[PATCH] scsi_debug: Thin provisioning support

[PATCH] scsi_debug: Thin provisioning support

[Martin K. Petersen]

This version fixes 64-bit modulo on 32-bit as well as inadvertent map
updates when TP was disabled.



Implement support for thin provisioning in scsi_debug.  No actual memory
de-allocation is taking place.  The intent is to emulate a thinly
provisioned storage device, not to be one.

There are four new module options:

 - unmap_granularity specifies the granularity at which to track mapped
   blocks (specified in number of logical blocks).  2048 (1 MB) is a
   realistic value for disk arrays although some may have a finer
   granularity.

 - unmap_alignment specifies the first LBA which is naturally aligned on
   an unmap_granularity boundary.

 - unmap_max_desc specifies the maximum number of ranges that can be
   unmapped using one UNMAP command.  If this is 0, only WRITE SAME is
   supported and UNMAP will cause a check condition.

 - unmap_max_blocks specifies the maximum number of blocks that can be
   unmapped using a single UNMAP command.  Default is 0xffffffff.

These parameters are reported in the new and extended block limits VPD.

If unmap_granularity is specified the device is tagged as thin
provisioning capable in READ CAPACITY(16).  A bitmap is allocated to
track whether blocks are mapped or not.  A WRITE request will cause a
block to be mapped.  So will WRITE SAME unless the UNMAP bit is set.

Blocks can be unmapped using either WRITE SAME or UNMAP.  No accounting
is done to track partial blocks.  This means that only whole blocks will
be marked free.  This is how the array people tell me their firmwares
work.

GET LBA STATUS is also supported.  This command reports whether a block
is mapped or not, and how long the adjoining mapped/unmapped extent is.

The block allocation bitmap can also be viewed from user space via:

	/sys/bus/pseudo/drivers/scsi_debug/map

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

---

diff --git a/drivers/scsi/scsi_debug.c b/drivers/scsi/scsi_debug.c
index c4103be..cb4bf16 100644
--- a/drivers/scsi/scsi_debug.c
+++ b/drivers/scsi/scsi_debug.c
@@ -44,6 +44,8 @@
 
 #include <net/checksum.h>
 
+#include <asm/unaligned.h>
+
 #include <scsi/scsi.h>
 #include <scsi/scsi_cmnd.h>
 #include <scsi/scsi_device.h>
@@ -105,6 +107,10 @@ static const char * scsi_debug_version_date = "20070104";
 #define DEF_ATO 1
 #define DEF_PHYSBLK_EXP 0
 #define DEF_LOWEST_ALIGNED 0
+#define DEF_UNMAP_MAX_BLOCKS 0
+#define DEF_UNMAP_MAX_DESC 0
+#define DEF_UNMAP_GRANULARITY 0
+#define DEF_UNMAP_ALIGNMENT 0
 
 /* bit mask values for scsi_debug_opts */
 #define SCSI_DEBUG_OPT_NOISE   1
@@ -162,6 +168,10 @@ static int scsi_debug_guard = DEF_GUARD;
 static int scsi_debug_ato = DEF_ATO;
 static int scsi_debug_physblk_exp = DEF_PHYSBLK_EXP;
 static int scsi_debug_lowest_aligned = DEF_LOWEST_ALIGNED;
+static int scsi_debug_unmap_max_desc = DEF_UNMAP_MAX_DESC;
+static int scsi_debug_unmap_max_blocks = DEF_UNMAP_MAX_BLOCKS;
+static int scsi_debug_unmap_granularity = DEF_UNMAP_GRANULARITY;
+static int scsi_debug_unmap_alignment = DEF_UNMAP_ALIGNMENT;
 
 static int scsi_debug_cmnd_count = 0;
 
@@ -223,7 +233,9 @@ static struct sdebug_queued_cmd queued_arr[SCSI_DEBUG_CANQUEUE];
 
 static unsigned char * fake_storep;	/* ramdisk storage */
 static unsigned char *dif_storep;	/* protection info */
+static void *map_storep;		/* provisioning map */
 
