[PATCH 0/7] Miscellaneous patches and bug fixes

[PATCH 0/7] Miscellaneous patches and bug fixes

[Uma Krishnan]

This patch series adds few improvements to the cxlflash driver and
it also contains couple of bug fixes.

This patch series is intended for 4.18 and is bisectable.

Matthew R. Ochs (1):
  cxlflash: Use local mutex for AFU serialization

Uma Krishnan (6):
  cxlflash: Yield to active send threads
  cxlflash: Limit the debug logs in the IO path
  cxlflash: Acquire semaphore before invoking ioctl services
  cxlflash: Add include guards to backend.h
  cxlflash: Abstract hardware dependent assignments
  cxlflash: Isolate external module dependencies

 drivers/scsi/cxlflash/Kconfig     |  2 +-
 drivers/scsi/cxlflash/Makefile    |  4 +++-
 drivers/scsi/cxlflash/backend.h   |  5 +++++
 drivers/scsi/cxlflash/common.h    |  1 +
 drivers/scsi/cxlflash/lunmgt.c    |  4 +++-
 drivers/scsi/cxlflash/main.c      | 21 +++++++++------------
 drivers/scsi/cxlflash/main.h      | 20 ++++++++++++++++++++
 drivers/scsi/cxlflash/superpipe.c |  9 ++++++++-
 drivers/scsi/cxlflash/vlun.c      |  3 ++-
 9 files changed, 52 insertions(+), 17 deletions(-)

-- 
2.1.0

[PATCH 0/7] Miscellaneous patches and bug fixes

[Uma Krishnan]

This patch series adds few improvements to the cxlflash driver and
it also contains couple of bug fixes.

This patch series is intended for 4.18 and is bisectable.

Matthew R. Ochs (1):
  cxlflash: Use local mutex for AFU serialization

Uma Krishnan (6):
  cxlflash: Yield to active send threads
  cxlflash: Limit the debug logs in the IO path
  cxlflash: Acquire semaphore before invoking ioctl services
  cxlflash: Add include guards to backend.h
  cxlflash: Abstract hardware dependent assignments
  cxlflash: Isolate external module dependencies

 drivers/scsi/cxlflash/Kconfig     |  2 +-
 drivers/scsi/cxlflash/Makefile    |  4 +++-
 drivers/scsi/cxlflash/backend.h   |  5 +++++
 drivers/scsi/cxlflash/common.h    |  1 +
 drivers/scsi/cxlflash/lunmgt.c    |  4 +++-
 drivers/scsi/cxlflash/main.c      | 21 +++++++++------------
 drivers/scsi/cxlflash/main.h      | 20 ++++++++++++++++++++
 drivers/scsi/cxlflash/superpipe.c |  9 ++++++++-
 drivers/scsi/cxlflash/vlun.c      |  3 ++-
 9 files changed, 52 insertions(+), 17 deletions(-)

-- 
2.1.0

[PATCH 1/7] cxlflash: Yield to active send threads

[Uma Krishnan]

The following Oops may be encountered if the device is reset, i.e. EEH
recovery, while there is heavy I/O traffic:

59:mon> t
[c000200db64bb680] c008000009264c40 cxlflash_queuecommand+0x3b8/0x500
					[cxlflash]
[c000200db64bb770] c00000000090d3b0 scsi_dispatch_cmd+0x130/0x2f0
[c000200db64bb7f0] c00000000090fdd8 scsi_request_fn+0x3c8/0x8d0
[c000200db64bb900] c00000000067f528 __blk_run_queue+0x68/0xb0
[c000200db64bb930] c00000000067ab80 __elv_add_request+0x140/0x3c0
[c000200db64bb9b0] c00000000068daac blk_execute_rq_nowait+0xec/0x1a0
[c000200db64bba00] c00000000068dbb0 blk_execute_rq+0x50/0xe0
[c000200db64bba50] c0000000006b2040 sg_io+0x1f0/0x520
[c000200db64bbaf0] c0000000006b2e94 scsi_cmd_ioctl+0x534/0x610
[c000200db64bbc20] c000000000926208 sd_ioctl+0x118/0x280
[c000200db64bbcc0] c00000000069f7ac blkdev_ioctl+0x7fc/0xe30
[c000200db64bbd20] c000000000439204 block_ioctl+0x84/0xa0
[c000200db64bbd40] c0000000003f8514 do_vfs_ioctl+0xd4/0xa00
[c000200db64bbde0] c0000000003f8f04 SyS_ioctl+0xc4/0x130
[c000200db64bbe30] c00000000000b184 system_call+0x58/0x6c

When there is no room to send the I/O request, the cached room is refreshed
by reading the memory mapped command room value from the AFU. The AFU
register mapping is refreshed during a reset, creating a race condition
that can lead to the Oops above.

