[PATCH v4 00/32] cxlflash: Miscellaneous bug fixes and corrections

[PATCH v4 00/32] cxlflash: Miscellaneous bug fixes and corrections

[Matthew R. Ochs]

This patch set contains various fixes and corrections for issues that
were found during test and code review. The series is based upon the
code upstreamed in 4.3 and is intended for the rc phase. The entire
set is bisectable. Please reference the changelog below for details
on what has been altered from previous versions of this patch set.

v4 Changes:
- Incorporate comments from Brian King
- Removed unnecessary check_state() parameter from "Fix to avoid CXL..."
- Added patch to fix potential deadlock on EEH
- Removed patch to avoid state change collision
- Changed fops initialization location in "Fix to avoid corrupting..."

v3 Changes:
- Rebased the series on top of patch by Dan Carpenter ("a couple off...")
- Incorporate comments from David Laight
- Incorporate comments from Tomas Henzl
- Incorporate comments from Brian King
- Removed patch to stop interrupt processing on remove
- Removed double scsi_device_put() from "Fix potential oops"
- Fixed usage of scnprintf() in "Refine host/device attributes"
- Removed unnecessary parenthesis from "Fix read capacity timeout"
- Added patch to use correct operator for doubling delay
- Changed location of cancel_work_sync() in "Fix to prevent workq..."
- Removed local mutex from cxlflash_afu_sync() in "Fix to avoid state..."
- Added patch to correctly identify a failed function in a trace
- Added patch to fix a fops corruption bug

v2 Changes:
- Incorporate comments from Ian Munsie
- Rework commit messages to be more descriptive
- Add state change serialization patch

Manoj Kumar (4):
  cxlflash: Fix to avoid invalid port_sel value
  cxlflash: Replace magic numbers with literals
  cxlflash: Fix read capacity timeout
  cxlflash: Fix to double the delay each time

Matthew R. Ochs (28):
  cxlflash: Fix potential oops following LUN removal
  cxlflash: Fix data corruption when vLUN used over multiple cards
  cxlflash: Fix to avoid sizeof(bool)
  cxlflash: Fix context encode mask width
  cxlflash: Fix to avoid CXL services during EEH
  cxlflash: Correct naming of limbo state and waitq
  cxlflash: Make functions static
  cxlflash: Refine host/device attributes
  cxlflash: Fix to avoid spamming the kernel log
  cxlflash: Fix to avoid stall while waiting on TMF
  cxlflash: Fix location of setting resid
  cxlflash: Fix host link up event handling
  cxlflash: Fix async interrupt bypass logic
  cxlflash: Remove dual port online dependency
  cxlflash: Fix AFU version access/storage and add check
  cxlflash: Correct usage of scsi_host_put()
  cxlflash: Fix to prevent workq from accessing freed memory
  cxlflash: Correct behavior in device reset handler following EEH
  cxlflash: Remove unnecessary scsi_block_requests
  cxlflash: Fix function prolog parameters and return codes
  cxlflash: Fix MMIO and endianness errors
  cxlflash: Fix to prevent EEH recovery failure
  cxlflash: Correct spelling, grammar, and alignment mistakes
  cxlflash: Fix to prevent stale AFU RRQ
  MAINTAINERS: Add cxlflash driver
  cxlflash: Fix to avoid corrupting adapter fops
  cxlflash: Correct trace string
  cxlflash: Fix to avoid potential deadlock on EEH

 MAINTAINERS                       |    9 +
 drivers/scsi/cxlflash/common.h    |   30 +-
 drivers/scsi/cxlflash/lunmgt.c    |    9 +-
 drivers/scsi/cxlflash/main.c      | 1529 ++++++++++++++++++++-----------------
 drivers/scsi/cxlflash/main.h      |    1 +
 drivers/scsi/cxlflash/sislite.h   |    8 +-
 drivers/scsi/cxlflash/superpipe.c |  204 +++--
 drivers/scsi/cxlflash/superpipe.h |   13 +-
 drivers/scsi/cxlflash/vlun.c      |   68 +-
 9 files changed, 1043 insertions(+), 828 deletions(-)

-- 
2.1.0

[PATCH v4 01/32] cxlflash: Fix to avoid invalid port_sel value

[Matthew R. Ochs]

From: Manoj Kumar <kumarmn@us.ibm.com>

If two concurrent MANAGE_LUN ioctls are issued with the same
WWID parameter, it would result in an incorrect value of port_sel.

This is because port_sel is modified without any locks being
held. If the first caller stalls after the return from
find_and_create_lun(), the value of port_sel will be set
incorrectly to indicate a single port, though in this case
it should have been set to both ports.

To fix, use the global mutex to serialize the lookup of the
WWID and the subsequent modification of port_sel.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/lunmgt.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/cxlflash/lunmgt.c b/drivers/scsi/cxlflash/lunmgt.c
index d98ad0f..8c372fc 100644
--- a/drivers/scsi/cxlflash/lunmgt.c
+++ b/drivers/scsi/cxlflash/lunmgt.c
@@ -120,7 +120,8 @@ static struct glun_info *lookup_global(u8 *wwid)
  *
  * The LUN is kept both in a local list (per adapter) and in a global list
  * (across all adapters). Certain attributes of the LUN are local to the
- * adapter (such as index, port selection mask etc.).
+ * adapter (such as index, port selection mask, etc.).
+ *
  * The block allocation map is shared across all adapters (i.e. associated
  * wih the global list). Since different attributes are associated with
  * the per adapter and global entries, allocate two separate structures for each
@@ -128,6 +129,8 @@ static struct glun_info *lookup_global(u8 *wwid)
  *
  * Keep a pointer back from the local to the global entry.
  *
+ * This routine assumes the caller holds the global mutex.
+ *
  * Return: Found/Allocated local lun_info structure on success, NULL on failure
  */
 static struct llun_info *find_and_create_lun(struct scsi_device *sdev, u8 *wwid)
@@ -137,7 +140,6 @@ static struct llun_info *find_and_create_lun(struct scsi_device *sdev, u8 *wwid)
 	struct Scsi_Host *shost = sdev->host;
 	struct cxlflash_cfg *cfg = shost_priv(shost);
 
-	mutex_lock(&global.mutex);
 	if (unlikely(!wwid))
 		goto out;
 
@@ -169,7 +171,6 @@ static struct llun_info *find_and_create_lun(struct scsi_device *sdev, u8 *wwid)
 	list_add(&gli->list, &global.gluns);
 
 out:
-	mutex_unlock(&global.mutex);
 	pr_debug("%s: returning %p\n", __func__, lli);
 	return lli;
 }
@@ -235,6 +236,7 @@ int cxlflash_manage_lun(struct scsi_device *sdev,
 	u64 flags = manage->hdr.flags;
 	u32 chan = sdev->channel;
 
+	mutex_lock(&global.mutex);
 	lli = find_and_create_lun(sdev, manage->wwid);
 	pr_debug("%s: ENTER: WWID = %016llX%016llX, flags = %016llX li = %p\n",
 		 __func__, get_unaligned_le64(&manage->wwid[0]),
@@ -261,6 +263,7 @@ int cxlflash_manage_lun(struct scsi_device *sdev,
 	}
 
 out:
+	mutex_unlock(&global.mutex);
 	pr_debug("%s: returning rc=%d\n", __func__, rc);
 	return rc;
 }
-- 
2.1.0

[PATCH v4 02/32] cxlflash: Replace magic numbers with literals

[Matthew R. Ochs]

From: Manoj Kumar <kumarmn@us.ibm.com>

Magic numbers are not meaningful and can create confusion. As a
remedy, replace them with descriptive literals.

Replace 512 with literal MAX_SECTOR_UNIT.
Replace 5 with literal CMD_RETRIES.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 6 ++++--
 drivers/scsi/cxlflash/superpipe.h | 3 +++
 drivers/scsi/cxlflash/vlun.c      | 3 ++-
 3 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 05e0ecf..4db15a4 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -315,7 +315,8 @@ retry:
 		retry_cnt ? "re" : "", scsi_cmd[0]);
 
 	result = scsi_execute(sdev, scsi_cmd, DMA_FROM_DEVICE, cmd_buf,
-			      CMD_BUFSIZE, sense_buf, tout, 5, 0, NULL);
+			      CMD_BUFSIZE, sense_buf, tout, CMD_RETRIES,
+			      0, NULL);
 
 	if (driver_byte(result) == DRIVER_SENSE) {
 		result &= ~(0xFF<<24); /* DRIVER_SENSE is not an error */
@@ -1375,7 +1376,8 @@ out_attach:
 	attach->block_size = gli->blk_len;
 	attach->mmio_size = sizeof(afu->afu_map->hosts[0].harea);
 	attach->last_lba = gli->max_lba;
-	attach->max_xfer = (sdev->host->max_sectors * 512) / gli->blk_len;
+	attach->max_xfer = (sdev->host->max_sectors * MAX_SECTOR_UNIT) /
+		gli->blk_len;
 
 out:
 	attach->adap_fd = fd;
diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index d7dc88b..3f7856b 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -29,6 +29,9 @@ extern struct cxlflash_global global;
 #define MC_CHUNK_SIZE     (1 << MC_RHT_NMASK)	/* in LBAs */
 
 #define MC_DISCOVERY_TIMEOUT 5  /* 5 secs */
+#define CMD_RETRIES 5   /* 5 retries for scsi_execute */
+
+#define MAX_SECTOR_UNIT  512 /* max_sector is in 512 byte multiples */
 
 #define CHAN2PORT(_x)	((_x) + 1)
 #define PORT2CHAN(_x)	((_x) - 1)
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 6155cb1..6d6608b 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -434,7 +434,8 @@ static int write_same16(struct scsi_device *sdev,
 				   &scsi_cmd[10]);
 
 		result = scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, cmd_buf,
-				      CMD_BUFSIZE, sense_buf, tout, 5, 0, NULL);
+				      CMD_BUFSIZE, sense_buf, tout, CMD_RETRIES,
+				      0, NULL);
 		if (result) {
 			dev_err_ratelimited(dev, "%s: command failed for "
 					    "offset %lld result=0x%x\n",
-- 
2.1.0

[PATCH v4 03/32] cxlflash: Fix read capacity timeout

[Matthew R. Ochs]

From: Manoj Kumar <kumarmn@us.ibm.com>

The timeout value for read capacity is too small. Certain devices
may take longer to respond and thus the command may prematurely
timeout. Additionally the literal used for the timeout is stale.

Update the timeout to 30 seconds (matches the value used in sd.c)
and rework the timeout literal to a more appropriate description.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 9 ++++-----
 drivers/scsi/cxlflash/superpipe.h | 2 +-
 drivers/scsi/cxlflash/vlun.c      | 4 ++--
 3 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 4db15a4..4e44a48 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -296,7 +296,7 @@ static int read_cap16(struct scsi_device *sdev, struct llun_info *lli)
 	int rc = 0;
 	int result = 0;
 	int retry_cnt = 0;
-	u32 tout = (MC_DISCOVERY_TIMEOUT * HZ);
+	u32 to = CMD_TIMEOUT * HZ;
 
 retry:
 	cmd_buf = kzalloc(CMD_BUFSIZE, GFP_KERNEL);
@@ -315,8 +315,7 @@ retry:
 		retry_cnt ? "re" : "", scsi_cmd[0]);
 
 	result = scsi_execute(sdev, scsi_cmd, DMA_FROM_DEVICE, cmd_buf,
-			      CMD_BUFSIZE, sense_buf, tout, CMD_RETRIES,
-			      0, NULL);
+			      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES, 0, NULL);
 
 	if (driver_byte(result) == DRIVER_SENSE) {
 		result &= ~(0xFF<<24); /* DRIVER_SENSE is not an error */
@@ -1376,8 +1375,8 @@ out_attach:
 	attach->block_size = gli->blk_len;
 	attach->mmio_size = sizeof(afu->afu_map->hosts[0].harea);
 	attach->last_lba = gli->max_lba;
-	attach->max_xfer = (sdev->host->max_sectors * MAX_SECTOR_UNIT) /
-		gli->blk_len;
+	attach->max_xfer = sdev->host->max_sectors * MAX_SECTOR_UNIT;
+	attach->max_xfer /= gli->blk_len;
 
 out:
 	attach->adap_fd = fd;
diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index 3f7856b..fffb179 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -28,7 +28,7 @@ extern struct cxlflash_global global;
 */
 #define MC_CHUNK_SIZE     (1 << MC_RHT_NMASK)	/* in LBAs */
 
-#define MC_DISCOVERY_TIMEOUT 5  /* 5 secs */
+#define CMD_TIMEOUT 30  /* 30 secs */
 #define CMD_RETRIES 5   /* 5 retries for scsi_execute */
 
 #define MAX_SECTOR_UNIT  512 /* max_sector is in 512 byte multiples */
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 6d6608b..68994c4 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -414,7 +414,7 @@ static int write_same16(struct scsi_device *sdev,
 	int ws_limit = SISLITE_MAX_WS_BLOCKS;
 	u64 offset = lba;
 	int left = nblks;
-	u32 tout = sdev->request_queue->rq_timeout;
+	u32 to = sdev->request_queue->rq_timeout;
 	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)sdev->host->hostdata;
 	struct device *dev = &cfg->dev->dev;
 
@@ -434,7 +434,7 @@ static int write_same16(struct scsi_device *sdev,
 				   &scsi_cmd[10]);
 
 		result = scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, cmd_buf,
-				      CMD_BUFSIZE, sense_buf, tout, CMD_RETRIES,
+				      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES,
 				      0, NULL);
 		if (result) {
 			dev_err_ratelimited(dev, "%s: command failed for "
-- 
2.1.0

[PATCH v4 04/32] cxlflash: Fix potential oops following LUN removal

[Matthew R. Ochs]

When a LUN is removed, the sdev that is associated with the LUN
remains intact until its reference count drops to 0. In order
to prevent an sdev from being removed while a context is still
associated with it, obtain an additional reference per-context
for each LUN attached to the context.

This resolves a potential Oops in the release handler when a
dealing with a LUN that has already been removed.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 35 +++++++++++++++++++++++------------
 1 file changed, 23 insertions(+), 12 deletions(-)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 4e44a48..ffa68cc 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -880,6 +880,9 @@ static int _cxlflash_disk_detach(struct scsi_device *sdev,
 			sys_close(lfd);
 	}
 
+	/* Release the sdev reference that bound this LUN to the context */
+	scsi_device_put(sdev);
+
 out:
 	if (put_ctx)
 		put_context(ctxi);
@@ -1287,11 +1290,17 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
 			}
 	}
 
+	rc = scsi_device_get(sdev);
+	if (unlikely(rc)) {
+		dev_err(dev, "%s: Unable to get sdev reference!\n", __func__);
+		goto out;
+	}
+
 	lun_access = kzalloc(sizeof(*lun_access), GFP_KERNEL);
 	if (unlikely(!lun_access)) {
 		dev_err(dev, "%s: Unable to allocate lun_access!\n", __func__);
 		rc = -ENOMEM;
-		goto out;
+		goto err0;
 	}
 
 	lun_access->lli = lli;
@@ -1311,21 +1320,21 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
 		dev_err(dev, "%s: Could not initialize context %p\n",
 			__func__, ctx);
 		rc = -ENODEV;
-		goto err0;
+		goto err1;
 	}
 
 	ctxid = cxl_process_element(ctx);
 	if (unlikely((ctxid >= MAX_CONTEXT) || (ctxid < 0))) {
 		dev_err(dev, "%s: ctxid (%d) invalid!\n", __func__, ctxid);
 		rc = -EPERM;
-		goto err1;
+		goto err2;
 	}
 
 	file = cxl_get_fd(ctx, &cfg->cxl_fops, &fd);
 	if (unlikely(fd < 0)) {
 		rc = -ENODEV;
 		dev_err(dev, "%s: Could not get file descriptor\n", __func__);
-		goto err1;
+		goto err2;
 	}
 
 	/* Translate read/write O_* flags from fcntl.h to AFU permission bits */
@@ -1335,7 +1344,7 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
 	if (unlikely(!ctxi)) {
 		dev_err(dev, "%s: Failed to create context! (%d)\n",
 			__func__, ctxid);
-		goto err2;
+		goto err3;
 	}
 
 	work = &ctxi->work;
@@ -1346,13 +1355,13 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
 	if (unlikely(rc)) {
 		dev_dbg(dev, "%s: Could not start context rc=%d\n",
 			__func__, rc);
-		goto err3;
+		goto err4;
 	}
 
 	rc = afu_attach(cfg, ctxi);
 	if (unlikely(rc)) {
 		dev_err(dev, "%s: Could not attach AFU rc %d\n", __func__, rc);
-		goto err4;
+		goto err5;
 	}
 
 	/*
@@ -1388,13 +1397,13 @@ out:
 		__func__, ctxid, fd, attach->block_size, rc, attach->last_lba);
 	return rc;
 
-err4:
+err5:
 	cxl_stop_context(ctx);
-err3:
+err4:
 	put_context(ctxi);
 	destroy_context(cfg, ctxi);
 	ctxi = NULL;
-err2:
+err3:
 	/*
 	 * Here, we're overriding the fops with a dummy all-NULL fops because
 	 * fput() calls the release fop, which will cause us to mistakenly
@@ -1406,10 +1415,12 @@ err2:
 	fput(file);
 	put_unused_fd(fd);
 	fd = -1;
-err1:
+err2:
 	cxl_release_context(ctx);
-err0:
+err1:
 	kfree(lun_access);
+err0:
+	scsi_device_put(sdev);
 	goto out;
 }
 
-- 
2.1.0

[PATCH v4 05/32] cxlflash: Fix data corruption when vLUN used over multiple cards

[Matthew R. Ochs]

If the same virtual LUN is accessed over multiple cards, only accesses
made over the first card will be valid. Accesses made over the second
card will go to the wrong LUN causing data corruption.

This is because the global LUN's mode word was being used to determine
whether the LUN table for that card needs to be programmed. The mode
word would be setup by the first card, causing the LUN table for the
second card to not be programmed.

By unconditionally initializing the LUN table (not depending on the
mode word), the problem is avoided.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/vlun.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 68994c4..96b074f 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -915,16 +915,9 @@ int cxlflash_disk_virtual_open(struct scsi_device *sdev, void *arg)
 
 	pr_debug("%s: ctxid=%llu ls=0x%llx\n", __func__, ctxid, lun_size);
 
+	/* Setup the LUNs block allocator on first call */
 	mutex_lock(&gli->mutex);
 	if (gli->mode == MODE_NONE) {
-		/* Setup the LUN table and block allocator on first call */
-		rc = init_luntable(cfg, lli);
-		if (rc) {
-			dev_err(dev, "%s: call to init_luntable failed "
-				"rc=%d!\n", __func__, rc);
-			goto err0;
-		}
-
 		rc = init_vlun(lli);
 		if (rc) {
 			dev_err(dev, "%s: call to init_vlun failed rc=%d!\n",
@@ -942,6 +935,13 @@ int cxlflash_disk_virtual_open(struct scsi_device *sdev, void *arg)
 	}
 	mutex_unlock(&gli->mutex);
 
+	rc = init_luntable(cfg, lli);
+	if (rc) {
+		dev_err(dev, "%s: call to init_luntable failed rc=%d!\n",
+			__func__, rc);
+		goto err1;
+	}
+
 	ctxi = get_context(cfg, rctxid, lli, 0);
 	if (unlikely(!ctxi)) {
 		dev_err(dev, "%s: Bad context! (%llu)\n", __func__, ctxid);
-- 
2.1.0

[PATCH v4 06/32] cxlflash: Fix to avoid sizeof(bool)

[Matthew R. Ochs]

Using sizeof(bool) is considered poor form for various reasons and
sparse warns us of that. Correct by changing type from bool to u8.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 2 +-
 drivers/scsi/cxlflash/superpipe.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index ffa68cc..28aa9d9 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -737,7 +737,7 @@ static struct ctx_info *create_context(struct cxlflash_cfg *cfg,
 	struct afu *afu = cfg->afu;
 	struct ctx_info *ctxi = NULL;
 	struct llun_info **lli = NULL;
-	bool *ws = NULL;
+	u8 *ws = NULL;
 	struct sisl_rht_entry *rhte;
 
 	ctxi = kzalloc(sizeof(*ctxi), GFP_KERNEL);
diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index fffb179..72d53cf 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -97,7 +97,7 @@ struct ctx_info {
 	u32 rht_out;		/* Number of checked out RHT entries */
 	u32 rht_perms;		/* User-defined permissions for RHT entries */
 	struct llun_info **rht_lun;       /* Mapping of RHT entries to LUNs */
-	bool *rht_needs_ws;	/* User-desired write-same function per RHTE */
+	u8 *rht_needs_ws;	/* User-desired write-same function per RHTE */
 
 	struct cxl_ioctl_start_work work;
 	u64 ctxid;
-- 
2.1.0

[PATCH v4 07/32] cxlflash: Fix context encode mask width

[Matthew R. Ochs]

The context encode mask covers more than 32-bits, making it
a long integer. This should be noted by appending the ULL
width suffix to the mask.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index 72d53cf..7947091 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -87,7 +87,7 @@ enum ctx_ctrl {
 	CTX_CTRL_FILE		= (1 << 5)
 };
 
-#define ENCODE_CTXID(_ctx, _id)	(((((u64)_ctx) & 0xFFFFFFFF0) << 28) | _id)
+#define ENCODE_CTXID(_ctx, _id)	(((((u64)_ctx) & 0xFFFFFFFF0ULL) << 28) | _id)
 #define DECODE_CTXID(_val)	(_val & 0xFFFFFFFF)
 
 struct ctx_info {
-- 
2.1.0

[PATCH v4 08/32] cxlflash: Fix to avoid CXL services during EEH

[Matthew R. Ochs]

During an EEH freeze event, certain CXL services should not be
called until after the hardware reset has taken place. Doing so
can result in unnecessary failures and possibly cause other ill
effects by triggering hardware accesses. This translates to a
requirement to quiesce all threads that may potentially use CXL
runtime service during this window. In particular, multiple ioctls
make use of the CXL services when acting on contexts on behalf of
the user. Thus, it is essential to 'drain' running ioctls _before_
proceeding with handling the EEH freeze event.

Create the ability to drain ioctls by wrapping the ioctl handler
call in a read semaphore and then implementing a small routine that
obtains the write semaphore, effectively creating a wait point for
all currently executing ioctls.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h    |  2 +
 drivers/scsi/cxlflash/main.c      | 18 +++++--
 drivers/scsi/cxlflash/superpipe.c | 98 ++++++++++++++++++++++++---------------
 3 files changed, 77 insertions(+), 41 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 1c56037..1abe4e0 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -16,6 +16,7 @@
 #define _CXLFLASH_COMMON_H
 
 #include <linux/list.h>
+#include <linux/rwsem.h>
 #include <linux/types.h>
 #include <scsi/scsi.h>
 #include <scsi/scsi_device.h>
@@ -110,6 +111,7 @@ struct cxlflash_cfg {
 	atomic_t recovery_threads;
 	struct mutex ctx_recovery_mutex;
 	struct mutex ctx_tbl_list_mutex;
+	struct rw_semaphore ioctl_rwsem;
 	struct ctx_info *ctx_tbl[MAX_CONTEXT];
 	struct list_head ctx_err_recovery; /* contexts w/ recovery pending */
 	struct file_operations cxl_fops;
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 3e3ccf1..6e85c77 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -2311,6 +2311,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->lr_port = -1;
 	mutex_init(&cfg->ctx_tbl_list_mutex);
 	mutex_init(&cfg->ctx_recovery_mutex);
+	init_rwsem(&cfg->ioctl_rwsem);
 	INIT_LIST_HEAD(&cfg->ctx_err_recovery);
 	INIT_LIST_HEAD(&cfg->lluns);
 
@@ -2365,6 +2366,19 @@ out_remove:
 }
 
 /**
+ * drain_ioctls() - wait until all currently executing ioctls have completed
+ * @cfg:	Internal structure associated with the host.
+ *
+ * Obtain write access to read/write semaphore that wraps ioctl
+ * handling to 'drain' ioctls currently executing.
+ */
+static void drain_ioctls(struct cxlflash_cfg *cfg)
+{
+	down_write(&cfg->ioctl_rwsem);
+	up_write(&cfg->ioctl_rwsem);
+}
+
+/**
  * cxlflash_pci_error_detected() - called when a PCI error is detected
  * @pdev:	PCI device struct.
  * @state:	PCI channel state.
@@ -2383,16 +2397,14 @@ static pci_ers_result_t cxlflash_pci_error_detected(struct pci_dev *pdev,
 	switch (state) {
 	case pci_channel_io_frozen:
 		cfg->state = STATE_LIMBO;
-
-		/* Turn off legacy I/O */
 		scsi_block_requests(cfg->host);
+		drain_ioctls(cfg);
 		rc = cxlflash_mark_contexts_error(cfg);
 		if (unlikely(rc))
 			dev_err(dev, "%s: Failed to mark user contexts!(%d)\n",
 				__func__, rc);
 		term_mc(cfg, UNDO_START);
 		stop_afu(cfg);
-
 		return PCI_ERS_RESULT_NEED_RESET;
 	case pci_channel_io_perm_failure:
 		cfg->state = STATE_FAILTERM;
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 28aa9d9..655cbf1 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -1214,6 +1214,46 @@ static const struct file_operations null_fops = {
 };
 
 /**
+ * check_state() - checks and responds to the current adapter state
+ * @cfg:	Internal structure associated with the host.
+ *
+ * This routine can block and should only be used on process context.
+ * It assumes that the caller is an ioctl thread and holding the ioctl
+ * read semaphore. This is temporarily let up across the wait to allow
+ * for draining actively running ioctls. Also note that when waking up
+ * from waiting in reset, the state is unknown and must be checked again
+ * before proceeding.
+ *
+ * Return: 0 on success, -errno on failure
+ */
+static int check_state(struct cxlflash_cfg *cfg)
+{
+	struct device *dev = &cfg->dev->dev;
+	int rc = 0;
+
+retry:
+	switch (cfg->state) {
+	case STATE_LIMBO:
+		dev_dbg(dev, "%s: Limbo state, going to wait...\n", __func__);
+		up_read(&cfg->ioctl_rwsem);
+		rc = wait_event_interruptible(cfg->limbo_waitq,
+					      cfg->state != STATE_LIMBO);
+		down_read(&cfg->ioctl_rwsem);
+		if (unlikely(rc))
+			break;
+		goto retry;
+	case STATE_FAILTERM:
+		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
+		rc = -ENODEV;
+		break;
+	default:
+		break;
+	}
+
+	return rc;
+}
+
+/**
  * cxlflash_disk_attach() - attach a LUN to a context
  * @sdev:	SCSI device associated with LUN.
  * @attach:	Attach ioctl data structure.
@@ -1523,41 +1563,6 @@ err1:
 }
 
 /**
- * check_state() - checks and responds to the current adapter state
- * @cfg:	Internal structure associated with the host.
- *
- * This routine can block and should only be used on process context.
- * Note that when waking up from waiting in limbo, the state is unknown
- * and must be checked again before proceeding.
- *
- * Return: 0 on success, -errno on failure
- */
-static int check_state(struct cxlflash_cfg *cfg)
-{
-	struct device *dev = &cfg->dev->dev;
-	int rc = 0;
-
-retry:
-	switch (cfg->state) {
-	case STATE_LIMBO:
-		dev_dbg(dev, "%s: Limbo, going to wait...\n", __func__);
-		rc = wait_event_interruptible(cfg->limbo_waitq,
-					      cfg->state != STATE_LIMBO);
-		if (unlikely(rc))
-			break;
-		goto retry;
-	case STATE_FAILTERM:
-		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
-		rc = -ENODEV;
-		break;
-	default:
-		break;
-	}
-
-	return rc;
-}
-
-/**
  * cxlflash_afu_recover() - initiates AFU recovery
  * @sdev:	SCSI device associated with LUN.
  * @recover:	Recover ioctl data structure.
@@ -1646,9 +1651,14 @@ retry_recover:
 	/* Test if in error state */
 	reg = readq_be(&afu->ctrl_map->mbox_r);
 	if (reg == -1) {
-		dev_dbg(dev, "%s: MMIO read fail! Wait for recovery...\n",
-			__func__);
-		mutex_unlock(&ctxi->mutex);
+		dev_dbg(dev, "%s: MMIO fail, wait for recovery.\n", __func__);
+
+		/*
+		 * Before checking the state, put back the context obtained with
+		 * get_context() as it is no longer needed and sleep for a short
+		 * period of time (see prolog notes).
+		 */
+		put_context(ctxi);
 		ctxi = NULL;
 		ssleep(1);
 		rc = check_state(cfg);
@@ -1967,6 +1977,14 @@ out:
  * @cmd:	IOCTL command.
  * @arg:	Userspace ioctl data structure.
  *
+ * A read/write semaphore is used to implement a 'drain' of currently
+ * running ioctls. The read semaphore is taken at the beginning of each
+ * ioctl thread and released upon concluding execution. Additionally the
+ * semaphore should be released and then reacquired in any ioctl execution
+ * path which will wait for an event to occur that is outside the scope of
+ * the ioctl (i.e. an adapter reset). To drain the ioctls currently running,
+ * a thread simply needs to acquire the write semaphore.
+ *
  * Return: 0 on success, -errno on failure
  */
 int cxlflash_ioctl(struct scsi_device *sdev, int cmd, void __user *arg)
@@ -2001,6 +2019,9 @@ int cxlflash_ioctl(struct scsi_device *sdev, int cmd, void __user *arg)
 	{sizeof(struct dk_cxlflash_clone), (sioctl)cxlflash_disk_clone},
 	};
 
+	/* Hold read semaphore so we can drain if needed */
+	down_read(&cfg->ioctl_rwsem);
+
 	/* Restrict command set to physical support only for internal LUN */
 	if (afu->internal_lun)
 		switch (cmd) {
@@ -2082,6 +2103,7 @@ int cxlflash_ioctl(struct scsi_device *sdev, int cmd, void __user *arg)
 	/* fall through to exit */
 
 cxlflash_ioctl_exit:
+	up_read(&cfg->ioctl_rwsem);
 	if (unlikely(rc && known_ioctl))
 		dev_err(dev, "%s: ioctl %s (%08X) on dev(%d/%d/%d/%llu) "
 			"returned rc %d\n", __func__,
-- 
2.1.0

[PATCH v4 09/32] cxlflash: Correct naming of limbo state and waitq

[Matthew R. Ochs]

Limbo is not an accurate representation of this state and is
also not consistent with the terminology that other drivers
use to represent this concept. Rename the state and and its
associated waitq to 'reset'.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h    |  4 ++--
 drivers/scsi/cxlflash/main.c      | 26 +++++++++++++-------------
 drivers/scsi/cxlflash/superpipe.c | 14 +++++++-------
 3 files changed, 22 insertions(+), 22 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 1abe4e0..11318de 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -79,7 +79,7 @@ enum cxlflash_init_state {
 
 enum cxlflash_state {
 	STATE_NORMAL,	/* Normal running state, everything good */
-	STATE_LIMBO,	/* Limbo running state, trying to reset/recover */
+	STATE_RESET,	/* Reset state, trying to reset/recover */
 	STATE_FAILTERM	/* Failed/terminating state, error out users/threads */
 };
 
