[PATCH v2 1/4] Enable support for Seagate HostAware drives

From: Shaun Tancheff
Date: Mon Aug 22 2016 - 00:32:12 EST

Seagate drives report a SAME code of 0 due to having:
- Zones of different types (CMR zones at the low LBA space).
- Zones of different size (A terminating 'runt' zone in the high
lba space).

Support loading the zone topology into the zone cache.

Signed-off-by: Shaun Tancheff <shaun.tancheff@xxxxxxxxxxx>
---
drivers/scsi/sd.c | 22 +++---
drivers/scsi/sd.h | 20 ++++--
drivers/scsi/sd_zbc.c | 183 +++++++++++++++++++++++++++++++++++--------------
include/linux/blkdev.h | 16 +++--
4 files changed, 170 insertions(+), 71 deletions(-)

diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 059a57f..7903e21 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -693,8 +693,13 @@ static void sd_config_discard(struct scsi_disk *sdkp, unsigned int mode)
break;

case SD_ZBC_RESET_WP:
- max_blocks = sdkp->unmap_granularity;
q->limits.discard_zeroes_data = 1;
+ q->limits.discard_granularity =
+ sd_zbc_discard_granularity(sdkp);
+
+ max_blocks = min_not_zero(sdkp->unmap_granularity,
+ q->limits.discard_granularity >>
+ ilog2(logical_block_size));
break;

case SD_LBP_ZERO:
@@ -1955,13 +1960,12 @@ static int sd_done(struct scsi_cmnd *SCpnt)
good_bytes = blk_rq_bytes(req);
scsi_set_resid(SCpnt, 0);
} else {
-#ifdef CONFIG_SCSI_ZBC
if (op == ZBC_OUT)
/* RESET WRITE POINTER failed */
sd_zbc_update_zones(sdkp,
blk_rq_pos(req),
- 512, true);
-#endif
+ 512, SD_ZBC_RESET_WP_ERR);
+
good_bytes = 0;
scsi_set_resid(SCpnt, blk_rq_bytes(req));
}
@@ -2034,7 +2038,6 @@ static int sd_done(struct scsi_cmnd *SCpnt)
good_bytes = blk_rq_bytes(req);
scsi_set_resid(SCpnt, 0);
}
-#ifdef CONFIG_SCSI_ZBC
/*
* ZBC: Unaligned write command.
* Write did not start a write pointer position.
@@ -2042,8 +2045,7 @@ static int sd_done(struct scsi_cmnd *SCpnt)
if (sshdr.ascq == 0x04)
sd_zbc_update_zones(sdkp,
blk_rq_pos(req),
- 512, true);
-#endif
+ 512, SD_ZBC_WRITE_ERR);
}
break;
default:
@@ -2270,7 +2272,7 @@ static void sd_read_zones(struct scsi_disk *sdkp, unsigned char *buffer)
* supports equal zone sizes.
*/
same = buffer[4] & 0xf;
- if (same == 0 || same > 3) {
+ if (same > 3) {
sd_printk(KERN_WARNING, sdkp,
"REPORT ZONES SAME type %d not supported\n", same);
return;
@@ -2282,9 +2284,9 @@ static void sd_read_zones(struct scsi_disk *sdkp, unsigned char *buffer)
sdkp->unmap_granularity = zone_len;
blk_queue_chunk_sectors(sdkp->disk->queue,
logical_to_sectors(sdkp->device, zone_len));
- sd_config_discard(sdkp, SD_ZBC_RESET_WP);

- sd_zbc_setup(sdkp, buffer, SD_BUF_SIZE);
+ sd_zbc_setup(sdkp, zone_len, buffer, SD_BUF_SIZE);
+ sd_config_discard(sdkp, SD_ZBC_RESET_WP);
}

static void read_capacity_error(struct scsi_disk *sdkp, struct scsi_device *sdp,
diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h
index 6ae4505..ef6c132 100644
--- a/drivers/scsi/sd.h
+++ b/drivers/scsi/sd.h
@@ -283,19 +283,24 @@ static inline void sd_dif_complete(struct scsi_cmnd *cmd, unsigned int a)

