Merge branch 'for-3.3/drivers' of git://git.kernel.dk/linux-block

* 'for-3.3/drivers' of git://git.kernel.dk/linux-block:
  mtip32xx: do rebuild monitoring asynchronously
  xen-blkfront: Use kcalloc instead of kzalloc to allocate array
  mtip32xx: uninitialized variable in mtip_quiesce_io()
  mtip32xx: updates based on feedback
  xen-blkback: convert hole punching to discard request on loop devices
  xen/blkback: Move processing of BLKIF_OP_DISCARD from dispatch_rw_block_io
  xen/blk[front|back]: Enhance discard support with secure erasing support.
  xen/blk[front|back]: Squash blkif_request_rw and blkif_request_discard together
  mtip32xx: update to new ->make_request() API
  mtip32xx: add module.h include to avoid conflict with moduleh tree
  mtip32xx: mark a few more items static
  mtip32xx: ensure that all local functions are static
  mtip32xx: cleanup compat ioctl handling
  mtip32xx: fix warnings/errors on 32-bit compiles
  block: Add driver for Micron RealSSD pcie flash cards

drivers/block/Kconfig

@@ -116,6 +116,8 @@ config PARIDE
 
 source "drivers/block/paride/Kconfig"
 
+source "drivers/block/mtip32xx/Kconfig"
+
 config BLK_CPQ_DA
 	tristate "Compaq SMART2 support"
 	depends on PCI && VIRT_TO_BUS

drivers/block/Makefile

@@ -39,5 +39,6 @@ obj-$(CONFIG_XEN_BLKDEV_FRONTEND)	+= xen-blkfront.o
 obj-$(CONFIG_XEN_BLKDEV_BACKEND)	+= xen-blkback/
 obj-$(CONFIG_BLK_DEV_DRBD)     += drbd/
 obj-$(CONFIG_BLK_DEV_RBD)     += rbd.o
+obj-$(CONFIG_BLK_DEV_PCIESSD_MTIP32XX)	+= mtip32xx/
 
 swim_mod-y	:= swim.o swim_asm.o

drivers/block/mtip32xx/Kconfig

+#
+# mtip32xx device driver configuration
+#
+
+config BLK_DEV_PCIESSD_MTIP32XX
+	tristate "Block Device Driver for Micron PCIe SSDs"
+	depends on HOTPLUG_PCI_PCIE
+	help
+          This enables the block driver for Micron PCIe SSDs.

drivers/block/mtip32xx/Makefile

+#
+# Makefile for  Block device driver for Micron PCIe SSD
+#
+
+obj-$(CONFIG_BLK_DEV_PCIESSD_MTIP32XX) += mtip32xx.o