@@ -125,7 +125,7 @@ struct cxlflash_cfg {
 
 	wait_queue_head_t tmf_waitq;
 	bool tmf_active;
-	wait_queue_head_t limbo_waitq;
+	wait_queue_head_t reset_waitq;
 	enum cxlflash_state state;
 };
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 6e85c77..8940336 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -382,8 +382,8 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
 
 	switch (cfg->state) {
-	case STATE_LIMBO:
-		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device in limbo!\n",
+	case STATE_RESET:
+		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device is in reset!\n",
 				    __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
@@ -479,8 +479,8 @@ static int cxlflash_eh_device_reset_handler(struct scsi_cmnd *scp)
 		if (unlikely(rcr))
 			rc = FAILED;
 		break;
-	case STATE_LIMBO:
-		wait_event(cfg->limbo_waitq, cfg->state != STATE_LIMBO);
+	case STATE_RESET:
+		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
 		if (cfg->state == STATE_NORMAL)
 			break;
 		/* fall through */
@@ -519,7 +519,7 @@ static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
 
 	switch (cfg->state) {
 	case STATE_NORMAL:
-		cfg->state = STATE_LIMBO;
+		cfg->state = STATE_RESET;
 		scsi_block_requests(cfg->host);
 		cxlflash_mark_contexts_error(cfg);
 		rcr = cxlflash_afu_reset(cfg);
@@ -528,11 +528,11 @@ static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
 			cfg->state = STATE_FAILTERM;
 		} else
 			cfg->state = STATE_NORMAL;
-		wake_up_all(&cfg->limbo_waitq);
+		wake_up_all(&cfg->reset_waitq);
 		scsi_unblock_requests(cfg->host);
 		break;
-	case STATE_LIMBO:
-		wait_event(cfg->limbo_waitq, cfg->state != STATE_LIMBO);
+	case STATE_RESET:
+		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
 		if (cfg->state == STATE_NORMAL)
 			break;
 		/* fall through */
@@ -705,7 +705,7 @@ static void cxlflash_wait_for_pci_err_recovery(struct cxlflash_cfg *cfg)
 	struct pci_dev *pdev = cfg->dev;
 
 	if (pci_channel_offline(pdev))
-		wait_event_timeout(cfg->limbo_waitq,
+		wait_event_timeout(cfg->reset_waitq,
 				   !pci_channel_offline(pdev),
 				   CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT);
 }
@@ -2304,7 +2304,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->mcctx = NULL;
 
 	init_waitqueue_head(&cfg->tmf_waitq);
-	init_waitqueue_head(&cfg->limbo_waitq);
+	init_waitqueue_head(&cfg->reset_waitq);
 
 	INIT_WORK(&cfg->work_q, cxlflash_worker_thread);
 	cfg->lr_state = LINK_RESET_INVALID;
@@ -2396,7 +2396,7 @@ static pci_ers_result_t cxlflash_pci_error_detected(struct pci_dev *pdev,
 
 	switch (state) {
 	case pci_channel_io_frozen:
-		cfg->state = STATE_LIMBO;
+		cfg->state = STATE_RESET;
 		scsi_block_requests(cfg->host);
 		drain_ioctls(cfg);
 		rc = cxlflash_mark_contexts_error(cfg);
@@ -2408,7 +2408,7 @@ static pci_ers_result_t cxlflash_pci_error_detected(struct pci_dev *pdev,
 		return PCI_ERS_RESULT_NEED_RESET;
 	case pci_channel_io_perm_failure:
 		cfg->state = STATE_FAILTERM;
-		wake_up_all(&cfg->limbo_waitq);
+		wake_up_all(&cfg->reset_waitq);
 		scsi_unblock_requests(cfg->host);
 		return PCI_ERS_RESULT_DISCONNECT;
 	default:
@@ -2455,7 +2455,7 @@ static void cxlflash_pci_resume(struct pci_dev *pdev)
 	dev_dbg(dev, "%s: pdev=%p\n", __func__, pdev);
 
 	cfg->state = STATE_NORMAL;
-	wake_up_all(&cfg->limbo_waitq);
+	wake_up_all(&cfg->reset_waitq);
 	scsi_unblock_requests(cfg->host);
 }
 
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 655cbf1..8a7ec5d 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -100,7 +100,7 @@ void cxlflash_stop_term_user_contexts(struct cxlflash_cfg *cfg)
 
 		dev_dbg(dev, "%s: Wait for user contexts to quiesce...\n",
 			__func__);
-		wake_up_all(&cfg->limbo_waitq);
+		wake_up_all(&cfg->reset_waitq);
 		ssleep(1);
 	}
 }
@@ -1233,11 +1233,11 @@ static int check_state(struct cxlflash_cfg *cfg)
 
 retry:
 	switch (cfg->state) {
-	case STATE_LIMBO:
-		dev_dbg(dev, "%s: Limbo state, going to wait...\n", __func__);
+	case STATE_RESET:
+		dev_dbg(dev, "%s: Reset state, going to wait...\n", __func__);
 		up_read(&cfg->ioctl_rwsem);
-		rc = wait_event_interruptible(cfg->limbo_waitq,
-					      cfg->state != STATE_LIMBO);
+		rc = wait_event_interruptible(cfg->reset_waitq,
+					      cfg->state != STATE_RESET);
 		down_read(&cfg->ioctl_rwsem);
 		if (unlikely(rc))
 			break;
@@ -1578,10 +1578,10 @@ err1:
  * quite possible for this routine to act as the kernel's EEH detection
  * source (MMIO read of mbox_r). Because of this, there is a window of
  * time where an EEH might have been detected but not yet 'serviced'
- * (callback invoked, causing the device to enter limbo state). To avoid
+ * (callback invoked, causing the device to enter reset state). To avoid
  * looping in this routine during that window, a 1 second sleep is in place
  * between the time the MMIO failure is detected and the time a wait on the
- * limbo wait queue is attempted via check_state().
+ * reset wait queue is attempted via check_state().
  *
  * Return: 0 on success, -errno on failure
  */
-- 
2.1.0

[PATCH v4 10/32] cxlflash: Make functions static

[Matthew R. Ochs]

Found during code inspection, that the following functions are not
being used outside of the file where they are defined. Make them static.

int cxlflash_send_cmd(struct afu *, struct afu_cmd *);
void cxlflash_wait_resp(struct afu *, struct afu_cmd *);
int cxlflash_afu_reset(struct cxlflash_cfg *);
struct afu_cmd *cxlflash_cmd_checkout(struct afu *);
void cxlflash_cmd_checkin(struct afu_cmd *);
void init_pcr(struct cxlflash_cfg *);
int init_global(struct cxlflash_cfg *);

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h |    5 -
 drivers/scsi/cxlflash/main.c   | 1018 ++++++++++++++++++++--------------------
 2 files changed, 509 insertions(+), 514 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 11318de..b038ac7 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -192,11 +192,6 @@ static inline u64 lun_to_lunid(u64 lun)
 	return swab64(lun_id);
 }
 
-int cxlflash_send_cmd(struct afu *, struct afu_cmd *);
-void cxlflash_wait_resp(struct afu *, struct afu_cmd *);
-int cxlflash_afu_reset(struct cxlflash_cfg *);
-struct afu_cmd *cxlflash_cmd_checkout(struct afu *);
-void cxlflash_cmd_checkin(struct afu_cmd *);
 int cxlflash_afu_sync(struct afu *, ctx_hndl_t, res_hndl_t, u8);
 void cxlflash_list_init(void);
 void cxlflash_term_global_luns(void);
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 8940336..226cefe 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -36,7 +36,7 @@ MODULE_LICENSE("GPL");
 
 
 /**
- * cxlflash_cmd_checkout() - checks out an AFU command
+ * cmd_checkout() - checks out an AFU command
  * @afu:	AFU to checkout from.
  *
  * Commands are checked out in a round-robin fashion. Note that since
@@ -47,7 +47,7 @@ MODULE_LICENSE("GPL");
  *
  * Return: The checked out command or NULL when command pool is empty.
  */
-struct afu_cmd *cxlflash_cmd_checkout(struct afu *afu)
+static struct afu_cmd *cmd_checkout(struct afu *afu)
 {
 	int k, dec = CXLFLASH_NUM_CMDS;
 	struct afu_cmd *cmd;
@@ -70,7 +70,7 @@ struct afu_cmd *cxlflash_cmd_checkout(struct afu *afu)
 }
 
 /**
- * cxlflash_cmd_checkin() - checks in an AFU command
+ * cmd_checkin() - checks in an AFU command
  * @cmd:	AFU command to checkin.
  *
  * Safe to pass commands that have already been checked in. Several
@@ -79,7 +79,7 @@ struct afu_cmd *cxlflash_cmd_checkout(struct afu *afu)
  * to avoid clobbering values in the event that the command is checked
  * out right away.
  */
-void cxlflash_cmd_checkin(struct afu_cmd *cmd)
+static void cmd_checkin(struct afu_cmd *cmd)
 {
 	cmd->rcb.scp = NULL;
 	cmd->rcb.timeout = 0;
@@ -238,7 +238,7 @@ static void cmd_complete(struct afu_cmd *cmd)
 
 		resid = cmd->sa.resid;
 		cmd_is_tmf = cmd->cmd_tmf;
-		cxlflash_cmd_checkin(cmd); /* Don't use cmd after here */
+		cmd_checkin(cmd); /* Don't use cmd after here */
 
 		pr_debug("%s: calling scsi_set_resid, scp=%p "
 			 "result=%X resid=%d\n", __func__,
@@ -260,6 +260,146 @@ static void cmd_complete(struct afu_cmd *cmd)
 }
 
 /**
+ * context_reset() - timeout handler for AFU commands
+ * @cmd:	AFU command that timed out.
+ *
+ * Sends a reset to the AFU.
+ */
+static void context_reset(struct afu_cmd *cmd)
+{
+	int nretry = 0;
+	u64 rrin = 0x1;
+	u64 room = 0;
+	struct afu *afu = cmd->parent;
+	ulong lock_flags;
+
+	pr_debug("%s: cmd=%p\n", __func__, cmd);
+
+	spin_lock_irqsave(&cmd->slock, lock_flags);
+
+	/* Already completed? */
+	if (cmd->sa.host_use_b[0] & B_DONE) {
+		spin_unlock_irqrestore(&cmd->slock, lock_flags);
+		return;
+	}
+
+	cmd->sa.host_use_b[0] |= (B_DONE | B_ERROR | B_TIMEOUT);
+	spin_unlock_irqrestore(&cmd->slock, lock_flags);
+
+	/*
+	 * We really want to send this reset at all costs, so spread
+	 * out wait time on successive retries for available room.
+	 */
+	do {
+		room = readq_be(&afu->host_map->cmd_room);
+		atomic64_set(&afu->room, room);
+		if (room)
+			goto write_rrin;
+		udelay(nretry);
+	} while (nretry++ < MC_ROOM_RETRY_CNT);
+
+	pr_err("%s: no cmd_room to send reset\n", __func__);
+	return;
+
+write_rrin:
+	nretry = 0;
+	writeq_be(rrin, &afu->host_map->ioarrin);
+	do {
+		rrin = readq_be(&afu->host_map->ioarrin);
+		if (rrin != 0x1)
+			break;
+		/* Double delay each time */
+		udelay(2 ^ nretry);
+	} while (nretry++ < MC_ROOM_RETRY_CNT);
+}
+
+/**
+ * send_cmd() - sends an AFU command
+ * @afu:	AFU associated with the host.
+ * @cmd:	AFU command to send.
+ *
+ * Return:
+ *	0 on success or SCSI_MLQUEUE_HOST_BUSY
+ */
+static int send_cmd(struct afu *afu, struct afu_cmd *cmd)
+{
+	struct cxlflash_cfg *cfg = afu->parent;
+	struct device *dev = &cfg->dev->dev;
+	int nretry = 0;
+	int rc = 0;
+	u64 room;
+	long newval;
+
+	/*
+	 * This routine is used by critical users such an AFU sync and to
+	 * send a task management function (TMF). Thus we want to retry a
+	 * bit before returning an error. To avoid the performance penalty
+	 * of MMIO, we spread the update of 'room' over multiple commands.
+	 */
+retry:
+	newval = atomic64_dec_if_positive(&afu->room);
+	if (!newval) {
+		do {
+			room = readq_be(&afu->host_map->cmd_room);
+			atomic64_set(&afu->room, room);
+			if (room)
+				goto write_ioarrin;
+			udelay(nretry);
+		} while (nretry++ < MC_ROOM_RETRY_CNT);
+
+		dev_err(dev, "%s: no cmd_room to send 0x%X\n",
+		       __func__, cmd->rcb.cdb[0]);
+
+		goto no_room;
+	} else if (unlikely(newval < 0)) {
+		/* This should be rare. i.e. Only if two threads race and
+		 * decrement before the MMIO read is done. In this case
+		 * just benefit from the other thread having updated
+		 * afu->room.
+		 */
+		if (nretry++ < MC_ROOM_RETRY_CNT) {
+			udelay(nretry);
+			goto retry;
+		}
+
+		goto no_room;
+	}
+
+write_ioarrin:
+	writeq_be((u64)&cmd->rcb, &afu->host_map->ioarrin);
+out:
+	pr_devel("%s: cmd=%p len=%d ea=%p rc=%d\n", __func__, cmd,
+		 cmd->rcb.data_len, (void *)cmd->rcb.data_ea, rc);
+	return rc;
+
+no_room:
+	afu->read_room = true;
+	schedule_work(&cfg->work_q);
+	rc = SCSI_MLQUEUE_HOST_BUSY;
+	goto out;
+}
+
+/**
+ * wait_resp() - polls for a response or timeout to a sent AFU command
+ * @afu:	AFU associated with the host.
+ * @cmd:	AFU command that was sent.
+ */
+static void wait_resp(struct afu *afu, struct afu_cmd *cmd)
+{
+	ulong timeout = msecs_to_jiffies(cmd->rcb.timeout * 2 * 1000);
+
+	timeout = wait_for_completion_timeout(&cmd->cevent, timeout);
+	if (!timeout)
+		context_reset(cmd);
+
+	if (unlikely(cmd->sa.ioasc != 0))
+		pr_err("%s: CMD 0x%X failed, IOASC: flags 0x%X, afu_rc 0x%X, "
+		       "scsi_rc 0x%X, fc_rc 0x%X\n", __func__, cmd->rcb.cdb[0],
+		       cmd->sa.rc.flags, cmd->sa.rc.afu_rc, cmd->sa.rc.scsi_rc,
+		       cmd->sa.rc.fc_rc);
+}
+
+/**
  * send_tmf() - sends a Task Management Function (TMF)
  * @afu:	AFU to checkout from.
  * @scp:	SCSI command from stack.
@@ -280,7 +420,7 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 	ulong lock_flags;
 	int rc = 0;
 
-	cmd = cxlflash_cmd_checkout(afu);
+	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
 		pr_err("%s: could not get a free command\n", __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
@@ -313,9 +453,9 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 	memcpy(cmd->rcb.cdb, &tmfcmd, sizeof(tmfcmd));
 
 	/* Send the command */
-	rc = cxlflash_send_cmd(afu, cmd);
+	rc = send_cmd(afu, cmd);
 	if (unlikely(rc)) {
-		cxlflash_cmd_checkin(cmd);
+		cmd_checkin(cmd);
 		spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
 		cfg->tmf_active = false;
 		spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
@@ -398,7 +538,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 		break;
 	}
 
-	cmd = cxlflash_cmd_checkout(afu);
+	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
 		pr_err("%s: could not get a free command\n", __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
@@ -438,9 +578,9 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	memcpy(cmd->rcb.cdb, scp->cmnd, sizeof(cmd->rcb.cdb));
 
 	/* Send the command */
-	rc = cxlflash_send_cmd(afu, cmd);
+	rc = send_cmd(afu, cmd);
 	if (unlikely(rc)) {
-		cxlflash_cmd_checkin(cmd);
+		cmd_checkin(cmd);
 		scsi_dma_unmap(scp);
 	}
 
@@ -449,369 +589,55 @@ out:
 }
 
 /**
- * cxlflash_eh_device_reset_handler() - reset a single LUN
- * @scp:	SCSI command to send.
- *
- * Return:
- *	SUCCESS as defined in scsi/scsi.h
- *	FAILED as defined in scsi/scsi.h
+ * cxlflash_wait_for_pci_err_recovery() - wait for error recovery during probe
+ * @cxlflash:	Internal structure associated with the host.
  */
-static int cxlflash_eh_device_reset_handler(struct scsi_cmnd *scp)
+static void cxlflash_wait_for_pci_err_recovery(struct cxlflash_cfg *cfg)
 {
-	int rc = SUCCESS;
-	struct Scsi_Host *host = scp->device->host;
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
-	struct afu *afu = cfg->afu;
-	int rcr = 0;
-
-	pr_debug("%s: (scp=%p) %d/%d/%d/%llu "
-		 "cdb=(%08X-%08X-%08X-%08X)\n", __func__, scp,
-		 host->host_no, scp->device->channel,
-		 scp->device->id, scp->device->lun,
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
-
-	switch (cfg->state) {
-	case STATE_NORMAL:
-		rcr = send_tmf(afu, scp, TMF_LUN_RESET);
-		if (unlikely(rcr))
-			rc = FAILED;
-		break;
-	case STATE_RESET:
-		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
-		if (cfg->state == STATE_NORMAL)
-			break;
-		/* fall through */
-	default:
-		rc = FAILED;
-		break;
-	}
+	struct pci_dev *pdev = cfg->dev;
 
-	pr_debug("%s: returning rc=%d\n", __func__, rc);
-	return rc;
+	if (pci_channel_offline(pdev))
+		wait_event_timeout(cfg->reset_waitq,
+				   !pci_channel_offline(pdev),
+				   CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT);
 }
 
 /**
- * cxlflash_eh_host_reset_handler() - reset the host adapter
- * @scp:	SCSI command from stack identifying host.
- *
- * Return:
- *	SUCCESS as defined in scsi/scsi.h
- *	FAILED as defined in scsi/scsi.h
+ * free_mem() - free memory associated with the AFU
+ * @cxlflash:	Internal structure associated with the host.
  */
-static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
+static void free_mem(struct cxlflash_cfg *cfg)
 {
-	int rc = SUCCESS;
-	int rcr = 0;
-	struct Scsi_Host *host = scp->device->host;
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
+	int i;
+	char *buf = NULL;
+	struct afu *afu = cfg->afu;
 
-	pr_debug("%s: (scp=%p) %d/%d/%d/%llu "
-		 "cdb=(%08X-%08X-%08X-%08X)\n", __func__, scp,
-		 host->host_no, scp->device->channel,
-		 scp->device->id, scp->device->lun,
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
+	if (cfg->afu) {
+		for (i = 0; i < CXLFLASH_NUM_CMDS; i++) {
+			buf = afu->cmd[i].buf;
+			if (!((u64)buf & (PAGE_SIZE - 1)))
+				free_page((ulong)buf);
+		}
 
-	switch (cfg->state) {
-	case STATE_NORMAL:
-		cfg->state = STATE_RESET;
-		scsi_block_requests(cfg->host);
-		cxlflash_mark_contexts_error(cfg);
-		rcr = cxlflash_afu_reset(cfg);
-		if (rcr) {
-			rc = FAILED;
-			cfg->state = STATE_FAILTERM;
-		} else
-			cfg->state = STATE_NORMAL;
-		wake_up_all(&cfg->reset_waitq);
-		scsi_unblock_requests(cfg->host);
-		break;
-	case STATE_RESET:
-		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
-		if (cfg->state == STATE_NORMAL)
-			break;
-		/* fall through */
-	default:
-		rc = FAILED;
-		break;
+		free_pages((ulong)afu, get_order(sizeof(struct afu)));
+		cfg->afu = NULL;
 	}
-
-	pr_debug("%s: returning rc=%d\n", __func__, rc);
-	return rc;
 }
 
 /**
- * cxlflash_change_queue_depth() - change the queue depth for the device
- * @sdev:	SCSI device destined for queue depth change.
- * @qdepth:	Requested queue depth value to set.
- *
- * The requested queue depth is capped to the maximum supported value.
+ * stop_afu() - stops the AFU command timers and unmaps the MMIO space
+ * @cxlflash:	Internal structure associated with the host.
  *
- * Return: The actual queue depth set.
+ * Safe to call with AFU in a partially allocated/initialized state.
  */
-static int cxlflash_change_queue_depth(struct scsi_device *sdev, int qdepth)
+static void stop_afu(struct cxlflash_cfg *cfg)
 {
+	int i;
+	struct afu *afu = cfg->afu;
 
-	if (qdepth > CXLFLASH_MAX_CMDS_PER_LUN)
-		qdepth = CXLFLASH_MAX_CMDS_PER_LUN;
-
-	scsi_change_queue_depth(sdev, qdepth);
-	return sdev->queue_depth;
-}
-
-/**
- * cxlflash_show_port_status() - queries and presents the current port status
- * @dev:	Generic device associated with the host owning the port.
- * @attr:	Device attribute representing the port.
- * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
- *
- * Return: The size of the ASCII string returned in @buf.
- */
-static ssize_t cxlflash_show_port_status(struct device *dev,
-					 struct device_attribute *attr,
-					 char *buf)
-{
-	struct Scsi_Host *shost = class_to_shost(dev);
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
-	struct afu *afu = cfg->afu;
-
-	char *disp_status;
-	int rc;
-	u32 port;
-	u64 status;
-	u64 *fc_regs;
-
-	rc = kstrtouint((attr->attr.name + 4), 10, &port);
-	if (rc || (port >= NUM_FC_PORTS))
-		return 0;
-
-	fc_regs = &afu->afu_map->global.fc_regs[port][0];
-	status =
-	    (readq_be(&fc_regs[FC_MTIP_STATUS / 8]) & FC_MTIP_STATUS_MASK);
-
-	if (status == FC_MTIP_STATUS_ONLINE)
-		disp_status = "online";
-	else if (status == FC_MTIP_STATUS_OFFLINE)
-		disp_status = "offline";
-	else
-		disp_status = "unknown";
-
-	return snprintf(buf, PAGE_SIZE, "%s\n", disp_status);
-}
-
-/**
- * cxlflash_show_lun_mode() - presents the current LUN mode of the host
- * @dev:	Generic device associated with the host.
- * @attr:	Device attribute representing the lun mode.
- * @buf:	Buffer of length PAGE_SIZE to report back the LUN mode in ASCII.
- *
- * Return: The size of the ASCII string returned in @buf.
- */
-static ssize_t cxlflash_show_lun_mode(struct device *dev,
-				      struct device_attribute *attr, char *buf)
-{
-	struct Scsi_Host *shost = class_to_shost(dev);
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
-	struct afu *afu = cfg->afu;
-
-	return snprintf(buf, PAGE_SIZE, "%u\n", afu->internal_lun);
-}
-
-/**
- * cxlflash_store_lun_mode() - sets the LUN mode of the host
- * @dev:	Generic device associated with the host.
- * @attr:	Device attribute representing the lun mode.
- * @buf:	Buffer of length PAGE_SIZE containing the LUN mode in ASCII.
- * @count:	Length of data resizing in @buf.
- *
- * The CXL Flash AFU supports a dummy LUN mode where the external
- * links and storage are not required. Space on the FPGA is used
- * to create 1 or 2 small LUNs which are presented to the system
- * as if they were a normal storage device. This feature is useful
- * during development and also provides manufacturing with a way
- * to test the AFU without an actual device.
- *
- * 0 = external LUN[s] (default)
- * 1 = internal LUN (1 x 64K, 512B blocks, id 0)
- * 2 = internal LUN (1 x 64K, 4K blocks, id 0)
- * 3 = internal LUN (2 x 32K, 512B blocks, ids 0,1)
- * 4 = internal LUN (2 x 32K, 4K blocks, ids 0,1)
- *
- * Return: The size of the ASCII string returned in @buf.
- */
-static ssize_t cxlflash_store_lun_mode(struct device *dev,
-				       struct device_attribute *attr,
-				       const char *buf, size_t count)
-{
-	struct Scsi_Host *shost = class_to_shost(dev);
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
-	struct afu *afu = cfg->afu;
-	int rc;
-	u32 lun_mode;
-
-	rc = kstrtouint(buf, 10, &lun_mode);
-	if (!rc && (lun_mode < 5) && (lun_mode != afu->internal_lun)) {
-		afu->internal_lun = lun_mode;
-		cxlflash_afu_reset(cfg);
-		scsi_scan_host(cfg->host);
-	}
-
-	return count;
-}
-
-/**
- * cxlflash_show_ioctl_version() - presents the current ioctl version of the host
- * @dev:	Generic device associated with the host.
- * @attr:	Device attribute representing the ioctl version.
- * @buf:	Buffer of length PAGE_SIZE to report back the ioctl version.
- *
- * Return: The size of the ASCII string returned in @buf.
- */
-static ssize_t cxlflash_show_ioctl_version(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
-{
-	return scnprintf(buf, PAGE_SIZE, "%u\n", DK_CXLFLASH_VERSION_0);
-}
-
-/**
- * cxlflash_show_dev_mode() - presents the current mode of the device
- * @dev:	Generic device associated with the device.
- * @attr:	Device attribute representing the device mode.
- * @buf:	Buffer of length PAGE_SIZE to report back the dev mode in ASCII.
- *
- * Return: The size of the ASCII string returned in @buf.
- */
-static ssize_t cxlflash_show_dev_mode(struct device *dev,
-				      struct device_attribute *attr, char *buf)
-{
-	struct scsi_device *sdev = to_scsi_device(dev);
-
-	return snprintf(buf, PAGE_SIZE, "%s\n",
-			sdev->hostdata ? "superpipe" : "legacy");
-}
-
-/**
- * cxlflash_wait_for_pci_err_recovery() - wait for error recovery during probe
- * @cxlflash:	Internal structure associated with the host.
- */
-static void cxlflash_wait_for_pci_err_recovery(struct cxlflash_cfg *cfg)
-{
-	struct pci_dev *pdev = cfg->dev;
-
-	if (pci_channel_offline(pdev))
-		wait_event_timeout(cfg->reset_waitq,
-				   !pci_channel_offline(pdev),
-				   CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT);
-}
-
-/*
- * Host attributes
- */
-static DEVICE_ATTR(port0, S_IRUGO, cxlflash_show_port_status, NULL);
-static DEVICE_ATTR(port1, S_IRUGO, cxlflash_show_port_status, NULL);
-static DEVICE_ATTR(lun_mode, S_IRUGO | S_IWUSR, cxlflash_show_lun_mode,
-		   cxlflash_store_lun_mode);
-static DEVICE_ATTR(ioctl_version, S_IRUGO, cxlflash_show_ioctl_version, NULL);
-
-static struct device_attribute *cxlflash_host_attrs[] = {
-	&dev_attr_port0,
-	&dev_attr_port1,
-	&dev_attr_lun_mode,
-	&dev_attr_ioctl_version,
-	NULL
-};
-
-/*
- * Device attributes
- */
-static DEVICE_ATTR(mode, S_IRUGO, cxlflash_show_dev_mode, NULL);
-
-static struct device_attribute *cxlflash_dev_attrs[] = {
-	&dev_attr_mode,
-	NULL
-};
-
-/*
- * Host template
- */
-static struct scsi_host_template driver_template = {
-	.module = THIS_MODULE,
-	.name = CXLFLASH_ADAPTER_NAME,
-	.info = cxlflash_driver_info,
-	.ioctl = cxlflash_ioctl,
-	.proc_name = CXLFLASH_NAME,
-	.queuecommand = cxlflash_queuecommand,
-	.eh_device_reset_handler = cxlflash_eh_device_reset_handler,
-	.eh_host_reset_handler = cxlflash_eh_host_reset_handler,
-	.change_queue_depth = cxlflash_change_queue_depth,
-	.cmd_per_lun = 16,
-	.can_queue = CXLFLASH_MAX_CMDS,
-	.this_id = -1,
-	.sg_tablesize = SG_NONE,	/* No scatter gather support. */
-	.max_sectors = CXLFLASH_MAX_SECTORS,
-	.use_clustering = ENABLE_CLUSTERING,
-	.shost_attrs = cxlflash_host_attrs,
-	.sdev_attrs = cxlflash_dev_attrs,
-};
-
-/*
- * Device dependent values
- */
-static struct dev_dependent_vals dev_corsa_vals = { CXLFLASH_MAX_SECTORS };
-
-/*
- * PCI device binding table
- */
-static struct pci_device_id cxlflash_pci_table[] = {
-	{PCI_VENDOR_ID_IBM, PCI_DEVICE_ID_IBM_CORSA,
-	 PCI_ANY_ID, PCI_ANY_ID, 0, 0, (kernel_ulong_t)&dev_corsa_vals},
-	{}
-};
-
-MODULE_DEVICE_TABLE(pci, cxlflash_pci_table);
-
-/**
- * free_mem() - free memory associated with the AFU
- * @cxlflash:	Internal structure associated with the host.
- */
-static void free_mem(struct cxlflash_cfg *cfg)
-{
-	int i;
-	char *buf = NULL;
-	struct afu *afu = cfg->afu;
-
-	if (cfg->afu) {
-		for (i = 0; i < CXLFLASH_NUM_CMDS; i++) {
-			buf = afu->cmd[i].buf;
-			if (!((u64)buf & (PAGE_SIZE - 1)))
-				free_page((ulong)buf);
-		}
-
-		free_pages((ulong)afu, get_order(sizeof(struct afu)));
-		cfg->afu = NULL;
-	}
-}
-
-/**
- * stop_afu() - stops the AFU command timers and unmaps the MMIO space
- * @cxlflash:	Internal structure associated with the host.
- *
- * Safe to call with AFU in a partially allocated/initialized state.
- */
-static void stop_afu(struct cxlflash_cfg *cfg)
-{
-	int i;
-	struct afu *afu = cfg->afu;
-
-	if (likely(afu)) {
-		for (i = 0; i < CXLFLASH_NUM_CMDS; i++)
-			complete(&afu->cmd[i].cevent);
+	if (likely(afu)) {
+		for (i = 0; i < CXLFLASH_NUM_CMDS; i++)
+			complete(&afu->cmd[i].cevent);
 
 		if (likely(afu->afu_map)) {
 			cxl_psa_unmap((void *)afu->afu_map);
@@ -1631,67 +1457,13 @@ out:
 }
 