During a device reset, the AFU should not be unmapped until all the active
send threads quiesce. An atomic counter, cmds_active, is currently used to
track internal AFU commands and quiesce during reset. This same counter can
also be used for the active send threads.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index a24d7e6..dad2be6 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -616,6 +616,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 		rc = 0;
 		goto out;
 	default:
+		atomic_inc(&afu->cmds_active);
 		break;
 	}
 
@@ -641,6 +642,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	memcpy(cmd->rcb.cdb, scp->cmnd, sizeof(cmd->rcb.cdb));
 
 	rc = afu->send_cmd(afu, cmd);
+	atomic_dec(&afu->cmds_active);
 out:
 	return rc;
 }
-- 
2.1.0

[PATCH 1/7] cxlflash: Yield to active send threads

[Uma Krishnan]

The following Oops may be encountered if the device is reset, i.e. EEH
recovery, while there is heavy I/O traffic:

59:mon> t
[c000200db64bb680] c008000009264c40 cxlflash_queuecommand+0x3b8/0x500
					[cxlflash]
[c000200db64bb770] c00000000090d3b0 scsi_dispatch_cmd+0x130/0x2f0
[c000200db64bb7f0] c00000000090fdd8 scsi_request_fn+0x3c8/0x8d0
[c000200db64bb900] c00000000067f528 __blk_run_queue+0x68/0xb0
[c000200db64bb930] c00000000067ab80 __elv_add_request+0x140/0x3c0
[c000200db64bb9b0] c00000000068daac blk_execute_rq_nowait+0xec/0x1a0
[c000200db64bba00] c00000000068dbb0 blk_execute_rq+0x50/0xe0
[c000200db64bba50] c0000000006b2040 sg_io+0x1f0/0x520
[c000200db64bbaf0] c0000000006b2e94 scsi_cmd_ioctl+0x534/0x610
[c000200db64bbc20] c000000000926208 sd_ioctl+0x118/0x280
[c000200db64bbcc0] c00000000069f7ac blkdev_ioctl+0x7fc/0xe30
[c000200db64bbd20] c000000000439204 block_ioctl+0x84/0xa0
[c000200db64bbd40] c0000000003f8514 do_vfs_ioctl+0xd4/0xa00
[c000200db64bbde0] c0000000003f8f04 SyS_ioctl+0xc4/0x130
[c000200db64bbe30] c00000000000b184 system_call+0x58/0x6c

When there is no room to send the I/O request, the cached room is refreshed
by reading the memory mapped command room value from the AFU. The AFU
register mapping is refreshed during a reset, creating a race condition
that can lead to the Oops above.

During a device reset, the AFU should not be unmapped until all the active
send threads quiesce. An atomic counter, cmds_active, is currently used to
track internal AFU commands and quiesce during reset. This same counter can
also be used for the active send threads.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index a24d7e6..dad2be6 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -616,6 +616,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 		rc = 0;
 		goto out;
 	default:
+		atomic_inc(&afu->cmds_active);
 		break;
 	}
 
@@ -641,6 +642,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp)
 	memcpy(cmd->rcb.cdb, scp->cmnd, sizeof(cmd->rcb.cdb));
 
 	rc = afu->send_cmd(afu, cmd);
+	atomic_dec(&afu->cmds_active);
 out:
 	return rc;
 }
-- 
2.1.0

Re: [PATCH 1/7] cxlflash: Yield to active send threads

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:04:46PM -0500, Uma Krishnan wrote:
> The following Oops may be encountered if the device is reset, i.e. EEH
> recovery, while there is heavy I/O traffic:
> 
> 59:mon> t
> [c000200db64bb680] c008000009264c40 cxlflash_queuecommand+0x3b8/0x500
> 					[cxlflash]
> [c000200db64bb770] c00000000090d3b0 scsi_dispatch_cmd+0x130/0x2f0
> [c000200db64bb7f0] c00000000090fdd8 scsi_request_fn+0x3c8/0x8d0
> [c000200db64bb900] c00000000067f528 __blk_run_queue+0x68/0xb0
> [c000200db64bb930] c00000000067ab80 __elv_add_request+0x140/0x3c0
> [c000200db64bb9b0] c00000000068daac blk_execute_rq_nowait+0xec/0x1a0
> [c000200db64bba00] c00000000068dbb0 blk_execute_rq+0x50/0xe0
> [c000200db64bba50] c0000000006b2040 sg_io+0x1f0/0x520
> [c000200db64bbaf0] c0000000006b2e94 scsi_cmd_ioctl+0x534/0x610
> [c000200db64bbc20] c000000000926208 sd_ioctl+0x118/0x280
> [c000200db64bbcc0] c00000000069f7ac blkdev_ioctl+0x7fc/0xe30
> [c000200db64bbd20] c000000000439204 block_ioctl+0x84/0xa0
> [c000200db64bbd40] c0000000003f8514 do_vfs_ioctl+0xd4/0xa00
> [c000200db64bbde0] c0000000003f8f04 SyS_ioctl+0xc4/0x130
> [c000200db64bbe30] c00000000000b184 system_call+0x58/0x6c
> 
> When there is no room to send the I/O request, the cached room is refreshed
> by reading the memory mapped command room value from the AFU. The AFU
> register mapping is refreshed during a reset, creating a race condition
> that can lead to the Oops above.
> 
> During a device reset, the AFU should not be unmapped until all the active
> send threads quiesce. An atomic counter, cmds_active, is currently used to
> track internal AFU commands and quiesce during reset. This same counter can
> also be used for the active send threads.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 1/7] cxlflash: Yield to active send threads