+static unsigned long map_size;
 static int num_aborts = 0;
 static int num_dev_resets = 0;
 static int num_bus_resets = 0;
@@ -317,6 +329,7 @@ static void get_data_transfer_info(unsigned char *cmd,
 			(u32)cmd[28] << 24;
 		break;
 
+	case WRITE_SAME_16:
 	case WRITE_16:
 	case READ_16:
 		*lba = (u64)cmd[9] | (u64)cmd[8] << 8 |
@@ -335,6 +348,7 @@ static void get_data_transfer_info(unsigned char *cmd,
 		*num = (u32)cmd[9] | (u32)cmd[8] << 8 | (u32)cmd[7] << 16 |
 			(u32)cmd[6] << 24;
 		break;
+	case WRITE_SAME:
 	case WRITE_10:
 	case READ_10:
 	case XDWRITEREAD_10:
@@ -691,6 +705,29 @@ static int inquiry_evpd_b0(unsigned char * arr)
 		arr[6] = (sdebug_store_sectors >> 8) & 0xff;
 		arr[7] = sdebug_store_sectors & 0xff;
 	}
+
+	if (scsi_debug_unmap_max_desc) {
+		unsigned int blocks;
+
+		if (scsi_debug_unmap_max_blocks)
+			blocks = scsi_debug_unmap_max_blocks;
+		else
+			blocks = 0xffffffff;
+
+		put_unaligned_be32(blocks, &arr[16]);
+		put_unaligned_be32(scsi_debug_unmap_max_desc, &arr[20]);
+	}
+
+	if (scsi_debug_unmap_alignment) {
+		put_unaligned_be32(scsi_debug_unmap_alignment, &arr[28]);
+		arr[28] |= 0x80; /* UGAVALID */
+	}
+
+	if (scsi_debug_unmap_granularity) {
+		put_unaligned_be32(scsi_debug_unmap_granularity, &arr[24]);
+		return 0x3c; /* Mandatory page length for thin provisioning */
+	}
+
 	return sizeof(vpdb0_data);
 }
 
@@ -974,6 +1011,10 @@ static int resp_readcap16(struct scsi_cmnd * scp,
 	arr[11] = scsi_debug_sector_size & 0xff;
 	arr[13] = scsi_debug_physblk_exp & 0xf;
 	arr[14] = (scsi_debug_lowest_aligned >> 8) & 0x3f;
+
+	if (scsi_debug_unmap_granularity)
+		arr[14] |= 0x80; /* TPE */
+
 	arr[15] = scsi_debug_lowest_aligned & 0xff;
 
 	if (scsi_debug_dif) {
@@ -1887,6 +1928,70 @@ out:
 	return ret;
 }
 
+static unsigned int map_state(sector_t lba, unsigned int *num)
+{
+	unsigned int granularity, alignment, mapped;
+	sector_t block, next, end;
+
+	granularity = scsi_debug_unmap_granularity;
+	alignment = granularity - scsi_debug_unmap_alignment;
+	block = lba + alignment;
+	do_div(block, granularity);
+
+	mapped = test_bit(block, map_storep);
+
+	if (mapped)
+		next = find_next_zero_bit(map_storep, map_size, block);
+	else
+		next = find_next_bit(map_storep, map_size, block);
+
+	end = next * granularity - scsi_debug_unmap_alignment;
+	*num = end - lba;
+
+	return mapped;
+}
+
+static void map_region(sector_t lba, unsigned int len)
+{
+	unsigned int granularity, alignment;
+	sector_t end = lba + len;
+
+	granularity = scsi_debug_unmap_granularity;
+	alignment = granularity - scsi_debug_unmap_alignment;
+
+	while (lba < end) {
+		sector_t block, rem;
+
+		block = lba + alignment;
+		rem = do_div(block, granularity);
+
+		set_bit(block, map_storep);
+
+		lba += granularity - rem;
+	}
+}
+
+static void unmap_region(sector_t lba, unsigned int len)
+{
+	unsigned int granularity, alignment;
+	sector_t end = lba + len;
+
+	granularity = scsi_debug_unmap_granularity;
+	alignment = granularity - scsi_debug_unmap_alignment;
+
+	while (lba < end) {
+		sector_t block, rem;
+
+		block = lba + alignment;
+		rem = do_div(block, granularity);
+
+		if (rem == 0 && lba + granularity <= end)
+			clear_bit(block, map_storep);
+
+		lba += granularity - rem;
+	}
+}
+
 static int resp_write(struct scsi_cmnd *SCpnt, unsigned long long lba,
 		      unsigned int num, struct sdebug_dev_info *devip,
 		      u32 ei_lba)
@@ -1910,6 +2015,8 @@ static int resp_write(struct scsi_cmnd *SCpnt, unsigned long long lba,
 