 /**
- * cxlflash_context_reset() - timeout handler for AFU commands
- * @cmd:	AFU command that timed out.
+ * init_pcr() - initialize the provisioning and control registers
+ * @cxlflash:	Internal structure associated with the host.
  *
- * Sends a reset to the AFU.
+ * Also sets up fast access to the mapped registers and initializes AFU
+ * command fields that never change.
  */
-void cxlflash_context_reset(struct afu_cmd *cmd)
-{
-	int nretry = 0;
-	u64 rrin = 0x1;
-	u64 room = 0;
-	struct afu *afu = cmd->parent;
-	ulong lock_flags;
-
-	pr_debug("%s: cmd=%p\n", __func__, cmd);
-
-	spin_lock_irqsave(&cmd->slock, lock_flags);
-
-	/* Already completed? */
-	if (cmd->sa.host_use_b[0] & B_DONE) {
-		spin_unlock_irqrestore(&cmd->slock, lock_flags);
-		return;
-	}
-
-	cmd->sa.host_use_b[0] |= (B_DONE | B_ERROR | B_TIMEOUT);
-	spin_unlock_irqrestore(&cmd->slock, lock_flags);
-
-	/*
-	 * We really want to send this reset at all costs, so spread
-	 * out wait time on successive retries for available room.
-	 */
-	do {
-		room = readq_be(&afu->host_map->cmd_room);
-		atomic64_set(&afu->room, room);
-		if (room)
-			goto write_rrin;
-		udelay(nretry);
-	} while (nretry++ < MC_ROOM_RETRY_CNT);
-
-	pr_err("%s: no cmd_room to send reset\n", __func__);
-	return;
-
-write_rrin:
-	nretry = 0;
-	writeq_be(rrin, &afu->host_map->ioarrin);
-	do {
-		rrin = readq_be(&afu->host_map->ioarrin);
-		if (rrin != 0x1)
-			break;
-		/* Double delay each time */
-		udelay(2 ^ nretry);
-	} while (nretry++ < MC_ROOM_RETRY_CNT);
-}
-
-/**
- * init_pcr() - initialize the provisioning and control registers
- * @cxlflash:	Internal structure associated with the host.
- *
- * Also sets up fast access to the mapped registers and initializes AFU
- * command fields that never change.
- */
-void init_pcr(struct cxlflash_cfg *cfg)
+static void init_pcr(struct cxlflash_cfg *cfg)
 {
 	struct afu *afu = cfg->afu;
 	struct sisl_ctrl_map *ctrl_map;
@@ -1727,7 +1499,7 @@ void init_pcr(struct cxlflash_cfg *cfg)
  * init_global() - initialize AFU global registers
  * @cxlflash:	Internal structure associated with the host.
  */
-int init_global(struct cxlflash_cfg *cfg)
+static int init_global(struct cxlflash_cfg *cfg)
 {
 	struct afu *afu = cfg->afu;
 	u64 wwpn[NUM_FC_PORTS];	/* wwpn of AFU ports */
@@ -1998,92 +1770,6 @@ err1:
 }
 
 /**
- * cxlflash_send_cmd() - sends an AFU command
- * @afu:	AFU associated with the host.
- * @cmd:	AFU command to send.
- *
- * Return:
- *	0 on success
- *	-1 on failure
- */
-int cxlflash_send_cmd(struct afu *afu, struct afu_cmd *cmd)
-{
-	struct cxlflash_cfg *cfg = afu->parent;
-	int nretry = 0;
-	int rc = 0;
-	u64 room;
-	long newval;
-
-	/*
-	 * This routine is used by critical users such an AFU sync and to
-	 * send a task management function (TMF). Thus we want to retry a
-	 * bit before returning an error. To avoid the performance penalty
-	 * of MMIO, we spread the update of 'room' over multiple commands.
-	 */
-retry:
-	newval = atomic64_dec_if_positive(&afu->room);
-	if (!newval) {
-		do {
-			room = readq_be(&afu->host_map->cmd_room);
-			atomic64_set(&afu->room, room);
-			if (room)
-				goto write_ioarrin;
-			udelay(nretry);
-		} while (nretry++ < MC_ROOM_RETRY_CNT);
-
-		pr_err("%s: no cmd_room to send 0x%X\n",
-		       __func__, cmd->rcb.cdb[0]);
-
-		goto no_room;
-	} else if (unlikely(newval < 0)) {
-		/* This should be rare. i.e. Only if two threads race and
-		 * decrement before the MMIO read is done. In this case
-		 * just benefit from the other thread having updated
-		 * afu->room.
-		 */
-		if (nretry++ < MC_ROOM_RETRY_CNT) {
-			udelay(nretry);
-			goto retry;
-		}
-
-		goto no_room;
-	}
-
-write_ioarrin:
-	writeq_be((u64)&cmd->rcb, &afu->host_map->ioarrin);
-out:
-	pr_debug("%s: cmd=%p len=%d ea=%p rc=%d\n", __func__, cmd,
-		 cmd->rcb.data_len, (void *)cmd->rcb.data_ea, rc);
-	return rc;
-
-no_room:
-	afu->read_room = true;
-	schedule_work(&cfg->work_q);
-	rc = SCSI_MLQUEUE_HOST_BUSY;
-	goto out;
-}
-
-/**
- * cxlflash_wait_resp() - polls for a response or timeout to a sent AFU command
- * @afu:	AFU associated with the host.
- * @cmd:	AFU command that was sent.
- */
-void cxlflash_wait_resp(struct afu *afu, struct afu_cmd *cmd)
-{
-	ulong timeout = jiffies + (cmd->rcb.timeout * 2 * HZ);
-
-	timeout = wait_for_completion_timeout(&cmd->cevent, timeout);
-	if (!timeout)
-		cxlflash_context_reset(cmd);
-
-	if (unlikely(cmd->sa.ioasc != 0))
-		pr_err("%s: CMD 0x%X failed, IOASC: flags 0x%X, afu_rc 0x%X, "
-		       "scsi_rc 0x%X, fc_rc 0x%X\n", __func__, cmd->rcb.cdb[0],
-		       cmd->sa.rc.flags, cmd->sa.rc.afu_rc, cmd->sa.rc.scsi_rc,
-		       cmd->sa.rc.fc_rc);
-}
-
-/**
  * cxlflash_afu_sync() - builds and sends an AFU sync command
  * @afu:	AFU associated with the host.
  * @ctx_hndl_u:	Identifies context requesting sync.
@@ -2121,7 +1807,7 @@ int cxlflash_afu_sync(struct afu *afu, ctx_hndl_t ctx_hndl_u,
 
 	mutex_lock(&sync_active);
 retry:
-	cmd = cxlflash_cmd_checkout(afu);
+	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
 		retry_cnt++;
 		udelay(1000 * retry_cnt);
@@ -2150,11 +1836,11 @@ retry:
 	*((u16 *)&cmd->rcb.cdb[2]) = swab16(ctx_hndl_u);
 	*((u32 *)&cmd->rcb.cdb[4]) = swab32(res_hndl_u);
 
-	rc = cxlflash_send_cmd(afu, cmd);
+	rc = send_cmd(afu, cmd);
 	if (unlikely(rc))
 		goto out;
 
-	cxlflash_wait_resp(afu, cmd);
+	wait_resp(afu, cmd);
 
 	/* set on timeout */
 	if (unlikely((cmd->sa.ioasc != 0) ||
@@ -2163,20 +1849,20 @@ retry:
 out:
 	mutex_unlock(&sync_active);
 	if (cmd)
-		cxlflash_cmd_checkin(cmd);
+		cmd_checkin(cmd);
 	pr_debug("%s: returning rc=%d\n", __func__, rc);
 	return rc;
 }
 
 /**
- * cxlflash_afu_reset() - resets the AFU
- * @cxlflash:	Internal structure associated with the host.
+ * afu_reset() - resets the AFU
+ * @cfg:	Internal structure associated with the host.
  *
  * Return:
  *	0 on success
  *	A failure value from internal services.
  */
-int cxlflash_afu_reset(struct cxlflash_cfg *cfg)
+static int afu_reset(struct cxlflash_cfg *cfg)
 {
 	int rc = 0;
 	/* Stop the context before the reset. Since the context is
@@ -2192,6 +1878,320 @@ int cxlflash_afu_reset(struct cxlflash_cfg *cfg)
 }
 
 /**
+ * cxlflash_eh_device_reset_handler() - reset a single LUN
+ * @scp:	SCSI command to send.
+ *
+ * Return:
+ *	SUCCESS as defined in scsi/scsi.h
+ *	FAILED as defined in scsi/scsi.h
+ */
+static int cxlflash_eh_device_reset_handler(struct scsi_cmnd *scp)
+{
+	int rc = SUCCESS;
+	struct Scsi_Host *host = scp->device->host;
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
+	struct afu *afu = cfg->afu;
+	int rcr = 0;
+
+	pr_debug("%s: (scp=%p) %d/%d/%d/%llu "
+		 "cdb=(%08X-%08X-%08X-%08X)\n", __func__, scp,
+		 host->host_no, scp->device->channel,
+		 scp->device->id, scp->device->lun,
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
+
+	switch (cfg->state) {
+	case STATE_NORMAL:
+		rcr = send_tmf(afu, scp, TMF_LUN_RESET);
+		if (unlikely(rcr))
+			rc = FAILED;
+		break;
+	case STATE_RESET:
+		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
+		if (cfg->state == STATE_NORMAL)
+			break;
+		/* fall through */
+	default:
+		rc = FAILED;
+		break;
+	}
+
+	pr_debug("%s: returning rc=%d\n", __func__, rc);
+	return rc;
+}
+
+/**
+ * cxlflash_eh_host_reset_handler() - reset the host adapter
+ * @scp:	SCSI command from stack identifying host.
+ *
+ * Return:
+ *	SUCCESS as defined in scsi/scsi.h
+ *	FAILED as defined in scsi/scsi.h
+ */
+static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
+{
+	int rc = SUCCESS;
+	int rcr = 0;
+	struct Scsi_Host *host = scp->device->host;
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
+
+	pr_debug("%s: (scp=%p) %d/%d/%d/%llu "
+		 "cdb=(%08X-%08X-%08X-%08X)\n", __func__, scp,
+		 host->host_no, scp->device->channel,
+		 scp->device->id, scp->device->lun,
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
+		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
+
+	switch (cfg->state) {
+	case STATE_NORMAL:
+		cfg->state = STATE_RESET;
+		scsi_block_requests(cfg->host);
+		cxlflash_mark_contexts_error(cfg);
+		rcr = afu_reset(cfg);
+		if (rcr) {
+			rc = FAILED;
+			cfg->state = STATE_FAILTERM;
+		} else
+			cfg->state = STATE_NORMAL;
+		wake_up_all(&cfg->reset_waitq);
+		scsi_unblock_requests(cfg->host);
+		break;
+	case STATE_RESET:
+		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
+		if (cfg->state == STATE_NORMAL)
+			break;
+		/* fall through */
+	default:
+		rc = FAILED;
+		break;
+	}
+
+	pr_debug("%s: returning rc=%d\n", __func__, rc);
+	return rc;
+}
+
+/**
+ * cxlflash_change_queue_depth() - change the queue depth for the device
+ * @sdev:	SCSI device destined for queue depth change.
+ * @qdepth:	Requested queue depth value to set.
+ *
+ * The requested queue depth is capped to the maximum supported value.
+ *
+ * Return: The actual queue depth set.
+ */
+static int cxlflash_change_queue_depth(struct scsi_device *sdev, int qdepth)
+{
+
+	if (qdepth > CXLFLASH_MAX_CMDS_PER_LUN)
+		qdepth = CXLFLASH_MAX_CMDS_PER_LUN;
+
+	scsi_change_queue_depth(sdev, qdepth);
+	return sdev->queue_depth;
+}
+
+/**
+ * cxlflash_show_port_status() - queries and presents the current port status
+ * @dev:	Generic device associated with the host owning the port.
+ * @attr:	Device attribute representing the port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_show_port_status(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	char *disp_status;
+	int rc;
+	u32 port;
+	u64 status;
+	u64 *fc_regs;
+
+	rc = kstrtouint((attr->attr.name + 4), 10, &port);
+	if (rc || (port >= NUM_FC_PORTS))
+		return 0;
+
+	fc_regs = &afu->afu_map->global.fc_regs[port][0];
+	status =
+	    (readq_be(&fc_regs[FC_MTIP_STATUS / 8]) & FC_MTIP_STATUS_MASK);
+
+	if (status == FC_MTIP_STATUS_ONLINE)
+		disp_status = "online";
+	else if (status == FC_MTIP_STATUS_OFFLINE)
+		disp_status = "offline";
+	else
+		disp_status = "unknown";
+
+	return snprintf(buf, PAGE_SIZE, "%s\n", disp_status);
+}
+
+/**
+ * cxlflash_show_lun_mode() - presents the current LUN mode of the host
+ * @dev:	Generic device associated with the host.
+ * @attr:	Device attribute representing the lun mode.
+ * @buf:	Buffer of length PAGE_SIZE to report back the LUN mode in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_show_lun_mode(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	return snprintf(buf, PAGE_SIZE, "%u\n", afu->internal_lun);
+}
+
+/**
+ * cxlflash_store_lun_mode() - sets the LUN mode of the host
+ * @dev:	Generic device associated with the host.
+ * @attr:	Device attribute representing the lun mode.
+ * @buf:	Buffer of length PAGE_SIZE containing the LUN mode in ASCII.
+ * @count:	Length of data resizing in @buf.
+ *
+ * The CXL Flash AFU supports a dummy LUN mode where the external
+ * links and storage are not required. Space on the FPGA is used
+ * to create 1 or 2 small LUNs which are presented to the system
+ * as if they were a normal storage device. This feature is useful
+ * during development and also provides manufacturing with a way
+ * to test the AFU without an actual device.
+ *
+ * 0 = external LUN[s] (default)
+ * 1 = internal LUN (1 x 64K, 512B blocks, id 0)
+ * 2 = internal LUN (1 x 64K, 4K blocks, id 0)
+ * 3 = internal LUN (2 x 32K, 512B blocks, ids 0,1)
+ * 4 = internal LUN (2 x 32K, 4K blocks, ids 0,1)
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_store_lun_mode(struct device *dev,
+				       struct device_attribute *attr,
+				       const char *buf, size_t count)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+	int rc;
+	u32 lun_mode;
+
+	rc = kstrtouint(buf, 10, &lun_mode);
+	if (!rc && (lun_mode < 5) && (lun_mode != afu->internal_lun)) {
+		afu->internal_lun = lun_mode;
+		afu_reset(cfg);
+		scsi_scan_host(cfg->host);
+	}
+
+	return count;
+}
+
+/**
+ * cxlflash_show_ioctl_version() - presents the hosts current ioctl version
+ * @dev:	Generic device associated with the host.
+ * @attr:	Device attribute representing the ioctl version.
+ * @buf:	Buffer of length PAGE_SIZE to report back the ioctl version.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_show_ioctl_version(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	return scnprintf(buf, PAGE_SIZE, "%u\n", DK_CXLFLASH_VERSION_0);
+}
+
+/**
+ * cxlflash_show_dev_mode() - presents the current mode of the device
+ * @dev:	Generic device associated with the device.
+ * @attr:	Device attribute representing the device mode.
+ * @buf:	Buffer of length PAGE_SIZE to report back the dev mode in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_show_dev_mode(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct scsi_device *sdev = to_scsi_device(dev);
+
+	return snprintf(buf, PAGE_SIZE, "%s\n",
+			sdev->hostdata ? "superpipe" : "legacy");
+}
+
+/*
+ * Host attributes
+ */
+static DEVICE_ATTR(port0, S_IRUGO, cxlflash_show_port_status, NULL);
+static DEVICE_ATTR(port1, S_IRUGO, cxlflash_show_port_status, NULL);
+static DEVICE_ATTR(lun_mode, S_IRUGO | S_IWUSR, cxlflash_show_lun_mode,
+		   cxlflash_store_lun_mode);
+static DEVICE_ATTR(ioctl_version, S_IRUGO, cxlflash_show_ioctl_version, NULL);
+
+static struct device_attribute *cxlflash_host_attrs[] = {
+	&dev_attr_port0,
+	&dev_attr_port1,
+	&dev_attr_lun_mode,
+	&dev_attr_ioctl_version,
+	NULL
+};
+
+/*
+ * Device attributes
+ */
+static DEVICE_ATTR(mode, S_IRUGO, cxlflash_show_dev_mode, NULL);
+
+static struct device_attribute *cxlflash_dev_attrs[] = {
+	&dev_attr_mode,
+	NULL
+};
+
+/*
+ * Host template
+ */
+static struct scsi_host_template driver_template = {
+	.module = THIS_MODULE,
+	.name = CXLFLASH_ADAPTER_NAME,
+	.info = cxlflash_driver_info,
+	.ioctl = cxlflash_ioctl,
+	.proc_name = CXLFLASH_NAME,
+	.queuecommand = cxlflash_queuecommand,
+	.eh_device_reset_handler = cxlflash_eh_device_reset_handler,
+	.eh_host_reset_handler = cxlflash_eh_host_reset_handler,
+	.change_queue_depth = cxlflash_change_queue_depth,
+	.cmd_per_lun = 16,
+	.can_queue = CXLFLASH_MAX_CMDS,
+	.this_id = -1,
+	.sg_tablesize = SG_NONE,	/* No scatter gather support. */
+	.max_sectors = CXLFLASH_MAX_SECTORS,
+	.use_clustering = ENABLE_CLUSTERING,
+	.shost_attrs = cxlflash_host_attrs,
+	.sdev_attrs = cxlflash_dev_attrs,
+};
+
+/*
+ * Device dependent values
+ */
+static struct dev_dependent_vals dev_corsa_vals = { CXLFLASH_MAX_SECTORS };
+
+/*
+ * PCI device binding table
+ */
+static struct pci_device_id cxlflash_pci_table[] = {
+	{PCI_VENDOR_ID_IBM, PCI_DEVICE_ID_IBM_CORSA,
+	 PCI_ANY_ID, PCI_ANY_ID, 0, 0, (kernel_ulong_t)&dev_corsa_vals},
+	{}
+};
+
+MODULE_DEVICE_TABLE(pci, cxlflash_pci_table);
+
+/**
  * cxlflash_worker_thread() - work thread handler for the AFU
  * @work:	Work structure contained within cxlflash associated with host.
  *
-- 
2.1.0

[PATCH v4 11/32] cxlflash: Refine host/device attributes

[Matthew R. Ochs]

Implement the following suggestions and add two new attributes
to allow for debugging the port LUN table.

 - use scnprintf() instead of snprintf()
 - use DEVICE_ATTR_RO and DEVICE_ATTR_RW

Suggested-by: Shane Seymour <shane.seymour@hp.com>
Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 180 +++++++++++++++++++++++++++++++++----------
 1 file changed, 138 insertions(+), 42 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 226cefe..b44212b 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1995,33 +1995,24 @@ static int cxlflash_change_queue_depth(struct scsi_device *sdev, int qdepth)
 
 /**
  * cxlflash_show_port_status() - queries and presents the current port status
- * @dev:	Generic device associated with the host owning the port.
- * @attr:	Device attribute representing the port.
+ * @port:	Desired port for status reporting.
+ * @afu:	AFU owning the specified port.
  * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
  *
  * Return: The size of the ASCII string returned in @buf.
  */
-static ssize_t cxlflash_show_port_status(struct device *dev,
-					 struct device_attribute *attr,
-					 char *buf)
+static ssize_t cxlflash_show_port_status(u32 port, struct afu *afu, char *buf)
 {
-	struct Scsi_Host *shost = class_to_shost(dev);
-	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
-	struct afu *afu = cfg->afu;
-
 	char *disp_status;
-	int rc;
-	u32 port;
 	u64 status;
-	u64 *fc_regs;
+	__be64 __iomem *fc_regs;
 
-	rc = kstrtouint((attr->attr.name + 4), 10, &port);
-	if (rc || (port >= NUM_FC_PORTS))
+	if (port >= NUM_FC_PORTS)
 		return 0;
 
 	fc_regs = &afu->afu_map->global.fc_regs[port][0];
-	status =
-	    (readq_be(&fc_regs[FC_MTIP_STATUS / 8]) & FC_MTIP_STATUS_MASK);
+	status = readq_be(&fc_regs[FC_MTIP_STATUS / 8]);
+	status &= FC_MTIP_STATUS_MASK;
 
 	if (status == FC_MTIP_STATUS_ONLINE)
 		disp_status = "online";
@@ -2030,31 +2021,69 @@ static ssize_t cxlflash_show_port_status(struct device *dev,
 	else
 		disp_status = "unknown";
 
-	return snprintf(buf, PAGE_SIZE, "%s\n", disp_status);
+	return scnprintf(buf, PAGE_SIZE, "%s\n", disp_status);
+}
+
+/**
+ * port0_show() - queries and presents the current status of port 0
+ * @dev:	Generic device associated with the host owning the port.
+ * @attr:	Device attribute representing the port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t port0_show(struct device *dev,
+			  struct device_attribute *attr,
+			  char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	return cxlflash_show_port_status(0, afu, buf);
 }
 
 /**
- * cxlflash_show_lun_mode() - presents the current LUN mode of the host
+ * port1_show() - queries and presents the current status of port 1
+ * @dev:	Generic device associated with the host owning the port.
+ * @attr:	Device attribute representing the port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t port1_show(struct device *dev,
+			  struct device_attribute *attr,
+			  char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	return cxlflash_show_port_status(1, afu, buf);
+}
+
+/**
+ * lun_mode_show() - presents the current LUN mode of the host
  * @dev:	Generic device associated with the host.
- * @attr:	Device attribute representing the lun mode.
+ * @attr:	Device attribute representing the LUN mode.
  * @buf:	Buffer of length PAGE_SIZE to report back the LUN mode in ASCII.
  *
  * Return: The size of the ASCII string returned in @buf.
  */
-static ssize_t cxlflash_show_lun_mode(struct device *dev,
-				      struct device_attribute *attr, char *buf)
+static ssize_t lun_mode_show(struct device *dev,
+			     struct device_attribute *attr, char *buf)
 {
 	struct Scsi_Host *shost = class_to_shost(dev);
 	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
 	struct afu *afu = cfg->afu;
 
-	return snprintf(buf, PAGE_SIZE, "%u\n", afu->internal_lun);
+	return scnprintf(buf, PAGE_SIZE, "%u\n", afu->internal_lun);
 }
 
 /**
- * cxlflash_store_lun_mode() - sets the LUN mode of the host
+ * lun_mode_store() - sets the LUN mode of the host
  * @dev:	Generic device associated with the host.
- * @attr:	Device attribute representing the lun mode.
+ * @attr:	Device attribute representing the LUN mode.
  * @buf:	Buffer of length PAGE_SIZE containing the LUN mode in ASCII.
  * @count:	Length of data resizing in @buf.
  *
@@ -2073,9 +2102,9 @@ static ssize_t cxlflash_show_lun_mode(struct device *dev,
  *
  * Return: The size of the ASCII string returned in @buf.
  */
-static ssize_t cxlflash_store_lun_mode(struct device *dev,
-				       struct device_attribute *attr,
-				       const char *buf, size_t count)
+static ssize_t lun_mode_store(struct device *dev,
+			      struct device_attribute *attr,
+			      const char *buf, size_t count)
 {
 	struct Scsi_Host *shost = class_to_shost(dev);
 	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
@@ -2094,58 +2123,125 @@ static ssize_t cxlflash_store_lun_mode(struct device *dev,
 }
 
 /**
- * cxlflash_show_ioctl_version() - presents the hosts current ioctl version
+ * ioctl_version_show() - presents the current ioctl version of the host
  * @dev:	Generic device associated with the host.
  * @attr:	Device attribute representing the ioctl version.
  * @buf:	Buffer of length PAGE_SIZE to report back the ioctl version.
  *
  * Return: The size of the ASCII string returned in @buf.
  */
-static ssize_t cxlflash_show_ioctl_version(struct device *dev,
-					   struct device_attribute *attr,
-					   char *buf)
+static ssize_t ioctl_version_show(struct device *dev,
+				  struct device_attribute *attr, char *buf)
 {
 	return scnprintf(buf, PAGE_SIZE, "%u\n", DK_CXLFLASH_VERSION_0);
 }
 
 /**
- * cxlflash_show_dev_mode() - presents the current mode of the device
+ * cxlflash_show_port_lun_table() - queries and presents the port LUN table
+ * @port:	Desired port for status reporting.
+ * @afu:	AFU owning the specified port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t cxlflash_show_port_lun_table(u32 port,
+					    struct afu *afu,
+					    char *buf)
+{
+	int i;
+	ssize_t bytes = 0;
+	__be64 __iomem *fc_port;
+
+	if (port >= NUM_FC_PORTS)
+		return 0;
+
+	fc_port = &afu->afu_map->global.fc_port[port][0];
+
+	for (i = 0; i < CXLFLASH_NUM_VLUNS; i++)
+		bytes += scnprintf(buf + bytes, PAGE_SIZE - bytes,
+				   "%03d: %016llX\n", i, readq_be(&fc_port[i]));
+	return bytes;
+}
+
+/**
+ * port0_lun_table_show() - presents the current LUN table of port 0
+ * @dev:	Generic device associated with the host owning the port.
+ * @attr:	Device attribute representing the port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t port0_lun_table_show(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	return cxlflash_show_port_lun_table(0, afu, buf);
+}
+
+/**
+ * port1_lun_table_show() - presents the current LUN table of port 1
+ * @dev:	Generic device associated with the host owning the port.
+ * @attr:	Device attribute representing the port.
+ * @buf:	Buffer of length PAGE_SIZE to report back port status in ASCII.
+ *
+ * Return: The size of the ASCII string returned in @buf.
+ */
+static ssize_t port1_lun_table_show(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct Scsi_Host *shost = class_to_shost(dev);
+	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)shost->hostdata;
+	struct afu *afu = cfg->afu;
+
+	return cxlflash_show_port_lun_table(1, afu, buf);
+}
+
+/**
+ * mode_show() - presents the current mode of the device
  * @dev:	Generic device associated with the device.
  * @attr:	Device attribute representing the device mode.
  * @buf:	Buffer of length PAGE_SIZE to report back the dev mode in ASCII.
  *
  * Return: The size of the ASCII string returned in @buf.
  */
-static ssize_t cxlflash_show_dev_mode(struct device *dev,
-				      struct device_attribute *attr, char *buf)
+static ssize_t mode_show(struct device *dev,
+			 struct device_attribute *attr, char *buf)
 {
 	struct scsi_device *sdev = to_scsi_device(dev);
 
-	return snprintf(buf, PAGE_SIZE, "%s\n",
-			sdev->hostdata ? "superpipe" : "legacy");
+	return scnprintf(buf, PAGE_SIZE, "%s\n",
+			 sdev->hostdata ? "superpipe" : "legacy");
 }
 
 /*
  * Host attributes
  */
-static DEVICE_ATTR(port0, S_IRUGO, cxlflash_show_port_status, NULL);
-static DEVICE_ATTR(port1, S_IRUGO, cxlflash_show_port_status, NULL);
-static DEVICE_ATTR(lun_mode, S_IRUGO | S_IWUSR, cxlflash_show_lun_mode,
-		   cxlflash_store_lun_mode);
-static DEVICE_ATTR(ioctl_version, S_IRUGO, cxlflash_show_ioctl_version, NULL);
+static DEVICE_ATTR_RO(port0);
+static DEVICE_ATTR_RO(port1);
+static DEVICE_ATTR_RW(lun_mode);
+static DEVICE_ATTR_RO(ioctl_version);
+static DEVICE_ATTR_RO(port0_lun_table);
+static DEVICE_ATTR_RO(port1_lun_table);
 
 static struct device_attribute *cxlflash_host_attrs[] = {
 	&dev_attr_port0,
 	&dev_attr_port1,
 	&dev_attr_lun_mode,
 	&dev_attr_ioctl_version,
+	&dev_attr_port0_lun_table,
+	&dev_attr_port1_lun_table,
 	NULL
 };
 
 /*
  * Device attributes
  */
-static DEVICE_ATTR(mode, S_IRUGO, cxlflash_show_dev_mode, NULL);
+static DEVICE_ATTR_RO(mode);
 
 static struct device_attribute *cxlflash_dev_attrs[] = {
 	&dev_attr_mode,
-- 
2.1.0

[PATCH v4 12/32] cxlflash: Fix to avoid spamming the kernel log

[Matthew R. Ochs]

During run-time the driver can be very chatty and spam the system
kernel log. Various print statements can be limited and/or moved
to development-only mode. Additionally, numerous prints can be
converted to trace the corresponding device.