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:04:46PM -0500, Uma Krishnan wrote:
> The following Oops may be encountered if the device is reset, i.e. EEH
> recovery, while there is heavy I/O traffic:
> 
> 59:mon> t
> [c000200db64bb680] c008000009264c40 cxlflash_queuecommand+0x3b8/0x500
> 					[cxlflash]
> [c000200db64bb770] c00000000090d3b0 scsi_dispatch_cmd+0x130/0x2f0
> [c000200db64bb7f0] c00000000090fdd8 scsi_request_fn+0x3c8/0x8d0
> [c000200db64bb900] c00000000067f528 __blk_run_queue+0x68/0xb0
> [c000200db64bb930] c00000000067ab80 __elv_add_request+0x140/0x3c0
> [c000200db64bb9b0] c00000000068daac blk_execute_rq_nowait+0xec/0x1a0
> [c000200db64bba00] c00000000068dbb0 blk_execute_rq+0x50/0xe0
> [c000200db64bba50] c0000000006b2040 sg_io+0x1f0/0x520
> [c000200db64bbaf0] c0000000006b2e94 scsi_cmd_ioctl+0x534/0x610
> [c000200db64bbc20] c000000000926208 sd_ioctl+0x118/0x280
> [c000200db64bbcc0] c00000000069f7ac blkdev_ioctl+0x7fc/0xe30
> [c000200db64bbd20] c000000000439204 block_ioctl+0x84/0xa0
> [c000200db64bbd40] c0000000003f8514 do_vfs_ioctl+0xd4/0xa00
> [c000200db64bbde0] c0000000003f8f04 SyS_ioctl+0xc4/0x130
> [c000200db64bbe30] c00000000000b184 system_call+0x58/0x6c
> 
> When there is no room to send the I/O request, the cached room is refreshed
> by reading the memory mapped command room value from the AFU. The AFU
> register mapping is refreshed during a reset, creating a race condition
> that can lead to the Oops above.
> 
> During a device reset, the AFU should not be unmapped until all the active
> send threads quiesce. An atomic counter, cmds_active, is currently used to
> track internal AFU commands and quiesce during reset. This same counter can
> also be used for the active send threads.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

[PATCH 2/7] cxlflash: Limit the debug logs in the IO path

[Uma Krishnan]

The kernel log can get filled with debug messages from send_cmd_ioarrin()
when dynamic debug is enabled for the cxlflash module and there is a lot
of legacy I/O traffic.

While these messages are necessary to debug issues that involve command
tracking, the abundance of data can overwrite other useful data in the
log. The best option available is to limit the messages that should
serve most of the common use cases.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index dad2be6..cf0b407 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -339,8 +339,8 @@ static int send_cmd_ioarrin(struct afu *afu, struct afu_cmd *cmd)
 	writeq_be((u64)&cmd->rcb, &hwq->host_map->ioarrin);
 out:
 	spin_unlock_irqrestore(&hwq->hsq_slock, lock_flags);
-	dev_dbg(dev, "%s: cmd=%p len=%u ea=%016llx rc=%d\n", __func__,
-		cmd, cmd->rcb.data_len, cmd->rcb.data_ea, rc);
+	dev_dbg_ratelimited(dev, "%s: cmd=%p len=%u ea=%016llx rc=%d\n",
+		__func__, cmd, cmd->rcb.data_len, cmd->rcb.data_ea, rc);
 	return rc;
 }
 
-- 
2.1.0

[PATCH 2/7] cxlflash: Limit the debug logs in the IO path

[Uma Krishnan]

The kernel log can get filled with debug messages from send_cmd_ioarrin()
when dynamic debug is enabled for the cxlflash module and there is a lot
of legacy I/O traffic.

