_state_free(struct bset_sort_state *);
+int bch_bset_sort_state_init(struct bset_sort_state *, unsigned);
+void bch_btree_sort_lazy(struct btree_keys *, struct bset_sort_state *);
+void bch_btree_sort_into(struct btree_keys *, struct btree_keys *,
+			 struct bset_sort_state *);
+void bch_btree_sort_and_fix_extents(struct btree_keys *, struct btree_iter *,
+				    struct bset_sort_state *);
+void bch_btree_sort_partial(struct btree_keys *, unsigned,
+			    struct bset_sort_state *);
+
+static inline void bch_btree_sort(struct btree_keys *b,
+				  struct bset_sort_state *state)
+{
+	bch_btree_sort_partial(b, 0, state);
+}
+
+struct bset_stats {
+	size_t sets_written, sets_unwritten;
+	size_t bytes_written, bytes_unwritten;
+	size_t floats, failed;
+};
+
+void bch_btree_keys_stats(struct btree_keys *, struct bset_stats *);
+
+/* Bkey utility code */
+
+#define bset_bkey_last(i)	bkey_idx((struct bkey *) (i)->d, (i)->keys)
+
+static inline struct bkey *bset_bkey_idx(struct bset *i, unsigned idx)
+{
+	return bkey_idx(i->start, idx);
+}
+
+static inline void bkey_init(struct bkey *k)
+{
+	*k = ZERO_KEY;
+}
+
 static __always_inline int64_t bkey_cmp(const struct bkey *l,
 					const struct bkey *r)
 {
@@ -196,6 +400,62 @@ static __always_inline int64_t bkey_cmp(const struct bkey *l,
 		: (int64_t) KEY_OFFSET(l) - (int64_t) KEY_OFFSET(r);
 }
 
+void bch_bkey_copy_single_ptr(struct bkey *, const struct bkey *,
+			      unsigned);
+bool __bch_cut_front(const struct bkey *, struct bkey *);
+bool __bch_cut_back(const struct bkey *, struct bkey *);
+
+static inline bool bch_cut_front(const struct bkey *where, struct bkey *k)
+{
+	BUG_ON(bkey_cmp(where, k) > 0);
+	return __bch_cut_front(where, k);
+}
+
+static inline bool bch_cut_back(const struct bkey *where, struct bkey *k)
+{
+	BUG_ON(bkey_cmp(where, &START_KEY(k)) < 0);
+	return __bch_cut_back(where, k);
+}
+
+#define PRECEDING_KEY(_k)					\
+({								\
+	struct bkey *_ret = NULL;				\
+								\
+	if (KEY_INODE(_k) || KEY_OFFSET(_k)) {			\
+		_ret = &KEY(KEY_INODE(_k), KEY_OFFSET(_k), 0);	\
+								\
+		if (!_ret->low)					\
+			_ret->high--;				\
+		_ret->low--;					\
+	}							\
+								\
+	_ret;							\
+})
+
+static inline bool bch_ptr_invalid(struct btree_keys *b, const struct bkey *k)
+{
+	return b->ops->key_invalid(b, k);
+}
+
+static inline bool bch_ptr_bad(struct btree_keys *b, const struct bkey *k)
+{
+	return b->ops->key_bad(b, k);
+}
+
+static inline void bch_bkey_to_text(struct btree_keys *b, char *buf,
+				    size_t size, const struct bkey *k)
+{
+	return b->ops->key_to_text(buf, size, k);
+}
+
+static inline bool bch_bkey_equal_header(const struct bkey *l,
+					 const struct bkey *r)
+{
+	return (KEY_DIRTY(l) == KEY_DIRTY(r) &&
+		KEY_PTRS(l) == KEY_PTRS(r) &&
+		KEY_CSUM(l) == KEY_CSUM(l));
+}
+
 /* Keylists */
 
 struct keylist {
@@ -257,136 +517,44 @@ static inline size_t bch_keylist_bytes(struct keylist *l)
 
 struct bkey *bch_keylist_pop(struct keylist *);
 void bch_keylist_pop_front(struct keylist *);
-int bch_keylist_realloc(struct keylist *, int, struct cache_set *);
+int __bch_keylist_realloc(struct keylist *, unsigned);
 