The following changes were made:
 - pr_debug to pr_devel
 - pr_debug to pr_debug_ratelimited
 - pr_err to dev_err
 - pr_debug to dev_dbg

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Conflicts:
	drivers/scsi/cxlflash/main.c
---
 drivers/scsi/cxlflash/main.c | 109 +++++++++++++++++++++++--------------------
 1 file changed, 59 insertions(+), 50 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index b44212b..527ff85 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -58,8 +58,8 @@ static struct afu_cmd *cmd_checkout(struct afu *afu)
 		cmd = &afu->cmd[k];
 
 		if (!atomic_dec_if_positive(&cmd->free)) {
-			pr_debug("%s: returning found index=%d\n",
-				 __func__, cmd->slot);
+			pr_devel("%s: returning found index=%d cmd=%p\n",
+				 __func__, cmd->slot, cmd);
 			memset(cmd->buf, 0, CMD_BUFSIZE);
 			memset(cmd->rcb.cdb, 0, sizeof(cmd->rcb.cdb));
 			return cmd;
@@ -93,7 +93,7 @@ static void cmd_checkin(struct afu_cmd *cmd)
 		return;
 	}
 
-	pr_debug("%s: released cmd %p index=%d\n", __func__, cmd, cmd->slot);
+	pr_devel("%s: released cmd %p index=%d\n", __func__, cmd, cmd->slot);
 }
 
 /**
@@ -127,7 +127,7 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 	}
 
 	pr_debug("%s: cmd failed afu_rc=%d scsi_rc=%d fc_rc=%d "
-		 "afu_extra=0x%X, scsi_entra=0x%X, fc_extra=0x%X\n",
+		 "afu_extra=0x%X, scsi_extra=0x%X, fc_extra=0x%X\n",
 		 __func__, ioasa->rc.afu_rc, ioasa->rc.scsi_rc,
 		 ioasa->rc.fc_rc, ioasa->afu_extra, ioasa->scsi_extra,
 		 ioasa->fc_extra);
@@ -240,9 +240,9 @@ static void cmd_complete(struct afu_cmd *cmd)
 		cmd_is_tmf = cmd->cmd_tmf;
 		cmd_checkin(cmd); /* Don't use cmd after here */
 
-		pr_debug("%s: calling scsi_set_resid, scp=%p "
-			 "result=%X resid=%d\n", __func__,
-			 scp, scp->result, resid);
+		pr_debug_ratelimited("%s: calling scsi_done scp=%p result=%X "
+				     "ioasc=%d\n", __func__, scp, scp->result,
+				     cmd->sa.ioasc);
 
 		scsi_set_resid(scp, resid);
 		scsi_dma_unmap(scp);
@@ -417,12 +417,13 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 	short lflag = 0;
 	struct Scsi_Host *host = scp->device->host;
 	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
+	struct device *dev = &cfg->dev->dev;
 	ulong lock_flags;
 	int rc = 0;
 
 	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
-		pr_err("%s: could not get a free command\n", __func__);
+		dev_err(dev, "%s: could not get a free command\n", __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
 	}
@@ -493,7 +494,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 {
 	struct cxlflash_cfg *cfg = (struct cxlflash_cfg *)host->hostdata;
 	struct afu *afu = cfg->afu;
-	struct pci_dev *pdev = cfg->dev;
+	struct device *dev = &cfg->dev->dev;
 	struct afu_cmd *cmd;
 	u32 port_sel = scp->device->channel + 1;
 	int nseg, i, ncount;
@@ -502,13 +503,14 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	short lflag = 0;
 	int rc = 0;
 
-	pr_debug("%s: (scp=%p) %d/%d/%d/%llu cdb=(%08X-%08X-%08X-%08X)\n",
-		 __func__, scp, host->host_no, scp->device->channel,
-		 scp->device->id, scp->device->lun,
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
-		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
+	dev_dbg_ratelimited(dev, "%s: (scp=%p) %d/%d/%d/%llu "
+			    "cdb=(%08X-%08X-%08X-%08X)\n",
+			    __func__, scp, host->host_no, scp->device->channel,
+			    scp->device->id, scp->device->lun,
+			    get_unaligned_be32(&((u32 *)scp->cmnd)[0]),
+			    get_unaligned_be32(&((u32 *)scp->cmnd)[1]),
+			    get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
+			    get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
 
 	/* If a Task Management Function is active, wait for it to complete
 	 * before continuing with regular commands.
@@ -523,13 +525,11 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 
 	switch (cfg->state) {
 	case STATE_RESET:
-		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device is in reset!\n",
-				    __func__);
+		dev_dbg_ratelimited(dev, "%s: device is in reset!\n", __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
 	case STATE_FAILTERM:
-		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device has failed!\n",
-				    __func__);
+		dev_dbg_ratelimited(dev, "%s: device has failed!\n", __func__);
 		scp->result = (DID_NO_CONNECT << 16);
 		scp->scsi_done(scp);
 		rc = 0;
@@ -540,7 +540,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 
 	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
-		pr_err("%s: could not get a free command\n", __func__);
+		dev_err(dev, "%s: could not get a free command\n", __func__);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
 	}
@@ -562,7 +562,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 
 	nseg = scsi_dma_map(scp);
 	if (unlikely(nseg < 0)) {
-		dev_err(&pdev->dev, "%s: Fail DMA map! nseg=%d\n",
+		dev_err(dev, "%s: Fail DMA map! nseg=%d\n",
 			__func__, nseg);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
@@ -585,6 +585,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	}
 
 out:
+	pr_devel("%s: returning rc=%d\n", __func__, rc);
 	return rc;
 }
 
@@ -657,9 +658,10 @@ static void term_mc(struct cxlflash_cfg *cfg, enum undo_level level)
 {
 	int rc = 0;
 	struct afu *afu = cfg->afu;
+	struct device *dev = &cfg->dev->dev;
 
 	if (!afu || !cfg->mcctx) {
-		pr_err("%s: returning from term_mc with NULL afu or MC\n",
+		dev_err(dev, "%s: returning from term_mc with NULL afu or MC\n",
 		       __func__);
 		return;
 	}
@@ -755,6 +757,7 @@ static int alloc_mem(struct cxlflash_cfg *cfg)
 	int rc = 0;
 	int i;
 	char *buf = NULL;
+	struct device *dev = &cfg->dev->dev;
 
 	/* This allocation is about 12K, i.e. only 1 64k page
 	 * and upto 4 4k pages
@@ -762,8 +765,8 @@ static int alloc_mem(struct cxlflash_cfg *cfg)
 	cfg->afu = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
 					    get_order(sizeof(struct afu)));
 	if (unlikely(!cfg->afu)) {
-		pr_err("%s: cannot get %d free pages\n",
-		       __func__, get_order(sizeof(struct afu)));
+		dev_err(dev, "%s: cannot get %d free pages\n",
+			__func__, get_order(sizeof(struct afu)));
 		rc = -ENOMEM;
 		goto out;
 	}
@@ -774,7 +777,8 @@ static int alloc_mem(struct cxlflash_cfg *cfg)
 		if (!((u64)buf & (PAGE_SIZE - 1))) {
 			buf = (void *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
 			if (unlikely(!buf)) {
-				pr_err("%s: Allocate command buffers fail!\n",
+				dev_err(dev,
+					"%s: Allocate command buffers fail!\n",
 				       __func__);
 				rc = -ENOMEM;
 				free_mem(cfg);
@@ -1280,7 +1284,8 @@ static irqreturn_t cxlflash_rrq_irq(int irq, void *data)
 static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 {
 	struct afu *afu = (struct afu *)data;
-	struct cxlflash_cfg *cfg;
+	struct cxlflash_cfg *cfg = afu->parent;
+	struct device *dev = &cfg->dev->dev;
 	u64 reg_unmasked;
 	const struct asyc_intr_info *info;
 	struct sisl_global_map *global = &afu->afu_map->global;
@@ -1288,14 +1293,12 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 	u8 port;
 	int i;
 
-	cfg = afu->parent;
-
 	reg = readq_be(&global->regs.aintr_status);
 	reg_unmasked = (reg & SISL_ASTATUS_UNMASK);
 
 	if (reg_unmasked == 0) {
-		pr_err("%s: spurious interrupt, aintr_status 0x%016llX\n",
-		       __func__, reg);
+		dev_err(dev, "%s: spurious interrupt, aintr_status 0x%016llX\n",
+			__func__, reg);
 		goto out;
 	}
 
@@ -1310,8 +1313,8 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 
 		port = info->port;
 
-		pr_err("%s: FC Port %d -> %s, fc_status 0x%08llX\n",
-		       __func__, port, info->desc,
+		dev_err(dev, "%s: FC Port %d -> %s, fc_status 0x%08llX\n",
+			__func__, port, info->desc,
 		       readq_be(&global->fc_regs[port][FC_STATUS / 8]));
 
 		/*
@@ -1319,8 +1322,8 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 		 * again if cleared before or w/o a reset
 		 */
 		if (info->action & LINK_RESET) {
-			pr_err("%s: FC Port %d: resetting link\n",
-			       __func__, port);
+			dev_err(dev, "%s: FC Port %d: resetting link\n",
+				__func__, port);
 			cfg->lr_state = LINK_RESET_REQUIRED;
 			cfg->lr_port = port;
 			schedule_work(&cfg->work_q);
@@ -1334,8 +1337,8 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 			 * should be the same and tracing one is sufficient.
 			 */
 
-			pr_err("%s: fc %d: clearing fc_error 0x%08llX\n",
-			       __func__, port, reg);
+			dev_err(dev, "%s: fc %d: clearing fc_error 0x%08llX\n",
+				__func__, port, reg);
 
 			writeq_be(reg, &global->fc_regs[port][FC_ERROR / 8]);
 			writeq_be(0, &global->fc_regs[port][FC_ERRCAP / 8]);
@@ -1343,7 +1346,7 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 	}
 
 out:
-	pr_debug("%s: returning rc=%d, afu=%p\n", __func__, IRQ_HANDLED, afu);
+	dev_dbg(dev, "%s: returning IRQ_HANDLED, afu=%p\n", __func__, afu);
 	return IRQ_HANDLED;
 }
 
@@ -1387,7 +1390,7 @@ static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 	/* Get the VPD data from the device */
 	vpd_size = pci_read_vpd(dev, 0, sizeof(vpd_data), vpd_data);
 	if (unlikely(vpd_size <= 0)) {
-		pr_err("%s: Unable to read VPD (size = %ld)\n",
+		dev_err(&dev->dev, "%s: Unable to read VPD (size = %ld)\n",
 		       __func__, vpd_size);
 		rc = -ENODEV;
 		goto out;
@@ -1397,7 +1400,8 @@ static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 	ro_start = pci_vpd_find_tag(vpd_data, 0, vpd_size,
 				    PCI_VPD_LRDT_RO_DATA);
 	if (unlikely(ro_start < 0)) {
-		pr_err("%s: VPD Read-only data not found\n", __func__);
+		dev_err(&dev->dev, "%s: VPD Read-only data not found\n",
+			__func__);
 		rc = -ENODEV;
 		goto out;
 	}
@@ -1426,8 +1430,8 @@ static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 
 		i = pci_vpd_find_info_keyword(vpd_data, i, j, wwpn_vpd_tags[k]);
 		if (unlikely(i < 0)) {
-			pr_err("%s: Port %d WWPN not found in VPD\n",
-			       __func__, k);
+			dev_err(&dev->dev, "%s: Port %d WWPN not found "
+				"in VPD\n", __func__, k);
 			rc = -ENODEV;
 			goto out;
 		}
@@ -1435,7 +1439,8 @@ static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 		j = pci_vpd_info_field_size(&vpd_data[i]);
 		i += PCI_VPD_INFO_FLD_HDR_SIZE;
 		if (unlikely((i + j > vpd_size) || (j != WWPN_LEN))) {
-			pr_err("%s: Port %d WWPN incomplete or VPD corrupt\n",
+			dev_err(&dev->dev, "%s: Port %d WWPN incomplete or "
+				"VPD corrupt\n",
 			       __func__, k);
 			rc = -ENODEV;
 			goto out;
@@ -1444,8 +1449,8 @@ static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 		memcpy(tmp_buf, &vpd_data[i], WWPN_LEN);
 		rc = kstrtoul(tmp_buf, WWPN_LEN, (ulong *)&wwpn[k]);
 		if (unlikely(rc)) {
-			pr_err("%s: Fail to convert port %d WWPN to integer\n",
-			       __func__, k);
+			dev_err(&dev->dev, "%s: Fail to convert port %d WWPN "
+				"to integer\n", __func__, k);
 			rc = -ENODEV;
 			goto out;
 		}
@@ -1502,6 +1507,7 @@ static void init_pcr(struct cxlflash_cfg *cfg)
 static int init_global(struct cxlflash_cfg *cfg)
 {
 	struct afu *afu = cfg->afu;
+	struct device *dev = &cfg->dev->dev;
 	u64 wwpn[NUM_FC_PORTS];	/* wwpn of AFU ports */
 	int i = 0, num_ports = 0;
 	int rc = 0;
@@ -1509,7 +1515,7 @@ static int init_global(struct cxlflash_cfg *cfg)
 
 	rc = read_vpd(cfg, &wwpn[0]);
 	if (rc) {
-		pr_err("%s: could not read vpd rc=%d\n", __func__, rc);
+		dev_err(dev, "%s: could not read vpd rc=%d\n", __func__, rc);
 		goto out;
 	}
 
@@ -1552,7 +1558,7 @@ static int init_global(struct cxlflash_cfg *cfg)
 		    afu_set_wwpn(afu, i,
 				 &afu->afu_map->global.fc_regs[i][0],
 				 wwpn[i])) {
-			pr_err("%s: failed to set WWPN on port %d\n",
+			dev_err(dev, "%s: failed to set WWPN on port %d\n",
 			       __func__, i);
 			rc = -EIO;
 			goto out;
@@ -1795,6 +1801,7 @@ int cxlflash_afu_sync(struct afu *afu, ctx_hndl_t ctx_hndl_u,
 		      res_hndl_t res_hndl_u, u8 mode)
 {
 	struct cxlflash_cfg *cfg = afu->parent;
+	struct device *dev = &cfg->dev->dev;
 	struct afu_cmd *cmd = NULL;
 	int rc = 0;
 	int retry_cnt = 0;
@@ -1813,7 +1820,7 @@ retry:
 		udelay(1000 * retry_cnt);
 		if (retry_cnt < MC_RETRY_CNT)
 			goto retry;
-		pr_err("%s: could not get a free command\n", __func__);
+		dev_err(dev, "%s: could not get a free command\n", __func__);
 		rc = -1;
 		goto out;
 	}
@@ -2301,6 +2308,7 @@ static void cxlflash_worker_thread(struct work_struct *work)
 	struct cxlflash_cfg *cfg = container_of(work, struct cxlflash_cfg,
 						work_q);
 	struct afu *afu = cfg->afu;
+	struct device *dev = &cfg->dev->dev;
 	int port;
 	ulong lock_flags;
 
@@ -2314,7 +2322,8 @@ static void cxlflash_worker_thread(struct work_struct *work)
 	if (cfg->lr_state == LINK_RESET_REQUIRED) {
 		port = cfg->lr_port;
 		if (port < 0)
-			pr_err("%s: invalid port index %d\n", __func__, port);
+			dev_err(dev, "%s: invalid port index %d\n",
+				__func__, port);
 		else {
 			spin_unlock_irqrestore(cfg->host->host_lock,
 					       lock_flags);
@@ -2419,7 +2428,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	 */
 	phys_dev = cxl_get_phys_dev(pdev);
 	if (!dev_is_pci(phys_dev)) {
-		pr_err("%s: not a pci dev\n", __func__);
+		dev_err(&pdev->dev, "%s: not a pci dev\n", __func__);
 		rc = -ENODEV;
 		goto out_remove;
 	}
-- 
2.1.0

[PATCH v4 13/32] cxlflash: Fix to avoid stall while waiting on TMF

[Matthew R. Ochs]

Borrowing the TMF waitq's spinlock causes a stall condition when
waiting for the TMF to complete. To remedy, introduce our own spin
lock to serialize TMF and use the appropriate wait services.

Also add a timeout while waiting for a TMF completion. When a TMF
times out, report back a failure such that a bigger hammer reset
can occur.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h |  1 +
 drivers/scsi/cxlflash/main.c   | 55 +++++++++++++++++++++++++-----------------
 2 files changed, 34 insertions(+), 22 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index b038ac7..7a0cb5c 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -124,6 +124,7 @@ struct cxlflash_cfg {
 	struct list_head lluns; /* list of llun_info structs */
 
 	wait_queue_head_t tmf_waitq;
+	spinlock_t tmf_slock;
 	bool tmf_active;
 	wait_queue_head_t reset_waitq;
 	enum cxlflash_state state;
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 527ff85..110037d 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -249,11 +249,10 @@ static void cmd_complete(struct afu_cmd *cmd)
 		scp->scsi_done(scp);
 
 		if (cmd_is_tmf) {
-			spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
+			spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
 			cfg->tmf_active = false;
 			wake_up_all_locked(&cfg->tmf_waitq);
-			spin_unlock_irqrestore(&cfg->tmf_waitq.lock,
-					       lock_flags);
+			spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 		}
 	} else
 		complete(&cmd->cevent);
@@ -420,6 +419,7 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 	struct device *dev = &cfg->dev->dev;
 	ulong lock_flags;
 	int rc = 0;
+	ulong to;
 
 	cmd = cmd_checkout(afu);
 	if (unlikely(!cmd)) {
@@ -428,15 +428,15 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 		goto out;
 	}
 
-	/* If a Task Management Function is active, do not send one more.
-	 */
-	spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
+	/* When Task Management Function is active do not send another */
+	spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
 	if (cfg->tmf_active)
-		wait_event_interruptible_locked_irq(cfg->tmf_waitq,
-						    !cfg->tmf_active);
+		wait_event_interruptible_lock_irq(cfg->tmf_waitq,
+						  !cfg->tmf_active,
+						  cfg->tmf_slock);
 	cfg->tmf_active = true;
 	cmd->cmd_tmf = true;
-	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+	spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 
 	cmd->rcb.ctx_id = afu->ctx_hndl;
 	cmd->rcb.port_sel = port_sel;
@@ -457,15 +457,24 @@ static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 	rc = send_cmd(afu, cmd);
 	if (unlikely(rc)) {
 		cmd_checkin(cmd);
-		spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
+		spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
 		cfg->tmf_active = false;
-		spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+		spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 		goto out;
 	}
 
-	spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
-	wait_event_interruptible_locked_irq(cfg->tmf_waitq, !cfg->tmf_active);
-	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+	spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
+	to = msecs_to_jiffies(5000);
+	to = wait_event_interruptible_lock_irq_timeout(cfg->tmf_waitq,
+						       !cfg->tmf_active,
+						       cfg->tmf_slock,
+						       to);
+	if (!to) {
+		cfg->tmf_active = false;
+		dev_err(dev, "%s: TMF timed out!\n", __func__);
+		rc = -1;
+	}
+	spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 out:
 	return rc;
 }
@@ -512,16 +521,17 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 			    get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
 			    get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
 
-	/* If a Task Management Function is active, wait for it to complete
+	/*
+	 * If a Task Management Function is active, wait for it to complete
 	 * before continuing with regular commands.
 	 */
-	spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
+	spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
 	if (cfg->tmf_active) {
-		spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+		spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 		rc = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
 	}
-	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+	spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 
 	switch (cfg->state) {
 	case STATE_RESET:
@@ -713,11 +723,12 @@ static void cxlflash_remove(struct pci_dev *pdev)
 	/* If a Task Management Function is active, wait for it to complete
 	 * before continuing with remove.
 	 */
-	spin_lock_irqsave(&cfg->tmf_waitq.lock, lock_flags);
+	spin_lock_irqsave(&cfg->tmf_slock, lock_flags);
 	if (cfg->tmf_active)
-		wait_event_interruptible_locked_irq(cfg->tmf_waitq,
-						    !cfg->tmf_active);
-	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
+		wait_event_interruptible_lock_irq(cfg->tmf_waitq,
+						  !cfg->tmf_active,
+						  cfg->tmf_slock);
+	spin_unlock_irqrestore(&cfg->tmf_slock, lock_flags);
 
 	cfg->state = STATE_FAILTERM;
 	cxlflash_stop_term_user_contexts(cfg);
-- 
2.1.0

[PATCH v4 14/32] cxlflash: Fix location of setting resid

[Matthew R. Ochs]

The resid is incorrectly set which can lead to unnecessary retry
attempts by the stack. This is due to resid _always_ being set
using a value returned from the adapter. Instead, the value
should only be interpreted and set when in an underrun scenario.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 110037d..5503a40 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -107,6 +107,7 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 {
 	struct sisl_ioarcb *ioarcb;
 	struct sisl_ioasa *ioasa;
+	u32 resid;
 
 	if (unlikely(!cmd))
 		return;
@@ -115,9 +116,10 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 	ioasa = &(cmd->sa);
 
 	if (ioasa->rc.flags & SISL_RC_FLAGS_UNDERRUN) {
-		pr_debug("%s: cmd underrun cmd = %p scp = %p\n",
-			 __func__, cmd, scp);
-		scp->result = (DID_ERROR << 16);
+		resid = ioasa->resid;
+		scsi_set_resid(scp, resid);
+		pr_debug("%s: cmd underrun cmd = %p scp = %p, resid = %d\n",
+			 __func__, cmd, scp, resid);
 	}
 
 	if (ioasa->rc.flags & SISL_RC_FLAGS_OVERRUN) {
@@ -158,8 +160,7 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 				/* If the SISL_RC_FLAGS_OVERRUN flag was set,
 				 * then we will handle this error else where.
 				 * If not then we must handle it here.
-				 * This is probably an AFU bug. We will
-				 * attempt a retry to see if that resolves it.
+				 * This is probably an AFU bug.
 				 */
 				scp->result = (DID_ERROR << 16);
 			}
@@ -183,7 +184,7 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 		/* We have an AFU error */
 		switch (ioasa->rc.afu_rc) {
 		case SISL_AFU_RC_NO_CHANNELS:
-			scp->result = (DID_MEDIUM_ERROR << 16);
+			scp->result = (DID_NO_CONNECT << 16);
 			break;
 		case SISL_AFU_RC_DATA_DMA_ERR:
 			switch (ioasa->afu_extra) {
@@ -217,7 +218,6 @@ static void process_cmd_err(struct afu_cmd *cmd, struct scsi_cmnd *scp)
 static void cmd_complete(struct afu_cmd *cmd)
 {
 	struct scsi_cmnd *scp;
-	u32 resid;
 	ulong lock_flags;
 	struct afu *afu = cmd->parent;
 	struct cxlflash_cfg *cfg = afu->parent;
@@ -229,14 +229,11 @@ static void cmd_complete(struct afu_cmd *cmd)
 
 	if (cmd->rcb.scp) {
 		scp = cmd->rcb.scp;
-		if (unlikely(cmd->sa.rc.afu_rc ||
-			     cmd->sa.rc.scsi_rc ||
-			     cmd->sa.rc.fc_rc))
+		if (unlikely(cmd->sa.ioasc))
 			process_cmd_err(cmd, scp);
 		else
 			scp->result = (DID_OK << 16);
 
-		resid = cmd->sa.resid;
 		cmd_is_tmf = cmd->cmd_tmf;
 		cmd_checkin(cmd); /* Don't use cmd after here */
 
@@ -244,7 +241,6 @@ static void cmd_complete(struct afu_cmd *cmd)
 				     "ioasc=%d\n", __func__, scp, scp->result,
 				     cmd->sa.ioasc);
 
-		scsi_set_resid(scp, resid);
 		scsi_dma_unmap(scp);
 		scp->scsi_done(scp);
 
-- 
2.1.0

[PATCH v4 15/32] cxlflash: Fix host link up event handling

[Matthew R. Ochs]

Following a link up event, the LUNs available to the host may
have changed. Without rescanning the host, the LUN topology is
unknown to the user. In such a state, the user would be unable
to locate provisioned resources.

To remedy, the host should be rescanned after a link up event.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Conflicts:
	drivers/scsi/cxlflash/common.h
---
 drivers/scsi/cxlflash/common.h |  1 +
 drivers/scsi/cxlflash/main.c   | 17 +++++++++++++----
 drivers/scsi/cxlflash/main.h   |  1 +
 3 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 7a0cb5c..faf7f56 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -102,6 +102,7 @@ struct cxlflash_cfg {
 	enum cxlflash_init_state init_state;
 	enum cxlflash_lr_state lr_state;
 	int lr_port;
+	atomic_t scan_host_needed;
 
 	struct cxl_afu *cxl_afu;
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 5503a40..98fdac1 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1119,17 +1119,17 @@ static const struct asyc_intr_info ainfo[] = {
 	{SISL_ASTATUS_FC0_CRC_T, "CRC threshold exceeded", 0, LINK_RESET},
 	{SISL_ASTATUS_FC0_LOGI_R, "login timed out, retrying", 0, 0},
 	{SISL_ASTATUS_FC0_LOGI_F, "login failed", 0, CLR_FC_ERROR},
-	{SISL_ASTATUS_FC0_LOGI_S, "login succeeded", 0, 0},
+	{SISL_ASTATUS_FC0_LOGI_S, "login succeeded", 0, SCAN_HOST},
 	{SISL_ASTATUS_FC0_LINK_DN, "link down", 0, 0},
-	{SISL_ASTATUS_FC0_LINK_UP, "link up", 0, 0},
+	{SISL_ASTATUS_FC0_LINK_UP, "link up", 0, SCAN_HOST},
 	{SISL_ASTATUS_FC1_OTHER, "other error", 1, CLR_FC_ERROR | LINK_RESET},
 	{SISL_ASTATUS_FC1_LOGO, "target initiated LOGO", 1, 0},
 	{SISL_ASTATUS_FC1_CRC_T, "CRC threshold exceeded", 1, LINK_RESET},
 	{SISL_ASTATUS_FC1_LOGI_R, "login timed out, retrying", 1, 0},
 	{SISL_ASTATUS_FC1_LOGI_F, "login failed", 1, CLR_FC_ERROR},
-	{SISL_ASTATUS_FC1_LOGI_S, "login succeeded", 1, 0},
+	{SISL_ASTATUS_FC1_LOGI_S, "login succeeded", 1, SCAN_HOST},
 	{SISL_ASTATUS_FC1_LINK_DN, "link down", 1, 0},
-	{SISL_ASTATUS_FC1_LINK_UP, "link up", 1, 0},
+	{SISL_ASTATUS_FC1_LINK_UP, "link up", 1, SCAN_HOST},
 	{0x0, "", 0, 0}		/* terminator */
 };
 
@@ -1350,6 +1350,11 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 			writeq_be(reg, &global->fc_regs[port][FC_ERROR / 8]);
 			writeq_be(0, &global->fc_regs[port][FC_ERRCAP / 8]);
 		}
+
+		if (info->action & SCAN_HOST) {
+			atomic_inc(&cfg->scan_host_needed);
+			schedule_work(&cfg->work_q);
+		}
 	}
 
 out:
@@ -2309,6 +2314,7 @@ MODULE_DEVICE_TABLE(pci, cxlflash_pci_table);
  * - Link reset which cannot be performed on interrupt context due to
  * blocking up to a few seconds
  * - Read AFU command room
+ * - Rescan the host
  */
 static void cxlflash_worker_thread(struct work_struct *work)
 {
@@ -2351,6 +2357,9 @@ static void cxlflash_worker_thread(struct work_struct *work)
 	}
 
 	spin_unlock_irqrestore(cfg->host->host_lock, lock_flags);
+
+	if (atomic_dec_if_positive(&cfg->scan_host_needed) >= 0)
+		scsi_scan_host(cfg->host);
 }
 
 /**
diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h
index cf0e809..6032456 100644
--- a/drivers/scsi/cxlflash/main.h
+++ b/drivers/scsi/cxlflash/main.h
@@ -99,6 +99,7 @@ struct asyc_intr_info {
 	u8 action;
 #define CLR_FC_ERROR	0x01
 #define LINK_RESET	0x02
+#define SCAN_HOST	0x04
 };
 
 #ifndef CONFIG_CXL_EEH
-- 
2.1.0

[PATCH v4 16/32] cxlflash: Fix async interrupt bypass logic

[Matthew R. Ochs]

A bug was introduced earlier in the development cycle when cleaning
up logic statements. Instead of skipping bits that are not set, set
bits are skipped, causing async interrupts to not be handled correctly.

To fix, simply add back in the proper evaluation for an unset bit.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 98fdac1..ed9fd8c 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1315,7 +1315,7 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 	/* check each bit that is on */
 	for (i = 0; reg_unmasked; i++, reg_unmasked = (reg_unmasked >> 1)) {
 		info = find_ainfo(1ULL << i);
-		if ((reg_unmasked & 0x1) || !info)
+		if (((reg_unmasked & 0x1) == 0) || !info)
 			continue;
 
 		port = info->port;
-- 
2.1.0

[PATCH v4 17/32] cxlflash: Remove dual port online dependency

[Matthew R. Ochs]

At present, both ports must be online for the device to
configure properly. Remove this dependency and the unnecessary
internal LUN override logic as well. Additionally, as a refactoring
measure, change the return code variable name to match that used
throughout the driver.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 23 ++++++++---------------
 1 file changed, 8 insertions(+), 15 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index ed9fd8c..d45388f 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1030,7 +1030,7 @@ static int wait_port_offline(u64 *fc_regs, u32 delay_us, u32 nretry)
  */
 static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 wwpn)
 {
-	int ret = 0;
+	int rc = 0;
 
 	set_port_offline(fc_regs);
 
@@ -1038,33 +1038,26 @@ static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 wwpn)
 			       FC_PORT_STATUS_RETRY_CNT)) {
 		pr_debug("%s: wait on port %d to go offline timed out\n",
 			 __func__, port);
-		ret = -1; /* but continue on to leave the port back online */
+		rc = -1; /* but continue on to leave the port back online */
 	}
 
-	if (ret == 0)
+	if (rc == 0)
 		writeq_be(wwpn, &fc_regs[FC_PNAME / 8]);
 
+	/* Always return success after programming WWPN */
+	rc = 0;
+
 	set_port_online(fc_regs);
 
 	if (!wait_port_online(fc_regs, FC_PORT_STATUS_RETRY_INTERVAL_US,
 			      FC_PORT_STATUS_RETRY_CNT)) {
 		pr_debug("%s: wait on port %d to go online timed out\n",
 			 __func__, port);
-		ret = -1;
-
-		/*
-		 * Override for internal lun!!!
-		 */
-		if (afu->internal_lun) {
-			pr_debug("%s: Overriding port %d online timeout!!!\n",
-				 __func__, port);
-			ret = 0;
-		}
 	}
 
-	pr_debug("%s: returning rc=%d\n", __func__, ret);
+	pr_debug("%s: returning rc=%d\n", __func__, rc);
 
-	return ret;
+	return rc;
 }
 
 /**
-- 
2.1.0

[PATCH v4 18/32] cxlflash: Fix AFU version access/storage and add check

[Matthew R. Ochs]

The AFU version is stored as a non-terminated string of bytes within
a 64-bit little-endian register. Presently the value is read directly
(no MMIO accessor) and is stored in a buffer that is not big enough
to contain a NULL terminator. Additionally the version obtained is not
evaluated against a known value to prevent usage with unsupported AFUs.
All of these deficiencies can lead to a variety of problems.

To remedy, use the correct MMIO accessor to read the version value into
a null-terminated buffer and add a check to prevent an incompatible AFU
from being used with this driver.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h  |  2 +-
 drivers/scsi/cxlflash/main.c    | 18 ++++++++++++------
 drivers/scsi/cxlflash/sislite.h |  2 +-
 3 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index faf7f56..3be5754 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -179,7 +179,7 @@ struct afu {
 	u32 cmd_couts;		/* Number of command checkouts */
 	u32 internal_lun;	/* User-desired LUN mode for this AFU */
 
-	char version[8];
+	char version[16];
 	u64 interface_version;
 
 	struct cxlflash_cfg *parent; /* Pointer back to parent cxlflash_cfg */
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index d45388f..b6a2584 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1751,14 +1751,20 @@ static int init_afu(struct cxlflash_cfg *cfg)
 		goto err1;
 	}
 
-	/* don't byte reverse on reading afu_version, else the string form */
-	/*     will be backwards */
-	reg = afu->afu_map->global.regs.afu_version;
-	memcpy(afu->version, &reg, 8);
+	/* No byte reverse on reading afu_version or string will be backwards */
+	reg = readq(&afu->afu_map->global.regs.afu_version);
+	memcpy(afu->version, &reg, sizeof(reg));
 	afu->interface_version =
 	    readq_be(&afu->afu_map->global.regs.interface_version);
-	pr_debug("%s: afu version %s, interface version 0x%llX\n",
-		 __func__, afu->version, afu->interface_version);
+	if ((afu->interface_version + 1) == 0) {
+		pr_err("Back level AFU, please upgrade. AFU version %s "
+		       "interface version 0x%llx\n", afu->version,
+		       afu->interface_version);
+		rc = -EINVAL;
+		goto err1;
+	} else
+		pr_debug("%s: afu version %s, interface version 0x%llX\n",
+			 __func__, afu->version, afu->interface_version);
 
 	rc = start_afu(cfg);
 	if (rc) {
diff --git a/drivers/scsi/cxlflash/sislite.h b/drivers/scsi/cxlflash/sislite.h
index 63bf394..8425d1a 100644
--- a/drivers/scsi/cxlflash/sislite.h
+++ b/drivers/scsi/cxlflash/sislite.h
@@ -340,7 +340,7 @@ struct sisl_global_regs {
 #define SISL_AFUCONF_MBOX_CLR_READ     0x0010ULL
 	__be64 afu_config;
 	__be64 rsvd[0xf8];
-	__be64 afu_version;
+	__le64 afu_version;
 	__be64 interface_version;
 };
 
-- 
2.1.0

[PATCH v4 19/32] cxlflash: Correct usage of scsi_host_put()

[Matthew R. Ochs]

Currently, scsi_host_put() is being called prematurely in the
remove path and is missing entirely in an error cleanup path.
The former can lead to memory being freed too early with
subsequent access potentially corrupting data whilst the former
would result in a memory leak.