While these messages are necessary to debug issues that involve command
tracking, the abundance of data can overwrite other useful data in the
log. The best option available is to limit the messages that should
serve most of the common use cases.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index dad2be6..cf0b407 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -339,8 +339,8 @@ static int send_cmd_ioarrin(struct afu *afu, struct afu_cmd *cmd)
 	writeq_be((u64)&cmd->rcb, &hwq->host_map->ioarrin);
 out:
 	spin_unlock_irqrestore(&hwq->hsq_slock, lock_flags);
-	dev_dbg(dev, "%s: cmd=%p len=%u ea=%016llx rc=%d\n", __func__,
-		cmd, cmd->rcb.data_len, cmd->rcb.data_ea, rc);
+	dev_dbg_ratelimited(dev, "%s: cmd=%p len=%u ea=%016llx rc=%d\n",
+		__func__, cmd, cmd->rcb.data_len, cmd->rcb.data_ea, rc);
 	return rc;
 }
 
-- 
2.1.0

Re: [PATCH 2/7] cxlflash: Limit the debug logs in the IO path

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:08PM -0500, Uma Krishnan wrote:
> The kernel log can get filled with debug messages from send_cmd_ioarrin()
> when dynamic debug is enabled for the cxlflash module and there is a lot
> of legacy I/O traffic.
> 
> While these messages are necessary to debug issues that involve command
> tracking, the abundance of data can overwrite other useful data in the
> log. The best option available is to limit the messages that should
> serve most of the common use cases.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 2/7] cxlflash: Limit the debug logs in the IO path

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:08PM -0500, Uma Krishnan wrote:
> The kernel log can get filled with debug messages from send_cmd_ioarrin()
> when dynamic debug is enabled for the cxlflash module and there is a lot
> of legacy I/O traffic.
> 
> While these messages are necessary to debug issues that involve command
> tracking, the abundance of data can overwrite other useful data in the
> log. The best option available is to limit the messages that should
> serve most of the common use cases.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

[PATCH 3/7] cxlflash: Acquire semaphore before invoking ioctl services

[Uma Krishnan]

When a superpipe process that makes use of virtual LUNs is terminated or
killed abruptly, there is a possibility that the cxlflash driver could
hang and deprive other operations on the adapter.

The release fop registered to be invoked on a context close, detaches
every LUN associated with the context. The underlying service to detach
the LUN assumes it has been called with the read semaphore held, and
releases the semaphore before any operation that could be time consuming.

When invoked without holding the read semaphore, an opportunity is created
for the semaphore's count to become negative when it is temporarily
released during one of these potential lengthy operations. This negative
count results in subsequent acquisition attempts taking forever, leading
to the hang.

To support the current design point of holding the semaphore on the
ioctl() paths, the release fop should acquire it before invoking any
ioctl services.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 04a3bf9..5ba6e62 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -988,6 +988,10 @@ static int cxlflash_disk_detach(struct scsi_device *sdev,
  * theoretically never occur), every call into this routine results
  * in a complete freeing of a context.
  *
+ * Detaching the LUN is typically an ioctl() operation and the underlying
+ * code assumes that ioctl_rwsem has been acquired as a reader. To support
+ * that design point, the semaphore is acquired and released around detach.
+ *
  * Return: 0 on success
  */
 static int cxlflash_cxl_release(struct inode *inode, struct file *file)
@@ -1026,9 +1030,11 @@ static int cxlflash_cxl_release(struct inode *inode, struct file *file)
 
 	dev_dbg(dev, "%s: close for ctxid=%d\n", __func__, ctxid);
 
+	down_read(&cfg->ioctl_rwsem);
 	detach.context_id = ctxi->ctxid;
 	list_for_each_entry_safe(lun_access, t, &ctxi->luns, list)
 		_cxlflash_disk_detach(lun_access->sdev, ctxi, &detach);
+	up_read(&cfg->ioctl_rwsem);
 out_release:
 	cfg->ops->fd_release(inode, file);
 out:
-- 
2.1.0

[PATCH 3/7] cxlflash: Acquire semaphore before invoking ioctl services

[Uma Krishnan]

When a superpipe process that makes use of virtual LUNs is terminated or
killed abruptly, there is a possibility that the cxlflash driver could
hang and deprive other operations on the adapter.

The release fop registered to be invoked on a context close, detaches
every LUN associated with the context. The underlying service to detach
the LUN assumes it has been called with the read semaphore held, and
releases the semaphore before any operation that could be time consuming.

When invoked without holding the read semaphore, an opportunity is created
for the semaphore's count to become negative when it is temporarily
released during one of these potential lengthy operations. This negative
count results in subsequent acquisition attempts taking forever, leading
to the hang.