-void bch_bkey_copy_single_ptr(struct bkey *, const struct bkey *,
-			      unsigned);
-bool __bch_cut_front(const struct bkey *, struct bkey *);
-bool __bch_cut_back(const struct bkey *, struct bkey *);
+/* Debug stuff */
 
-static inline bool bch_cut_front(const struct bkey *where, struct bkey *k)
-{
-	BUG_ON(bkey_cmp(where, k) > 0);
-	return __bch_cut_front(where, k);
-}
+#ifdef CONFIG_BCACHE_DEBUG
 
-static inline bool bch_cut_back(const struct bkey *where, struct bkey *k)
-{
-	BUG_ON(bkey_cmp(where, &START_KEY(k)) < 0);
-	return __bch_cut_back(where, k);
-}
+int __bch_count_data(struct btree_keys *);
+void __bch_check_keys(struct btree_keys *, const char *, ...);
+void bch_dump_bset(struct btree_keys *, struct bset *, unsigned);
+void bch_dump_bucket(struct btree_keys *);
 
-const char *bch_ptr_status(struct cache_set *, const struct bkey *);
-bool bch_btree_ptr_invalid(struct cache_set *, const struct bkey *);
-bool bch_extent_ptr_invalid(struct cache_set *, const struct bkey *);
+#else
 
-bool bch_ptr_bad(struct btree *, const struct bkey *);
+static inline int __bch_count_data(struct btree_keys *b) { return -1; }
+static inline void __bch_check_keys(struct btree_keys *b, const char *fmt, ...) {}
+static inline void bch_dump_bucket(struct btree_keys *b) {}
+void bch_dump_bset(struct btree_keys *, struct bset *, unsigned);
 
-static inline uint8_t gen_after(uint8_t a, uint8_t b)
-{
-	uint8_t r = a - b;
-	return r > 128U ? 0 : r;
-}
-
-static inline uint8_t ptr_stale(struct cache_set *c, const struct bkey *k,
-				unsigned i)
-{
-	return gen_after(PTR_BUCKET(c, k, i)->gen, PTR_GEN(k, i));
-}
-
-static inline bool ptr_available(struct cache_set *c, const struct bkey *k,
-				 unsigned i)
-{
-	return (PTR_DEV(k, i) < MAX_CACHES_PER_SET) && PTR_CACHE(c, k, i);
-}
-
-
-typedef bool (*ptr_filter_fn)(struct btree *, const struct bkey *);
-
-struct bkey *bch_btree_iter_next(struct btree_iter *);
-struct bkey *bch_btree_iter_next_filter(struct btree_iter *,
-					struct btree *, ptr_filter_fn);
+#endif
 