 	write_lock_irqsave(&atomic_rw, iflags);
 	ret = do_device_access(SCpnt, devip, lba, num, 1);
+	if (scsi_debug_unmap_granularity)
+		map_region(lba, num);
 	write_unlock_irqrestore(&atomic_rw, iflags);
 	if (-1 == ret)
 		return (DID_ERROR << 16);
@@ -1917,9 +2024,143 @@ static int resp_write(struct scsi_cmnd *SCpnt, unsigned long long lba,
 		 (SCSI_DEBUG_OPT_NOISE & scsi_debug_opts))
 		printk(KERN_INFO "scsi_debug: write: cdb indicated=%u, "
 		       " IO sent=%d bytes\n", num * scsi_debug_sector_size, ret);
+
+	return 0;
+}
+
+static int resp_write_same(struct scsi_cmnd *scmd, unsigned long long lba,
+		      unsigned int num, struct sdebug_dev_info *devip,
+			   u32 ei_lba, unsigned int unmap)
+{
+	unsigned long iflags;
+	unsigned long long i;
+	int ret;
+
+	ret = check_device_access_params(devip, lba, num);
+	if (ret)
+		return ret;
+
+	write_lock_irqsave(&atomic_rw, iflags);
+
+	if (unmap && scsi_debug_unmap_granularity) {
+		unmap_region(lba, num);
+		goto out;
+	}
+
+	/* Else fetch one logical block */
+	ret = fetch_to_dev_buffer(scmd,
+				  fake_storep + (lba * scsi_debug_sector_size),
+				  scsi_debug_sector_size);
+
+	if (-1 == ret) {
+		write_unlock_irqrestore(&atomic_rw, iflags);
+		return (DID_ERROR << 16);
+	} else if ((ret < (num * scsi_debug_sector_size)) &&
+		 (SCSI_DEBUG_OPT_NOISE & scsi_debug_opts))
+		printk(KERN_INFO "scsi_debug: write same: cdb indicated=%u, "
+		       " IO sent=%d bytes\n", num * scsi_debug_sector_size, ret);
+
+	/* Copy first sector to remaining blocks */
+	for (i = 1 ; i < num ; i++)
+		memcpy(fake_storep + ((lba + i) * scsi_debug_sector_size),
+		       fake_storep + (lba * scsi_debug_sector_size),
+		       scsi_debug_sector_size);
+
+	if (scsi_debug_unmap_granularity)
+		map_region(lba, num);
+out:
+	write_unlock_irqrestore(&atomic_rw, iflags);
+
 	return 0;
 }
 