To support the current design point of holding the semaphore on the
ioctl() paths, the release fop should acquire it before invoking any
ioctl services.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/superpipe.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 04a3bf9..5ba6e62 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -988,6 +988,10 @@ static int cxlflash_disk_detach(struct scsi_device *sdev,
  * theoretically never occur), every call into this routine results
  * in a complete freeing of a context.
  *
+ * Detaching the LUN is typically an ioctl() operation and the underlying
+ * code assumes that ioctl_rwsem has been acquired as a reader. To support
+ * that design point, the semaphore is acquired and released around detach.
+ *
  * Return: 0 on success
  */
 static int cxlflash_cxl_release(struct inode *inode, struct file *file)
@@ -1026,9 +1030,11 @@ static int cxlflash_cxl_release(struct inode *inode, struct file *file)
 
 	dev_dbg(dev, "%s: close for ctxid=%d\n", __func__, ctxid);
 
+	down_read(&cfg->ioctl_rwsem);
 	detach.context_id = ctxi->ctxid;
 	list_for_each_entry_safe(lun_access, t, &ctxi->luns, list)
 		_cxlflash_disk_detach(lun_access->sdev, ctxi, &detach);
+	up_read(&cfg->ioctl_rwsem);
 out_release:
 	cfg->ops->fd_release(inode, file);
 out:
-- 
2.1.0

Re: [PATCH 3/7] cxlflash: Acquire semaphore before invoking ioctl services

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:22PM -0500, Uma Krishnan wrote:
> When a superpipe process that makes use of virtual LUNs is terminated or
> killed abruptly, there is a possibility that the cxlflash driver could
> hang and deprive other operations on the adapter.
> 
> The release fop registered to be invoked on a context close, detaches
> every LUN associated with the context. The underlying service to detach
> the LUN assumes it has been called with the read semaphore held, and
> releases the semaphore before any operation that could be time consuming.
> 
> When invoked without holding the read semaphore, an opportunity is created
> for the semaphore's count to become negative when it is temporarily
> released during one of these potential lengthy operations. This negative
> count results in subsequent acquisition attempts taking forever, leading
> to the hang.
> 
> To support the current design point of holding the semaphore on the
> ioctl() paths, the release fop should acquire it before invoking any
> ioctl services.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 3/7] cxlflash: Acquire semaphore before invoking ioctl services

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:22PM -0500, Uma Krishnan wrote:
> When a superpipe process that makes use of virtual LUNs is terminated or
> killed abruptly, there is a possibility that the cxlflash driver could
> hang and deprive other operations on the adapter.
> 
> The release fop registered to be invoked on a context close, detaches
> every LUN associated with the context. The underlying service to detach
> the LUN assumes it has been called with the read semaphore held, and
> releases the semaphore before any operation that could be time consuming.
> 
> When invoked without holding the read semaphore, an opportunity is created
> for the semaphore's count to become negative when it is temporarily
> released during one of these potential lengthy operations. This negative
> count results in subsequent acquisition attempts taking forever, leading
> to the hang.
> 
> To support the current design point of holding the semaphore on the
> ioctl() paths, the release fop should acquire it before invoking any
> ioctl services.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

[PATCH 4/7] cxlflash: Use local mutex for AFU serialization

[Uma Krishnan]

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

AFUs can only process a single AFU command at a time. This is enforced
with a global mutex situated within the AFU send routine. As this mutex
has a global scope, it has the potential to unnecessarily block commands
destined for other AFUs.

Instead of using a global mutex, transition the mutex to be per-AFU. This
will allow commands to only be blocked by siblings of the same AFU.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h | 1 +
 drivers/scsi/cxlflash/main.c   | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 3556b1d..dfcbecb 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -240,6 +240,7 @@ struct afu {
 	struct cxlflash_afu_map __iomem *afu_map;	/* entire MMIO map */
 
 	atomic_t cmds_active;	/* Number of currently active AFU commands */
+	struct mutex sync_active;	/* Mutex to serialize AFU commands */
 	u64 hb;
 	u32 internal_lun;	/* User-desired LUN mode for this AFU */
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index cf0b407..c91e912 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -2126,6 +2126,7 @@ static int init_afu(struct cxlflash_cfg *cfg)
 
 	cfg->ops->perst_reloads_same_image(cfg->afu_cookie, true);
 
+	mutex_init(&afu->sync_active);
 	afu->num_hwqs = afu->desired_hwqs;
 	for (i = 0; i < afu->num_hwqs; i++) {
 		rc = init_mc(cfg, i);
@@ -2309,7 +2310,6 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 	char *buf = NULL;
 	int rc = 0;
 	int nretry = 0;
-	static DEFINE_MUTEX(sync_active);
 
 	if (cfg->state != STATE_NORMAL) {
 		dev_dbg(dev, "%s: Sync not required state=%u\n",
@@ -2317,7 +2317,7 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 		return 0;
 	}
 
-	mutex_lock(&sync_active);
+	mutex_lock(&afu->sync_active);
 	atomic_inc(&afu->cmds_active);
 	buf = kmalloc(sizeof(*cmd) + __alignof__(*cmd) - 1, GFP_KERNEL);
 	if (unlikely(!buf)) {
@@ -2372,7 +2372,7 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 		*rcb->ioasa = cmd->sa;
 out:
 	atomic_dec(&afu->cmds_active);
-	mutex_unlock(&sync_active);
+	mutex_unlock(&afu->sync_active);
 	kfree(buf);
 	dev_dbg(dev, "%s: returning rc=%d\n", __func__, rc);
 	return rc;
-- 
2.1.0

[PATCH 4/7] cxlflash: Use local mutex for AFU serialization

[Uma Krishnan]

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

AFUs can only process a single AFU command at a time. This is enforced
with a global mutex situated within the AFU send routine. As this mutex
has a global scope, it has the potential to unnecessarily block commands
destined for other AFUs.