Move the usage on remove to be the last cleanup action taken
and introduce a call to scsi_host_put() in the one initialization
error path that does not use remove to cleanup.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index b6a2584..0bb6629 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -733,7 +733,6 @@ static void cxlflash_remove(struct pci_dev *pdev)
 	case INIT_STATE_SCSI:
 		cxlflash_term_local_luns(cfg);
 		scsi_remove_host(cfg->host);
-		scsi_host_put(cfg->host);
 		/* Fall through */
 	case INIT_STATE_AFU:
 		term_afu(cfg);
@@ -743,6 +742,7 @@ static void cxlflash_remove(struct pci_dev *pdev)
 	case INIT_STATE_NONE:
 		flush_work(&cfg->work_q);
 		free_mem(cfg);
+		scsi_host_put(cfg->host);
 		break;
 	}
 
@@ -2404,6 +2404,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 		dev_err(&pdev->dev, "%s: call to scsi_host_alloc failed!\n",
 			__func__);
 		rc = -ENOMEM;
+		scsi_host_put(cfg->host);
 		goto out;
 	}
 
-- 
2.1.0

[PATCH v4 20/32] cxlflash: Fix to prevent workq from accessing freed memory

[Matthew R. Ochs]

The workq can process work in parallel with a remove event, leading
to a condition where the workq handler can access freed memory.

To remedy, the workq should be terminated prior to freeing memory. Move
the termination call earlier in remove and use cancel_work_sync() instead
of flush_work() as there is not a need to process any scheduled work when
shutting down.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 0bb6629..fada355 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -736,11 +736,11 @@ static void cxlflash_remove(struct pci_dev *pdev)
 		/* Fall through */
 	case INIT_STATE_AFU:
 		term_afu(cfg);
+		cancel_work_sync(&cfg->work_q);
 	case INIT_STATE_PCI:
 		pci_release_regions(cfg->dev);
 		pci_disable_device(pdev);
 	case INIT_STATE_NONE:
-		flush_work(&cfg->work_q);
 		free_mem(cfg);
 		scsi_host_put(cfg->host);
 		break;
-- 
2.1.0

[PATCH v4 21/32] cxlflash: Correct behavior in device reset handler following EEH

[Matthew R. Ochs]

When the device reset handler is entered while a reset operation
is taking place, the handler exits without actually sending a
reset (TMF) to the targeted device. This behavior is incorrect
as the device is not reset. Further complicating matters is the
fact that a success is returned even when the TMF was not sent.

To fix, the state is rechecked after coming out of the reset
state. When the state is normal, a TMF will be sent out.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index fada355..cf6a067 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1920,6 +1920,7 @@ static int cxlflash_eh_device_reset_handler(struct scsi_cmnd *scp)
 		 get_unaligned_be32(&((u32 *)scp->cmnd)[2]),
 		 get_unaligned_be32(&((u32 *)scp->cmnd)[3]));
 
+retry:
 	switch (cfg->state) {
 	case STATE_NORMAL:
 		rcr = send_tmf(afu, scp, TMF_LUN_RESET);
@@ -1928,9 +1929,7 @@ static int cxlflash_eh_device_reset_handler(struct scsi_cmnd *scp)
 		break;
 	case STATE_RESET:
 		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
-		if (cfg->state == STATE_NORMAL)
-			break;
-		/* fall through */
+		goto retry;
 	default:
 		rc = FAILED;
 		break;
-- 
2.1.0

[PATCH v4 22/32] cxlflash: Remove unnecessary scsi_block_requests

[Matthew R. Ochs]

The host reset handler is called with I/O already blocked, thus
there is no need to explicitly block and unblock I/O in the handler.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index cf6a067..729f742 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1966,7 +1966,6 @@ static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
 	switch (cfg->state) {
 	case STATE_NORMAL:
 		cfg->state = STATE_RESET;
-		scsi_block_requests(cfg->host);
 		cxlflash_mark_contexts_error(cfg);
 		rcr = afu_reset(cfg);
 		if (rcr) {
@@ -1975,7 +1974,6 @@ static int cxlflash_eh_host_reset_handler(struct scsi_cmnd *scp)
 		} else
 			cfg->state = STATE_NORMAL;
 		wake_up_all(&cfg->reset_waitq);
-		scsi_unblock_requests(cfg->host);
 		break;
 	case STATE_RESET:
 		wait_event(cfg->reset_waitq, cfg->state != STATE_RESET);
-- 
2.1.0

[PATCH v4 23/32] cxlflash: Fix function prolog parameters and return codes

[Matthew R. Ochs]

Several function prologs have incorrect parameter names and return
code descriptions. This can lead to confusion when reviewing the
source and creates inaccurate documentation.

To remedy, update the function prologs to properly reflect parameter
names and return codes.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 68 ++++++++++++++++----------------------------
 1 file changed, 25 insertions(+), 43 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 729f742..7c76227 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -401,8 +401,7 @@ static void wait_resp(struct afu *afu, struct afu_cmd *cmd)
  * @tmfcmd:	TMF command to send.
  *
  * Return:
- *	0 on success
- *	SCSI_MLQUEUE_HOST_BUSY when host is busy
+ *	0 on success or SCSI_MLQUEUE_HOST_BUSY
  */
 static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
 {
@@ -491,9 +490,7 @@ static const char *cxlflash_driver_info(struct Scsi_Host *host)
  * @host:	SCSI host associated with device.
  * @scp:	SCSI command to send.
  *
- * Return:
- *	0 on success
- *	SCSI_MLQUEUE_HOST_BUSY when host is busy
+ * Return: 0 on success or SCSI_MLQUEUE_HOST_BUSY
  */
 static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 {
@@ -597,7 +594,7 @@ out:
 
 /**
  * cxlflash_wait_for_pci_err_recovery() - wait for error recovery during probe
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  */
 static void cxlflash_wait_for_pci_err_recovery(struct cxlflash_cfg *cfg)
 {
@@ -611,7 +608,7 @@ static void cxlflash_wait_for_pci_err_recovery(struct cxlflash_cfg *cfg)
 
 /**
  * free_mem() - free memory associated with the AFU
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  */
 static void free_mem(struct cxlflash_cfg *cfg)
 {
@@ -633,7 +630,7 @@ static void free_mem(struct cxlflash_cfg *cfg)
 
 /**
  * stop_afu() - stops the AFU command timers and unmaps the MMIO space
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * Safe to call with AFU in a partially allocated/initialized state.
  */
@@ -655,7 +652,7 @@ static void stop_afu(struct cxlflash_cfg *cfg)
 
 /**
  * term_mc() - terminates the master context
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  * @level:	Depth of allocation, where to begin waterfall tear down.
  *
  * Safe to call with AFU/MC in partially allocated/initialized state.
@@ -691,7 +688,7 @@ static void term_mc(struct cxlflash_cfg *cfg, enum undo_level level)
 
 /**
  * term_afu() - terminates the AFU
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * Safe to call with AFU/MC in partially allocated/initialized state.
  */
@@ -751,7 +748,7 @@ static void cxlflash_remove(struct pci_dev *pdev)
 
 /**
  * alloc_mem() - allocates the AFU and its command pool
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * A partially allocated state remains on failure.
  *
@@ -804,12 +801,9 @@ out:
 
 /**
  * init_pci() - initializes the host as a PCI device
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
- * Return:
- *	0 on success
- *	-EIO on unable to communicate with device
- *	A return code from the PCI sub-routines
+ * Return: 0 on success, -errno on failure
  */
 static int init_pci(struct cxlflash_cfg *cfg)
 {
@@ -889,11 +883,9 @@ out_release_regions:
 
 /**
  * init_scsi() - adds the host to the SCSI stack and kicks off host scan
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
- * Return:
- *	0 on success
- *	A return code from adding the host
+ * Return: 0 on success, -errno on failure
  */
 static int init_scsi(struct cxlflash_cfg *cfg)
 {
@@ -1357,7 +1349,7 @@ out:
 
 /**
  * start_context() - starts the master context
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * Return: A success or failure value from CXL services.
  */
@@ -1375,12 +1367,10 @@ static int start_context(struct cxlflash_cfg *cfg)
 
 /**
  * read_vpd() - obtains the WWPNs from VPD
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  * @wwpn:	Array of size NUM_FC_PORTS to pass back WWPNs
  *
- * Return:
- *	0 on success
- *	-ENODEV when VPD or WWPN keywords not found
+ * Return: 0 on success, -errno on failure
  */
 static int read_vpd(struct cxlflash_cfg *cfg, u64 wwpn[])
 {
@@ -1468,7 +1458,7 @@ out:
 
 /**
  * init_pcr() - initialize the provisioning and control registers
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * Also sets up fast access to the mapped registers and initializes AFU
  * command fields that never change.
@@ -1507,7 +1497,7 @@ static void init_pcr(struct cxlflash_cfg *cfg)
 
 /**
  * init_global() - initialize AFU global registers
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  */
 static int init_global(struct cxlflash_cfg *cfg)
 {
@@ -1592,7 +1582,7 @@ out:
 
 /**
  * start_afu() - initializes and starts the AFU
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  */
 static int start_afu(struct cxlflash_cfg *cfg)
 {
@@ -1626,12 +1616,9 @@ static int start_afu(struct cxlflash_cfg *cfg)
 
 /**
  * init_mc() - create and register as the master context
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
- * Return:
- *	0 on success
- *	-ENOMEM when unable to obtain a context from CXL services
- *	A failure value from CXL services.
+ * Return: 0 on success, -errno on failure
  */
 static int init_mc(struct cxlflash_cfg *cfg)
 {
@@ -1715,15 +1702,12 @@ out:
 
 /**
  * init_afu() - setup as master context and start AFU
- * @cxlflash:	Internal structure associated with the host.
+ * @cfg:	Internal structure associated with the host.
  *
  * This routine is a higher level of control for configuring the
  * AFU on probe and reset paths.
  *
- * Return:
- *	0 on success
- *	-ENOMEM when unable to map the AFU MMIO space
- *	A failure value from internal services.
+ * Return: 0 on success, -errno on failure
  */
 static int init_afu(struct cxlflash_cfg *cfg)
 {
@@ -1876,9 +1860,7 @@ out:
  * afu_reset() - resets the AFU
  * @cfg:	Internal structure associated with the host.
  *
- * Return:
- *	0 on success
- *	A failure value from internal services.
+ * Return: 0 on success, -errno on failure
  */
 static int afu_reset(struct cxlflash_cfg *cfg)
 {
@@ -2363,7 +2345,7 @@ static void cxlflash_worker_thread(struct work_struct *work)
  * @pdev:	PCI device associated with the host.
  * @dev_id:	PCI device id associated with device.
  *
- * Return: 0 on success / non-zero on failure
+ * Return: 0 on success, -errno on failure
  */
 static int cxlflash_probe(struct pci_dev *pdev,
 			  const struct pci_device_id *dev_id)
@@ -2597,7 +2579,7 @@ static struct pci_driver cxlflash_driver = {
 /**
  * init_cxlflash() - module entry point
  *
- * Return: 0 on success / non-zero on failure
+ * Return: 0 on success, -errno on failure
  */
 static int __init init_cxlflash(void)
 {
-- 
2.1.0

[PATCH v4 24/32] cxlflash: Fix MMIO and endianness errors

[Matthew R. Ochs]

Sparse uncovered several errors with MMIO operations (accessing
directly) and handling endianness. These can cause issues when
running in different environments.

Introduce __iomem and proper endianness tags/swaps where
appropriate to make driver sparse clean.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h    | 10 +++++-----
 drivers/scsi/cxlflash/main.c      | 25 +++++++++++++------------
 drivers/scsi/cxlflash/superpipe.c |  6 +++---
 drivers/scsi/cxlflash/superpipe.h |  2 +-
 drivers/scsi/cxlflash/vlun.c      |  4 ++--
 5 files changed, 24 insertions(+), 23 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 3be5754..a810585 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -164,9 +164,9 @@ struct afu {
 
 	/* AFU HW */
 	struct cxl_ioctl_start_work work;
-	struct cxlflash_afu_map *afu_map;	/* entire MMIO map */
-	struct sisl_host_map *host_map;		/* MC host map */
-	struct sisl_ctrl_map *ctrl_map;		/* MC control map */
+	struct cxlflash_afu_map __iomem *afu_map;	/* entire MMIO map */
+	struct sisl_host_map __iomem *host_map;		/* MC host map */
+	struct sisl_ctrl_map __iomem *ctrl_map;		/* MC control map */
 
 	ctx_hndl_t ctx_hndl;	/* master's context handle */
 	u64 *hrrq_start;
@@ -188,10 +188,10 @@ struct afu {
 
 static inline u64 lun_to_lunid(u64 lun)
 {
-	u64 lun_id;
+	__be64 lun_id;
 
 	int_to_scsilun(lun, (struct scsi_lun *)&lun_id);
-	return swab64(lun_id);
+	return be64_to_cpu(lun_id);
 }
 
 int cxlflash_afu_sync(struct afu *, ctx_hndl_t, res_hndl_t, u8);
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 7c76227..4893516 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -644,7 +644,7 @@ static void stop_afu(struct cxlflash_cfg *cfg)
 			complete(&afu->cmd[i].cevent);
 
 		if (likely(afu->afu_map)) {
-			cxl_psa_unmap((void *)afu->afu_map);
+			cxl_psa_unmap((void __iomem *)afu->afu_map);
 			afu->afu_map = NULL;
 		}
 	}
@@ -914,7 +914,7 @@ out:
  * that the FC link layer has synced, completed the handshaking process, and
  * is ready for login to start.
  */
-static void set_port_online(u64 *fc_regs)
+static void set_port_online(__be64 __iomem *fc_regs)
 {
 	u64 cmdcfg;
 
@@ -930,7 +930,7 @@ static void set_port_online(u64 *fc_regs)
  *
  * The provided MMIO region must be mapped prior to call.
  */
-static void set_port_offline(u64 *fc_regs)
+static void set_port_offline(__be64 __iomem *fc_regs)
 {
 	u64 cmdcfg;
 
@@ -954,7 +954,7 @@ static void set_port_offline(u64 *fc_regs)
  *	FALSE (0) when the specified port fails to come online after timeout
  *	-EINVAL when @delay_us is less than 1000
  */
-static int wait_port_online(u64 *fc_regs, u32 delay_us, u32 nretry)
+static int wait_port_online(__be64 __iomem *fc_regs, u32 delay_us, u32 nretry)
 {
 	u64 status;
 
@@ -985,7 +985,7 @@ static int wait_port_online(u64 *fc_regs, u32 delay_us, u32 nretry)
  *	FALSE (0) when the specified port fails to go offline after timeout
  *	-EINVAL when @delay_us is less than 1000
  */
-static int wait_port_offline(u64 *fc_regs, u32 delay_us, u32 nretry)
+static int wait_port_offline(__be64 __iomem *fc_regs, u32 delay_us, u32 nretry)
 {
 	u64 status;
 
@@ -1020,7 +1020,8 @@ static int wait_port_offline(u64 *fc_regs, u32 delay_us, u32 nretry)
  *	0 when the WWPN is successfully written and the port comes back online
  *	-1 when the port fails to go offline or come back up online
  */
-static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 wwpn)
+static int afu_set_wwpn(struct afu *afu, int port, __be64 __iomem *fc_regs,
+			u64 wwpn)
 {
 	int rc = 0;
 
@@ -1065,7 +1066,7 @@ static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 wwpn)
  * the alternate port exclusively while the reset takes place.
  * failure to come online is overridden.
  */
-static void afu_link_reset(struct afu *afu, int port, u64 *fc_regs)
+static void afu_link_reset(struct afu *afu, int port, __be64 __iomem *fc_regs)
 {
 	u64 port_sel;
 
@@ -1280,7 +1281,7 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 	struct device *dev = &cfg->dev->dev;
 	u64 reg_unmasked;
 	const struct asyc_intr_info *info;
-	struct sisl_global_map *global = &afu->afu_map->global;
+	struct sisl_global_map __iomem *global = &afu->afu_map->global;
 	u64 reg;
 	u8 port;
 	int i;
@@ -1466,7 +1467,7 @@ out:
 static void init_pcr(struct cxlflash_cfg *cfg)
 {
 	struct afu *afu = cfg->afu;
-	struct sisl_ctrl_map *ctrl_map;
+	struct sisl_ctrl_map __iomem *ctrl_map;
 	int i;
 
 	for (i = 0; i < MAX_CONTEXT; i++) {
@@ -1755,7 +1756,7 @@ static int init_afu(struct cxlflash_cfg *cfg)
 		dev_err(dev, "%s: call to start_afu failed, rc=%d!\n",
 			__func__, rc);
 		term_mc(cfg, UNDO_START);
-		cxl_psa_unmap((void *)afu->afu_map);
+		cxl_psa_unmap((void __iomem *)afu->afu_map);
 		afu->afu_map = NULL;
 		goto err1;
 	}
@@ -1835,8 +1836,8 @@ retry:
 	cmd->rcb.cdb[1] = mode;
 
 	/* The cdb is aligned, no unaligned accessors required */
-	*((u16 *)&cmd->rcb.cdb[2]) = swab16(ctx_hndl_u);
-	*((u32 *)&cmd->rcb.cdb[4]) = swab32(res_hndl_u);
+	*((__be16 *)&cmd->rcb.cdb[2]) = cpu_to_be16(ctx_hndl_u);
+	*((__be32 *)&cmd->rcb.cdb[4]) = cpu_to_be32(res_hndl_u);
 
 	rc = send_cmd(afu, cmd);
 	if (unlikely(rc))
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 8a7ec5d..a6316f5 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -253,7 +253,7 @@ static int afu_attach(struct cxlflash_cfg *cfg, struct ctx_info *ctxi)
 {
 	struct device *dev = &cfg->dev->dev;
 	struct afu *afu = cfg->afu;
-	struct sisl_ctrl_map *ctrl_map = ctxi->ctrl_map;
+	struct sisl_ctrl_map __iomem *ctrl_map = ctxi->ctrl_map;
 	int rc = 0;
 	u64 val;
 
@@ -365,8 +365,8 @@ retry:
 	 * as the buffer is allocated on an aligned boundary.
 	 */
 	mutex_lock(&gli->mutex);
-	gli->max_lba = be64_to_cpu(*((u64 *)&cmd_buf[0]));
-	gli->blk_len = be32_to_cpu(*((u32 *)&cmd_buf[8]));
+	gli->max_lba = be64_to_cpu(*((__be64 *)&cmd_buf[0]));
+	gli->blk_len = be32_to_cpu(*((__be32 *)&cmd_buf[8]));
 	mutex_unlock(&gli->mutex);
 
 out:
diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index 7947091..7df88ee 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -91,7 +91,7 @@ enum ctx_ctrl {
 #define DECODE_CTXID(_val)	(_val & 0xFFFFFFFF)
 
 struct ctx_info {
-	struct sisl_ctrl_map *ctrl_map; /* initialized at startup */
+	struct sisl_ctrl_map __iomem *ctrl_map; /* initialized at startup */
 	struct sisl_rht_entry *rht_start; /* 1 page (req'd for alignment),
 					     alloc/free on attach/detach */
 	u32 rht_out;		/* Number of checked out RHT entries */
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 96b074f..f91b5b3 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -786,7 +786,7 @@ void cxlflash_restore_luntable(struct cxlflash_cfg *cfg)
 	u32 chan;
 	u32 lind;
 	struct afu *afu = cfg->afu;
-	struct sisl_global_map *agm = &afu->afu_map->global;
+	struct sisl_global_map __iomem *agm = &afu->afu_map->global;
 
 	mutex_lock(&global.mutex);
 
@@ -831,7 +831,7 @@ static int init_luntable(struct cxlflash_cfg *cfg, struct llun_info *lli)
 	u32 lind;
 	int rc = 0;
 	struct afu *afu = cfg->afu;
-	struct sisl_global_map *agm = &afu->afu_map->global;
+	struct sisl_global_map __iomem *agm = &afu->afu_map->global;
 
 	mutex_lock(&global.mutex);
 
-- 
2.1.0

[PATCH v4 25/32] cxlflash: Fix to prevent EEH recovery failure

[Matthew R. Ochs]

The process_sense() routine can perform a read capacity which
can take some time to complete. If an EEH occurs while waiting
on the read capacity, the EEH handler is unable to obtain the
context's mutex in order to put the context in an error state.
The EEH handler will sit and wait until the context is free,
but this wait can last longer than the EEH handler tolerates,
leading to a failed recovery.

To address this issue, make the context unavailable to new,
non-system owned threads and release the context while calling
into process_sense(). After returning from process_sense() the
context mutex is reacquired and the context is made available
again. The context can be safely moved to the error state if
needed during the unavailable window as no other threads will
hold its reference.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index a6316f5..7283e83 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -1787,12 +1787,21 @@ static int cxlflash_disk_verify(struct scsi_device *sdev,
 	 * inquiry (i.e. the Unit attention is due to the WWN changing).
 	 */
 	if (verify->hint & DK_CXLFLASH_VERIFY_HINT_SENSE) {
+		/* Can't hold mutex across process_sense/read_cap16,
+		 * since we could have an intervening EEH event.
+		 */
+		ctxi->unavail = true;
+		mutex_unlock(&ctxi->mutex);
 		rc = process_sense(sdev, verify);
 		if (unlikely(rc)) {
 			dev_err(dev, "%s: Failed to validate sense data (%d)\n",
 				__func__, rc);
+			mutex_lock(&ctxi->mutex);
+			ctxi->unavail = false;
 			goto out;
 		}
+		mutex_lock(&ctxi->mutex);
+		ctxi->unavail = false;
 	}
 
 	switch (gli->mode) {
-- 
2.1.0

[PATCH v4 26/32] cxlflash: Correct spelling, grammar, and alignment mistakes

[Matthew R. Ochs]

There are several spelling and grammar mistakes throughout the
driver. Additionally there are a handful of places where there
are extra lines and unnecessary variables/statements. These are
a nuisance and pollute the driver.

Fix spelling and grammar issues. Update some comments for clarity and
consistency. Remove extra lines and a few unneeded variables/statements.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h    |  2 --
 drivers/scsi/cxlflash/main.c      | 62 +++++++++++++++++----------------------
 drivers/scsi/cxlflash/sislite.h   |  6 ++--
 drivers/scsi/cxlflash/superpipe.c |  2 +-
 drivers/scsi/cxlflash/vlun.c      | 14 ++++-----
 5 files changed, 38 insertions(+), 48 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index a810585..bbfe711 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -105,8 +105,6 @@ struct cxlflash_cfg {
 	atomic_t scan_host_needed;
 
 	struct cxl_afu *cxl_afu;
-
-	struct pci_pool *cxlflash_cmd_pool;
 	struct pci_dev *parent_dev;
 
 	atomic_t recovery_threads;
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 4893516..24aedfb 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -34,7 +34,6 @@ MODULE_AUTHOR("Manoj N. Kumar <manoj@linux.vnet.ibm.com>");
 MODULE_AUTHOR("Matthew R. Ochs <mrochs@linux.vnet.ibm.com>");
 MODULE_LICENSE("GPL");
 
-
 /**
  * cmd_checkout() - checks out an AFU command
  * @afu:	AFU to checkout from.
@@ -730,7 +729,7 @@ static void cxlflash_remove(struct pci_dev *pdev)
 	case INIT_STATE_SCSI:
 		cxlflash_term_local_luns(cfg);
 		scsi_remove_host(cfg->host);
-		/* Fall through */
+		/* fall through */
 	case INIT_STATE_AFU:
 		term_afu(cfg);
 		cancel_work_sync(&cfg->work_q);
@@ -763,9 +762,7 @@ static int alloc_mem(struct cxlflash_cfg *cfg)
 	char *buf = NULL;
 	struct device *dev = &cfg->dev->dev;
 
-	/* This allocation is about 12K, i.e. only 1 64k page
-	 * and upto 4 4k pages
-	 */
+	/* AFU is ~12k, i.e. only one 64k page or up to four 4k pages */
 	cfg->afu = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
 					    get_order(sizeof(struct afu)));
 	if (unlikely(!cfg->afu)) {
@@ -1295,10 +1292,10 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 		goto out;
 	}
 
-	/* it is OK to clear AFU status before FC_ERROR */
+	/* FYI, it is 'okay' to clear AFU status before FC_ERROR */
 	writeq_be(reg_unmasked, &global->regs.aintr_clear);
 
-	/* check each bit that is on */
+	/* Check each bit that is on */
 	for (i = 0; reg_unmasked; i++, reg_unmasked = (reg_unmasked >> 1)) {
 		info = find_ainfo(1ULL << i);
 		if (((reg_unmasked & 0x1) == 0) || !info)
@@ -1311,7 +1308,7 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 		       readq_be(&global->fc_regs[port][FC_STATUS / 8]));
 
 		/*
-		 * do link reset first, some OTHER errors will set FC_ERROR
+		 * Do link reset first, some OTHER errors will set FC_ERROR
 		 * again if cleared before or w/o a reset
 		 */
 		if (info->action & LINK_RESET) {
@@ -1326,7 +1323,7 @@ static irqreturn_t cxlflash_async_err_irq(int irq, void *data)
 			reg = readq_be(&global->fc_regs[port][FC_ERROR / 8]);
 
 			/*
-			 * since all errors are unmasked, FC_ERROR and FC_ERRCAP
+			 * Since all errors are unmasked, FC_ERROR and FC_ERRCAP
 			 * should be the same and tracing one is sufficient.
 			 */
 
@@ -1472,23 +1469,22 @@ static void init_pcr(struct cxlflash_cfg *cfg)
 
 	for (i = 0; i < MAX_CONTEXT; i++) {
 		ctrl_map = &afu->afu_map->ctrls[i].ctrl;
-		/* disrupt any clients that could be running */
-		/* e. g. clients that survived a master restart */
+		/* Disrupt any clients that could be running */
+		/* e.g. clients that survived a master restart */
 		writeq_be(0, &ctrl_map->rht_start);
 		writeq_be(0, &ctrl_map->rht_cnt_id);
 		writeq_be(0, &ctrl_map->ctx_cap);
 	}
 
-	/* copy frequently used fields into afu */
+	/* Copy frequently used fields into afu */
 	afu->ctx_hndl = (u16) cxl_process_element(cfg->mcctx);
-	/* ctx_hndl is 16 bits in CAIA */
 	afu->host_map = &afu->afu_map->hosts[afu->ctx_hndl].host;
 	afu->ctrl_map = &afu->afu_map->ctrls[afu->ctx_hndl].ctrl;
 
 	/* Program the Endian Control for the master context */
 	writeq_be(SISL_ENDIAN_CTRL, &afu->host_map->endian_ctrl);
 
-	/* initialize cmd fields that never change */
+	/* Initialize cmd fields that never change */
 	for (i = 0; i < CXLFLASH_NUM_CMDS; i++) {
 		afu->cmd[i].rcb.ctx_id = afu->ctx_hndl;
 		afu->cmd[i].rcb.msi = SISL_MSI_RRQ_UPDATED;
@@ -1517,7 +1513,7 @@ static int init_global(struct cxlflash_cfg *cfg)
 
 	pr_debug("%s: wwpn0=0x%llX wwpn1=0x%llX\n", __func__, wwpn[0], wwpn[1]);
 
-	/* set up RRQ in AFU for master issued cmds */
+	/* Set up RRQ in AFU for master issued cmds */
 	writeq_be((u64) afu->hrrq_start, &afu->host_map->rrq_start);
 	writeq_be((u64) afu->hrrq_end, &afu->host_map->rrq_end);
 
@@ -1530,9 +1526,9 @@ static int init_global(struct cxlflash_cfg *cfg)
 	/* checker on if dual afu */
 	writeq_be(reg, &afu->afu_map->global.regs.afu_config);
 
-	/* global port select: select either port */
+	/* Global port select: select either port */
 	if (afu->internal_lun) {
-		/* only use port 0 */
+		/* Only use port 0 */
 		writeq_be(PORT0, &afu->afu_map->global.regs.afu_port_sel);
 		num_ports = NUM_FC_PORTS - 1;
 	} else {
@@ -1541,15 +1537,15 @@ static int init_global(struct cxlflash_cfg *cfg)
 	}
 
 	for (i = 0; i < num_ports; i++) {
-		/* unmask all errors (but they are still masked at AFU) */
+		/* Unmask all errors (but they are still masked at AFU) */
 		writeq_be(0, &afu->afu_map->global.fc_regs[i][FC_ERRMSK / 8]);
-		/* clear CRC error cnt & set a threshold */
+		/* Clear CRC error cnt & set a threshold */
 		(void)readq_be(&afu->afu_map->global.
 			       fc_regs[i][FC_CNT_CRCERR / 8]);
 		writeq_be(MC_CRC_THRESH, &afu->afu_map->global.fc_regs[i]
 			  [FC_CRC_THRESH / 8]);
 
-		/* set WWPNs. If already programmed, wwpn[i] is 0 */
+		/* Set WWPNs. If already programmed, wwpn[i] is 0 */
 		if (wwpn[i] != 0 &&
 		    afu_set_wwpn(afu, i,
 				 &afu->afu_map->global.fc_regs[i][0],
@@ -1563,18 +1559,17 @@ static int init_global(struct cxlflash_cfg *cfg)
 		 * offline/online transitions and a PLOGI
 		 */
 		msleep(100);
-
 	}
 
-	/* set up master's own CTX_CAP to allow real mode, host translation */
-	/* tbls, afu cmds and read/write GSCSI cmds. */
+	/* Set up master's own CTX_CAP to allow real mode, host translation */
+	/* tables, afu cmds and read/write GSCSI cmds. */
 	/* First, unlock ctx_cap write by reading mbox */
 	(void)readq_be(&afu->ctrl_map->mbox_r);	/* unlock ctx_cap */
 	writeq_be((SISL_CTX_CAP_REAL_MODE | SISL_CTX_CAP_HOST_XLATE |
 		   SISL_CTX_CAP_READ_CMD | SISL_CTX_CAP_WRITE_CMD |
 		   SISL_CTX_CAP_AFU_CMD | SISL_CTX_CAP_GSCSI_CMD),
 		  &afu->ctrl_map->ctx_cap);
-	/* init heartbeat */
+	/* Initialize heartbeat */
 	afu->hb = readq_be(&afu->afu_map->global.regs.afu_hb);
 
 out:
@@ -1603,7 +1598,7 @@ static int start_afu(struct cxlflash_cfg *cfg)
 
 	init_pcr(cfg);
 