-void bch_btree_iter_push(struct btree_iter *, struct bkey *, struct bkey *);
-struct bkey *__bch_btree_iter_init(struct btree *, struct btree_iter *,
-				   struct bkey *, struct bset_tree *);
-
-/* 32 bits total: */
-#define BKEY_MID_BITS		3
-#define BKEY_EXPONENT_BITS	7
-#define BKEY_MANTISSA_BITS	22
-#define BKEY_MANTISSA_MASK	((1 << BKEY_MANTISSA_BITS) - 1)
-
-struct bkey_float {
-	unsigned	exponent:BKEY_EXPONENT_BITS;
-	unsigned	m:BKEY_MID_BITS;
-	unsigned	mantissa:BKEY_MANTISSA_BITS;
-} __packed;
-
-/*
- * BSET_CACHELINE was originally intended to match the hardware cacheline size -
- * it used to be 64, but I realized the lookup code would touch slightly less
- * memory if it was 128.
- *
- * It definites the number of bytes (in struct bset) per struct bkey_float in
- * the auxiliar search tree - when we're done searching the bset_float tree we
- * have this many bytes left that we do a linear search over.
- *
- * Since (after level 5) every level of the bset_tree is on a new cacheline,
- * we're touching one fewer cacheline in the bset tree in exchange for one more
- * cacheline in the linear search - but the linear search might stop before it
- * gets to the second cacheline.
- */
-
-#define BSET_CACHELINE		128
-#define bset_tree_space(b)	(btree_data_space(b) / BSET_CACHELINE)
-
-#define bset_tree_bytes(b)	(bset_tree_space(b) * sizeof(struct bkey_float))
-#define bset_prev_bytes(b)	(bset_tree_space(b) * sizeof(uint8_t))
-
-void bch_bset_init_next(struct btree *);
-
-void bch_bset_fix_invalidated_key(struct btree *, struct bkey *);
-void bch_bset_fix_lookup_table(struct btree *, struct bkey *);
-
-struct bkey *__bch_bset_search(struct btree *, struct bset_tree *,
-			   const struct bkey *);
-
-/*
- * Returns the first key that is strictly greater than search
- */
-static inline struct bkey *bch_bset_search(struct btree *b, struct bset_tree *t,
-					   const struct bkey *search)
+static inline bool btree_keys_expensive_checks(struct btree_keys *b)
 {
-	return search ? __bch_bset_search(b, t, search) : t->data->start;
+#ifdef CONFIG_BCACHE_DEBUG
+	return *b->expensive_debug_checks;
+#else
+	return false;
+#endif
 }
-
-#define PRECEDING_KEY(_k)					\
-({								\
-	struct bkey *_ret = NULL;				\
-								\
-	if (KEY_INODE(_k) || KEY_OFFSET(_k)) {			\
-		_ret = &KEY(KEY_INODE(_k), KEY_OFFSET(_k), 0);	\
-								\
-		if (!_ret->low)					\
-			_ret->high--;				\
-		_ret->low--;					\
-	}							\
-								\
-	_ret;							\
-})
-
-bool bch_bkey_try_merge(struct btree *, struct bkey *, struct bkey *);
-void bch_btree_sort_lazy(struct btree *);
-void bch_btree_sort_into(struct btree *, struct btree *);
-void bch_btree_sort_and_fix_extents(struct btree *, struct btree_iter *);
-void bch_btree_sort_partial(struct btree *, unsigned);
 
-static inline void bch_btree_sort(struct btree *b)
+static inline int bch_count_data(struct btree_keys *b)
 {
-	bch_btree_sort_partial(b, 0);
+	return btree_keys_expensive_checks(b) ? __bch_count_data(b) : -1;
 }
 
-int bch_bset_print_stats(struct cache_set *, char *);
+#define bch_check_keys(b, ...)						\
+do {									\
+	if (btree_keys_expensive_checks(b))				\
+		__bch_check_keys(b, __VA_ARGS__);			\
+} while (0)
 
 #endif

drivers/md/bcache/btree.h

@@ -130,20 +130,12 @@ struct btree {
 	unsigned long		flags;
 	uint16_t		written;	/* would be nice to kill */
 	uint8_t			level;
-	uint8_t			nsets;
-	uint8_t			page_order;
 
-	/*
-	 * Set of sorted keys - the real btree node - plus a binary search tree
-	 *
-	 * sets[0] is special; set[0]->tree, set[0]->prev and set[0]->data point
-	 * to the memory we have allocated for this btree node. Additionally,
-	 * set[0]->data points to the entire btree node as it exists on disk.
-	 */
-	struct bset_tree	sets[MAX_BSETS];
+	struct btree_keys	keys;
 
 	/* For outstanding btree writes, used as a lock - protects write_idx */
-	struct closure_with_waitlist	io;
+	struct closure		io;
+	struct semaphore	io_mutex;
 
 	struct list_head	list;
 	struct delayed_work	work;
@@ -179,24 +171,19 @@ static inline struct btree_write *btree_prev_write(struct btree *b)
 	return b->writes + (btree_node_write_idx(b) ^ 1);
 }
 
-static inline unsigned bset_offset(struct btree *b, struct bset *i)
+static inline struct bset *btree_bset_first(struct btree *b)
 {
-	return (((size_t) i) - ((size_t) b->sets->data)) >> 9;
+	return b->keys.set->data;
 }
 