+struct unmap_block_desc {
+	__be64	lba;
+	__be32	blocks;
+	__be32	__reserved;
+};
+
+static int resp_unmap(struct scsi_cmnd * scmd, struct sdebug_dev_info * devip)
+{
+	unsigned char *buf;
+	struct unmap_block_desc *desc;
+	unsigned int i, payload_len, descriptors;
+	int ret;
+
+	ret = check_readiness(scmd, 1, devip);
+	if (ret)
+		return ret;
+
+	payload_len = get_unaligned_be16(&scmd->cmnd[7]);
+	BUG_ON(scsi_bufflen(scmd) != payload_len);
+
+	descriptors = (payload_len - 8) / 16;
+
+	buf = kmalloc(scsi_bufflen(scmd), GFP_ATOMIC);
+	if (!buf)
+		return check_condition_result;
+
+	scsi_sg_copy_to_buffer(scmd, buf, scsi_bufflen(scmd));
+
+	BUG_ON(get_unaligned_be16(&buf[0]) != payload_len - 2);
+	BUG_ON(get_unaligned_be16(&buf[2]) != descriptors * 16);
+
+	desc = (void *)&buf[8];
+
+	for (i = 0 ; i < descriptors ; i++) {
+		unsigned long long lba = get_unaligned_be64(&desc[i].lba);
+		unsigned int num = get_unaligned_be32(&desc[i].blocks);
+
+		ret = check_device_access_params(devip, lba, num);
+		if (ret)
+			goto out;
+
+		unmap_region(lba, num);
+	}
+
+	ret = 0;
+
+out:
+	kfree(buf);
+
+	return ret;
+}
+
+#define SDEBUG_GET_LBA_STATUS_LEN 32
+
+static int resp_get_lba_status(struct scsi_cmnd * scmd,
+			       struct sdebug_dev_info * devip)
+{
+	unsigned long long lba;
+	unsigned int alloc_len, mapped, num;
+	unsigned char arr[SDEBUG_GET_LBA_STATUS_LEN];
+	int ret;
+
+	ret = check_readiness(scmd, 1, devip);
+	if (ret)
+		return ret;
+
+	lba = get_unaligned_be64(&scmd->cmnd[2]);
+	alloc_len = get_unaligned_be32(&scmd->cmnd[10]);
+
+	if (alloc_len < 24)
+		return 0;
+
+	ret = check_device_access_params(devip, lba, 1);
+	if (ret)
+		return ret;
+
+	mapped = map_state(lba, &num);
+
+	memset(arr, 0, SDEBUG_GET_LBA_STATUS_LEN);
+	put_unaligned_be32(16, &arr[0]);	/* Parameter Data Length */
+	put_unaligned_be64(lba, &arr[8]);	/* LBA */
+	put_unaligned_be32(num, &arr[16]);	/* Number of blocks */
+	arr[20] = !mapped;			/* mapped = 0, unmapped = 1 */
+
+	return fill_from_dev_buffer(scmd, arr, SDEBUG_GET_LBA_STATUS_LEN);
+}
+
 #define SDEBUG_RLUN_ARR_SZ 256
 
 static int resp_report_luns(struct scsi_cmnd * scp,
@@ -2430,6 +2671,10 @@ module_param_named(guard, scsi_debug_guard, int, S_IRUGO);
 module_param_named(ato, scsi_debug_ato, int, S_IRUGO);
 module_param_named(physblk_exp, scsi_debug_physblk_exp, int, S_IRUGO);
 module_param_named(lowest_aligned, scsi_debug_lowest_aligned, int, S_IRUGO);
+module_param_named(unmap_max_blocks, scsi_debug_unmap_max_blocks, int, S_IRUGO);
+module_param_named(unmap_max_desc, scsi_debug_unmap_max_desc, int, S_IRUGO);
+module_param_named(unmap_granularity, scsi_debug_unmap_granularity, int, S_IRUGO);
+module_param_named(unmap_alignment, scsi_debug_unmap_alignment, int, S_IRUGO);
 
 MODULE_AUTHOR("Eric Youngdale + Douglas Gilbert");
 MODULE_DESCRIPTION("SCSI debug adapter driver");
@@ -2458,6 +2703,10 @@ MODULE_PARM_DESC(dix, "data integrity extensions mask (def=0)");
 MODULE_PARM_DESC(dif, "data integrity field type: 0-3 (def=0)");
 MODULE_PARM_DESC(guard, "protection checksum: 0=crc, 1=ip (def=0)");
 MODULE_PARM_DESC(ato, "application tag ownership: 0=disk 1=host (def=1)");
+MODULE_PARM_DESC(unmap_max_blocks, "max # of blocks can be unmapped in one cmd (def=0)");
+MODULE_PARM_DESC(unmap_max_desc, "max # of ranges that can be unmapped in one cmd (def=0)");
+MODULE_PARM_DESC(unmap_granularity, "thin provisioning granularity in blocks (def=0)");
+MODULE_PARM_DESC(unmap_alignment, "lowest aligned thin provisioning lba (def=0)");
 
 static char sdebug_info[256];
 