drivers/block/mtip32xx/mtip32xx.h

+/*
+ * mtip32xx.h - Header file for the P320 SSD Block Driver
+ *   Copyright (C) 2011 Micron Technology, Inc.
+ *
+ * Portions of this code were derived from works subjected to the
+ * following copyright:
+ *    Copyright (C) 2009 Integrated Device Technology, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#ifndef __MTIP32XX_H__
+#define __MTIP32XX_H__
+
+#include <linux/spinlock.h>
+#include <linux/rwsem.h>
+#include <linux/ata.h>
+#include <linux/interrupt.h>
+#include <linux/genhd.h>
+#include <linux/version.h>
+
+/* Offset of Subsystem Device ID in pci confoguration space */
+#define PCI_SUBSYSTEM_DEVICEID	0x2E
+
+/* offset of Device Control register in PCIe extended capabilites space */
+#define PCIE_CONFIG_EXT_DEVICE_CONTROL_OFFSET	0x48
+
+/* # of times to retry timed out IOs */
+#define MTIP_MAX_RETRIES	5
+
+/* Various timeout values in ms */
+#define MTIP_NCQ_COMMAND_TIMEOUT_MS       5000
+#define MTIP_IOCTL_COMMAND_TIMEOUT_MS     5000
+#define MTIP_INTERNAL_COMMAND_TIMEOUT_MS  5000
+
+/* check for timeouts every 500ms */
+#define MTIP_TIMEOUT_CHECK_PERIOD	500
+
+/* ftl rebuild */
+#define MTIP_FTL_REBUILD_OFFSET		142
+#define MTIP_FTL_REBUILD_MAGIC		0xED51
+#define MTIP_FTL_REBUILD_TIMEOUT_MS	2400000
+
+/* Macro to extract the tag bit number from a tag value. */
+#define MTIP_TAG_BIT(tag)	(tag & 0x1F)
+
+/*
+ * Macro to extract the tag index from a tag value. The index
+ * is used to access the correct s_active/Command Issue register based
+ * on the tag value.
+ */
+#define MTIP_TAG_INDEX(tag)	(tag >> 5)
+
+/*
+ * Maximum number of scatter gather entries
+ * a single command may have.
+ */
+#define MTIP_MAX_SG		128
+
+/*
+ * Maximum number of slot groups (Command Issue & s_active registers)
+ * NOTE: This is the driver maximum; check dd->slot_groups for actual value.
+ */
+#define MTIP_MAX_SLOT_GROUPS	8
+
+/* Internal command tag. */
+#define MTIP_TAG_INTERNAL	0
+
+/* Micron Vendor ID & P320x SSD Device ID */
+#define PCI_VENDOR_ID_MICRON    0x1344
+#define P320_DEVICE_ID		0x5150
+
+/* Driver name and version strings */
+#define MTIP_DRV_NAME		"mtip32xx"
+#define MTIP_DRV_VERSION	"1.2.6os3"
+
+/* Maximum number of minor device numbers per device. */
+#define MTIP_MAX_MINORS		16
+
+/* Maximum number of supported command slots. */
+#define MTIP_MAX_COMMAND_SLOTS	(MTIP_MAX_SLOT_GROUPS * 32)
+
+/*
+ * Per-tag bitfield size in longs.
+ * Linux bit manipulation functions
+ * (i.e. test_and_set_bit, find_next_zero_bit)
+ * manipulate memory in longs, so we try to make the math work.
+ * take the slot groups and find the number of longs, rounding up.
+ * Careful! i386 and x86_64 use different size longs!
+ */
+#define U32_PER_LONG	(sizeof(long) / sizeof(u32))
+#define SLOTBITS_IN_LONGS ((MTIP_MAX_SLOT_GROUPS + \
+					(U32_PER_LONG-1))/U32_PER_LONG)
+
+/* BAR number used to access the HBA registers. */
+#define MTIP_ABAR		5
+
+/* Forced Unit Access Bit */
+#define FUA_BIT			0x80
+
+#ifdef DEBUG
+ #define dbg_printk(format, arg...)	\
+	printk(pr_fmt(format), ##arg);
+#else
+ #define dbg_printk(format, arg...)
+#endif
+
+#define __force_bit2int (unsigned int __force)
+
+/* below are bit numbers in 'flags' defined in mtip_port */
+#define MTIP_FLAG_IC_ACTIVE_BIT			0
+#define MTIP_FLAG_EH_ACTIVE_BIT			1
+#define MTIP_FLAG_SVC_THD_ACTIVE_BIT		2
+#define MTIP_FLAG_ISSUE_CMDS_BIT		4
+#define MTIP_FLAG_REBUILD_BIT			5
+#define MTIP_FLAG_SVC_THD_SHOULD_STOP_BIT	8
+
+/* Register Frame Information Structure (FIS), host to device. */
+struct host_to_dev_fis {
+	/*
+	 * FIS type.
+	 * - 27h Register FIS, host to device.
+	 * - 34h Register FIS, device to host.
+	 * - 39h DMA Activate FIS, device to host.
+	 * - 41h DMA Setup FIS, bi-directional.
+	 * - 46h Data FIS, bi-directional.
+	 * - 58h BIST Activate FIS, bi-directional.
+	 * - 5Fh PIO Setup FIS, device to host.
+	 * - A1h Set Device Bits FIS, device to host.
+	 */
+	unsigned char type;
+	unsigned char opts;
+	unsigned char command;
+	unsigned char features;
+
+	union {
+		unsigned char lba_low;
+		unsigned char sector;
+	};
+	union {
+		unsigned char lba_mid;
+		unsigned char cyl_low;
+	};
+	union {
+		unsigned char lba_hi;
+		unsigned char cyl_hi;
+	};
+	union {
+		unsigned char device;
+		unsigned char head;
+	};
+
+	union {
+		unsigned char lba_low_ex;
+		unsigned char sector_ex;
+	};
+	union {
+		unsigned char lba_mid_ex;
+		unsigned char cyl_low_ex;
+	};
+	union {
+		unsigned char lba_hi_ex;
+		unsigned char cyl_hi_ex;
+	};
+	unsigned char features_ex;
+
+	unsigned char sect_count;
+	unsigned char sect_cnt_ex;
+	unsigned char res2;
+	unsigned char control;
+
+	unsigned int res3;
+};
+
+/* Command header structure. */
+struct mtip_cmd_hdr {
+	/*
+	 * Command options.
+	 * - Bits 31:16 Number of PRD entries.
+	 * - Bits 15:8 Unused in this implementation.
+	 * - Bit 7 Prefetch bit, informs the drive to prefetch PRD entries.
+	 * - Bit 6 Write bit, should be set when writing data to the device.
+	 * - Bit 5 Unused in this implementation.
+	 * - Bits 4:0 Length of the command FIS in DWords (DWord = 4 bytes).
+	 */
+	unsigned int opts;
+	/* This field is unsed when using NCQ. */
+	union {
+		unsigned int byte_count;
+		unsigned int status;
+	};
+	/*
+	 * Lower 32 bits of the command table address associated with this
+	 * header. The command table addresses must be 128 byte aligned.
+	 */
+	unsigned int ctba;
+	/*
+	 * If 64 bit addressing is used this field is the upper 32 bits
+	 * of the command table address associated with this command.
+	 */
+	unsigned int ctbau;
+	/* Reserved and unused. */
+	unsigned int res[4];
+};
+
+/* Command scatter gather structure (PRD). */
+struct mtip_cmd_sg {
+	/*
+	 * Low 32 bits of the data buffer address. For P320 this
+	 * address must be 8 byte aligned signified by bits 2:0 being
+	 * set to 0.
+	 */
+	unsigned int dba;
+	/*
+	 * When 64 bit addressing is used this field is the upper