Instead of using a global mutex, transition the mutex to be per-AFU. This
will allow commands to only be blocked by siblings of the same AFU.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/common.h | 1 +
 drivers/scsi/cxlflash/main.c   | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/cxlflash/common.h b/drivers/scsi/cxlflash/common.h
index 3556b1d..dfcbecb 100644
--- a/drivers/scsi/cxlflash/common.h
+++ b/drivers/scsi/cxlflash/common.h
@@ -240,6 +240,7 @@ struct afu {
 	struct cxlflash_afu_map __iomem *afu_map;	/* entire MMIO map */
 
 	atomic_t cmds_active;	/* Number of currently active AFU commands */
+	struct mutex sync_active;	/* Mutex to serialize AFU commands */
 	u64 hb;
 	u32 internal_lun;	/* User-desired LUN mode for this AFU */
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index cf0b407..c91e912 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -2126,6 +2126,7 @@ static int init_afu(struct cxlflash_cfg *cfg)
 
 	cfg->ops->perst_reloads_same_image(cfg->afu_cookie, true);
 
+	mutex_init(&afu->sync_active);
 	afu->num_hwqs = afu->desired_hwqs;
 	for (i = 0; i < afu->num_hwqs; i++) {
 		rc = init_mc(cfg, i);
@@ -2309,7 +2310,6 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 	char *buf = NULL;
 	int rc = 0;
 	int nretry = 0;
-	static DEFINE_MUTEX(sync_active);
 
 	if (cfg->state != STATE_NORMAL) {
 		dev_dbg(dev, "%s: Sync not required state=%u\n",
@@ -2317,7 +2317,7 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 		return 0;
 	}
 
-	mutex_lock(&sync_active);
+	mutex_lock(&afu->sync_active);
 	atomic_inc(&afu->cmds_active);
 	buf = kmalloc(sizeof(*cmd) + __alignof__(*cmd) - 1, GFP_KERNEL);
 	if (unlikely(!buf)) {
@@ -2372,7 +2372,7 @@ static int send_afu_cmd(struct afu *afu, struct sisl_ioarcb *rcb)
 		*rcb->ioasa = cmd->sa;
 out:
 	atomic_dec(&afu->cmds_active);
-	mutex_unlock(&sync_active);
+	mutex_unlock(&afu->sync_active);
 	kfree(buf);
 	dev_dbg(dev, "%s: returning rc=%d\n", __func__, rc);
 	return rc;
-- 
2.1.0

[PATCH 5/7] cxlflash: Add include guards to backend.h

[Uma Krishnan]

The new header file, backend.h, that was recently added is missing
the include guards. This commit adds the guards.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/backend.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/scsi/cxlflash/backend.h b/drivers/scsi/cxlflash/backend.h
index bcd8a6c..55638d1 100644
--- a/drivers/scsi/cxlflash/backend.h
+++ b/drivers/scsi/cxlflash/backend.h
@@ -12,6 +12,9 @@
  * 2 of the License, or (at your option) any later version.
  */
 
+#ifndef _CXLFLASH_BACKEND_H
+#define _CXLFLASH_BACKEND_H
+
 extern const struct cxlflash_backend_ops cxlflash_cxl_ops;
 extern const struct cxlflash_backend_ops cxlflash_ocxl_ops;
 
@@ -45,3 +48,5 @@ struct cxlflash_backend_ops {
 	int (*fd_mmap)(struct file *file, struct vm_area_struct *vm);
 	int (*fd_release)(struct inode *inode, struct file *file);
 };
+
+#endif /* _CXLFLASH_BACKEND_H */
-- 
2.1.0

[PATCH 5/7] cxlflash: Add include guards to backend.h

[Uma Krishnan]

The new header file, backend.h, that was recently added is missing
the include guards. This commit adds the guards.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/backend.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/scsi/cxlflash/backend.h b/drivers/scsi/cxlflash/backend.h
index bcd8a6c..55638d1 100644
--- a/drivers/scsi/cxlflash/backend.h
+++ b/drivers/scsi/cxlflash/backend.h
@@ -12,6 +12,9 @@
  * 2 of the License, or (at your option) any later version.
  */
 