@@ -2816,6 +3065,23 @@ static ssize_t sdebug_ato_show(struct device_driver *ddp, char *buf)
 }
 DRIVER_ATTR(ato, S_IRUGO, sdebug_ato_show, NULL);
 
+static ssize_t sdebug_map_show(struct device_driver *ddp, char *buf)
+{
+	ssize_t count;
+
+	if (scsi_debug_unmap_granularity == 0)
+		return scnprintf(buf, PAGE_SIZE, "0-%u\n",
+				 sdebug_store_sectors);
+
+	count = bitmap_scnlistprintf(buf, PAGE_SIZE, map_storep, map_size);
+
+	buf[count++] = '\n';
+	buf[count++] = 0;
+
+	return count;
+}
+DRIVER_ATTR(map, S_IRUGO, sdebug_map_show, NULL);
+
 
 /* Note: The following function creates attribute files in the
    /sys/bus/pseudo/drivers/scsi_debug directory. The advantage of these
@@ -2847,11 +3113,13 @@ static int do_create_driverfs_files(void)
 	ret |= driver_create_file(&sdebug_driverfs_driver, &driver_attr_dif);
 	ret |= driver_create_file(&sdebug_driverfs_driver, &driver_attr_guard);
 	ret |= driver_create_file(&sdebug_driverfs_driver, &driver_attr_ato);
+	ret |= driver_create_file(&sdebug_driverfs_driver, &driver_attr_map);
 	return ret;
 }
 
 static void do_remove_driverfs_files(void)
 {
+	driver_remove_file(&sdebug_driverfs_driver, &driver_attr_map);
 	driver_remove_file(&sdebug_driverfs_driver, &driver_attr_ato);
 	driver_remove_file(&sdebug_driverfs_driver, &driver_attr_guard);
 	driver_remove_file(&sdebug_driverfs_driver, &driver_attr_dif);
@@ -2989,6 +3257,36 @@ static int __init scsi_debug_init(void)
 		memset(dif_storep, 0xff, dif_size);
 	}
 
+	if (scsi_debug_unmap_granularity) {
+		unsigned int map_bytes;
+
+		if (scsi_debug_unmap_granularity < scsi_debug_unmap_alignment) {
+			printk(KERN_ERR
+			       "%s: ERR: unmap_granularity < unmap_alignment\n",
+			       __func__);
+			return -EINVAL;
+		}
+
+		map_size = (sdebug_store_sectors / scsi_debug_unmap_granularity);
+		map_bytes = map_size >> 3;
+		map_storep = vmalloc(map_bytes);
+
+		printk(KERN_INFO "scsi_debug_init: %lu provisioning blocks\n",
+		       map_size);
+
+		if (map_storep == NULL) {
+			printk(KERN_ERR "scsi_debug_init: out of mem. (MAP)\n");
+			ret = -ENOMEM;
+			goto free_vm;
+		}
+
+		memset(map_storep, 0x0, map_bytes);
+
+		/* Map first 1KB for partition table */
+		if (scsi_debug_num_parts)
+			map_region(0, 2);
+	}
+
 	ret = device_register(&pseudo_primary);
 	if (ret < 0) {
 		printk(KERN_WARNING "scsi_debug: device_register error: %d\n",
@@ -3041,6 +3339,8 @@ bus_unreg:
 dev_unreg:
 	device_unregister(&pseudo_primary);
 free_vm:
+	if (map_storep)
+		vfree(map_storep);
 	if (dif_storep)
 		vfree(dif_storep);
 	vfree(fake_storep);
@@ -3167,6 +3467,7 @@ int scsi_debug_queuecommand(struct scsi_cmnd *SCpnt, done_funct_t done)
 	int inj_dif = 0;
 	int inj_dix = 0;
 	int delay_override = 0;
+	int unmap = 0;
 