+	 * 32 bits of the data buffer address.
+	 */
+	unsigned int dba_upper;
+	/* Unused. */
+	unsigned int reserved;
+	/*
+	 * Bit 31: interrupt when this data block has been transferred.
+	 * Bits 30..22: reserved
+	 * Bits 21..0: byte count (minus 1).  For P320 the byte count must be
+	 * 8 byte aligned signified by bits 2:0 being set to 1.
+	 */
+	unsigned int info;
+};
+struct mtip_port;
+
+/* Structure used to describe a command. */
+struct mtip_cmd {
+
+	struct mtip_cmd_hdr *command_header; /* ptr to command header entry */
+
+	dma_addr_t command_header_dma; /* corresponding physical address */
+
+	void *command; /* ptr to command table entry */
+
+	dma_addr_t command_dma; /* corresponding physical address */
+
+	void *comp_data; /* data passed to completion function comp_func() */
+	/*
+	 * Completion function called by the ISR upon completion of
+	 * a command.
+	 */
+	void (*comp_func)(struct mtip_port *port,
+				int tag,
+				void *data,
+				int status);
+	/* Additional callback function that may be called by comp_func() */
+	void (*async_callback)(void *data, int status);
+
+	void *async_data; /* Addl. data passed to async_callback() */
+
+	int scatter_ents; /* Number of scatter list entries used */
+
+	struct scatterlist sg[MTIP_MAX_SG]; /* Scatter list entries */
+
+	int retries; /* The number of retries left for this command. */
+
+	int direction; /* Data transfer direction */
+
+	unsigned long comp_time; /* command completion time, in jiffies */
+
+	atomic_t active; /* declares if this command sent to the drive. */
+};
+
+/* Structure used to describe a port. */
+struct mtip_port {
+	/* Pointer back to the driver data for this port. */
+	struct driver_data *dd;
+	/*
+	 * Used to determine if the data pointed to by the
+	 * identify field is valid.
+	 */
+	unsigned long identify_valid;
+	/* Base address of the memory mapped IO for the port. */
+	void __iomem *mmio;
+	/* Array of pointers to the memory mapped s_active registers. */
+	void __iomem *s_active[MTIP_MAX_SLOT_GROUPS];
+	/* Array of pointers to the memory mapped completed registers. */
+	void __iomem *completed[MTIP_MAX_SLOT_GROUPS];
+	/* Array of pointers to the memory mapped Command Issue registers. */
+	void __iomem *cmd_issue[MTIP_MAX_SLOT_GROUPS];
+	/*
+	 * Pointer to the beginning of the command header memory as used
+	 * by the driver.
+	 */
+	void *command_list;
+	/*
+	 * Pointer to the beginning of the command header memory as used
+	 * by the DMA.
+	 */
+	dma_addr_t command_list_dma;
+	/*
+	 * Pointer to the beginning of the RX FIS memory as used
+	 * by the driver.
+	 */
+	void *rxfis;
+	/*
+	 * Pointer to the beginning of the RX FIS memory as used
+	 * by the DMA.
+	 */
+	dma_addr_t rxfis_dma;
+	/*
+	 * Pointer to the beginning of the command table memory as used
+	 * by the driver.
+	 */
+	void *command_table;
+	/*
+	 * Pointer to the beginning of the command table memory as used
+	 * by the DMA.
+	 */
+	dma_addr_t command_tbl_dma;
+	/*
+	 * Pointer to the beginning of the identify data memory as used
+	 * by the driver.
+	 */
+	u16 *identify;
+	/*
+	 * Pointer to the beginning of the identify data memory as used
+	 * by the DMA.
+	 */
+	dma_addr_t identify_dma;
+	/*
+	 * Pointer to the beginning of a sector buffer that is used
+	 * by the driver when issuing internal commands.
+	 */
+	u16 *sector_buffer;
+	/*
+	 * Pointer to the beginning of a sector buffer that is used
+	 * by the DMA when the driver issues internal commands.
+	 */
+	dma_addr_t sector_buffer_dma;
+	/*
+	 * Bit significant, used to determine if a command slot has
+	 * been allocated. i.e. the slot is in use.  Bits are cleared
+	 * when the command slot and all associated data structures
+	 * are no longer needed.
+	 */
+	unsigned long allocated[SLOTBITS_IN_LONGS];
+	/*
+	 * used to queue commands when an internal command is in progress
+	 * or error handling is active
+	 */
+	unsigned long cmds_to_issue[SLOTBITS_IN_LONGS];
+	/*
+	 * Array of command slots. Structure includes pointers to the
+	 * command header and command table, and completion function and data
+	 * pointers.
+	 */
+	struct mtip_cmd commands[MTIP_MAX_COMMAND_SLOTS];
+	/* Used by mtip_service_thread to wait for an event */
+	wait_queue_head_t svc_wait;
+	/*
+	 * indicates the state of the port. Also, helps the service thread
+	 * to determine its action on wake up.
+	 */
+	unsigned long flags;
+	/*
+	 * Timer used to complete commands that have been active for too long.
+	 */
+	struct timer_list cmd_timer;
+	/*
+	 * Semaphore used to block threads if there are no
+	 * command slots available.
+	 */
+	struct semaphore cmd_slot;
+	/* Spinlock for working around command-issue bug. */
+	spinlock_t cmd_issue_lock;
+};
+
+/*
+ * Driver private data structure.
+ *
+ * One structure is allocated per probed device.
+ */
+struct driver_data {
+	void __iomem *mmio; /* Base address of the HBA registers. */
+
+	int major; /* Major device number. */
+
+	int instance; /* Instance number. First device probed is 0, ... */
+
+	struct gendisk *disk; /* Pointer to our gendisk structure. */
+
+	struct pci_dev *pdev; /* Pointer to the PCI device structure. */
+
+	struct request_queue *queue; /* Our request queue. */
+
+	struct mtip_port *port; /* Pointer to the port data structure. */
+
+	/* Tasklet used to process the bottom half of the ISR. */
+	struct tasklet_struct tasklet;
+
+	unsigned product_type; /* magic value declaring the product type */
+
+	unsigned slot_groups; /* number of slot groups the product supports */
+
+	atomic_t drv_cleanup_done; /* Atomic variable for SRSI */
+
+	unsigned long index; /* Index to determine the disk name */
+
+	unsigned int ftlrebuildflag; /* FTL rebuild flag */
+
+	atomic_t resumeflag; /* Atomic variable to track suspend/resume */
+
+	atomic_t eh_active; /* Flag for error handling tracking */
+
+	struct task_struct *mtip_svc_handler; /* task_struct of svc thd */
+};
+
+#endif