-static inline struct bset *write_block(struct btree *b)
+static inline struct bset *btree_bset_last(struct btree *b)
 {
-	return ((void *) b->sets[0].data) + b->written * block_bytes(b->c);
+	return bset_tree_last(&b->keys)->data;
 }
 
-static inline bool bset_written(struct btree *b, struct bset_tree *t)
+static inline unsigned bset_block_offset(struct btree *b, struct bset *i)
 {
-	return t->data < write_block(b);
-}
-
-static inline bool bkey_written(struct btree *b, struct bkey *k)
-{
-	return k < write_block(b)->start;
+	return bset_sector_offset(&b->keys, i) >> b->c->block_bits;
 }
 
 static inline void set_gc_sectors(struct cache_set *c)
@@ -204,21 +191,6 @@ static inline void set_gc_sectors(struct cache_set *c)
 	atomic_set(&c->sectors_to_gc, c->sb.bucket_size * c->nbuckets / 16);
 }
 
-static inline struct bkey *bch_btree_iter_init(struct btree *b,
-					       struct btree_iter *iter,
-					       struct bkey *search)
-{
-	return __bch_btree_iter_init(b, iter, search, b->sets);
-}
-
-static inline bool bch_ptr_invalid(struct btree *b, const struct bkey *k)
-{
-	if (b->level)
-		return bch_btree_ptr_invalid(b->c, k);
-	else
-		return bch_extent_ptr_invalid(b->c, k);
-}
-
 void bkey_put(struct cache_set *c, struct bkey *k);
 
 /* Looping macros */
@@ -229,17 +201,12 @@ void bkey_put(struct cache_set *c, struct bkey *k);
 	     iter++)							\
 		hlist_for_each_entry_rcu((b), (c)->bucket_hash + iter, hash)
 
-#define for_each_key_filter(b, k, iter, filter)				\
-	for (bch_btree_iter_init((b), (iter), NULL);			\
-	     ((k) = bch_btree_iter_next_filter((iter), b, filter));)
-
-#define for_each_key(b, k, iter)					\
-	for (bch_btree_iter_init((b), (iter), NULL);			\
-	     ((k) = bch_btree_iter_next(iter));)
-
 /* Recursing down the btree */
 
 struct btree_op {
+	/* for waiting on btree reserve in btree_split() */
+	wait_queue_t		wait;
+
 	/* Btree level at which we start taking write locks */
 	short			lock;
 
@@ -249,6 +216,7 @@ struct btree_op {
 static inline void bch_btree_op_init(struct btree_op *op, int write_lock_level)
 {
 	memset(op, 0, sizeof(struct btree_op));
+	init_wait(&op->wait);
 	op->lock = write_lock_level;
 }
 
@@ -267,7 +235,7 @@ static inline void rw_unlock(bool w, struct btree *b)
 	(w ? up_write : up_read)(&b->lock);
 }
 
-void bch_btree_node_read(struct btree *);
+void bch_btree_node_read_done(struct btree *);
 void bch_btree_node_write(struct btree *, struct closure *);
 
 void bch_btree_set_root(struct btree *);

drivers/md/bcache/closure.c

@@ -11,19 +11,6 @@
 
 #include "closure.h"
 
-#define CL_FIELD(type, field)					\
-	case TYPE_ ## type:					\
-	return &container_of(cl, struct type, cl)->field
-
-static struct closure_waitlist *closure_waitlist(struct closure *cl)
-{
-	switch (cl->type) {
-		CL_FIELD(closure_with_waitlist, wait);
-	default:
-		return NULL;
-	}
-}
-
 static inline void closure_put_after_sub(struct closure *cl, int flags)
 {
 	int r = flags & CLOSURE_REMAINING_MASK;
@@ -42,17 +29,10 @@ static inline void closure_put_after_sub(struct closure *cl, int flags)
 			closure_queue(cl);
 		} else {
 			struct closure *parent = cl->parent;
-			struct closure_waitlist *wait = closure_waitlist(cl);
 			closure_fn *destructor = cl->fn;
 
 			closure_debug_destroy(cl);
 