 	scsi_set_resid(SCpnt, 0);
 	if ((SCSI_DEBUG_OPT_NOISE & scsi_debug_opts) && cmd) {
@@ -3272,13 +3573,21 @@ int scsi_debug_queuecommand(struct scsi_cmnd *SCpnt, done_funct_t done)
 		errsts = resp_readcap(SCpnt, devip);
 		break;
 	case SERVICE_ACTION_IN:
-		if (SAI_READ_CAPACITY_16 != cmd[1]) {
+		if (cmd[1] == SAI_READ_CAPACITY_16)
+			errsts = resp_readcap16(SCpnt, devip);
+		else if (cmd[1] == SAI_GET_LBA_STATUS) {
+
+			if (scsi_debug_unmap_max_desc == 0) {
+				mk_sense_buffer(devip, ILLEGAL_REQUEST,
+						INVALID_COMMAND_OPCODE, 0);
+				errsts = check_condition_result;
+			} else
+				errsts = resp_get_lba_status(SCpnt, devip);
+		} else {
 			mk_sense_buffer(devip, ILLEGAL_REQUEST,
 					INVALID_OPCODE, 0);
 			errsts = check_condition_result;
-			break;
 		}
-		errsts = resp_readcap16(SCpnt, devip);
 		break;
 	case MAINTENANCE_IN:
 		if (MI_REPORT_TARGET_PGS != cmd[1]) {
@@ -3378,6 +3687,29 @@ write:
 			errsts = illegal_condition_result;
 		}
 		break;
+	case WRITE_SAME_16:
+		if (cmd[1] & 0x8)
+			unmap = 1;
+		/* fall through */
+	case WRITE_SAME:
+		errsts = check_readiness(SCpnt, 0, devip);
+		if (errsts)
+			break;
+		get_data_transfer_info(cmd, &lba, &num, &ei_lba);
+		errsts = resp_write_same(SCpnt, lba, num, devip, ei_lba, unmap);
+		break;
+	case UNMAP:
+		errsts = check_readiness(SCpnt, 0, devip);
+		if (errsts)
+			break;
+
+		if (scsi_debug_unmap_max_desc == 0) {
+			mk_sense_buffer(devip, ILLEGAL_REQUEST,
+					INVALID_COMMAND_OPCODE, 0);
+			errsts = check_condition_result;
+		} else
+			errsts = resp_unmap(SCpnt, devip);
+		break;
 	case MODE_SENSE:
 	case MODE_SENSE_10:
 		errsts = resp_mode_sense(SCpnt, target, devip);

Re: [PATCH] scsi_debug: Thin provisioning support

[Douglas Gilbert]

[-- Attachment #1: Type: text/plain, Size: 2384 bytes --]

Martin K. Petersen wrote:
> This version fixes 64-bit modulo on 32-bit as well as inadvertent map
> updates when TP was disabled.
> 
> 
> 
> Implement support for thin provisioning in scsi_debug.  No actual memory
> de-allocation is taking place.  The intent is to emulate a thinly
> provisioned storage device, not to be one.
> 
> There are four new module options:
> 
>  - unmap_granularity specifies the granularity at which to track mapped
>    blocks (specified in number of logical blocks).  2048 (1 MB) is a
>    realistic value for disk arrays although some may have a finer
>    granularity.
> 
>  - unmap_alignment specifies the first LBA which is naturally aligned on
>    an unmap_granularity boundary.
> 
>  - unmap_max_desc specifies the maximum number of ranges that can be
>    unmapped using one UNMAP command.  If this is 0, only WRITE SAME is
>    supported and UNMAP will cause a check condition.
> 
>  - unmap_max_blocks specifies the maximum number of blocks that can be
>    unmapped using a single UNMAP command.  Default is 0xffffffff.
> 
> These parameters are reported in the new and extended block limits VPD.
> 
> If unmap_granularity is specified the device is tagged as thin
> provisioning capable in READ CAPACITY(16).  A bitmap is allocated to
> track whether blocks are mapped or not.  A WRITE request will cause a
> block to be mapped.  So will WRITE SAME unless the UNMAP bit is set.
> 
> Blocks can be unmapped using either WRITE SAME or UNMAP.  No accounting
> is done to track partial blocks.  This means that only whole blocks will
> be marked free.  This is how the array people tell me their firmwares
> work.
> 
> GET LBA STATUS is also supported.  This command reports whether a block
> is mapped or not, and how long the adjoining mapped/unmapped extent is.
> 
> The block allocation bitmap can also be viewed from user space via:
> 
> 	/sys/bus/pseudo/drivers/scsi_debug/map
> 
> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Douglas Gilbert <dgilbert@interlog.com>