drivers/block/xen-blkback/blkback.c

@@ -39,9 +39,6 @@
 #include <linux/list.h>
 #include <linux/delay.h>
 #include <linux/freezer.h>
-#include <linux/loop.h>
-#include <linux/falloc.h>
-#include <linux/fs.h>
 
 #include <xen/events.h>
 #include <xen/page.h>
@@ -362,7 +359,7 @@ static int xen_blkbk_map(struct blkif_request *req,
 {
 	struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 	int i;
-	int nseg = req->nr_segments;
+	int nseg = req->u.rw.nr_segments;
 	int ret = 0;
 
 	/*
@@ -416,30 +413,25 @@ static int xen_blkbk_map(struct blkif_request *req,
 	return ret;
 }
 
-static void xen_blk_discard(struct xen_blkif *blkif, struct blkif_request *req)
+static int dispatch_discard_io(struct xen_blkif *blkif,
+				struct blkif_request *req)
 {
 	int err = 0;
 	int status = BLKIF_RSP_OKAY;
 	struct block_device *bdev = blkif->vbd.bdev;
 
-	if (blkif->blk_backend_type == BLKIF_BACKEND_PHY)
-		/* just forward the discard request */
+	blkif->st_ds_req++;
+
+	xen_blkif_get(blkif);
+	if (blkif->blk_backend_type == BLKIF_BACKEND_PHY ||
+	    blkif->blk_backend_type == BLKIF_BACKEND_FILE) {
+		unsigned long secure = (blkif->vbd.discard_secure &&
+			(req->u.discard.flag & BLKIF_DISCARD_SECURE)) ?
+			BLKDEV_DISCARD_SECURE : 0;
 		err = blkdev_issue_discard(bdev,
 				req->u.discard.sector_number,
 				req->u.discard.nr_sectors,
-				GFP_KERNEL, 0);
-	else if (blkif->blk_backend_type == BLKIF_BACKEND_FILE) {
-		/* punch a hole in the backing file */
-		struct loop_device *lo = bdev->bd_disk->private_data;
-		struct file *file = lo->lo_backing_file;
-
-		if (file->f_op->fallocate)
-			err = file->f_op->fallocate(file,
-				FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
-				req->u.discard.sector_number << 9,
-				req->u.discard.nr_sectors << 9);
-		else
-			err = -EOPNOTSUPP;
+				GFP_KERNEL, secure);
 	} else
 		err = -EOPNOTSUPP;
 
@@ -449,7 +441,9 @@ static void xen_blk_discard(struct xen_blkif *blkif, struct blkif_request *req)
 	} else if (err)
 		status = BLKIF_RSP_ERROR;
 
-	make_response(blkif, req->id, req->operation, status);
+	make_response(blkif, req->u.discard.id, req->operation, status);
+	xen_blkif_put(blkif);
+	return err;
 }
 
 static void xen_blk_drain_io(struct xen_blkif *blkif)
@@ -573,8 +567,11 @@ __do_block_io_op(struct xen_blkif *blkif)
 
 		/* Apply all sanity checks to /private copy/ of request. */
 		barrier();
-
-		if (dispatch_rw_block_io(blkif, &req, pending_req))
+		if (unlikely(req.operation == BLKIF_OP_DISCARD)) {
+			free_req(pending_req);
+			if (dispatch_discard_io(blkif, &req))
+				break;
+		} else if (dispatch_rw_block_io(blkif, &req, pending_req))
 			break;
 
 		/* Yield point for this unbounded loop. */
@@ -633,10 +630,6 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		blkif->st_f_req++;
 		operation = WRITE_FLUSH;
 		break;
-	case BLKIF_OP_DISCARD:
-		blkif->st_ds_req++;
-		operation = REQ_DISCARD;
-		break;
 	default:
 		operation = 0; /* make gcc happy */
 		goto fail_response;
@@ -644,9 +637,9 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 	}
 
 	/* Check that the number of segments is sane. */
-	nseg = req->nr_segments;
-	if (unlikely(nseg == 0 && operation != WRITE_FLUSH &&
-				operation != REQ_DISCARD) ||
+	nseg = req->u.rw.nr_segments;
+
+	if (unlikely(nseg == 0 && operation != WRITE_FLUSH) ||
 	    unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
 		pr_debug(DRV_PFX "Bad number of segments in request (%d)\n",
 			 nseg);
@@ -654,12 +647,12 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		goto fail_response;
 	}
 