#endif /* CONFIG_BLK_DEV_INTEGRITY */

+
+#define SD_ZBC_INIT 0
+#define SD_ZBC_RESET_WP_ERR 1
+#define SD_ZBC_WRITE_ERR 2
+
#ifdef CONFIG_SCSI_ZBC

extern int sd_zbc_report_zones(struct scsi_disk *, unsigned char *, int,
sector_t, enum zbc_zone_reporting_options, bool);
-extern int sd_zbc_setup(struct scsi_disk *, char *, int);
+extern int sd_zbc_setup(struct scsi_disk *, u64 zlen, char *buf, int buf_len);
extern void sd_zbc_remove(struct scsi_disk *);
extern void sd_zbc_reset_zones(struct scsi_disk *);
extern int sd_zbc_setup_discard(struct scsi_disk *, struct request *,
sector_t, unsigned int);
extern int sd_zbc_setup_read_write(struct scsi_disk *, struct request *,
sector_t, unsigned int *);
-extern void sd_zbc_update_zones(struct scsi_disk *, sector_t, int, bool);
-extern void sd_zbc_refresh_zone_work(struct work_struct *);
+extern void sd_zbc_update_zones(struct scsi_disk *, sector_t, int, int reason);
+extern unsigned int sd_zbc_discard_granularity(struct scsi_disk *sdkp);

#else /* CONFIG_SCSI_ZBC */

@@ -308,7 +313,7 @@ static inline int sd_zbc_report_zones(struct scsi_disk *sdkp,
return -EOPNOTSUPP;
}

-static inline int sd_zbc_setup(struct scsi_disk *sdkp,
+static inline int sd_zbc_setup(struct scsi_disk *sdkp, u64 zlen,
unsigned char *buf, int buf_len)
{
return 0;
@@ -328,6 +333,13 @@ static inline int sd_zbc_setup_read_write(struct scsi_disk *sdkp,
return BLKPREP_OK;
}

+static inline unsigned int sd_zbc_discard_granularity(struct scsi_disk *sdkp)
+{
+ return sdkp->device->sector_size;
+}
+
+static inline void sd_zbc_update_zones(struct scsi_disk *sdkp, sector_t s,
+ int buf_sz, int reason) {}
static inline void sd_zbc_remove(struct scsi_disk *sdkp) {}
#endif /* CONFIG_SCSI_ZBC */

diff --git a/drivers/scsi/sd_zbc.c b/drivers/scsi/sd_zbc.c
index f953d16..17414fb 100644
--- a/drivers/scsi/sd_zbc.c
+++ b/drivers/scsi/sd_zbc.c
@@ -36,17 +36,6 @@
#include "sd.h"
#include "scsi_priv.h"

-enum zbc_zone_cond {
- ZBC_ZONE_COND_NO_WP,
- ZBC_ZONE_COND_EMPTY,
- ZBC_ZONE_COND_IMPLICIT_OPEN,
- ZBC_ZONE_COND_EXPLICIT_OPEN,
- ZBC_ZONE_COND_CLOSED,
- ZBC_ZONE_COND_READONLY = 0xd,
- ZBC_ZONE_COND_FULL,
- ZBC_ZONE_COND_OFFLINE,
-};
-
#define SD_ZBC_BUF_SIZE 524288

#define sd_zbc_debug(sdkp, fmt, args...) \
@@ -69,10 +58,10 @@ struct zbc_update_work {
char zone_buf[0];
};

+static
struct blk_zone *zbc_desc_to_zone(struct scsi_disk *sdkp, unsigned char *rec)
{
struct blk_zone *zone;
- enum zbc_zone_cond zone_cond;
sector_t wp = (sector_t)-1;

zone = kzalloc(sizeof(struct blk_zone), GFP_KERNEL);
@@ -81,37 +70,27 @@ struct blk_zone *zbc_desc_to_zone(struct scsi_disk *sdkp, unsigned char *rec)

spin_lock_init(&zone->lock);
zone->type = rec[0] & 0xf;
- zone_cond = (rec[1] >> 4) & 0xf;
+ zone->state = (rec[1] >> 4) & 0xf;
zone->len = logical_to_sectors(sdkp->device,
get_unaligned_be64(&rec[8]));
zone->start = logical_to_sectors(sdkp->device,
get_unaligned_be64(&rec[16]));

- if (blk_zone_is_smr(zone)) {
+ if (blk_zone_is_smr(zone))
wp = logical_to_sectors(sdkp->device,
get_unaligned_be64(&rec[24]));
- if (zone_cond == ZBC_ZONE_COND_READONLY) {
- zone->state = BLK_ZONE_READONLY;
- } else if (zone_cond == ZBC_ZONE_COND_OFFLINE) {
- zone->state = BLK_ZONE_OFFLINE;
- } else {
- zone->state = BLK_ZONE_OPEN;
- }
- } else
- zone->state = BLK_ZONE_NO_WP;
-
zone->wp = wp;
/*
* Fixup block zone state
*/
- if (zone_cond == ZBC_ZONE_COND_EMPTY &&
+ if (zone->state == BLK_ZONE_EMPTY &&
zone->wp != zone->start) {
sd_zbc_debug(sdkp,
"zone %zu state EMPTY wp %zu: adjust wp\n",
zone->start, zone->wp);
zone->wp = zone->start;
}
- if (zone_cond == ZBC_ZONE_COND_FULL &&
+ if (zone->state == BLK_ZONE_FULL &&
zone->wp != zone->start + zone->len) {
sd_zbc_debug(sdkp,
"zone %zu state FULL wp %zu: adjust wp\n",
@@ -122,7 +101,8 @@ struct blk_zone *zbc_desc_to_zone(struct scsi_disk *sdkp, unsigned char *rec)
return zone;
}