+#ifndef _CXLFLASH_BACKEND_H
+#define _CXLFLASH_BACKEND_H
+
 extern const struct cxlflash_backend_ops cxlflash_cxl_ops;
 extern const struct cxlflash_backend_ops cxlflash_ocxl_ops;
 
@@ -45,3 +48,5 @@ struct cxlflash_backend_ops {
 	int (*fd_mmap)(struct file *file, struct vm_area_struct *vm);
 	int (*fd_release)(struct inode *inode, struct file *file);
 };
+
+#endif /* _CXLFLASH_BACKEND_H */
-- 
2.1.0

Re: [PATCH 5/7] cxlflash: Add include guards to backend.h

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:51PM -0500, Uma Krishnan wrote:
> The new header file, backend.h, that was recently added is missing
> the include guards. This commit adds the guards.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 5/7] cxlflash: Add include guards to backend.h

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:05:51PM -0500, Uma Krishnan wrote:
> The new header file, backend.h, that was recently added is missing
> the include guards. This commit adds the guards.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

[PATCH 6/7] cxlflash: Abstract hardware dependent assignments

[Uma Krishnan]

As a staging cleanup to support transport specific builds of the cxlflash
module, relocate device dependent assignments to header files. This will
avoid littering the core driver with conditional compilation logic.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c |  7 ++-----
 drivers/scsi/cxlflash/main.h | 15 +++++++++++++++
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index c91e912..cd7dcc5 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -3708,11 +3708,8 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->init_state = INIT_STATE_NONE;
 	cfg->dev = pdev;
 	cfg->cxl_fops = cxlflash_cxl_fops;
-
-	if (ddv->flags & CXLFLASH_OCXL_DEV)
-		cfg->ops = &cxlflash_ocxl_ops;
-	else
-		cfg->ops = &cxlflash_cxl_ops;
+	cfg->ops = cxlflash_assign_ops(ddv);
+	WARN_ON_ONCE(!cfg->ops);
 
 	/*
 	 * Promoted LUNs move to the top of the LUN table. The rest stay on
diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h
index 6f1be62..ed4908e 100644
--- a/drivers/scsi/cxlflash/main.h
+++ b/drivers/scsi/cxlflash/main.h
@@ -20,6 +20,8 @@
 #include <scsi/scsi.h>
 #include <scsi/scsi_device.h>
 
+#include "backend.h"
+
 #define CXLFLASH_NAME		"cxlflash"
 #define CXLFLASH_ADAPTER_NAME	"IBM POWER CXL Flash Adapter"
 #define CXLFLASH_MAX_ADAPTERS	32
@@ -100,6 +102,19 @@ struct dev_dependent_vals {
 #define CXLFLASH_OCXL_DEV		0x0000000000000004ULL
 };
 
+static inline const struct cxlflash_backend_ops *
+cxlflash_assign_ops(struct dev_dependent_vals *ddv)
+{
+	const struct cxlflash_backend_ops *ops = NULL;
+
+	if (ddv->flags & CXLFLASH_OCXL_DEV)
+		ops = &cxlflash_ocxl_ops;
+	if (!(ddv->flags & CXLFLASH_OCXL_DEV))
+		ops = &cxlflash_cxl_ops;
+
+	return ops;
+}
+
 struct asyc_intr_info {
 	u64 status;
 	char *desc;
-- 
2.1.0

[PATCH 6/7] cxlflash: Abstract hardware dependent assignments

[Uma Krishnan]

As a staging cleanup to support transport specific builds of the cxlflash
module, relocate device dependent assignments to header files. This will
avoid littering the core driver with conditional compilation logic.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/main.c |  7 ++-----
 drivers/scsi/cxlflash/main.h | 15 +++++++++++++++
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index c91e912..cd7dcc5 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -3708,11 +3708,8 @@ static int cxlflash_probe(struct pci_dev *pdev,
 	cfg->init_state = INIT_STATE_NONE;
 	cfg->dev = pdev;
 	cfg->cxl_fops = cxlflash_cxl_fops;
-
-	if (ddv->flags & CXLFLASH_OCXL_DEV)
-		cfg->ops = &cxlflash_ocxl_ops;
-	else
-		cfg->ops = &cxlflash_cxl_ops;
+	cfg->ops = cxlflash_assign_ops(ddv);
+	WARN_ON_ONCE(!cfg->ops);
 
 	/*
 	 * Promoted LUNs move to the top of the LUN table. The rest stay on
diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h
index 6f1be62..ed4908e 100644
--- a/drivers/scsi/cxlflash/main.h
+++ b/drivers/scsi/cxlflash/main.h
@@ -20,6 +20,8 @@
 #include <scsi/scsi.h>
 #include <scsi/scsi_device.h>
 