-	preq.dev           = req->handle;
+	preq.dev           = req->u.rw.handle;
 	preq.sector_number = req->u.rw.sector_number;
 	preq.nr_sects      = 0;
 
 	pending_req->blkif     = blkif;
-	pending_req->id        = req->id;
+	pending_req->id        = req->u.rw.id;
 	pending_req->operation = req->operation;
 	pending_req->status    = BLKIF_RSP_OKAY;
 	pending_req->nr_pages  = nseg;
@@ -707,7 +700,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 	 * the hypercall to unmap the grants - that is all done in
 	 * xen_blkbk_unmap.
 	 */
-	if (operation != REQ_DISCARD && xen_blkbk_map(req, pending_req, seg))
+	if (xen_blkbk_map(req, pending_req, seg))
 		goto fail_flush;
 
 	/*
@@ -739,9 +732,8 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 
 	/* This will be hit if the operation was a flush or discard. */
 	if (!bio) {
-		BUG_ON(operation != WRITE_FLUSH && operation != REQ_DISCARD);
+		BUG_ON(operation != WRITE_FLUSH);
 
-		if (operation == WRITE_FLUSH) {
 		bio = bio_alloc(GFP_KERNEL, 0);
 		if (unlikely(bio == NULL))
 			goto fail_put_bio;
@@ -750,12 +742,6 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 		bio->bi_bdev    = preq.bdev;
 		bio->bi_private = pending_req;
 		bio->bi_end_io  = end_block_io_op;
-		} else if (operation == REQ_DISCARD) {
-			xen_blk_discard(blkif, req);
-			xen_blkif_put(blkif);
-			free_req(pending_req);
-			return 0;
-		}
 	}
 
 	/*
@@ -784,7 +770,7 @@ static int dispatch_rw_block_io(struct xen_blkif *blkif,
 	xen_blkbk_unmap(pending_req);
  fail_response:
 	/* Haven't submitted any bio's yet. */
-	make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
+	make_response(blkif, req->u.rw.id, req->operation, BLKIF_RSP_ERROR);
 	free_req(pending_req);
 	msleep(1); /* back off a bit */
 	return -EIO;

drivers/block/xen-blkback/common.h

@@ -60,58 +60,66 @@ struct blkif_common_response {
 	char dummy;
 };
 
-/* i386 protocol version */
-#pragma pack(push, 4)
-
 struct blkif_x86_32_request_rw {
+	uint8_t        nr_segments;  /* number of segments                   */
+	blkif_vdev_t   handle;       /* only for read/write requests         */
+	uint64_t       id;           /* private guest value, echoed in resp  */
 	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
 	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-};
+} __attribute__((__packed__));
 
 struct blkif_x86_32_request_discard {
+	uint8_t        flag;         /* BLKIF_DISCARD_SECURE or zero         */
+	blkif_vdev_t   _pad1;        /* was "handle" for read/write requests */
+	uint64_t       id;           /* private guest value, echoed in resp  */
 	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
 	uint64_t       nr_sectors;
-};
+} __attribute__((__packed__));
 
 struct blkif_x86_32_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
-	uint8_t        nr_segments;  /* number of segments                   */
-	blkif_vdev_t   handle;       /* only for read/write requests         */
-	uint64_t       id;           /* private guest value, echoed in resp  */
 	union {
 		struct blkif_x86_32_request_rw rw;
 		struct blkif_x86_32_request_discard discard;
 	} u;
-};
+} __attribute__((__packed__));
+
+/* i386 protocol version */
+#pragma pack(push, 4)
 struct blkif_x86_32_response {
 	uint64_t        id;              /* copied from request */
 	uint8_t         operation;       /* copied from request */
 	int16_t         status;          /* BLKIF_RSP_???       */
 };
 #pragma pack(pop)
-
 /* x86_64 protocol version */
 
 struct blkif_x86_64_request_rw {
+	uint8_t        nr_segments;  /* number of segments                   */
+	blkif_vdev_t   handle;       /* only for read/write requests         */
+	uint32_t       _pad1;        /* offsetof(blkif_reqest..,u.rw.id)==8  */
+	uint64_t       id;
 	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
 	struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-};
+} __attribute__((__packed__));
 
 struct blkif_x86_64_request_discard {
+	uint8_t        flag;         /* BLKIF_DISCARD_SECURE or zero         */
+	blkif_vdev_t   _pad1;        /* was "handle" for read/write requests */
+        uint32_t       _pad2;        /* offsetof(blkif_..,u.discard.id)==8   */
+	uint64_t       id;
 	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
 	uint64_t       nr_sectors;
-};
+} __attribute__((__packed__));
 
 struct blkif_x86_64_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
-	uint8_t        nr_segments;  /* number of segments                   */
-	blkif_vdev_t   handle;       /* only for read/write requests         */
-	uint64_t       __attribute__((__aligned__(8))) id;
 	union {
 		struct blkif_x86_64_request_rw rw;
 		struct blkif_x86_64_request_discard discard;
 	} u;
-};
+} __attribute__((__packed__));
+
 struct blkif_x86_64_response {
 	uint64_t       __attribute__((__aligned__(8))) id;
 	uint8_t         operation;       /* copied from request */
@@ -156,6 +164,7 @@ struct xen_vbd {
 	/* Cached size parameter. */
 	sector_t		size;
 	bool			flush_support;
+	bool			discard_secure;
 };
 