-sector_t zbc_parse_zones(struct scsi_disk *sdkp, unsigned char *buf,
+static
+sector_t zbc_parse_zones(struct scsi_disk *sdkp, u64 zlen, unsigned char *buf,
unsigned int buf_len)
{
struct request_queue *q = sdkp->disk->queue;
@@ -149,6 +129,11 @@ sector_t zbc_parse_zones(struct scsi_disk *sdkp, unsigned char *buf,
if (!this)
break;

+ if (same == 0 && this->len != zlen) {
+ next_sector = this->start + this->len;
+ break;
+ }
+
next_sector = this->start + this->len;
old = blk_insert_zone(q, this);
if (old) {
@@ -171,29 +156,58 @@ sector_t zbc_parse_zones(struct scsi_disk *sdkp, unsigned char *buf,
return next_sector;
}

-void sd_zbc_refresh_zone_work(struct work_struct *work)
+static void sd_zbc_refresh_zone_work(struct work_struct *work)
{
struct zbc_update_work *zbc_work =
container_of(work, struct zbc_update_work, zone_work);
struct scsi_disk *sdkp = zbc_work->sdkp;
struct request_queue *q = sdkp->disk->queue;
- unsigned int zone_buflen;
+ unsigned char *zone_buf = zbc_work->zone_buf;
+ unsigned int zone_buflen = zbc_work->zone_buflen;
int ret;
+ u8 same;
+ u64 zlen = 0;
sector_t last_sector;
sector_t capacity = logical_to_sectors(sdkp->device, sdkp->capacity);

- zone_buflen = zbc_work->zone_buflen;
- ret = sd_zbc_report_zones(sdkp, zbc_work->zone_buf, zone_buflen,
+ ret = sd_zbc_report_zones(sdkp, zone_buf, zone_buflen,
zbc_work->zone_sector,
ZBC_ZONE_REPORTING_OPTION_ALL, true);
if (ret)
goto done_free;

- last_sector = zbc_parse_zones(sdkp, zbc_work->zone_buf, zone_buflen);
+ /* this whole path is unlikely so extra reports shouldn't be a
+ * large impact */
+ same = zone_buf[4] & 0xf;
+ if (same == 0) {
+ unsigned char *desc = &zone_buf[64];
+ unsigned int blen = zone_buflen;
+
+ /* just pull the first zone */
+ if (blen > 512)
+ blen = 512;
+ ret = sd_zbc_report_zones(sdkp, zone_buf, blen, 0,
+ ZBC_ZONE_REPORTING_OPTION_ALL, true);
+ if (ret)
+ goto done_free;
+
+ /* Read the zone length from the first zone descriptor */
+ zlen = logical_to_sectors(sdkp->device,
+ get_unaligned_be64(&desc[8]));
+
+ ret = sd_zbc_report_zones(sdkp, zone_buf, zone_buflen,
+ zbc_work->zone_sector,
+ ZBC_ZONE_REPORTING_OPTION_ALL, true);
+ if (ret)
+ goto done_free;
+ }
+
+ last_sector = zbc_parse_zones(sdkp, zlen, zone_buf, zone_buflen);
+ capacity = logical_to_sectors(sdkp->device, sdkp->capacity);
if (last_sector != -1 && last_sector < capacity) {
if (test_bit(SD_ZBC_ZONE_RESET, &sdkp->zone_flags)) {
sd_zbc_debug(sdkp,
- "zones in reset, cancelling refresh\n");
+ "zones in reset, canceling refresh\n");
ret = -EAGAIN;
goto done_free;
}
@@ -207,7 +221,7 @@ done_free:
kfree(zbc_work);
if (test_and_clear_bit(SD_ZBC_ZONE_INIT, &sdkp->zone_flags) && ret) {
sd_zbc_debug(sdkp,
- "Cancelling zone initialisation\n");
+ "Canceling zone initialization\n");
}
done_start_queue:
if (q->mq_ops)
@@ -226,10 +240,10 @@ done_start_queue:
* @sdkp: SCSI disk for which the zone information needs to be updated
* @sector: sector to be updated
* @bufsize: buffersize to be allocated
- * @update: true if existing zones should be updated
+ * @reason: non-zero if existing zones should be updated
*/
void sd_zbc_update_zones(struct scsi_disk *sdkp, sector_t sector, int bufsize,
- bool update)
+ int reason)
{
struct request_queue *q = sdkp->disk->queue;
struct zbc_update_work *zbc_work;
@@ -240,13 +254,24 @@ void sd_zbc_update_zones(struct scsi_disk *sdkp, sector_t sector, int bufsize,