-			smp_mb();
-			atomic_set(&cl->remaining, -1);
-
-			if (wait)
-				closure_wake_up(wait);
-
 			if (destructor)
 				destructor(cl);
 
@@ -69,19 +49,18 @@ void closure_sub(struct closure *cl, int v)
 }
 EXPORT_SYMBOL(closure_sub);
 
+/**
+ * closure_put - decrement a closure's refcount
+ */
 void closure_put(struct closure *cl)
 {
 	closure_put_after_sub(cl, atomic_dec_return(&cl->remaining));
 }
 EXPORT_SYMBOL(closure_put);
 
-static void set_waiting(struct closure *cl, unsigned long f)
-{
-#ifdef CONFIG_BCACHE_CLOSURES_DEBUG
-	cl->waiting_on = f;
-#endif
-}
-
+/**
+ * closure_wake_up - wake up all closures on a wait list, without memory barrier
+ */
 void __closure_wake_up(struct closure_waitlist *wait_list)
 {
 	struct llist_node *list;
@@ -106,27 +85,34 @@ void __closure_wake_up(struct closure_waitlist *wait_list)
 		cl = container_of(reverse, struct closure, list);
 		reverse = llist_next(reverse);
 
-		set_waiting(cl, 0);
+		closure_set_waiting(cl, 0);
 		closure_sub(cl, CLOSURE_WAITING + 1);
 	}
 }
 EXPORT_SYMBOL(__closure_wake_up);
 
-bool closure_wait(struct closure_waitlist *list, struct closure *cl)
+/**
+ * closure_wait - add a closure to a waitlist
+ *
+ * @waitlist will own a ref on @cl, which will be released when
+ * closure_wake_up() is called on @waitlist.
+ *
+ */
+bool closure_wait(struct closure_waitlist *waitlist, struct closure *cl)
 {
 	if (atomic_read(&cl->remaining) & CLOSURE_WAITING)
 		return false;
 
-	set_waiting(cl, _RET_IP_);
+	closure_set_waiting(cl, _RET_IP_);
 	atomic_add(CLOSURE_WAITING + 1, &cl->remaining);
-	llist_add(&cl->list, &list->list);
+	llist_add(&cl->list, &waitlist->list);
 
 	return true;
 }
 EXPORT_SYMBOL(closure_wait);
 
 /**
- * closure_sync() - sleep until a closure a closure has nothing left to wait on
+ * closure_sync - sleep until a closure a closure has nothing left to wait on
  *
  * Sleeps until the refcount hits 1 - the thread that's running the closure owns
  * the last refcount.
@@ -148,46 +134,6 @@ void closure_sync(struct closure *cl)
 }
 EXPORT_SYMBOL(closure_sync);
 
-/**
- * closure_trylock() - try to acquire the closure, without waiting
- * @cl:		closure to lock
- *
- * Returns true if the closure was succesfully locked.
- */
-bool closure_trylock(struct closure *cl, struct closure *parent)
-{
-	if (atomic_cmpxchg(&cl->remaining, -1,
-			   CLOSURE_REMAINING_INITIALIZER) != -1)
-		return false;
-
-	smp_mb();
-
-	cl->parent = parent;
-	if (parent)
-		closure_get(parent);
-
-	closure_set_ret_ip(cl);
-	closure_debug_create(cl);
-	return true;
-}
-EXPORT_SYMBOL(closure_trylock);
-
-void __closure_lock(struct closure *cl, struct closure *parent,
-		    struct closure_waitlist *wait_list)
-{
-	struct closure wait;
-	closure_init_stack(&wait);
-
-	while (1) {
-		if (closure_trylock(cl, parent))
-			return;
-
-		closure_wait_event(wait_list, &wait,
-				   atomic_read(&cl->remaining) == -1);
-	}
-}
-EXPORT_SYMBOL(__closure_lock);
-
 #ifdef CONFIG_BCACHE_CLOSURES_DEBUG
 
 static LIST_HEAD(closure_list);