 struct backend_info;
@@ -237,22 +246,23 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 {
 	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST;
 	dst->operation = src->operation;
-	dst->nr_segments = src->nr_segments;
-	dst->handle = src->handle;
-	dst->id = src->id;
 	switch (src->operation) {
 	case BLKIF_OP_READ:
 	case BLKIF_OP_WRITE:
 	case BLKIF_OP_WRITE_BARRIER:
 	case BLKIF_OP_FLUSH_DISKCACHE:
+		dst->u.rw.nr_segments = src->u.rw.nr_segments;
+		dst->u.rw.handle = src->u.rw.handle;
+		dst->u.rw.id = src->u.rw.id;
 		dst->u.rw.sector_number = src->u.rw.sector_number;
 		barrier();
-		if (n > dst->nr_segments)
-			n = dst->nr_segments;
+		if (n > dst->u.rw.nr_segments)
+			n = dst->u.rw.nr_segments;
 		for (i = 0; i < n; i++)
 			dst->u.rw.seg[i] = src->u.rw.seg[i];
 		break;
 	case BLKIF_OP_DISCARD:
+		dst->u.discard.flag = src->u.discard.flag;
 		dst->u.discard.sector_number = src->u.discard.sector_number;
 		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
 		break;
@@ -266,22 +276,23 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst,
 {
 	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST;
 	dst->operation = src->operation;
-	dst->nr_segments = src->nr_segments;
-	dst->handle = src->handle;
-	dst->id = src->id;
 	switch (src->operation) {
 	case BLKIF_OP_READ:
 	case BLKIF_OP_WRITE:
 	case BLKIF_OP_WRITE_BARRIER:
 	case BLKIF_OP_FLUSH_DISKCACHE:
+		dst->u.rw.nr_segments = src->u.rw.nr_segments;
+		dst->u.rw.handle = src->u.rw.handle;
+		dst->u.rw.id = src->u.rw.id;
 		dst->u.rw.sector_number = src->u.rw.sector_number;
 		barrier();
-		if (n > dst->nr_segments)
-			n = dst->nr_segments;
+		if (n > dst->u.rw.nr_segments)
+			n = dst->u.rw.nr_segments;
 		for (i = 0; i < n; i++)
 			dst->u.rw.seg[i] = src->u.rw.seg[i];
 		break;
 	case BLKIF_OP_DISCARD:
+		dst->u.discard.flag = src->u.discard.flag;
 		dst->u.discard.sector_number = src->u.discard.sector_number;
 		dst->u.discard.nr_sectors = src->u.discard.nr_sectors;
 		break;

drivers/block/xen-blkback/xenbus.c

@@ -338,6 +338,9 @@ static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle,
 	if (q && q->flush_flags)
 		vbd->flush_support = true;
 
+	if (q && blk_queue_secdiscard(q))
+		vbd->discard_secure = true;
+
 	DPRINTK("Successful creation of handle=%04x (dom=%u)\n",
 		handle, blkif->domid);
 	return 0;
@@ -420,6 +423,15 @@ int xen_blkbk_discard(struct xenbus_transaction xbt, struct backend_info *be)
 				state = 1;
 				blkif->blk_backend_type = BLKIF_BACKEND_PHY;
 			}
+			/* Optional. */
+			err = xenbus_printf(xbt, dev->nodename,
+				"discard-secure", "%d",
+				blkif->vbd.discard_secure);
+			if (err) {
+				xenbus_dev_fatal(dev, err,
+					"writting discard-secure");
+				goto kfree;
+			}
 		}
 	} else {
 		err = PTR_ERR(type);

drivers/block/xen-blkfront.c

@@ -98,7 +98,8 @@ struct blkfront_info
 	unsigned long shadow_free;
 	unsigned int feature_flush;
 	unsigned int flush_op;
-	unsigned int feature_discard;
+	unsigned int feature_discard:1;
+	unsigned int feature_secdiscard:1;
 	unsigned int discard_granularity;
 	unsigned int discard_alignment;
 	int is_ready;
@@ -135,15 +136,15 @@ static int get_id_from_freelist(struct blkfront_info *info)
 {
 	unsigned long free = info->shadow_free;
 	BUG_ON(free >= BLK_RING_SIZE);
-	info->shadow_free = info->shadow[free].req.id;
-	info->shadow[free].req.id = 0x0fffffee; /* debug */
+	info->shadow_free = info->shadow[free].req.u.rw.id;
+	info->shadow[free].req.u.rw.id = 0x0fffffee; /* debug */
 	return free;
 }
 
 static void add_id_to_freelist(struct blkfront_info *info,
 			       unsigned long id)
 {
-	info->shadow[id].req.id  = info->shadow_free;
+	info->shadow[id].req.u.rw.id  = info->shadow_free;
 	info->shadow[id].request = NULL;
 	info->shadow_free = id;
 }
@@ -156,7 +157,7 @@ static int xlbd_reserve_minors(unsigned int minor, unsigned int nr)
 	if (end > nr_minors) {
 		unsigned long *bitmap, *old;
 
-		bitmap = kzalloc(BITS_TO_LONGS(end) * sizeof(*bitmap),
+		bitmap = kcalloc(BITS_TO_LONGS(end), sizeof(*bitmap),
 				 GFP_KERNEL);
 		if (bitmap == NULL)
 			return -ENOMEM;
@@ -287,9 +288,9 @@ static int blkif_queue_request(struct request *req)
 	id = get_id_from_freelist(info);
 	info->shadow[id].request = req;
 