if (test_bit(SD_ZBC_ZONE_RESET, &sdkp->zone_flags)) {
sd_zbc_debug(sdkp,
- "zones in reset, not starting update\n");
+ "zones in reset, not starting reason\n");
return;
}

+ if (reason != SD_ZBC_INIT) {
+ /* lookup sector, is zone pref? then ignore */
+ struct blk_zone *zone = blk_lookup_zone(q, sector);
+
+ if (reason == SD_ZBC_RESET_WP)
+ sd_zbc_debug(sdkp, "RESET WP failed %lx\n", sector);
+
+ if (zone && blk_zone_is_seq_pref(zone))
+ return;
+ }
+
retry:
zbc_work = kzalloc(sizeof(struct zbc_update_work) + bufsize,
- update ? GFP_NOWAIT : GFP_KERNEL);
+ reason != SD_ZBC_INIT ? GFP_NOWAIT : GFP_KERNEL);
if (!zbc_work) {
if (bufsize > 512) {
sd_zbc_debug(sdkp,
@@ -256,7 +281,7 @@ retry:
}
sd_zbc_debug(sdkp,
"failed to allocate %d bytes\n", bufsize);
- if (!update)
+ if (reason == SD_ZBC_INIT)
clear_bit(SD_ZBC_ZONE_INIT, &sdkp->zone_flags);
return;
}
@@ -269,7 +294,7 @@ retry:
/*
* Mark zones under update as BUSY
*/
- if (update) {
+ if (reason != SD_ZBC_INIT) {
for (node = rb_first(&q->zones); node; node = rb_next(node)) {
unsigned long flags;

@@ -333,8 +358,7 @@ int sd_zbc_report_zones(struct scsi_disk *sdkp, unsigned char *buffer,
if (!scsi_device_online(sdp))
return -ENODEV;

- sd_zbc_debug(sdkp, "REPORT ZONES lba %zu len %d\n",
- start_lba, bufflen);
+ sd_zbc_debug(sdkp, "REPORT ZONES lba %zu len %d\n", start_lba, bufflen);

memset(cmd, 0, 16);
cmd[0] = ZBC_IN;
@@ -460,9 +484,37 @@ int sd_zbc_setup_read_write(struct scsi_disk *sdkp, struct request *rq,
goto out;
}

- if (req_op(rq) == REQ_OP_WRITE || req_op(rq) == REQ_OP_WRITE_SAME) {
- if (zone->type != BLK_ZONE_TYPE_SEQWRITE_REQ)
- goto out;
+ if (blk_zone_is_cmr(zone))
+ goto out;
+
+ if (blk_zone_is_seq_pref(zone)) {
+ if (op_is_write(req_op(rq))) {
+ u64 nwp = sector + sectors;
+
+ while (nwp > (zone->start + zone->len)) {
+ struct rb_node *node = rb_next(&zone->node);
+
+ zone->wp = zone->start + zone->len;
+ sector = zone->wp;
+ sectors = nwp - zone->wp;
+ spin_unlock_irqrestore(&zone->lock, flags);
+
+ if (!node)
+ return BLKPREP_OK;
+ zone = rb_entry(node, struct blk_zone, node);
+ if (!zone)
+ return BLKPREP_OK;
+
+ spin_lock_irqsave(&zone->lock, flags);
+ nwp = sector + sectors;
+ }
+ if (nwp > zone->wp)
+ zone->wp = nwp;
+ }
+ goto out;
+ }
+
+ if (op_is_write(req_op(rq))) {
if (zone->state == BLK_ZONE_READONLY)
goto out;
if (blk_zone_is_full(zone)) {
@@ -480,8 +532,7 @@ int sd_zbc_setup_read_write(struct scsi_disk *sdkp, struct request *rq,
goto out;
}
zone->wp += sectors;
- } else if (zone->type == BLK_ZONE_TYPE_SEQWRITE_REQ &&
- zone->wp <= sector + sectors) {
+ } else if (zone->wp <= sector + sectors) {
if (zone->wp <= sector) {
/* Read beyond WP: clear request buffer */
struct req_iterator iter;
@@ -513,14 +564,18 @@ out:
return ret;
}