While testing scsi_debug with these patches I found a
problem with the Block Limits VPD page function. The
length returned by the inquiry_evpd_b0() function was
too short. A patch to fix that and a cosmetic change
(that the form factor of scsi_debug is less than 1.8
inches) is attached.

Signed-of-by: Douglas Gilbert <dgilbert@interlog.com>


[-- Attachment #2: sdebug2631dg1.patch --]
[-- Type: text/x-patch, Size: 864 bytes --]

--- linux/drivers/scsi/scsi_debug.c2631mp2	2009-10-14 11:41:17.000000000 -0400
+++ linux/drivers/scsi/scsi_debug.c	2009-10-15 14:52:08.000000000 -0400
@@ -685,10 +685,12 @@
 }
 
 
+/* Block limits VPD page (SBC-3) */
 static unsigned char vpdb0_data[] = {
-	/* from 4th byte */ 0,0,0,4,
-	0,0,0x4,0,
-	0,0,0,64,
+	/* from 4th byte */ 0,0,0,4, 0,0,0x4,0, 0,0,0,64,
+	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+	0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
 };
 
 static int inquiry_evpd_b0(unsigned char * arr)
@@ -731,11 +733,14 @@
 	return sizeof(vpdb0_data);
 }
 
+/* Block device characteristics VPD page (SBC-3) */
 static int inquiry_evpd_b1(unsigned char *arr)
 {
 	memset(arr, 0, 0x3c);
 	arr[0] = 0;
-	arr[1] = 1;
+	arr[1] = 1;	/* non rotating medium (e.g. solid state) */
+	arr[2] = 0;
+	arr[3] = 5;	/* less than 1.8" */
 
 	return 0x3c;
 }

Re: [PATCH] scsi_debug: Thin provisioning support

[Christoph Hellwig]

On Thu, Oct 15, 2009 at 02:45:27PM -0400, Martin K. Petersen wrote:
> 
> This version fixes 64-bit modulo on 32-bit as well as inadvertent map
> updates when TP was disabled.


Just started testing this my fs discard support and I noticed that you
track the mapping state globally.  Given that scsi_debug supports
multiple devices that seems rather suboptimal.

Re: [PATCH] scsi_debug: Thin provisioning support

[Martin K. Petersen]

>>>>> "Christoph" == Christoph Hellwig <hch@infradead.org> writes:

Christoph> Just started testing this my fs discard support and I noticed
Christoph> that you track the mapping state globally.  Given that
Christoph> scsi_debug supports multiple devices that seems rather
Christoph> suboptimal.

Well, that's how scsi_debug works -- the devices all use the same
buffer.  Having a per-device backing store is on my todo list.

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH] scsi_debug: Thin provisioning support

[Christoph Hellwig]

And a crash when using unmap_granularity=1 when using the discard-enable
mkfs.xfs binary from xfsprogs-dev.git:

qemu1:~/xfsprogs-dev# ./mkfs/mkfs.xfs -f /dev/sdb
[   53.562041] BUG: unable to handle kernel paging request at fcc80000
[   53.563812] IP: [<c070fbb7>] unmap_region+0xc7/0x120
[   53.564091] *pde = 36fd0067 *pte = 00000000 
[   53.564091] Oops: 0002 [#1] SMP 
[   53.564091] last sysfs file: /sys/block/sr0/dev
[   53.564091] Modules linked in:
[   53.564091] 
[   53.564091] Pid: 184, comm: kblockd/0 Not tainted (2.6.32-rc5-xfs
#414) 
[   53.564091] EIP: 0060:[<c070fbb7>] EFLAGS: 00010046 CPU: 0
[   53.564091] EIP is at unmap_region+0xc7/0x120
[   53.564091] EAX: 00020000 EBX: 0001ffff ECX: 00020000 EDX: fcc7c000
[   53.564091] ESI: 00000000 EDI: 00000001 EBP: f7959ca4 ESP: f7959c54
[   53.564091]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[   53.564091] Process kblockd/0 (pid: 184, ti=f7958000 task=f7956f40
task.ti=f7958000)
[   53.564091] Stack:
[   53.564091]  00020000 00000000 00000000 00000000 00000000 00000000
00020000 00000000
[   53.564091] <0> 00020000 00000000 fcc7c000 00000001 00000000 00020000
00000000 00000001
[   53.564091] <0> 00000000 00000001 00000000 00000000 f7959eb0 c0715156
00020000 f792c200
[   53.564091] Call Trace:
[   53.564091]  [<c0715156>] ? scsi_debug_queuecommand+0x1786/0x19b0
[   53.564091]  [<c0131f58>] ? sched_clock+0x8/0x10
[   53.564091]  [<c0182412>] ? sched_clock_local+0xd2/0x170
[   53.564091]  [<c06e8c80>] ? scsi_done+0x0/0x10
[   53.564091]  [<c0182412>] ? sched_clock_local+0xd2/0x170
[   53.564091]  [<c01825d8>] ? sched_clock_cpu+0x128/0x170
[   53.564091]  [<c0191023>] ? __lock_acquire+0x2e3/0x1270
[   53.564091]  [<c0191023>] ? __lock_acquire+0x2e3/0x1270
[   53.564091]  [<c06e8e44>] ? scsi_dispatch_cmd+0xa4/0x250
[   53.564091]  [<c0893af6>] ? _spin_lock_irqsave+0x76/0x90
[   53.564091]  [<c06e8e89>] ? scsi_dispatch_cmd+0xe9/0x250
[   53.564091]  [<c06eed5e>] ? scsi_request_fn+0x38e/0x4d0
[   53.564091]  [<c0651739>] ? __generic_unplug_device+0x29/0x30
[   53.564091]  [<c0651761>] ? generic_unplug_device+0x21/0x30
[   53.564091]  [<c06499a5>] ? blk_unplug_work+0x35/0x40
[   53.564091]  [<c0178785>] ? worker_thread+0x195/0x290
[   53.564091]  [<c0178720>] ? worker_thread+0x130/0x290
[   53.564091]  [<c0649970>] ? blk_unplug_work+0x0/0x40
[   53.564091]  [<c017c810>] ? autoremove_wake_function+0x0/0x50
[   53.564091]  [<c01785f0>] ? worker_thread+0x0/0x290
[   53.564091]  [<c017c504>] ? kthread+0x74/0x80
[   53.564091]  [<c017c490>] ? kthread+0x0/0x80
[   53.564091]  [<c012d4c7>] ? kernel_thread_helper+0x7/0x10

Re: [PATCH] scsi_debug: Thin provisioning support

[Martin K. Petersen]

>>>>> "Christoph" == Christoph Hellwig <hch@infradead.org> writes:

Christoph> And a crash when using unmap_granularity=1 when using the
Christoph> discard-enable mkfs.xfs binary from xfsprogs-dev.git:

I was unable to reproduce.  I also tried with xfs_trim.  Both work fine.
I guess I'll try with your two latest patches in the mix next.

i386, x86_64, ppc?


Christoph> qemu1:~/xfsprogs-dev# ./mkfs/mkfs.xfs -f /dev/sdb [
Christoph> 53.562041] BUG: unable to handle kernel paging request at
Christoph> fcc80000 [ 53.563812] IP: [<c070fbb7>]
Christoph> unmap_region+0xc7/0x120
           ^^^^^^^^^^^^^^^^^

That's pretty odd.  It's way beyond the end of unmap_region here.

-- 
Martin K. Petersen	Oracle Linux Engineering