-	ring_req->id = id;
+	ring_req->u.rw.id = id;
 	ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req);
-	ring_req->handle = info->handle;
+	ring_req->u.rw.handle = info->handle;
 
 	ring_req->operation = rq_data_dir(req) ?
 		BLKIF_OP_WRITE : BLKIF_OP_READ;
@@ -305,16 +306,21 @@ static int blkif_queue_request(struct request *req)
 		ring_req->operation = info->flush_op;
 	}
 
-	if (unlikely(req->cmd_flags & REQ_DISCARD)) {
+	if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) {
 		/* id, sector_number and handle are set above. */
 		ring_req->operation = BLKIF_OP_DISCARD;
-		ring_req->nr_segments = 0;
 		ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
+		if ((req->cmd_flags & REQ_SECURE) && info->feature_secdiscard)
+			ring_req->u.discard.flag = BLKIF_DISCARD_SECURE;
+		else
+			ring_req->u.discard.flag = 0;
 	} else {
-		ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
-		BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
+		ring_req->u.rw.nr_segments = blk_rq_map_sg(req->q, req,
+							   info->sg);
+		BUG_ON(ring_req->u.rw.nr_segments >
+		       BLKIF_MAX_SEGMENTS_PER_REQUEST);
 
-		for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
+		for_each_sg(info->sg, sg, ring_req->u.rw.nr_segments, i) {
 			buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg)));
 			fsect = sg->offset >> 9;
 			lsect = fsect + (sg->length >> 9) - 1;
@@ -424,6 +430,8 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 		blk_queue_max_discard_sectors(rq, get_capacity(gd));
 		rq->limits.discard_granularity = info->discard_granularity;
 		rq->limits.discard_alignment = info->discard_alignment;
+		if (info->feature_secdiscard)
+			queue_flag_set_unlocked(QUEUE_FLAG_SECDISCARD, rq);
 	}
 
 	/* Hard sector size and max sectors impersonate the equiv. hardware. */
@@ -705,7 +713,9 @@ static void blkif_free(struct blkfront_info *info, int suspend)
 static void blkif_completion(struct blk_shadow *s)
 {
 	int i;
-	for (i = 0; i < s->req.nr_segments; i++)
+	/* Do not let BLKIF_OP_DISCARD as nr_segment is in the same place
+	 * flag. */
+	for (i = 0; i < s->req.u.rw.nr_segments; i++)
 		gnttab_end_foreign_access(s->req.u.rw.seg[i].gref, 0, 0UL);
 }
 
@@ -736,6 +746,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 		id   = bret->id;
 		req  = info->shadow[id].request;
 
+		if (bret->operation != BLKIF_OP_DISCARD)
 			blkif_completion(&info->shadow[id]);
 
 		add_id_to_freelist(info, id);
@@ -749,7 +760,9 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 					   info->gd->disk_name);
 				error = -EOPNOTSUPP;
 				info->feature_discard = 0;
+				info->feature_secdiscard = 0;
 				queue_flag_clear(QUEUE_FLAG_DISCARD, rq);
+				queue_flag_clear(QUEUE_FLAG_SECDISCARD, rq);
 			}
 			__blk_end_request_all(req, error);
 			break;
@@ -763,7 +776,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 				error = -EOPNOTSUPP;
 			}
 			if (unlikely(bret->status == BLKIF_RSP_ERROR &&
-				     info->shadow[id].req.nr_segments == 0)) {
+				     info->shadow[id].req.u.rw.nr_segments == 0)) {
 				printk(KERN_WARNING "blkfront: %s: empty write %s op failed\n",
 				       info->flush_op == BLKIF_OP_WRITE_BARRIER ?
 				       "barrier" :  "flush disk cache",
@@ -984,8 +997,8 @@ static int blkfront_probe(struct xenbus_device *dev,
 	INIT_WORK(&info->work, blkif_restart_queue);
 
 	for (i = 0; i < BLK_RING_SIZE; i++)
-		info->shadow[i].req.id = i+1;
-	info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
+		info->shadow[i].req.u.rw.id = i+1;
+	info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
 
 	/* Front end dir is a number, which is used as the id. */
 	info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
@@ -1019,9 +1032,9 @@ static int blkif_recover(struct blkfront_info *info)
 	/* Stage 2: Set up free list. */
 	memset(&info->shadow, 0, sizeof(info->shadow));
 	for (i = 0; i < BLK_RING_SIZE; i++)
-		info->shadow[i].req.id = i+1;
+		info->shadow[i].req.u.rw.id = i+1;
 	info->shadow_free = info->ring.req_prod_pvt;
-	info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
+	info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
 
 	/* Stage 3: Find pending requests and requeue them. */
 	for (i = 0; i < BLK_RING_SIZE; i++) {
@@ -1034,17 +1047,19 @@ static int blkif_recover(struct blkfront_info *info)
 		*req = copy[i].req;
 
 		/* We get a new request id, and must reset the shadow state. */
-		req->id = get_id_from_freelist(info);
-		memcpy(&info->shadow[req->id], &copy[i], sizeof(copy[i]));
+		req->u.rw.id = get_id_from_freelist(info);
+		memcpy(&info->shadow[req->u.rw.id], &copy[i], sizeof(copy[i]));
 