-int sd_zbc_setup(struct scsi_disk *sdkp, char *buf, int buf_len)
+/**
+ * sd_zbc_setup - Load zones of matching zlen size into rb tree.
+ *
+ */
+int sd_zbc_setup(struct scsi_disk *sdkp, u64 zlen, char *buf, int buf_len)
{
sector_t capacity = logical_to_sectors(sdkp->device, sdkp->capacity);
sector_t last_sector;

if (test_and_set_bit(SD_ZBC_ZONE_INIT, &sdkp->zone_flags)) {
sdev_printk(KERN_WARNING, sdkp->device,
- "zone initialisation already running\n");
+ "zone initialization already running\n");
return 0;
}

@@ -539,15 +594,20 @@ int sd_zbc_setup(struct scsi_disk *sdkp, char *buf, int buf_len)
clear_bit(SD_ZBC_ZONE_RESET, &sdkp->zone_flags);
}

- last_sector = zbc_parse_zones(sdkp, buf, buf_len);
+ last_sector = zbc_parse_zones(sdkp, zlen, buf, buf_len);
+ capacity = logical_to_sectors(sdkp->device, sdkp->capacity);
if (last_sector != -1 && last_sector < capacity) {
- sd_zbc_update_zones(sdkp, last_sector, SD_ZBC_BUF_SIZE, false);
+ sd_zbc_update_zones(sdkp, last_sector,
+ SD_ZBC_BUF_SIZE, SD_ZBC_INIT);
} else
clear_bit(SD_ZBC_ZONE_INIT, &sdkp->zone_flags);

return 0;
}

+/**
+ * sd_zbc_remove -
+ */
void sd_zbc_remove(struct scsi_disk *sdkp)
{
if (sdkp->zone_work_q) {
@@ -557,3 +617,24 @@ void sd_zbc_remove(struct scsi_disk *sdkp)
destroy_workqueue(sdkp->zone_work_q);
}
}
+/**
+ * sd_zbc_discard_granularity - Determine discard granularity.
+ * @sdkp: SCSI disk used to calculate discard granularity.
+ *
+ * Discard granularity should match the (maximum non-CMR) zone
+ * size reported on the drive.
+ */
+unsigned int sd_zbc_discard_granularity(struct scsi_disk *sdkp)
+{
+ unsigned int bytes = 1;
+ struct request_queue *q = sdkp->disk->queue;
+ struct rb_node *node = rb_first(&q->zones);
+
+ if (node) {
+ struct blk_zone *zone = rb_entry(node, struct blk_zone, node);
+
+ bytes = zone->len;
+ }
+ bytes <<= ilog2(sdkp->device->sector_size);
+ return bytes;
+}
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 9084a9e..68198eb 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -272,12 +272,16 @@ enum blk_zone_type {
};

enum blk_zone_state {
- BLK_ZONE_UNKNOWN,
BLK_ZONE_NO_WP,
+ BLK_ZONE_EMPTY,
BLK_ZONE_OPEN,
- BLK_ZONE_READONLY,
+ BLK_ZONE_OPEN_EXPLICIT,
+ BLK_ZONE_CLOSED,
+ BLK_ZONE_UNKNOWN = 5,
+ BLK_ZONE_READONLY = 0xd,
+ BLK_ZONE_FULL,
BLK_ZONE_OFFLINE,
- BLK_ZONE_BUSY,
+ BLK_ZONE_BUSY = 0x20,
};

struct blk_zone {
@@ -291,9 +295,9 @@ struct blk_zone {
void *private_data;
};

-#define blk_zone_is_smr(z) ((z)->type == BLK_ZONE_TYPE_SEQWRITE_REQ || \
- (z)->type == BLK_ZONE_TYPE_SEQWRITE_PREF)
-
+#define blk_zone_is_seq_req(z) ((z)->type == BLK_ZONE_TYPE_SEQWRITE_REQ)
+#define blk_zone_is_seq_pref(z) ((z)->type == BLK_ZONE_TYPE_SEQWRITE_PREF)
+#define blk_zone_is_smr(z) (blk_zone_is_seq_req(z) || blk_zone_is_seq_pref(z))
#define blk_zone_is_cmr(z) ((z)->type == BLK_ZONE_TYPE_CONVENTIONAL)
#define blk_zone_is_full(z) ((z)->wp == (z)->start + (z)->len)
#define blk_zone_is_empty(z) ((z)->wp == (z)->start)
--
2.9.3