-	/* initialize RRQ pointers */
+	/* Initialize RRQ pointers */
 	afu->hrrq_start = &afu->rrq_entry[0];
 	afu->hrrq_end = &afu->rrq_entry[NUM_RRQ_ENTRY - 1];
 	afu->hrrq_curr = afu->hrrq_start;
@@ -1726,8 +1721,7 @@ static int init_afu(struct cxlflash_cfg *cfg)
 		goto err1;
 	}
 
-	/* Map the entire MMIO space of the AFU.
-	 */
+	/* Map the entire MMIO space of the AFU */
 	afu->afu_map = cxl_psa_map(cfg->mcctx);
 	if (!afu->afu_map) {
 		rc = -ENOMEM;
@@ -1779,7 +1773,7 @@ err1:
  * @mode:	Type of sync to issue (lightweight, heavyweight, global).
  *
  * The AFU can only take 1 sync command at a time. This routine enforces this
- * limitation by using a mutex to provide exlusive access to the AFU during
+ * limitation by using a mutex to provide exclusive access to the AFU during
  * the sync. This design point requires calling threads to not be on interrupt
  * context due to the possibility of sleeping during concurrent sync operations.
  *
@@ -1845,7 +1839,7 @@ retry:
 
 	wait_resp(afu, cmd);
 
-	/* set on timeout */
+	/* Set on timeout */
 	if (unlikely((cmd->sa.ioasc != 0) ||
 		     (cmd->sa.host_use_b[0] & B_ERROR)))
 		rc = -1;
@@ -2262,7 +2256,7 @@ static struct scsi_host_template driver_template = {
 	.cmd_per_lun = 16,
 	.can_queue = CXLFLASH_MAX_CMDS,
 	.this_id = -1,
-	.sg_tablesize = SG_NONE,	/* No scatter gather support. */
+	.sg_tablesize = SG_NONE,	/* No scatter gather support */
 	.max_sectors = CXLFLASH_MAX_SECTORS,
 	.use_clustering = ENABLE_CLUSTERING,
 	.shost_attrs = cxlflash_host_attrs,
@@ -2322,8 +2316,7 @@ static void cxlflash_worker_thread(struct work_struct *work)
 
 			/* The reset can block... */
 			afu_link_reset(afu, port,
-				       &afu->afu_map->
-				       global.fc_regs[port][0]);
+				       &afu->afu_map->global.fc_regs[port][0]);
 			spin_lock_irqsave(cfg->host->host_lock, lock_flags);
 		}
 
@@ -2402,7 +2395,6 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->last_lun_index[1] = CXLFLASH_NUM_VLUNS/2 - 1;
 
 	cfg->dev_id = (struct pci_device_id *)dev_id;
-	cfg->mcctx = NULL;
 
 	init_waitqueue_head(&cfg->tmf_waitq);
 	init_waitqueue_head(&cfg->reset_waitq);
@@ -2418,7 +2410,8 @@ static int cxlflash_probe(struct pci_dev *pdev,
 
 	pci_set_drvdata(pdev, cfg);
 
-	/* Use the special service provided to look up the physical
+	/*
+	 * Use the special service provided to look up the physical
 	 * PCI device, since we are called on the probe of the virtual
 	 * PCI host bus (vphb)
 	 */
@@ -2448,7 +2441,6 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	}
 	cfg->init_state = INIT_STATE_AFU;
 
-
 	rc = init_scsi(cfg);
 	if (rc) {
 		dev_err(&pdev->dev, "%s: call to init_scsi "
diff --git a/drivers/scsi/cxlflash/sislite.h b/drivers/scsi/cxlflash/sislite.h
index 8425d1a..0b3366f 100644
--- a/drivers/scsi/cxlflash/sislite.h
+++ b/drivers/scsi/cxlflash/sislite.h
@@ -146,7 +146,7 @@ struct sisl_rc {
 #define SISL_FC_RC_ABORTFAIL	0x59	/* pending abort completed w/fail */
 #define SISL_FC_RC_RESID	0x5A	/* ioasa underrun/overrun flags set */
 #define SISL_FC_RC_RESIDERR	0x5B	/* actual data len does not match SCSI
-					   reported len, possbly due to dropped
+					   reported len, possibly due to dropped
 					   frames */
 #define SISL_FC_RC_TGTABORT	0x5C	/* command aborted by target */
 };
@@ -258,7 +258,7 @@ struct sisl_host_map {
 	__be64 rrq_start;	/* start & end are both inclusive */
 	__be64 rrq_end;		/* write sequence: start followed by end */
 	__be64 cmd_room;
-	__be64 ctx_ctrl;	/* least signiifcant byte or b56:63 is LISN# */
+	__be64 ctx_ctrl;	/* least significant byte or b56:63 is LISN# */
 	__be64 mbox_w;		/* restricted use */
 };
 
@@ -290,7 +290,7 @@ struct sisl_global_regs {
 #define SISL_ASTATUS_FC0_LOGO    0x4000ULL /* b49, target sent FLOGI/PLOGI/LOGO
 						   while logged in */
 #define SISL_ASTATUS_FC0_CRC_T   0x2000ULL /* b50, CRC threshold exceeded */
-#define SISL_ASTATUS_FC0_LOGI_R  0x1000ULL /* b51, login state mechine timed out
+#define SISL_ASTATUS_FC0_LOGI_R  0x1000ULL /* b51, login state machine timed out
 						   and retrying */
 #define SISL_ASTATUS_FC0_LOGI_F  0x0800ULL /* b52, login failed,
 					      FC_ERROR[19:0] */
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 7283e83..3cc8609 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -76,7 +76,7 @@ void cxlflash_free_errpage(void)
  *
  * When the host needs to go down, all users must be quiesced and their
  * memory freed. This is accomplished by putting the contexts in error
- * state which will notify the user and let them 'drive' the tear-down.
+ * state which will notify the user and let them 'drive' the tear down.
  * Meanwhile, this routine camps until all user contexts have been removed.
  */
 void cxlflash_stop_term_user_contexts(struct cxlflash_cfg *cfg)
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index f91b5b3..b0eaf55 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -132,7 +132,7 @@ static int ba_init(struct ba_lun *ba_lun)
 		return -ENOMEM;
 	}
 
-	/* Pass the allocated lun info as a handle to the user */
+	/* Pass the allocated LUN info as a handle to the user */
 	ba_lun->ba_lun_handle = bali;
 
 	pr_debug("%s: Successfully initialized the LUN: "
@@ -165,7 +165,7 @@ static int find_free_range(u32 low,
 			num_bits = (sizeof(*lam) * BITS_PER_BYTE);
 			bit_pos = find_first_bit(lam, num_bits);
 
-			pr_devel("%s: Found free bit %llX in lun "
+			pr_devel("%s: Found free bit %llX in LUN "
 				 "map entry %llX at bitmap index = %X\n",
 				 __func__, bit_pos, bali->lun_alloc_map[i],
 				 i);
@@ -682,14 +682,14 @@ out:
 }
 
 /**
- * _cxlflash_vlun_resize() - changes the size of a virtual lun
+ * _cxlflash_vlun_resize() - changes the size of a virtual LUN
  * @sdev:	SCSI device associated with LUN owning virtual LUN.
  * @ctxi:	Context owning resources.
  * @resize:	Resize ioctl data structure.
  *
  * On successful return, the user is informed of the new size (in blocks)
- * of the virtual lun in last LBA format. When the size of the virtual
- * lun is zero, the last LBA is reflected as -1. See comment in the
+ * of the virtual LUN in last LBA format. When the size of the virtual
+ * LUN is zero, the last LBA is reflected as -1. See comment in the
  * prologue for _cxlflash_disk_release() regarding AFU syncs and contexts
  * on the error recovery list.
  *
@@ -886,8 +886,8 @@ out:
  * @arg:	UVirtual ioctl data structure.
  *
  * On successful return, the user is informed of the resource handle
- * to be used to identify the virtual lun and the size (in blocks) of
- * the virtual lun in last LBA format. When the size of the virtual lun
+ * to be used to identify the virtual LUN and the size (in blocks) of
+ * the virtual LUN in last LBA format. When the size of the virtual LUN
  * is zero, the last LBA is reflected as -1.
  *
  * Return: 0 on success, -errno on failure
-- 
2.1.0

[PATCH v4 27/32] cxlflash: Fix to prevent stale AFU RRQ

[Matthew R. Ochs]

Following an adapter reset, the AFU RRQ that resides in host memory
holds stale data. This can lead to a condition where the RRQ interrupt
handler tries to process stale entries and/or endlessly loops due to an
out of sync generation bit.

To fix, the AFU RRQ in host memory needs to be cleared after each reset.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 24aedfb..ab11ee6 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -1598,6 +1598,9 @@ static int start_afu(struct cxlflash_cfg *cfg)
 
 	init_pcr(cfg);
 
+	/* After an AFU reset, RRQ entries are stale, clear them */
+	memset(&afu->rrq_entry, 0, sizeof(afu->rrq_entry));
+
 	/* Initialize RRQ pointers */
 	afu->hrrq_start = &afu->rrq_entry[0];
 	afu->hrrq_end = &afu->rrq_entry[NUM_RRQ_ENTRY - 1];
-- 
2.1.0

[PATCH v4 28/32] MAINTAINERS: Add cxlflash driver

[Matthew R. Ochs]

Add stanza for cxlflash SCSI driver.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
---
 MAINTAINERS | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 274f854..f2f3046 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3153,6 +3153,15 @@ F:	Documentation/powerpc/cxl.txt
 F:	Documentation/powerpc/cxl.txt
 F:	Documentation/ABI/testing/sysfs-class-cxl
 
+CXLFLASH (IBM Coherent Accelerator Processor Interface CAPI Flash) SCSI DRIVER
+M:	Manoj N. Kumar <manoj@linux.vnet.ibm.com>
+M:	Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
+L:	linux-scsi@vger.kernel.org
+S:	Supported
+F:	drivers/scsi/cxlflash/
+F:	include/uapi/scsi/cxlflash_ioctls.h
+F:	Documentation/powerpc/cxlflash.txt
+
 STMMAC ETHERNET DRIVER
 M:	Giuseppe Cavallaro <peppe.cavallaro@st.com>
 L:	netdev@vger.kernel.org
-- 
2.1.0

[PATCH v4 29/32] cxlflash: Fix to double the delay each time

[Matthew R. Ochs]

From: Manoj Kumar <manoj@linux.vnet.ibm.com>

The operator used to double the delay is incorrect and
does not result in delay doubling.

To fix, use a left shift instead of the XOR operator.

Reported-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index ab11ee6..be78906 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -303,7 +303,7 @@ write_rrin:
 		if (rrin != 0x1)
 			break;
 		/* Double delay each time */
-		udelay(2 ^ nretry);
+		udelay(2 << nretry);
 	} while (nretry++ < MC_ROOM_RETRY_CNT);
 }
 
-- 
2.1.0

[PATCH v4 30/32] cxlflash: Fix to avoid corrupting adapter fops

[Matthew R. Ochs]

The fops owned by the adapter can be corrupted in certain scenarios,
opening a window where certain fops are temporarily NULLed before being
reset to their proper value. This can potentially lead software to make
incorrect decisions, leaving the user with the inability to function as
intended.

An example of this behavior can be observed when there are a number of
users with a high rate of turn around (attach to LUN, perform an I/O,
detach from LUN, repeat). Every so often a user is given a valid
context and adapter file descriptor, but the file associated with the
descriptor lacks the correct read permission bit (FMODE_CAN_READ) and
thus the read system call bails before calling the valid read fop.

Background:

The fops is stored in the adapter structure to provide the ability to
lookup the adapter structure from within the fop handler. CXL services
use the file's private_data and at present, the CXL context does not
have a private section. In an effort to limit areas of the cxlflash
driver with code specific the superpipe function, a design choice was
made to keep the details of the fops situated away from the legacy
portions of the driver. This drove the behavior that the adapter fops
is set at the beginning of the disk attach ioctl handler when there
are no users present.

The corruption that this fix remedies is due to the fact that the fops
is initially defaulted to values found within a static structure. When
the fops is handed down to the CXL services later in the attach path,
certain services are patched. The fops structure remains correct until
the user count drops to 0 and the fops is reset, triggering the process
to repeat again. The user counts are tightly coupled with the creation
and deletion of the user context. If multiple users perform a disk
attach at the same time, when the user count is currently 0, some users
can be in the middle of obtaining a file descriptor and have not yet
reached the context creation code that [in addition to creating the
context] increments the user count. Subsequent users coming in to
perform the attach see that the user count is still 0, and reinitialize
the fops, temporarily removing the patched fops. The users that are in
the middle obtaining their file descriptor may then receive an invalid
descriptor.

The fix simply removes the user count altogether and moves the fops
initialization to probe time such that it is only performed one time
for the life of the adapter. In the future, if the CXL services adopt
a private member for their context, that could be used to store the
adapter structure reference and cxlflash could revert to a model that
does not require an embedded fops.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h    |  3 +--
 drivers/scsi/cxlflash/main.c      |  1 +
 drivers/scsi/cxlflash/superpipe.c | 11 +----------
 3 files changed, 3 insertions(+), 12 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index bbfe711..c11cd19 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -21,6 +21,7 @@
 #include <scsi/scsi.h>
 #include <scsi/scsi_device.h>
 
+extern const struct file_operations cxlflash_cxl_fops;
 
 #define MAX_CONTEXT  CXLFLASH_MAX_CONTEXT       /* num contexts per afu */
 
@@ -115,8 +116,6 @@ struct cxlflash_cfg {
 	struct list_head ctx_err_recovery; /* contexts w/ recovery pending */
 	struct file_operations cxl_fops;
 
-	atomic_t num_user_contexts;
-
 	/* Parameters that are LUN table related */
 	int last_lun_index[CXLFLASH_NUM_FC_PORTS];
 	int promote_lun_index;
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index be78906..38e7edc 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -2386,6 +2386,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 
 	cfg->init_state = INIT_STATE_NONE;
 	cfg->dev = pdev;
+	cfg->cxl_fops = cxlflash_cxl_fops;
 
 	/*
 	 * The promoted LUNs move to the top of the LUN table. The rest stay
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 3cc8609..f625e07 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -712,7 +712,6 @@ static void destroy_context(struct cxlflash_cfg *cfg,
 	kfree(ctxi->rht_needs_ws);
 	kfree(ctxi->rht_lun);
 	kfree(ctxi);
-	atomic_dec_if_positive(&cfg->num_user_contexts);
 }
 
 /**
@@ -769,7 +768,6 @@ static struct ctx_info *create_context(struct cxlflash_cfg *cfg,
 	INIT_LIST_HEAD(&ctxi->luns);
 	INIT_LIST_HEAD(&ctxi->list); /* initialize for list_empty() */
 
-	atomic_inc(&cfg->num_user_contexts);
 	mutex_lock(&ctxi->mutex);
 out:
 	return ctxi;
@@ -1164,10 +1162,7 @@ out:
 	return rc;
 }
 
-/*
- * Local fops for adapter file descriptor
- */
-static const struct file_operations cxlflash_cxl_fops = {
+const struct file_operations cxlflash_cxl_fops = {
 	.owner = THIS_MODULE,
 	.mmap = cxlflash_cxl_mmap,
 	.release = cxlflash_cxl_release,
@@ -1286,10 +1281,6 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
 
 	int fd = -1;
 
-	/* On first attach set fileops */
-	if (atomic_read(&cfg->num_user_contexts) == 0)
-		cfg->cxl_fops = cxlflash_cxl_fops;
-
 	if (attach->num_interrupts > 4) {
 		dev_dbg(dev, "%s: Cannot support this many interrupts %llu\n",
 			__func__, attach->num_interrupts);
-- 
2.1.0

[PATCH v4 31/32] cxlflash: Correct trace string

[Matthew R. Ochs]

The trace following the failure of alloc_mem() incorrectly identifies
which function failed. This can lead to misdiagnosing a failure.

Fix the string to correctly indicate that alloc_mem() failed.

Reported-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index 38e7edc..e8f8c7e 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -2377,7 +2377,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->host = host;
 	rc = alloc_mem(cfg);
 	if (rc) {
-		dev_err(&pdev->dev, "%s: call to scsi_host_alloc failed!\n",
+		dev_err(&pdev->dev, "%s: call to alloc_mem failed!\n",
 			__func__);
 		rc = -ENOMEM;
 		scsi_host_put(cfg->host);
-- 
2.1.0

[PATCH v4 32/32] cxlflash: Fix to avoid potential deadlock on EEH

[Matthew R. Ochs]

Ioctl threads that use scsi_execute() can run for an excessive amount
of time due to the fact that they have lengthy timeouts and retry logic
built in. Under normal operation this is not an issue. However, once EEH
enters the picture, a long execution time coupled with the possibility
that a timeout can trigger entry to the driver via registered reset
callbacks becomes a liability.

In particular, a deadlock can occur when an EEH event is encountered
while in running in scsi_execute(). As part of the recovery, the EEH
handler drains all currently running ioctls, waiting until they have
completed before proceeding with a reset. As the scsi_execute()'s are
situated on the ioctl path, the EEH handler will wait until they (and
the remainder of the ioctl handler they're associated with) have
completed. Normally this would not be much of an issue aside from the
longer recovery period. Unfortunately, the scsi_execute() triggers a
reset when it times out. The reset handler will see that the device is
already being reset and wait until that reset completed. This creates
a condition where the EEH handler becomes stuck, infinitely waiting for
the ioctl thread to complete.

To avoid this behavior, temporarily unmark the scsi_execute() threads
as an ioctl thread by releasing the ioctl read semaphore. This allows
the EEH handler to proceed with a recovery while the thread is still
running. Once the scsi_execute() returns, the ioctl read semaphore is
reacquired and the adapter state is rechecked in case it changed while
inside of scsi_execute(). The state check will wait if the adapter is
still being recovered or returns a failure if the recovery failed. In
the event that the adapter reset failed, the failure is simply returned
as the ioctl would be unable to continue.

Reported-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 30 +++++++++++++++++++++++++++++-
 drivers/scsi/cxlflash/superpipe.h |  2 ++
 drivers/scsi/cxlflash/vlun.c      | 29 +++++++++++++++++++++++++++++
 3 files changed, 60 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index f625e07..8af7cdc 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -283,6 +283,24 @@ out:
  * @sdev:	SCSI device associated with LUN.
  * @lli:	LUN destined for capacity request.
  *
+ * The READ_CAP16 can take quite a while to complete. Should an EEH occur while
+ * in scsi_execute(), the EEH handler will attempt to recover. As part of the
+ * recovery, the handler drains all currently running ioctls, waiting until they
+ * have completed before proceeding with a reset. As this routine is used on the
+ * ioctl path, this can create a condition where the EEH handler becomes stuck,
+ * infinitely waiting for this ioctl thread. To avoid this behavior, temporarily
+ * unmark this thread as an ioctl thread by releasing the ioctl read semaphore.
+ * This will allow the EEH handler to proceed with a recovery while this thread
+ * is still running. Once the scsi_execute() returns, reacquire the ioctl read
+ * semaphore and check the adapter state in case it changed while inside of
+ * scsi_execute(). The state check will wait if the adapter is still being
+ * recovered or return a failure if the recovery failed. In the event that the
+ * adapter reset failed, simply return the failure as the ioctl would be unable
+ * to continue.
+ *
+ * Note that the above puts a requirement on this routine to only be called on
+ * an ioctl thread.
+ *
  * Return: 0 on success, -errno on failure
  */
 static int read_cap16(struct scsi_device *sdev, struct llun_info *lli)
@@ -314,8 +332,18 @@ retry:
 	dev_dbg(dev, "%s: %ssending cmd(0x%x)\n", __func__,
 		retry_cnt ? "re" : "", scsi_cmd[0]);
 
+	/* Drop the ioctl read semahpore across lengthy call */
+	up_read(&cfg->ioctl_rwsem);
 	result = scsi_execute(sdev, scsi_cmd, DMA_FROM_DEVICE, cmd_buf,
 			      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES, 0, NULL);
+	down_read(&cfg->ioctl_rwsem);
+	rc = check_state(cfg);
+	if (rc) {
+		dev_err(dev, "%s: Failed state! result=0x08%X\n",
+			__func__, result);
+		rc = -ENODEV;
+		goto out;
+	}
 
 	if (driver_byte(result) == DRIVER_SENSE) {
 		result &= ~(0xFF<<24); /* DRIVER_SENSE is not an error */
@@ -1221,7 +1249,7 @@ static const struct file_operations null_fops = {
  *
  * Return: 0 on success, -errno on failure
  */
-static int check_state(struct cxlflash_cfg *cfg)
+int check_state(struct cxlflash_cfg *cfg)
 {
 	struct device *dev = &cfg->dev->dev;
 	int rc = 0;
diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
index 7df88ee..06a805a 100644
--- a/drivers/scsi/cxlflash/superpipe.h
+++ b/drivers/scsi/cxlflash/superpipe.h
@@ -147,4 +147,6 @@ void cxlflash_ba_terminate(struct ba_lun *);
 
 int cxlflash_manage_lun(struct scsi_device *, struct dk_cxlflash_manage_lun *);
 
+int check_state(struct cxlflash_cfg *);
+
 #endif /* ifndef _CXLFLASH_SUPERPIPE_H */
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index b0eaf55..a53f583 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -400,6 +400,24 @@ static int init_vlun(struct llun_info *lli)
  * @lba:	Logical block address to start write same.
  * @nblks:	Number of logical blocks to write same.
  *
+ * The SCSI WRITE_SAME16 can take quite a while to complete. Should an EEH occur
+ * while in scsi_execute(), the EEH handler will attempt to recover. As part of
+ * the recovery, the handler drains all currently running ioctls, waiting until
+ * they have completed before proceeding with a reset. As this routine is used
+ * on the ioctl path, this can create a condition where the EEH handler becomes
+ * stuck, infinitely waiting for this ioctl thread. To avoid this behavior,
+ * temporarily unmark this thread as an ioctl thread by releasing the ioctl read
+ * semaphore. This will allow the EEH handler to proceed with a recovery while
+ * this thread is still running. Once the scsi_execute() returns, reacquire the
+ * ioctl read semaphore and check the adapter state in case it changed while
+ * inside of scsi_execute(). The state check will wait if the adapter is still
+ * being recovered or return a failure if the recovery failed. In the event that
+ * the adapter reset failed, simply return the failure as the ioctl would be
+ * unable to continue.
+ *
+ * Note that the above puts a requirement on this routine to only be called on
+ * an ioctl thread.
+ *
  * Return: 0 on success, -errno on failure
  */
 static int write_same16(struct scsi_device *sdev,
@@ -433,9 +451,20 @@ static int write_same16(struct scsi_device *sdev,
 		put_unaligned_be32(ws_limit < left ? ws_limit : left,
 				   &scsi_cmd[10]);
 
+		/* Drop the ioctl read semahpore across lengthy call */
+		up_read(&cfg->ioctl_rwsem);
 		result = scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, cmd_buf,
 				      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES,
 				      0, NULL);
+		down_read(&cfg->ioctl_rwsem);
+		rc = check_state(cfg);
+		if (rc) {
+			dev_err(dev, "%s: Failed state! result=0x08%X\n",
+				__func__, result);
+			rc = -ENODEV;
+			goto out;
+		}
+
 		if (result) {
 			dev_err_ratelimited(dev, "%s: command failed for "
 					    "offset %lld result=0x%x\n",
-- 
2.1.0

Re: [PATCH v4 08/32] cxlflash: Fix to avoid CXL services during EEH

[Brian King]

Reviewed-by: Brian King <brking@linux.vnet.ibm.com>


-- 
Brian King
Power Linux I/O
IBM Linux Technology Center

Re: [PATCH v4 30/32] cxlflash: Fix to avoid corrupting adapter fops

[Brian King]

Reviewed-by: Brian King <brking@linux.vnet.ibm.com>


-- 
Brian King
Power Linux I/O
IBM Linux Technology Center

Re: [PATCH v4 06/32] cxlflash: Fix to avoid sizeof(bool)

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512


Looks good to me.

Reviewed-by: Daniel Axtens <dja@axtens.net>

Regards,
Daniel

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> Using sizeof(bool) is considered poor form for various reasons and
> sparse warns us of that. Correct by changing type from bool to u8.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/superpipe.c | 2 +-
>  drivers/scsi/cxlflash/superpipe.h | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/su=
perpipe.c
> index ffa68cc..28aa9d9 100644
> --- a/drivers/scsi/cxlflash/superpipe.c
> +++ b/drivers/scsi/cxlflash/superpipe.c
> @@ -737,7 +737,7 @@ static struct ctx_info *create_context(struct cxlflas=
h_cfg *cfg,
>  	struct afu *afu =3D cfg->afu;
>  	struct ctx_info *ctxi =3D NULL;
>  	struct llun_info **lli =3D NULL;
> -	bool *ws =3D NULL;
> +	u8 *ws =3D NULL;
>  	struct sisl_rht_entry *rhte;
>=20=20
>  	ctxi =3D kzalloc(sizeof(*ctxi), GFP_KERNEL);
> diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/su=
perpipe.h
> index fffb179..72d53cf 100644
> --- a/drivers/scsi/cxlflash/superpipe.h
> +++ b/drivers/scsi/cxlflash/superpipe.h
> @@ -97,7 +97,7 @@ struct ctx_info {
>  	u32 rht_out;		/* Number of checked out RHT entries */
>  	u32 rht_perms;		/* User-defined permissions for RHT entries */
>  	struct llun_info **rht_lun;       /* Mapping of RHT entries to LUNs */
> -	bool *rht_needs_ws;	/* User-desired write-same function per RHTE */
> +	u8 *rht_needs_ws;	/* User-desired write-same function per RHTE */
>=20=20
>  	struct cxl_ioctl_start_work work;
>  	u64 ctxid;
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DRI1B
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 07/32] cxlflash: Fix context encode mask width

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512


Looks good to me.

Reviewed-by: Daniel Axtens <dja@axtens.net>

Regards,
Daniel

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> The context encode mask covers more than 32-bits, making it
> a long integer. This should be noted by appending the ULL
> width suffix to the mask.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/superpipe.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/su=
perpipe.h
> index 72d53cf..7947091 100644
> --- a/drivers/scsi/cxlflash/superpipe.h
> +++ b/drivers/scsi/cxlflash/superpipe.h
> @@ -87,7 +87,7 @@ enum ctx_ctrl {
>  	CTX_CTRL_FILE		=3D (1 << 5)
>  };
>=20=20
> -#define ENCODE_CTXID(_ctx, _id)	(((((u64)_ctx) & 0xFFFFFFFF0) << 28) | _=
id)
> +#define ENCODE_CTXID(_ctx, _id)	(((((u64)_ctx) & 0xFFFFFFFF0ULL) << 28) =
| _id)
>  #define DECODE_CTXID(_val)	(_val & 0xFFFFFFFF)
>=20=20
>  struct ctx_info {
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DM1UV
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 08/32] cxlflash: Fix to avoid CXL services during EEH

[Daniel Axtens]

You have two versions of check_state() below, which is a bit
confusing. It looks like you've moved the function and also added the
up/down of the read semaphore. I assume that's all that changed?

>  
>  /**
> + * check_state() - checks and responds to the current adapter state
> + * @cfg:	Internal structure associated with the host.
> + *
> + * This routine can block and should only be used on process context.
> + * It assumes that the caller is an ioctl thread and holding the ioctl
> + * read semaphore. This is temporarily let up across the wait to allow
> + * for draining actively running ioctls. Also note that when waking up
> + * from waiting in reset, the state is unknown and must be checked again
> + * before proceeding.
> + *
> + * Return: 0 on success, -errno on failure
> + */
> +static int check_state(struct cxlflash_cfg *cfg)
> +{
> +	struct device *dev = &cfg->dev->dev;
> +	int rc = 0;
> +
> +retry:
> +	switch (cfg->state) {
> +	case STATE_LIMBO:
> +		dev_dbg(dev, "%s: Limbo state, going to wait...\n", __func__);
> +		up_read(&cfg->ioctl_rwsem);
> +		rc = wait_event_interruptible(cfg->limbo_waitq,
> +					      cfg->state != STATE_LIMBO);
> +		down_read(&cfg->ioctl_rwsem);
> +		if (unlikely(rc))
> +			break;
> +		goto retry;
> +	case STATE_FAILTERM:
> +		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
> +		rc = -ENODEV;
> +		break;
> +	default:
> +		break;
> +	}
> +
> +	return rc;
> +}
> +
> +/**
>   * cxlflash_disk_attach() - attach a LUN to a context
>   * @sdev:	SCSI device associated with LUN.
>   * @attach:	Attach ioctl data structure.
> @@ -1523,41 +1563,6 @@ err1:
>  }
>  
>  /**
> - * check_state() - checks and responds to the current adapter state
> - * @cfg:	Internal structure associated with the host.
> - *
> - * This routine can block and should only be used on process context.
> - * Note that when waking up from waiting in limbo, the state is unknown
> - * and must be checked again before proceeding.
> - *
> - * Return: 0 on success, -errno on failure
> - */
> -static int check_state(struct cxlflash_cfg *cfg)
> -{
> -	struct device *dev = &cfg->dev->dev;
> -	int rc = 0;
> -
> -retry:
> -	switch (cfg->state) {
> -	case STATE_LIMBO:
> -		dev_dbg(dev, "%s: Limbo, going to wait...\n", __func__);
> -		rc = wait_event_interruptible(cfg->limbo_waitq,
> -					      cfg->state != STATE_LIMBO);
> -		if (unlikely(rc))
> -			break;
> -		goto retry;
> -	case STATE_FAILTERM:
> -		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
> -		rc = -ENODEV;
> -		break;
> -	default:
> -		break;
> -	}
> -
> -	return rc;
> -}
> -
> -/**
>   * cxlflash_afu_recover() - initiates AFU recovery
>   * @sdev:	SCSI device associated with LUN.
>   * @recover:	Recover ioctl data structure.
> @@ -1646,9 +1651,14 @@ retry_recover:
>  	/* Test if in error state */
>  	reg = readq_be(&afu->ctrl_map->mbox_r);
>  	if (reg == -1) {
> -		dev_dbg(dev, "%s: MMIO read fail! Wait for recovery...\n",
> -			__func__);
> -		mutex_unlock(&ctxi->mutex);
> +		dev_dbg(dev, "%s: MMIO fail, wait for recovery.\n", __func__);
> +
> +		/*
> +		 * Before checking the state, put back the context obtained with
> +		 * get_context() as it is no longer needed and sleep for a short
> +		 * period of time (see prolog notes).
> +		 */
> +		put_context(ctxi);

Is this needed for the drain to work? It looks like it fixes a
refcounting bug in the function, but I'm not sure I understand how it
interacts with the rest of this patch.