+		if (req->operation != BLKIF_OP_DISCARD) {
 		/* Rewrite any grant references invalidated by susp/resume. */
-		for (j = 0; j < req->nr_segments; j++)
+			for (j = 0; j < req->u.rw.nr_segments; j++)
 				gnttab_grant_foreign_access_ref(
 					req->u.rw.seg[j].gref,
 					info->xbdev->otherend_id,
-				pfn_to_mfn(info->shadow[req->id].frame[j]),
-				rq_data_dir(info->shadow[req->id].request));
-		info->shadow[req->id].req = *req;
+					pfn_to_mfn(info->shadow[req->u.rw.id].frame[j]),
+					rq_data_dir(info->shadow[req->u.rw.id].request));
+		}
+		info->shadow[req->u.rw.id].req = *req;
 
 		info->ring.req_prod_pvt++;
 	}
@@ -1135,11 +1150,13 @@ static void blkfront_setup_discard(struct blkfront_info *info)
 	char *type;
 	unsigned int discard_granularity;
 	unsigned int discard_alignment;
+	unsigned int discard_secure;
 
 	type = xenbus_read(XBT_NIL, info->xbdev->otherend, "type", NULL);
 	if (IS_ERR(type))
 		return;
 
+	info->feature_secdiscard = 0;
 	if (strncmp(type, "phy", 3) == 0) {
 		err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
 			"discard-granularity", "%u", &discard_granularity,
@@ -1150,6 +1167,12 @@ static void blkfront_setup_discard(struct blkfront_info *info)
 			info->discard_granularity = discard_granularity;
 			info->discard_alignment = discard_alignment;
 		}
+		err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
+			    "discard-secure", "%d", &discard_secure,
+			    NULL);
+		if (!err)
+			info->feature_secdiscard = discard_secure;
+
 	} else if (strncmp(type, "file", 4) == 0)
 		info->feature_discard = 1;
 

include/xen/interface/io/blkif.h

@@ -84,6 +84,21 @@ typedef uint64_t blkif_sector_t;
  *     e07154r6-Data_Set_Management_Proposal_for_ATA-ACS2.doc
  * http://www.seagate.com/staticfiles/support/disc/manuals/
  *     Interface%20manuals/100293068c.pdf
+ * The backend can optionally provide three extra XenBus attributes to
+ * further optimize the discard functionality:
+ * 'discard-aligment' - Devices that support discard functionality may
+ * internally allocate space in units that are bigger than the exported
+ * logical block size. The discard-alignment parameter indicates how many bytes
+ * the beginning of the partition is offset from the internal allocation unit's
+ * natural alignment.
+ * 'discard-granularity'  - Devices that support discard functionality may
+ * internally allocate space using units that are bigger than the logical block
+ * size. The discard-granularity parameter indicates the size of the internal
+ * allocation unit in bytes if reported by the device. Otherwise the
+ * discard-granularity will be set to match the device's physical block size.
+ * 'discard-secure' - All copies of the discarded sectors (potentially created
+ * by garbage collection) must also be erased.  To use this feature, the flag
+ * BLKIF_DISCARD_SECURE must be set in the blkif_request_trim.
  */
 #define BLKIF_OP_DISCARD           5
 
@@ -95,6 +110,12 @@ typedef uint64_t blkif_sector_t;
 #define BLKIF_MAX_SEGMENTS_PER_REQUEST 11
 
 struct blkif_request_rw {
+	uint8_t        nr_segments;  /* number of segments                   */
+	blkif_vdev_t   handle;       /* only for read/write requests         */
+#ifdef CONFIG_X86_64
+	uint32_t       _pad1;	     /* offsetof(blkif_request,u.rw.id) == 8 */
+#endif
+	uint64_t       id;           /* private guest value, echoed in resp  */
 	blkif_sector_t sector_number;/* start sector idx on disk (r/w only)  */
 	struct blkif_request_segment {
 		grant_ref_t gref;        /* reference to I/O buffer frame        */
@@ -102,23 +123,28 @@ struct blkif_request_rw {
 		/* @last_sect: last sector in frame to transfer (inclusive).     */
 		uint8_t     first_sect, last_sect;
 	} seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
-};
+} __attribute__((__packed__));
 
 struct blkif_request_discard {
+	uint8_t        flag;         /* BLKIF_DISCARD_SECURE or zero.        */
+#define BLKIF_DISCARD_SECURE (1<<0)  /* ignored if discard-secure=0          */
+	blkif_vdev_t   _pad1;        /* only for read/write requests         */
+#ifdef CONFIG_X86_64
+	uint32_t       _pad2;        /* offsetof(blkif_req..,u.discard.id)==8*/
+#endif
+	uint64_t       id;           /* private guest value, echoed in resp  */
 	blkif_sector_t sector_number;
 	uint64_t       nr_sectors;
-};
+	uint8_t        _pad3;
+} __attribute__((__packed__));
 
 struct blkif_request {
 	uint8_t        operation;    /* BLKIF_OP_???                         */
-	uint8_t        nr_segments;  /* number of segments                   */
-	blkif_vdev_t   handle;       /* only for read/write requests         */
-	uint64_t       id;           /* private guest value, echoed in resp  */
 	union {
 		struct blkif_request_rw rw;
 		struct blkif_request_discard discard;
 	} u;
-};
+} __attribute__((__packed__));
 
 struct blkif_response {
 	uint64_t        id;              /* copied from request */