Anyway, the patch overall looks good to me, and makes your driver
interact with CXL's EEH support in the way I intended when I wrote it.

Reviewed-by: Daniel Axtens <dja@axtens.net>

Regards,
Daniel

Re: [PATCH v4 09/32] cxlflash: Correct naming of limbo state and waitq

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

Looks good from an EEH point of view: in an error situation, your driver
asks to be reset and then is waiting for CXL and EEH to carry that out,
so 'reset' matches with that as well.

Reviewed-by: Daniel Axtens <dja@axtens.net>

> Limbo is not an accurate representation of this state and is
> also not consistent with the terminology that other drivers
> use to represent this concept. Rename the state and and its
> associated waitq to 'reset'.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/common.h    |  4 ++--
>  drivers/scsi/cxlflash/main.c      | 26 +++++++++++++-------------
>  drivers/scsi/cxlflash/superpipe.c | 14 +++++++-------
>  3 files changed, 22 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/commo=
n.h
> index 1abe4e0..11318de 100644
> --- a/drivers/scsi/cxlflash/common.h
> +++ b/drivers/scsi/cxlflash/common.h
> @@ -79,7 +79,7 @@ enum cxlflash_init_state {
>=20=20
>  enum cxlflash_state {
>  	STATE_NORMAL,	/* Normal running state, everything good */
> -	STATE_LIMBO,	/* Limbo running state, trying to reset/recover */
> +	STATE_RESET,	/* Reset state, trying to reset/recover */
>  	STATE_FAILTERM	/* Failed/terminating state, error out users/threads */
>  };
>=20=20
> @@ -125,7 +125,7 @@ struct cxlflash_cfg {
>=20=20
>  	wait_queue_head_t tmf_waitq;
>  	bool tmf_active;
> -	wait_queue_head_t limbo_waitq;
> +	wait_queue_head_t reset_waitq;
>  	enum cxlflash_state state;
>  };
>=20=20
> diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
> index 6e85c77..8940336 100644
> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -382,8 +382,8 @@ static int cxlflash_queuecommand(struct Scsi_Host *ho=
st, struct scsi_cmnd *scp)
>  	spin_unlock_irqrestore(&cfg->tmf_waitq.lock, lock_flags);
>=20=20
>  	switch (cfg->state) {
> -	case STATE_LIMBO:
> -		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device in limbo!\n",
> +	case STATE_RESET:
> +		dev_dbg_ratelimited(&cfg->dev->dev, "%s: device is in reset!\n",
>  				    __func__);
>  		rc =3D SCSI_MLQUEUE_HOST_BUSY;
>  		goto out;
> @@ -479,8 +479,8 @@ static int cxlflash_eh_device_reset_handler(struct sc=
si_cmnd *scp)
>  		if (unlikely(rcr))
>  			rc =3D FAILED;
>  		break;
> -	case STATE_LIMBO:
> -		wait_event(cfg->limbo_waitq, cfg->state !=3D STATE_LIMBO);
> +	case STATE_RESET:
> +		wait_event(cfg->reset_waitq, cfg->state !=3D STATE_RESET);
>  		if (cfg->state =3D=3D STATE_NORMAL)
>  			break;
>  		/* fall through */
> @@ -519,7 +519,7 @@ static int cxlflash_eh_host_reset_handler(struct scsi=
_cmnd *scp)
>=20=20
>  	switch (cfg->state) {
>  	case STATE_NORMAL:
> -		cfg->state =3D STATE_LIMBO;
> +		cfg->state =3D STATE_RESET;
>  		scsi_block_requests(cfg->host);
>  		cxlflash_mark_contexts_error(cfg);
>  		rcr =3D cxlflash_afu_reset(cfg);
> @@ -528,11 +528,11 @@ static int cxlflash_eh_host_reset_handler(struct sc=
si_cmnd *scp)
>  			cfg->state =3D STATE_FAILTERM;
>  		} else
>  			cfg->state =3D STATE_NORMAL;
> -		wake_up_all(&cfg->limbo_waitq);
> +		wake_up_all(&cfg->reset_waitq);
>  		scsi_unblock_requests(cfg->host);
>  		break;
> -	case STATE_LIMBO:
> -		wait_event(cfg->limbo_waitq, cfg->state !=3D STATE_LIMBO);
> +	case STATE_RESET:
> +		wait_event(cfg->reset_waitq, cfg->state !=3D STATE_RESET);
>  		if (cfg->state =3D=3D STATE_NORMAL)
>  			break;
>  		/* fall through */
> @@ -705,7 +705,7 @@ static void cxlflash_wait_for_pci_err_recovery(struct=
 cxlflash_cfg *cfg)
>  	struct pci_dev *pdev =3D cfg->dev;
>=20=20
>  	if (pci_channel_offline(pdev))
> -		wait_event_timeout(cfg->limbo_waitq,
> +		wait_event_timeout(cfg->reset_waitq,
>  				   !pci_channel_offline(pdev),
>  				   CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT);
>  }
> @@ -2304,7 +2304,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
>  	cfg->mcctx =3D NULL;
>=20=20
>  	init_waitqueue_head(&cfg->tmf_waitq);
> -	init_waitqueue_head(&cfg->limbo_waitq);
> +	init_waitqueue_head(&cfg->reset_waitq);
>=20=20
>  	INIT_WORK(&cfg->work_q, cxlflash_worker_thread);
>  	cfg->lr_state =3D LINK_RESET_INVALID;
> @@ -2396,7 +2396,7 @@ static pci_ers_result_t cxlflash_pci_error_detected=
(struct pci_dev *pdev,
>=20=20
>  	switch (state) {
>  	case pci_channel_io_frozen:
> -		cfg->state =3D STATE_LIMBO;
> +		cfg->state =3D STATE_RESET;
>  		scsi_block_requests(cfg->host);
>  		drain_ioctls(cfg);
>  		rc =3D cxlflash_mark_contexts_error(cfg);
> @@ -2408,7 +2408,7 @@ static pci_ers_result_t cxlflash_pci_error_detected=
(struct pci_dev *pdev,
>  		return PCI_ERS_RESULT_NEED_RESET;
>  	case pci_channel_io_perm_failure:
>  		cfg->state =3D STATE_FAILTERM;
> -		wake_up_all(&cfg->limbo_waitq);
> +		wake_up_all(&cfg->reset_waitq);
>  		scsi_unblock_requests(cfg->host);
>  		return PCI_ERS_RESULT_DISCONNECT;
>  	default:
> @@ -2455,7 +2455,7 @@ static void cxlflash_pci_resume(struct pci_dev *pde=
v)
>  	dev_dbg(dev, "%s: pdev=3D%p\n", __func__, pdev);
>=20=20
>  	cfg->state =3D STATE_NORMAL;
> -	wake_up_all(&cfg->limbo_waitq);
> +	wake_up_all(&cfg->reset_waitq);
>  	scsi_unblock_requests(cfg->host);
>  }
>=20=20
> diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/su=
perpipe.c
> index 655cbf1..8a7ec5d 100644
> --- a/drivers/scsi/cxlflash/superpipe.c
> +++ b/drivers/scsi/cxlflash/superpipe.c
> @@ -100,7 +100,7 @@ void cxlflash_stop_term_user_contexts(struct cxlflash=
_cfg *cfg)
>=20=20
>  		dev_dbg(dev, "%s: Wait for user contexts to quiesce...\n",
>  			__func__);
> -		wake_up_all(&cfg->limbo_waitq);
> +		wake_up_all(&cfg->reset_waitq);
>  		ssleep(1);
>  	}
>  }
> @@ -1233,11 +1233,11 @@ static int check_state(struct cxlflash_cfg *cfg)
>=20=20
>  retry:
>  	switch (cfg->state) {
> -	case STATE_LIMBO:
> -		dev_dbg(dev, "%s: Limbo state, going to wait...\n", __func__);
> +	case STATE_RESET:
> +		dev_dbg(dev, "%s: Reset state, going to wait...\n", __func__);
>  		up_read(&cfg->ioctl_rwsem);
> -		rc =3D wait_event_interruptible(cfg->limbo_waitq,
> -					      cfg->state !=3D STATE_LIMBO);
> +		rc =3D wait_event_interruptible(cfg->reset_waitq,
> +					      cfg->state !=3D STATE_RESET);
>  		down_read(&cfg->ioctl_rwsem);
>  		if (unlikely(rc))
>  			break;
> @@ -1578,10 +1578,10 @@ err1:
>   * quite possible for this routine to act as the kernel's EEH detection
>   * source (MMIO read of mbox_r). Because of this, there is a window of
>   * time where an EEH might have been detected but not yet 'serviced'
> - * (callback invoked, causing the device to enter limbo state). To avoid
> + * (callback invoked, causing the device to enter reset state). To avoid
>   * looping in this routine during that window, a 1 second sleep is in pl=
ace
>   * between the time the MMIO failure is detected and the time a wait on =
the
> - * limbo wait queue is attempted via check_state().
> + * reset wait queue is attempted via check_state().
>   *
>   * Return: 0 on success, -errno on failure
>   */
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DD6L8
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 17/32] cxlflash: Remove dual port online dependency

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Hi,

>  static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 wwp=
n)
>  {
> -	int ret =3D 0;
> +	int rc =3D 0;
I realise it's nice to have things consistent, but making this change
now makes the rest of the patch quite difficult to follow.

>=20=20
>  	set_port_offline(fc_regs);
>=20=20
> @@ -1038,33 +1038,26 @@ static int afu_set_wwpn(struct afu *afu, int port=
, u64 *fc_regs, u64 wwpn)
>  			       FC_PORT_STATUS_RETRY_CNT)) {
>  		pr_debug("%s: wait on port %d to go offline timed out\n",
>  			 __func__, port);
> -		ret =3D -1; /* but continue on to leave the port back online */
> +		rc =3D -1; /* but continue on to leave the port back online */
>  	}
>=20=20
> -	if (ret =3D=3D 0)
> +	if (rc =3D=3D 0)
>  		writeq_be(wwpn, &fc_regs[FC_PNAME / 8]);
>=20=20
> +	/* Always return success after programming WWPN */
> +	rc =3D 0;
> +
>  	set_port_online(fc_regs);
>=20=20
>  	if (!wait_port_online(fc_regs, FC_PORT_STATUS_RETRY_INTERVAL_US,
>  			      FC_PORT_STATUS_RETRY_CNT)) {
>  		pr_debug("%s: wait on port %d to go online timed out\n",
>  			 __func__, port);
> -		ret =3D -1;
> -
> -		/*
> -		 * Override for internal lun!!!
> -		 */
> -		if (afu->internal_lun) {
> -			pr_debug("%s: Overriding port %d online timeout!!!\n",
> -				 __func__, port);
> -			ret =3D 0;
> -		}
>  	}
>=20=20
> -	pr_debug("%s: returning rc=3D%d\n", __func__, ret);
> +	pr_debug("%s: returning rc=3D%d\n", __func__, rc);
I'm not sure I fully understand the flow of this function, but it looks
like you set rc=3D0 regardless of how things actually go: is this ever
going to print a return value other than zero?

Regards,
Daniel
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DcZRI
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 32/32] cxlflash: Fix to avoid potential deadlock on EEH

[Brian King]

On 09/25/2015 06:19 PM, Matthew R. Ochs wrote:
>  static int write_same16(struct scsi_device *sdev,
> @@ -433,9 +451,20 @@ static int write_same16(struct scsi_device *sdev,
>  		put_unaligned_be32(ws_limit < left ? ws_limit : left,
>  				   &scsi_cmd[10]);
> 
> +		/* Drop the ioctl read semahpore across lengthy call */
> +		up_read(&cfg->ioctl_rwsem);
>  		result = scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, cmd_buf,
>  				      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES,
>  				      0, NULL);
> +		down_read(&cfg->ioctl_rwsem);
> +		rc = check_state(cfg);
> +		if (rc) {
> +			dev_err(dev, "%s: Failed state! result=0x08%X\n",
> +				__func__, result);
> +			rc = -ENODEV;

Since check_state only returns 0 or -ENODEV, this is a bit redundant, but not worth redoing the
patch in my mind.

Reviewed-by: Brian King <brking@linux.vnet.ibm.com>




-- 
Brian King
Power Linux I/O
IBM Linux Technology Center

Re: [PATCH v4 30/32] cxlflash: Fix to avoid corrupting adapter fops

[Andrew Donnellan]

On 26/09/15 09:19, Matthew R. Ochs wrote:
> The fops owned by the adapter can be corrupted in certain scenarios,
> opening a window where certain fops are temporarily NULLed before being
> reset to their proper value. This can potentially lead software to make
> incorrect decisions, leaving the user with the inability to function as
> intended.
>
> An example of this behavior can be observed when there are a number of
> users with a high rate of turn around (attach to LUN, perform an I/O,
> detach from LUN, repeat). Every so often a user is given a valid
> context and adapter file descriptor, but the file associated with the
> descriptor lacks the correct read permission bit (FMODE_CAN_READ) and
> thus the read system call bails before calling the valid read fop.
>
> Background:
>
> The fops is stored in the adapter structure to provide the ability to
> lookup the adapter structure from within the fop handler. CXL services
> use the file's private_data and at present, the CXL context does not
> have a private section. In an effort to limit areas of the cxlflash
> driver with code specific the superpipe function, a design choice was
> made to keep the details of the fops situated away from the legacy
> portions of the driver. This drove the behavior that the adapter fops
> is set at the beginning of the disk attach ioctl handler when there
> are no users present.
>
> The corruption that this fix remedies is due to the fact that the fops
> is initially defaulted to values found within a static structure. When
> the fops is handed down to the CXL services later in the attach path,
> certain services are patched. The fops structure remains correct until
> the user count drops to 0 and the fops is reset, triggering the process
> to repeat again. The user counts are tightly coupled with the creation
> and deletion of the user context. If multiple users perform a disk
> attach at the same time, when the user count is currently 0, some users
> can be in the middle of obtaining a file descriptor and have not yet
> reached the context creation code that [in addition to creating the
> context] increments the user count. Subsequent users coming in to
> perform the attach see that the user count is still 0, and reinitialize
> the fops, temporarily removing the patched fops. The users that are in
> the middle obtaining their file descriptor may then receive an invalid
> descriptor.
>
> The fix simply removes the user count altogether and moves the fops
> initialization to probe time such that it is only performed one time
> for the life of the adapter. In the future, if the CXL services adopt
> a private member for their context, that could be used to store the
> adapter structure reference and cxlflash could revert to a model that
> does not require an embedded fops.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 26/32] cxlflash: Correct spelling, grammar, and alignment mistakes

[Andrew Donnellan]

On 26/09/15 09:18, Matthew R. Ochs wrote:
> There are several spelling and grammar mistakes throughout the
> driver. Additionally there are a handful of places where there
> are extra lines and unnecessary variables/statements. These are
> a nuisance and pollute the driver.
>
> Fix spelling and grammar issues. Update some comments for clarity and
> consistency. Remove extra lines and a few unneeded variables/statements.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 29/32] cxlflash: Fix to double the delay each time

[Andrew Donnellan]

On 26/09/15 09:19, Matthew R. Ochs wrote:
> From: Manoj Kumar <manoj@linux.vnet.ibm.com>
>
> The operator used to double the delay is incorrect and
> does not result in delay doubling.
>
> To fix, use a left shift instead of the XOR operator.
>
> Reported-by: Tomas Henzl <thenzl@redhat.com>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 31/32] cxlflash: Correct trace string

[Andrew Donnellan]

On 26/09/15 09:19, Matthew R. Ochs wrote:
> The trace following the failure of alloc_mem() incorrectly identifies
> which function failed. This can lead to misdiagnosing a failure.
>
> Fix the string to correctly indicate that alloc_mem() failed.
>
> Reported-by: Brian King <brking@linux.vnet.ibm.com>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 25/32] cxlflash: Fix to prevent EEH recovery failure

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:


> The process_sense() routine can perform a read capacity which
> can take some time to complete. If an EEH occurs while waiting
> on the read capacity, the EEH handler is unable to obtain the
> context's mutex in order to put the context in an error state.
> The EEH handler will sit and wait until the context is free,
> but this wait can last longer than the EEH handler tolerates,
> leading to a failed recovery.

I'm not quite clear on what you mean by the EEH handler timing
out. AFAIK there's nothing in eehd and the EEH core that times out if a
driver doesn't respond - indeed, it's pretty easy to hang eehd with a
misbehaving driver.

Are you referring to your own internal timeouts?
cxlflash_wait_for_pci_err_recovery and anything else that uses
CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT?

Regards,
Daniel

>
> To address this issue, make the context unavailable to new,
> non-system owned threads and release the context while calling
> into process_sense(). After returning from process_sense() the
> context mutex is reacquired and the context is made available
> again. The context can be safely moved to the error state if
> needed during the unavailable window as no other threads will
> hold its reference.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/superpipe.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
>
> diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/su=
perpipe.c
> index a6316f5..7283e83 100644
> --- a/drivers/scsi/cxlflash/superpipe.c
> +++ b/drivers/scsi/cxlflash/superpipe.c
> @@ -1787,12 +1787,21 @@ static int cxlflash_disk_verify(struct scsi_devic=
e *sdev,
>  	 * inquiry (i.e. the Unit attention is due to the WWN changing).
>  	 */
>  	if (verify->hint & DK_CXLFLASH_VERIFY_HINT_SENSE) {
> +		/* Can't hold mutex across process_sense/read_cap16,
> +		 * since we could have an intervening EEH event.
> +		 */
> +		ctxi->unavail =3D true;
> +		mutex_unlock(&ctxi->mutex);
>  		rc =3D process_sense(sdev, verify);
>  		if (unlikely(rc)) {
>  			dev_err(dev, "%s: Failed to validate sense data (%d)\n",
>  				__func__, rc);
> +			mutex_lock(&ctxi->mutex);
> +			ctxi->unavail =3D false;
>  			goto out;
>  		}
> +		mutex_lock(&ctxi->mutex);
> +		ctxi->unavail =3D false;
>  	}
>=20=20
>  	switch (gli->mode) {
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3D3BcU
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 27/32] cxlflash: Fix to prevent stale AFU RRQ

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> Following an adapter reset, the AFU RRQ that resides in host memory
> holds stale data. This can lead to a condition where the RRQ interrupt
> handler tries to process stale entries and/or endlessly loops due to an
> out of sync generation bit.
>
> To fix, the AFU RRQ in host memory needs to be cleared after each reset.

This looks good. Do you need anything to bail out of cxlflash_rrq_irq if
the data goes stale or to all Fs while that function is running?

Daniel

>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/main.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
> index 24aedfb..ab11ee6 100644
> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -1598,6 +1598,9 @@ static int start_afu(struct cxlflash_cfg *cfg)
>=20=20
>  	init_pcr(cfg);
>=20=20
> +	/* After an AFU reset, RRQ entries are stale, clear them */
> +	memset(&afu->rrq_entry, 0, sizeof(afu->rrq_entry));
> +
>  	/* Initialize RRQ pointers */
>  	afu->hrrq_start =3D &afu->rrq_entry[0];
>  	afu->hrrq_end =3D &afu->rrq_entry[NUM_RRQ_ENTRY - 1];
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DO6Zk
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 29/32] cxlflash: Fix to double the delay each time

[Daniel Axtens]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> From: Manoj Kumar <manoj@linux.vnet.ibm.com>
>
> The operator used to double the delay is incorrect and
> does not result in delay doubling.
>
> To fix, use a left shift instead of the XOR operator.
>
I can see that the patch is correct, but this commit message is a bit
confusing. What delay? In what circumstances are you doubling it? Why?

Regards,
Daniel

> Reported-by: Tomas Henzl <thenzl@redhat.com>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/main.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
> index ab11ee6..be78906 100644
> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -303,7 +303,7 @@ write_rrin:
>  		if (rrin !=3D 0x1)
>  			break;
>  		/* Double delay each time */
> -		udelay(2 ^ nretry);
> +		udelay(2 << nretry);
>  	} while (nretry++ < MC_ROOM_RETRY_CNT);
>  }
>=20=20
> --=20
> 2.1.0
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: GPGTools - https://gpgtools.org
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=3DKm/z
=2D----END PGP SIGNATURE-----

Re: [PATCH v4 24/32] cxlflash: Fix MMIO and endianness errors

[Andrew Donnellan]

On 26/09/15 09:18, Matthew R. Ochs wrote:
> Sparse uncovered several errors with MMIO operations (accessing
> directly) and handling endianness. These can cause issues when
> running in different environments.
>
> Introduce __iomem and proper endianness tags/swaps where
> appropriate to make driver sparse clean.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 11/32] cxlflash: Refine host/device attributes

[Andrew Donnellan]

On 26/09/15 09:14, Matthew R. Ochs wrote:
> Implement the following suggestions and add two new attributes
> to allow for debugging the port LUN table.
>
>   - use scnprintf() instead of snprintf()
>   - use DEVICE_ATTR_RO and DEVICE_ATTR_RW
>
> Suggested-by: Shane Seymour <shane.seymour@hp.com>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

The commit message could explain the patch a bit better.


Andrew

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 23/32] cxlflash: Fix function prolog parameters and return codes

[Andrew Donnellan]

On 26/09/15 09:18, Matthew R. Ochs wrote:
> Several function prologs have incorrect parameter names and return
> code descriptions. This can lead to confusion when reviewing the
> source and creates inaccurate documentation.
>
> To remedy, update the function prologs to properly reflect parameter
> names and return codes.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

See further comments below.

> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -401,8 +401,7 @@ static void wait_resp(struct afu *afu, struct afu_cmd *cmd)
>    * @tmfcmd:	TMF command to send.
>    *
>    * Return:
> - *	0 on success
> - *	SCSI_MLQUEUE_HOST_BUSY when host is busy
> + *	0 on success or SCSI_MLQUEUE_HOST_BUSY
>    */
>   static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 tmfcmd)
>   {
> @@ -491,9 +490,7 @@ static const char *cxlflash_driver_info(struct Scsi_Host *host)
>    * @host:	SCSI host associated with device.
>    * @scp:	SCSI command to send.
>    *
> - * Return:
> - *	0 on success
> - *	SCSI_MLQUEUE_HOST_BUSY when host is busy
> + * Return: 0 on success or SCSI_MLQUEUE_HOST_BUSY
>    */

I'd prefer it to say "SCSI_MLQUEUE_HOST_BUSY on failure". (Aesthetically 
I prefer having it on a separate line, but that's just personal preference.)

As an aside, while checking the correctness of this, I found that the 
comment for cxlflash_send_cmd() states that it returns -1 on failure, 
when the only error value it actually returns is SCSI_MLQUEUE_HOST_BUSY. 
If you send a v5 you might want to fix this.


Andrew

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 12/32] cxlflash: Fix to avoid spamming the kernel log

[Andrew Donnellan]

On 26/09/15 09:15, Matthew R. Ochs wrote:
> During run-time the driver can be very chatty and spam the system
> kernel log. Various print statements can be limited and/or moved
> to development-only mode. Additionally, numerous prints can be
> converted to trace the corresponding device.
>
> The following changes were made:
>   - pr_debug to pr_devel
>   - pr_debug to pr_debug_ratelimited
>   - pr_err to dev_err
>   - pr_debug to dev_dbg
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

Changes mostly look fine, further comments below.

> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -58,8 +58,8 @@ static struct afu_cmd *cmd_checkout(struct afu *afu)
>   		cmd = &afu->cmd[k];
>
>   		if (!atomic_dec_if_positive(&cmd->free)) {
> -			pr_debug("%s: returning found index=%d\n",
> -				 __func__, cmd->slot);
> +			pr_devel("%s: returning found index=%d cmd=%p\n",
> +				 __func__, cmd->slot, cmd);

>   	pr_debug("%s: cmd failed afu_rc=%d scsi_rc=%d fc_rc=%d "
> -		 "afu_extra=0x%X, scsi_entra=0x%X, fc_extra=0x%X\n",
> +		 "afu_extra=0x%X, scsi_extra=0x%X, fc_extra=0x%X\n",
>   		 __func__, ioasa->rc.afu_rc, ioasa->rc.scsi_rc,
>   		 ioasa->rc.fc_rc, ioasa->afu_extra, ioasa->scsi_extra,
>   		 ioasa->fc_extra);

Minor nitpicking: mention that you fix these in the commit message.

> @@ -240,9 +240,9 @@ static void cmd_complete(struct afu_cmd *cmd)
>   		cmd_is_tmf = cmd->cmd_tmf;
>   		cmd_checkin(cmd); /* Don't use cmd after here */
>
> -		pr_debug("%s: calling scsi_set_resid, scp=%p "
> -			 "result=%X resid=%d\n", __func__,
> -			 scp, scp->result, resid);
> +		pr_debug_ratelimited("%s: calling scsi_done scp=%p result=%X "
> +				     "ioasc=%d\n", __func__, scp, scp->result,
> +				     cmd->sa.ioasc);
>
>   		scsi_set_resid(scp, resid);
>   		scsi_dma_unmap(scp);

Why has the message changed from scsi_set_resid to scsi_done, and should 
the message be moved to immediately before the scsi_done call?


Andrew

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 02/32] cxlflash: Replace magic numbers with literals

[Andrew Donnellan]

On 26/09/15 09:12, Matthew R. Ochs wrote:
> From: Manoj Kumar <kumarmn@us.ibm.com>
>
> Magic numbers are not meaningful and can create confusion. As a
> remedy, replace them with descriptive literals.
>
> Replace 512 with literal MAX_SECTOR_UNIT.
> Replace 5 with literal CMD_RETRIES.
>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>

Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>

-- 
Andrew Donnellan              Software Engineer, OzLabs
andrew.donnellan@au1.ibm.com  Australia Development Lab, Canberra
+61 2 6201 8874 (work)        IBM Australia Limited

Re: [PATCH v4 08/32] cxlflash: Fix to avoid CXL services during EEH

[Matthew R. Ochs]

> On Sep 28, 2015, at 6:05 PM, Daniel Axtens <dja@axtens.net> wrote:
>=20
> You have two versions of check_state() below, which is a bit
> confusing. It looks like you've moved the function and also added the
> up/down of the read semaphore. I assume that's all that changed?

Correct.

It was originally moved to meet a dependency due to it being defined =
statically.

>=20
>>=20
>> /**
>> + * check_state() - checks and responds to the current adapter state
>> + * @cfg:	Internal structure associated with the host.
>> + *
>> + * This routine can block and should only be used on process =
context.
>> + * It assumes that the caller is an ioctl thread and holding the =
ioctl
>> + * read semaphore. This is temporarily let up across the wait to =
allow
>> + * for draining actively running ioctls. Also note that when waking =
up
>> + * from waiting in reset, the state is unknown and must be checked =
again
>> + * before proceeding.
>> + *
>> + * Return: 0 on success, -errno on failure
>> + */
>> +static int check_state(struct cxlflash_cfg *cfg)
>> +{
>> +	struct device *dev =3D &cfg->dev->dev;
>> +	int rc =3D 0;
>> +
>> +retry:
>> +	switch (cfg->state) {
>> +	case STATE_LIMBO:
>> +		dev_dbg(dev, "%s: Limbo state, going to wait...\n", =
__func__);
>> +		up_read(&cfg->ioctl_rwsem);
>> +		rc =3D wait_event_interruptible(cfg->limbo_waitq,
>> +					      cfg->state !=3D =
STATE_LIMBO);
>> +		down_read(&cfg->ioctl_rwsem);
>> +		if (unlikely(rc))
>> +			break;
>> +		goto retry;
>> +	case STATE_FAILTERM:
>> +		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
>> +		rc =3D -ENODEV;
>> +		break;
>> +	default:
>> +		break;
>> +	}
>> +
>> +	return rc;
>> +}
>> +
>> +/**
>>  * cxlflash_disk_attach() - attach a LUN to a context
>>  * @sdev:	SCSI device associated with LUN.
>>  * @attach:	Attach ioctl data structure.
>> @@ -1523,41 +1563,6 @@ err1:
>> }
>>=20
>> /**
>> - * check_state() - checks and responds to the current adapter state
>> - * @cfg:	Internal structure associated with the host.
>> - *
>> - * This routine can block and should only be used on process =
context.
>> - * Note that when waking up from waiting in limbo, the state is =
unknown
>> - * and must be checked again before proceeding.
>> - *
>> - * Return: 0 on success, -errno on failure
>> - */
>> -static int check_state(struct cxlflash_cfg *cfg)
>> -{
>> -	struct device *dev =3D &cfg->dev->dev;
>> -	int rc =3D 0;
>> -
>> -retry:
>> -	switch (cfg->state) {
>> -	case STATE_LIMBO:
>> -		dev_dbg(dev, "%s: Limbo, going to wait...\n", __func__);
>> -		rc =3D wait_event_interruptible(cfg->limbo_waitq,
>> -					      cfg->state !=3D =
STATE_LIMBO);
>> -		if (unlikely(rc))
>> -			break;
>> -		goto retry;
>> -	case STATE_FAILTERM:
>> -		dev_dbg(dev, "%s: Failed/Terminating!\n", __func__);
>> -		rc =3D -ENODEV;
>> -		break;
>> -	default:
>> -		break;
>> -	}
>> -
>> -	return rc;
>> -}
>> -
>> -/**
>>  * cxlflash_afu_recover() - initiates AFU recovery
>>  * @sdev:	SCSI device associated with LUN.
>>  * @recover:	Recover ioctl data structure.
>> @@ -1646,9 +1651,14 @@ retry_recover:
>> 	/* Test if in error state */
>> 	reg =3D readq_be(&afu->ctrl_map->mbox_r);
>> 	if (reg =3D=3D -1) {
>> -		dev_dbg(dev, "%s: MMIO read fail! Wait for =
recovery...\n",
>> -			__func__);
>> -		mutex_unlock(&ctxi->mutex);
>> +		dev_dbg(dev, "%s: MMIO fail, wait for recovery.\n", =
__func__);
>> +
>> +		/*
>> +		 * Before checking the state, put back the context =
obtained with
>> +		 * get_context() as it is no longer needed and sleep for =
a short
>> +		 * period of time (see prolog notes).
>> +		 */
>> +		put_context(ctxi);
>=20
> Is this needed for the drain to work? It looks like it fixes a
> refcounting bug in the function, but I'm not sure I understand how it
> interacts with the rest of this patch.

This was simply some "while I'm here" refactoring as the commit =
originally
included a change here. The main point of this change was to replace the
mutex_unlock() with put_context(), which is a wrapper around the =
unlocking
of the context's mutex.
>=20
> Anyway, the patch overall looks good to me, and makes your driver
> interact with CXL's EEH support in the way I intended when I wrote it.

Thanks for reviewing.

Re: [PATCH v4 17/32] cxlflash: Remove dual port online dependency

[Matthew R. Ochs]

> On Sep 28, 2015, at 6:37 PM, Daniel Axtens <dja@axtens.net> wrote:
>=20
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
>=20
> Hi,
>=20
>> static int afu_set_wwpn(struct afu *afu, int port, u64 *fc_regs, u64 =
wwpn)
>> {
>> -	int ret =3D 0;
>> +	int rc =3D 0;
> I realise it's nice to have things consistent, but making this change
> now makes the rest of the patch quite difficult to follow.

Next time I will try to separate out a change like this.

>=20
>>=20
>> 	set_port_offline(fc_regs);
>>=20
>> @@ -1038,33 +1038,26 @@ static int afu_set_wwpn(struct afu *afu, int =
port, u64 *fc_regs, u64 wwpn)
>> 			       FC_PORT_STATUS_RETRY_CNT)) {
>> 		pr_debug("%s: wait on port %d to go offline timed =
out\n",
>> 			 __func__, port);
>> -		ret =3D -1; /* but continue on to leave the port back =
online */
>> +		rc =3D -1; /* but continue on to leave the port back =
online */
>> 	}
>>=20
>> -	if (ret =3D=3D 0)
>> +	if (rc =3D=3D 0)
>> 		writeq_be(wwpn, &fc_regs[FC_PNAME / 8]);
>>=20
>> +	/* Always return success after programming WWPN */
>> +	rc =3D 0;
>> +
>> 	set_port_online(fc_regs);
>>=20
>> 	if (!wait_port_online(fc_regs, FC_PORT_STATUS_RETRY_INTERVAL_US,
>> 			      FC_PORT_STATUS_RETRY_CNT)) {
>> 		pr_debug("%s: wait on port %d to go online timed out\n",
>> 			 __func__, port);
>> -		ret =3D -1;
>> -
>> -		/*
>> -		 * Override for internal lun!!!
>> -		 */
>> -		if (afu->internal_lun) {
>> -			pr_debug("%s: Overriding port %d online =
timeout!!!\n",
>> -				 __func__, port);
>> -			ret =3D 0;
>> -		}
>> 	}
>>=20
>> -	pr_debug("%s: returning rc=3D%d\n", __func__, ret);
>> +	pr_debug("%s: returning rc=3D%d\n", __func__, rc);
> I'm not sure I fully understand the flow of this function, but it =
looks
> like you set rc=3D0 regardless of how things actually go: is this ever
> going to print a return value other than zero?

Correct, this function behaves more like a void for the time being. The
overall goal of this is to allow a card to configure even when the link =
is
down. At some later point when the link is transitioned to 'up', a link =
state
change interrupt will trigger the port configuration. I left this with a =
return
code for right now in case we need to alter the behavior again (based
upon testing) and actually return a value other than 0.

Re: [PATCH v4 32/32] cxlflash: Fix to avoid potential deadlock on EEH

[Matthew R. Ochs]

> On Sep 28, 2015, at 6:41 PM, Brian King <brking@linux.vnet.ibm.com> =
wrote:
> On 09/25/2015 06:19 PM, Matthew R. Ochs wrote:
>> static int write_same16(struct scsi_device *sdev,
>> @@ -433,9 +451,20 @@ static int write_same16(struct scsi_device =
*sdev,
>> 		put_unaligned_be32(ws_limit < left ? ws_limit : left,
>> 				   &scsi_cmd[10]);
>>=20
>> +		/* Drop the ioctl read semahpore across lengthy call */
>> +		up_read(&cfg->ioctl_rwsem);
>> 		result =3D scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, =
cmd_buf,
>> 				      CMD_BUFSIZE, sense_buf, to, =
CMD_RETRIES,
>> 				      0, NULL);
>> +		down_read(&cfg->ioctl_rwsem);
>> +		rc =3D check_state(cfg);
>> +		if (rc) {
>> +			dev_err(dev, "%s: Failed state! =
result=3D0x08%X\n",
>> +				__func__, result);
>> +			rc =3D -ENODEV;
>=20
> Since check_state only returns 0 or -ENODEV, this is a bit redundant, =
but not worth redoing the
> patch in my mind.

Agreed. This occurred to me the other day after submitting this patch =
when I was
reviewing the state locking code. Will look at revising in a future =
patch.

Thanks again for reviewing.

Re: [PATCH v4 25/32] cxlflash: Fix to prevent EEH recovery failure

[Matthew R. Ochs]

> On Sep 28, 2015, at 8:25 PM, Daniel Axtens <dja@axtens.net> wrote:
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
> 
> "Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:
> 
> 
>> The process_sense() routine can perform a read capacity which
>> can take some time to complete. If an EEH occurs while waiting
>> on the read capacity, the EEH handler is unable to obtain the
>> context's mutex in order to put the context in an error state.
>> The EEH handler will sit and wait until the context is free,
>> but this wait can last longer than the EEH handler tolerates,
>> leading to a failed recovery.
> 
> I'm not quite clear on what you mean by the EEH handler timing
> out. AFAIK there's nothing in eehd and the EEH core that times out if a
> driver doesn't respond - indeed, it's pretty easy to hang eehd with a
> misbehaving driver.
> 
> Are you referring to your own internal timeouts?
> cxlflash_wait_for_pci_err_recovery and anything else that uses
> CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT?

Reading through this again I can see how this is misleading. This is
actually similar and related to the deadlock scenario described in
"Fix to avoid potential deadlock on EEH". Without this fix, you'd end
up in a similar situation but deadlocked on the context mutex instead
of the ioctl semaphore.

Re: [PATCH v4 27/32] cxlflash: Fix to prevent stale AFU RRQ

[Matthew R. Ochs]

> On Sep 28, 2015, at 8:36 PM, Daniel Axtens <dja@axtens.net> wrote:
>=20
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
>=20
> "Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:
>=20
>> Following an adapter reset, the AFU RRQ that resides in host memory
>> holds stale data. This can lead to a condition where the RRQ =
interrupt
>> handler tries to process stale entries and/or endlessly loops due to =
an
>> out of sync generation bit.
>>=20
>> To fix, the AFU RRQ in host memory needs to be cleared after each =
reset.
>=20
> This looks good. Do you need anything to bail out of cxlflash_rrq_irq =
if
> the data goes stale or to all Fs while that function is running?

We're not performing an MMIO here, so I'm not sure how the all Fs check
would apply. We're also protected fairly well by the generation bit. I =
suppose
we could look at adding some type of 'max iterations' count to protect =
against
a runaway handler but that would be in a future patch.

Re: [PATCH v4 29/32] cxlflash: Fix to double the delay each time

[Matthew R. Ochs]

> On Sep 28, 2015, at 8:40 PM, Daniel Axtens <dja@axtens.net> wrote:
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
> 
> "Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:
> 
>> From: Manoj Kumar <manoj@linux.vnet.ibm.com>
>> 
>> The operator used to double the delay is incorrect and
>> does not result in delay doubling.
>> 
>> To fix, use a left shift instead of the XOR operator.
>> 
> I can see that the patch is correct, but this commit message is a bit
> confusing. What delay? In what circumstances are you doubling it? Why?

This is the response delay while resetting the master context. The reset
is performed by writing a bit and then waiting for it to clear. While waiting
for it to clear, the code relaxes the delta between MMIO reads.

Re: [PATCH v4 23/32] cxlflash: Fix function prolog parameters and return codes

[Matthew R. Ochs]

> On Sep 28, 2015, at 11:36 PM, Andrew Donnellan =
<andrew.donnellan@au1.ibm.com> wrote:
> On 26/09/15 09:18, Matthew R. Ochs wrote:
>>=20
>>   */
>>  static int send_tmf(struct afu *afu, struct scsi_cmnd *scp, u64 =
tmfcmd)
>>  {
>> @@ -491,9 +490,7 @@ static const char *cxlflash_driver_info(struct =
Scsi_Host *host)
>>   * @host:	SCSI host associated with device.
>>   * @scp:	SCSI command to send.
>>   *
>> - * Return:
>> - *	0 on success
>> - *	SCSI_MLQUEUE_HOST_BUSY when host is busy
>> + * Return: 0 on success or SCSI_MLQUEUE_HOST_BUSY
>>   */
>=20
> I'd prefer it to say "SCSI_MLQUEUE_HOST_BUSY on failure". =
(Aesthetically I prefer having it on a separate line, but that's just =
personal preference.)
>=20
> As an aside, while checking the correctness of this, I found that the =
comment for cxlflash_send_cmd() states that it returns -1 on failure, =
when the only error value it actually returns is SCSI_MLQUEUE_HOST_BUSY. =
If you send a v5 you might want to fix this.

I'll make a note of this.

Re: [PATCH v4 12/32] cxlflash: Fix to avoid spamming the kernel log

[Matthew R. Ochs]

> On Sep 29, 2015, at 12:05 AM, Andrew Donnellan =
<andrew.donnellan@au1.ibm.com> wrote:
> On 26/09/15 09:15, Matthew R. Ochs wrote:
>> During run-time the driver can be very chatty and spam the system
>> kernel log. Various print statements can be limited and/or moved
>> to development-only mode. Additionally, numerous prints can be
>> converted to trace the corresponding device.
>>=20
>> The following changes were made:
>>  - pr_debug to pr_devel
>>  - pr_debug to pr_debug_ratelimited
>>  - pr_err to dev_err
>>  - pr_debug to dev_dbg
>>=20
>> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
>> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
>> Reviewed-by: Brian King <brking@linux.vnet.ibm.com>
>=20
> Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
>=20
> Changes mostly look fine, further comments below.
>=20
>> --- a/drivers/scsi/cxlflash/main.c
>> +++ b/drivers/scsi/cxlflash/main.c
>> @@ -58,8 +58,8 @@ static struct afu_cmd *cmd_checkout(struct afu =
*afu)
>>  		cmd =3D &afu->cmd[k];
>>=20
>>  		if (!atomic_dec_if_positive(&cmd->free)) {
>> -			pr_debug("%s: returning found index=3D%d\n",
>> -				 __func__, cmd->slot);
>> +			pr_devel("%s: returning found index=3D%d =
cmd=3D%p\n",
>> +				 __func__, cmd->slot, cmd);
>=20
>>  	pr_debug("%s: cmd failed afu_rc=3D%d scsi_rc=3D%d fc_rc=3D%d "
>> -		 "afu_extra=3D0x%X, scsi_entra=3D0x%X, fc_extra=3D0x%X\n",=

>> +		 "afu_extra=3D0x%X, scsi_extra=3D0x%X, fc_extra=3D0x%X\n",=

>>  		 __func__, ioasa->rc.afu_rc, ioasa->rc.scsi_rc,
>>  		 ioasa->rc.fc_rc, ioasa->afu_extra, ioasa->scsi_extra,
>>  		 ioasa->fc_extra);
>=20
> Minor nitpicking: mention that you fix these in the commit message.

Noted for the future.

>> @@ -240,9 +240,9 @@ static void cmd_complete(struct afu_cmd *cmd)
>>  		cmd_is_tmf =3D cmd->cmd_tmf;
>>  		cmd_checkin(cmd); /* Don't use cmd after here */
>>=20
>> -		pr_debug("%s: calling scsi_set_resid, scp=3D%p "
>> -			 "result=3D%X resid=3D%d\n", __func__,
>> -			 scp, scp->result, resid);
>> +		pr_debug_ratelimited("%s: calling scsi_done scp=3D%p =
result=3D%X "
>> +				     "ioasc=3D%d\n", __func__, scp, =
scp->result,
>> +				     cmd->sa.ioasc);
>>=20
>>  		scsi_set_resid(scp, resid);
>>  		scsi_dma_unmap(scp);
>=20
> Why has the message changed from scsi_set_resid to scsi_done, and =
should the message be moved to immediately before the scsi_done call?

In a later patch in the series the scsi_set_resid() is actually moved.

Re: [PATCH v4 29/32] cxlflash: Fix to double the delay each time

[Daniel Axtens]

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

>> On Sep 28, 2015, at 8:40 PM, Daniel Axtens <dja@axtens.net> wrote:
>> 
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA512
>> 
>> "Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:
>> 
>>> From: Manoj Kumar <manoj@linux.vnet.ibm.com>
>>> 
>>> The operator used to double the delay is incorrect and
>>> does not result in delay doubling.
>>> 
>>> To fix, use a left shift instead of the XOR operator.
>>> 
>> I can see that the patch is correct, but this commit message is a bit
>> confusing. What delay? In what circumstances are you doubling it? Why?
>
> This is the response delay while resetting the master context. The reset
> is performed by writing a bit and then waiting for it to clear. While waiting
> for it to clear, the code relaxes the delta between MMIO reads.

OK. If you do a v5, please include this in the commit message.

Regards,
Daniel

Re: [PATCH v4 30/32] cxlflash: Fix to avoid corrupting adapter fops

[Daniel Axtens]

[-- Attachment #1: Type: text/plain, Size: 4726 bytes --]

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> The corruption that this fix remedies is due to the fact that the fops
> is initially defaulted to values found within a static structure. When
> the fops is handed down to the CXL services later in the attach path,
> certain services are patched. The fops structure remains correct until
> the user count drops to 0 and the fops is reset, triggering the process
> to repeat again. The user counts are tightly coupled with the creation
> and deletion of the user context. If multiple users perform a disk
> attach at the same time, when the user count is currently 0, some users
> can be in the middle of obtaining a file descriptor and have not yet
> reached the context creation code that [in addition to creating the
> context] increments the user count. Subsequent users coming in to
> perform the attach see that the user count is still 0, and reinitialize
> the fops, temporarily removing the patched fops. The users that are in
> the middle obtaining their file descriptor may then receive an invalid
> descriptor.
>
> The fix simply removes the user count altogether and moves the fops
> initialization to probe time such that it is only performed one time
> for the life of the adapter. In the future, if the CXL services adopt
> a private member for their context, that could be used to store the
> adapter structure reference and cxlflash could revert to a model that
> does not require an embedded fops.

Yep, this looks good.

We have discussed adding a private data field to a cxl context, and will
no doubt revisit the question at some point in the future :)

Reviewed-by: Daniel Axtens <dja@axtens.net>

>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/common.h    |  3 +--
>  drivers/scsi/cxlflash/main.c      |  1 +
>  drivers/scsi/cxlflash/superpipe.c | 11 +----------
>  3 files changed, 3 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
> index bbfe711..c11cd19 100644
> --- a/drivers/scsi/cxlflash/common.h
> +++ b/drivers/scsi/cxlflash/common.h
> @@ -21,6 +21,7 @@
>  #include <scsi/scsi.h>
>  #include <scsi/scsi_device.h>
>  
> +extern const struct file_operations cxlflash_cxl_fops;
>  
>  #define MAX_CONTEXT  CXLFLASH_MAX_CONTEXT       /* num contexts per afu */
>  
> @@ -115,8 +116,6 @@ struct cxlflash_cfg {
>  	struct list_head ctx_err_recovery; /* contexts w/ recovery pending */
>  	struct file_operations cxl_fops;
>  
> -	atomic_t num_user_contexts;
> -
>  	/* Parameters that are LUN table related */
>  	int last_lun_index[CXLFLASH_NUM_FC_PORTS];
>  	int promote_lun_index;
> diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
> index be78906..38e7edc 100644
> --- a/drivers/scsi/cxlflash/main.c
> +++ b/drivers/scsi/cxlflash/main.c
> @@ -2386,6 +2386,7 @@ static int cxlflash_probe(struct pci_dev *pdev,
>  
>  	cfg->init_state = INIT_STATE_NONE;
>  	cfg->dev = pdev;
> +	cfg->cxl_fops = cxlflash_cxl_fops;
>  
>  	/*
>  	 * The promoted LUNs move to the top of the LUN table. The rest stay
> diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
> index 3cc8609..f625e07 100644
> --- a/drivers/scsi/cxlflash/superpipe.c
> +++ b/drivers/scsi/cxlflash/superpipe.c
> @@ -712,7 +712,6 @@ static void destroy_context(struct cxlflash_cfg *cfg,
>  	kfree(ctxi->rht_needs_ws);
>  	kfree(ctxi->rht_lun);
>  	kfree(ctxi);
> -	atomic_dec_if_positive(&cfg->num_user_contexts);
>  }
>  
>  /**
> @@ -769,7 +768,6 @@ static struct ctx_info *create_context(struct cxlflash_cfg *cfg,
>  	INIT_LIST_HEAD(&ctxi->luns);
>  	INIT_LIST_HEAD(&ctxi->list); /* initialize for list_empty() */
>  
> -	atomic_inc(&cfg->num_user_contexts);
>  	mutex_lock(&ctxi->mutex);
>  out:
>  	return ctxi;
> @@ -1164,10 +1162,7 @@ out:
>  	return rc;
>  }
>  
> -/*
> - * Local fops for adapter file descriptor
> - */
> -static const struct file_operations cxlflash_cxl_fops = {
> +const struct file_operations cxlflash_cxl_fops = {
>  	.owner = THIS_MODULE,
>  	.mmap = cxlflash_cxl_mmap,
>  	.release = cxlflash_cxl_release,
> @@ -1286,10 +1281,6 @@ static int cxlflash_disk_attach(struct scsi_device *sdev,
>  
>  	int fd = -1;
>  
> -	/* On first attach set fileops */
> -	if (atomic_read(&cfg->num_user_contexts) == 0)
> -		cfg->cxl_fops = cxlflash_cxl_fops;
> -
>  	if (attach->num_interrupts > 4) {
>  		dev_dbg(dev, "%s: Cannot support this many interrupts %llu\n",
>  			__func__, attach->num_interrupts);
> -- 
> 2.1.0

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 859 bytes --]

Re: [PATCH v4 32/32] cxlflash: Fix to avoid potential deadlock on EEH

[Daniel Axtens]

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

> Ioctl threads that use scsi_execute() can run for an excessive amount
> of time due to the fact that they have lengthy timeouts and retry logic
> built in. Under normal operation this is not an issue. However, once EEH
> enters the picture, a long execution time coupled with the possibility
> that a timeout can trigger entry to the driver via registered reset
> callbacks becomes a liability.
>
> In particular, a deadlock can occur when an EEH event is encountered
> while in running in scsi_execute(). As part of the recovery, the EEH
> handler drains all currently running ioctls, waiting until they have
> completed before proceeding with a reset. As the scsi_execute()'s are
> situated on the ioctl path, the EEH handler will wait until they (and
> the remainder of the ioctl handler they're associated with) have
> completed. Normally this would not be much of an issue aside from the
> longer recovery period. Unfortunately, the scsi_execute() triggers a
> reset when it times out. The reset handler will see that the device is
> already being reset and wait until that reset completed. This creates
> a condition where the EEH handler becomes stuck, infinitely waiting for
> the ioctl thread to complete.
>
> To avoid this behavior, temporarily unmark the scsi_execute() threads
> as an ioctl thread by releasing the ioctl read semaphore. This allows
> the EEH handler to proceed with a recovery while the thread is still
> running. Once the scsi_execute() returns, the ioctl read semaphore is
> reacquired and the adapter state is rechecked in case it changed while
> inside of scsi_execute(). The state check will wait if the adapter is
> still being recovered or returns a failure if the recovery failed. In
> the event that the adapter reset failed, the failure is simply returned
> as the ioctl would be unable to continue.

Yep, looks good.

Reviewed-by: Daniel Axtens <dja@axtens.net>

>
> Reported-by: Brian King <brking@linux.vnet.ibm.com>
> Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
> Signed-off-by: Manoj N. Kumar <manoj@linux.vnet.ibm.com>
> ---
>  drivers/scsi/cxlflash/superpipe.c | 30 +++++++++++++++++++++++++++++-
>  drivers/scsi/cxlflash/superpipe.h |  2 ++
>  drivers/scsi/cxlflash/vlun.c      | 29 +++++++++++++++++++++++++++++
>  3 files changed, 60 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
> index f625e07..8af7cdc 100644
> --- a/drivers/scsi/cxlflash/superpipe.c
> +++ b/drivers/scsi/cxlflash/superpipe.c
> @@ -283,6 +283,24 @@ out:
>   * @sdev:	SCSI device associated with LUN.
>   * @lli:	LUN destined for capacity request.
>   *
> + * The READ_CAP16 can take quite a while to complete. Should an EEH occur while
> + * in scsi_execute(), the EEH handler will attempt to recover. As part of the
> + * recovery, the handler drains all currently running ioctls, waiting until they
> + * have completed before proceeding with a reset. As this routine is used on the
> + * ioctl path, this can create a condition where the EEH handler becomes stuck,
> + * infinitely waiting for this ioctl thread. To avoid this behavior, temporarily
> + * unmark this thread as an ioctl thread by releasing the ioctl read semaphore.
> + * This will allow the EEH handler to proceed with a recovery while this thread
> + * is still running. Once the scsi_execute() returns, reacquire the ioctl read
> + * semaphore and check the adapter state in case it changed while inside of
> + * scsi_execute(). The state check will wait if the adapter is still being
> + * recovered or return a failure if the recovery failed. In the event that the
> + * adapter reset failed, simply return the failure as the ioctl would be unable
> + * to continue.
> + *
> + * Note that the above puts a requirement on this routine to only be called on
> + * an ioctl thread.
> + *
>   * Return: 0 on success, -errno on failure
>   */
>  static int read_cap16(struct scsi_device *sdev, struct llun_info *lli)
> @@ -314,8 +332,18 @@ retry:
>  	dev_dbg(dev, "%s: %ssending cmd(0x%x)\n", __func__,
>  		retry_cnt ? "re" : "", scsi_cmd[0]);
>  
> +	/* Drop the ioctl read semahpore across lengthy call */
> +	up_read(&cfg->ioctl_rwsem);
>  	result = scsi_execute(sdev, scsi_cmd, DMA_FROM_DEVICE, cmd_buf,
>  			      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES, 0, NULL);
> +	down_read(&cfg->ioctl_rwsem);
> +	rc = check_state(cfg);
> +	if (rc) {
> +		dev_err(dev, "%s: Failed state! result=0x08%X\n",
> +			__func__, result);
> +		rc = -ENODEV;
> +		goto out;
> +	}
>  
>  	if (driver_byte(result) == DRIVER_SENSE) {
>  		result &= ~(0xFF<<24); /* DRIVER_SENSE is not an error */
> @@ -1221,7 +1249,7 @@ static const struct file_operations null_fops = {
>   *
>   * Return: 0 on success, -errno on failure
>   */
> -static int check_state(struct cxlflash_cfg *cfg)
> +int check_state(struct cxlflash_cfg *cfg)
>  {
>  	struct device *dev = &cfg->dev->dev;
>  	int rc = 0;
> diff --git a/drivers/scsi/cxlflash/superpipe.h b/drivers/scsi/cxlflash/superpipe.h
> index 7df88ee..06a805a 100644
> --- a/drivers/scsi/cxlflash/superpipe.h
> +++ b/drivers/scsi/cxlflash/superpipe.h
> @@ -147,4 +147,6 @@ void cxlflash_ba_terminate(struct ba_lun *);
>  
>  int cxlflash_manage_lun(struct scsi_device *, struct dk_cxlflash_manage_lun *);
>  
> +int check_state(struct cxlflash_cfg *);
> +
>  #endif /* ifndef _CXLFLASH_SUPERPIPE_H */
> diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
> index b0eaf55..a53f583 100644
> --- a/drivers/scsi/cxlflash/vlun.c
> +++ b/drivers/scsi/cxlflash/vlun.c
> @@ -400,6 +400,24 @@ static int init_vlun(struct llun_info *lli)
>   * @lba:	Logical block address to start write same.
>   * @nblks:	Number of logical blocks to write same.
>   *
> + * The SCSI WRITE_SAME16 can take quite a while to complete. Should an EEH occur
> + * while in scsi_execute(), the EEH handler will attempt to recover. As part of
> + * the recovery, the handler drains all currently running ioctls, waiting until
> + * they have completed before proceeding with a reset. As this routine is used
> + * on the ioctl path, this can create a condition where the EEH handler becomes
> + * stuck, infinitely waiting for this ioctl thread. To avoid this behavior,
> + * temporarily unmark this thread as an ioctl thread by releasing the ioctl read
> + * semaphore. This will allow the EEH handler to proceed with a recovery while
> + * this thread is still running. Once the scsi_execute() returns, reacquire the
> + * ioctl read semaphore and check the adapter state in case it changed while
> + * inside of scsi_execute(). The state check will wait if the adapter is still
> + * being recovered or return a failure if the recovery failed. In the event that
> + * the adapter reset failed, simply return the failure as the ioctl would be
> + * unable to continue.
> + *
> + * Note that the above puts a requirement on this routine to only be called on
> + * an ioctl thread.
> + *
>   * Return: 0 on success, -errno on failure
>   */
>  static int write_same16(struct scsi_device *sdev,
> @@ -433,9 +451,20 @@ static int write_same16(struct scsi_device *sdev,
>  		put_unaligned_be32(ws_limit < left ? ws_limit : left,
>  				   &scsi_cmd[10]);
>  
> +		/* Drop the ioctl read semahpore across lengthy call */
> +		up_read(&cfg->ioctl_rwsem);
>  		result = scsi_execute(sdev, scsi_cmd, DMA_TO_DEVICE, cmd_buf,
>  				      CMD_BUFSIZE, sense_buf, to, CMD_RETRIES,
>  				      0, NULL);
> +		down_read(&cfg->ioctl_rwsem);
> +		rc = check_state(cfg);
> +		if (rc) {
> +			dev_err(dev, "%s: Failed state! result=0x08%X\n",
> +				__func__, result);
> +			rc = -ENODEV;
> +			goto out;
> +		}
> +
>  		if (result) {
>  			dev_err_ratelimited(dev, "%s: command failed for "
>  					    "offset %lld result=0x%x\n",
> -- 
> 2.1.0

Re: [PATCH v4 17/32] cxlflash: Remove dual port online dependency

[Daniel Axtens]

(resending to the list this time, apologies!)

>> I'm not sure I fully understand the flow of this function, but it looks
>> like you set rc=0 regardless of how things actually go: is this ever
>> going to print a return value other than zero?
>
> Correct, this function behaves more like a void for the time being. The
> overall goal of this is to allow a card to configure even when the link is
> down. At some later point when the link is transitioned to 'up', a link state
> change interrupt will trigger the port configuration. I left this with a return
> code for right now in case we need to alter the behavior again (based
> upon testing) and actually return a value other than 0.

OK. That makes more sense - it wasn't clear to me how it could be
correct to proceed if the links were down but now I understand how that
works. I think that explanation should go in the commit message.

>>> 	if (!wait_port_online(fc_regs, FC_PORT_STATUS_RETRY_INTERVAL_US,
>>> 			      FC_PORT_STATUS_RETRY_CNT)) {
>>> 		pr_debug("%s: wait on port %d to go online timed out\n",
>>> 			 __func__, port);

As an aside, should this be a bit noisier? It seems like something
a user would probably want to know - especially in the case where
something has actually gone wrong so there's no link state change
interrupt forthcoming regardless of how long you wait.

Regards,
Daniel

Re: [PATCH v4 27/32] cxlflash: Fix to prevent stale AFU RRQ

[Daniel Axtens]

>>> Following an adapter reset, the AFU RRQ that resides in host memory
>>> holds stale data. This can lead to a condition where the RRQ interrupt
>>> handler tries to process stale entries and/or endlessly loops due to an
>>> out of sync generation bit.
>>> 
>>> To fix, the AFU RRQ in host memory needs to be cleared after each reset.
>> 
>> This looks good. Do you need anything to bail out of cxlflash_rrq_irq if
>> the data goes stale or to all Fs while that function is running?
>
> We're not performing an MMIO here, so I'm not sure how the all Fs check
> would apply. We're also protected fairly well by the generation bit. I suppose
> we could look at adding some type of 'max iterations' count to protect against
> a runaway handler but that would be in a future patch.

Ah, right you are. I had confused all Fs with UEs.

Reviewed-by: Daniel Axtens <dja@axtens.net>

Regards,
Daniel

Re: [PATCH v4 25/32] cxlflash: Fix to prevent EEH recovery failure

[Daniel Axtens]

"Matthew R. Ochs" <mrochs@linux.vnet.ibm.com> writes:

>>> The process_sense() routine can perform a read capacity which
>>> can take some time to complete. If an EEH occurs while waiting
>>> on the read capacity, the EEH handler is unable to obtain the
>>> context's mutex in order to put the context in an error state.
>>> The EEH handler will sit and wait until the context is free,
>>> but this wait can last longer than the EEH handler tolerates,
>>> leading to a failed recovery.
>> 
>> I'm not quite clear on what you mean by the EEH handler timing
>> out. AFAIK there's nothing in eehd and the EEH core that times out if a
>> driver doesn't respond - indeed, it's pretty easy to hang eehd with a
>> misbehaving driver.
>> 
>> Are you referring to your own internal timeouts?
>> cxlflash_wait_for_pci_err_recovery and anything else that uses
>> CXLFLASH_PCI_ERROR_RECOVERY_TIMEOUT?
>
> Reading through this again I can see how this is misleading. This is
> actually similar and related to the deadlock scenario described in
> "Fix to avoid potential deadlock on EEH". Without this fix, you'd end
> up in a similar situation but deadlocked on the context mutex instead
> of the ioctl semaphore.

That makes _much_ more sense. If you could please revise the commit
message to explain that, you can include this in the next version:
Reviewed-by: Daniel Axtens <dja@axtens.net>

Regards,
Daniel

Re: [PATCH v4 17/32] cxlflash: Remove dual port online dependency

[Matthew R. Ochs]

> On Sep 30, 2015, at 6:50 PM, Daniel Axtens <dja@axtens.net> wrote:
> (resending to the list this time, apologies!)
>=20
>>> I'm not sure I fully understand the flow of this function, but it =
looks
>>> like you set rc=3D0 regardless of how things actually go: is this =
ever
>>> going to print a return value other than zero?
>>=20
>> Correct, this function behaves more like a void for the time being. =
The
>> overall goal of this is to allow a card to configure even when the =
link is
>> down. At some later point when the link is transitioned to 'up', a =
link state
>> change interrupt will trigger the port configuration. I left this =
with a return
>> code for right now in case we need to alter the behavior again (based
>> upon testing) and actually return a value other than 0.
>=20
> OK. That makes more sense - it wasn't clear to me how it could be
> correct to proceed if the links were down but now I understand how =
that
> works. I think that explanation should go in the commit message.
>=20
>>>> 	if (!wait_port_online(fc_regs, FC_PORT_STATUS_RETRY_INTERVAL_US,
>>>> 			      FC_PORT_STATUS_RETRY_CNT)) {
>>>> 		pr_debug("%s: wait on port %d to go online timed out\n",
>>>> 			 __func__, port);
>=20
> As an aside, should this be a bit noisier? It seems like something
> a user would probably want to know - especially in the case where
> something has actually gone wrong so there's no link state change
> interrupt forthcoming regardless of how long you wait.

You bring up a good point. There is another place where we are noisier
with respect to the link being down, so we'll do the same here. I'll =
include
this in v5 along with an updated commit message as requested.