+#include "backend.h"
+
 #define CXLFLASH_NAME		"cxlflash"
 #define CXLFLASH_ADAPTER_NAME	"IBM POWER CXL Flash Adapter"
 #define CXLFLASH_MAX_ADAPTERS	32
@@ -100,6 +102,19 @@ struct dev_dependent_vals {
 #define CXLFLASH_OCXL_DEV		0x0000000000000004ULL
 };
 
+static inline const struct cxlflash_backend_ops *
+cxlflash_assign_ops(struct dev_dependent_vals *ddv)
+{
+	const struct cxlflash_backend_ops *ops = NULL;
+
+	if (ddv->flags & CXLFLASH_OCXL_DEV)
+		ops = &cxlflash_ocxl_ops;
+	if (!(ddv->flags & CXLFLASH_OCXL_DEV))
+		ops = &cxlflash_cxl_ops;
+
+	return ops;
+}
+
 struct asyc_intr_info {
 	u64 status;
 	char *desc;
-- 
2.1.0

Re: [PATCH 6/7] cxlflash: Abstract hardware dependent assignments

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:06:05PM -0500, Uma Krishnan wrote:
> As a staging cleanup to support transport specific builds of the cxlflash
> module, relocate device dependent assignments to header files. This will
> avoid littering the core driver with conditional compilation logic.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 6/7] cxlflash: Abstract hardware dependent assignments

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:06:05PM -0500, Uma Krishnan wrote:
> As a staging cleanup to support transport specific builds of the cxlflash
> module, relocate device dependent assignments to header files. This will
> avoid littering the core driver with conditional compilation logic.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

[PATCH 7/7] cxlflash: Isolate external module dependencies

[Uma Krishnan]

Depending on the underlying transport, cxlflash has a dependency on either
the CXL or OCXL drivers, which are enabled via their Kconfig option.
Instead of having a module wide dependency on these config options, it is
better to isolate the object modules that are dependent on the CXL and OCXL
drivers and adjust the module dependencies accordingly.

This commit isolates the object files that are dependent on CXL and/or
OCXL. The cxl/ocxl fops used in the core driver are tucked under an ifdef
to avoid compilation errors.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/Kconfig     | 2 +-
 drivers/scsi/cxlflash/Makefile    | 4 +++-
 drivers/scsi/cxlflash/lunmgt.c    | 4 +++-
 drivers/scsi/cxlflash/main.c      | 2 --
 drivers/scsi/cxlflash/main.h      | 5 +++++
 drivers/scsi/cxlflash/superpipe.c | 3 ++-
 drivers/scsi/cxlflash/vlun.c      | 3 ++-
 7 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/drivers/scsi/cxlflash/Kconfig b/drivers/scsi/cxlflash/Kconfig
index e2a3a1b..f1b17e3 100644
--- a/drivers/scsi/cxlflash/Kconfig
+++ b/drivers/scsi/cxlflash/Kconfig
@@ -4,7 +4,7 @@
 
 config CXLFLASH
 	tristate "Support for IBM CAPI Flash"
-	depends on PCI && SCSI && CXL && OCXL && EEH
+	depends on PCI && SCSI && (CXL || OCXL) && EEH
 	select IRQ_POLL
 	default m
 	help
diff --git a/drivers/scsi/cxlflash/Makefile b/drivers/scsi/cxlflash/Makefile
index 5124c68..283377d 100644
--- a/drivers/scsi/cxlflash/Makefile
+++ b/drivers/scsi/cxlflash/Makefile
@@ -1,2 +1,4 @@
 obj-$(CONFIG_CXLFLASH) += cxlflash.o
-cxlflash-y += main.o superpipe.o lunmgt.o vlun.o cxl_hw.o ocxl_hw.o
+cxlflash-y += main.o superpipe.o lunmgt.o vlun.o
+cxlflash-$(CONFIG_CXL) += cxl_hw.o
+cxlflash-$(CONFIG_OCXL) += ocxl_hw.o
diff --git a/drivers/scsi/cxlflash/lunmgt.c b/drivers/scsi/cxlflash/lunmgt.c
index 4d232e2..edea125 100644
--- a/drivers/scsi/cxlflash/lunmgt.c
+++ b/drivers/scsi/cxlflash/lunmgt.c
@@ -12,9 +12,11 @@
  * 2 of the License, or (at your option) any later version.
  */
 
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+
 #include <scsi/scsi_host.h>
 #include <uapi/scsi/cxlflash_ioctl.h>
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index cd7dcc5..6637116 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -19,8 +19,6 @@
 
 #include <asm/unaligned.h>
 
-#include <misc/cxl.h>
-
 #include <scsi/scsi_cmnd.h>
 #include <scsi/scsi_host.h>
 #include <uapi/scsi/cxlflash_ioctl.h>
diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h
index ed4908e..2a39778 100644
--- a/drivers/scsi/cxlflash/main.h
+++ b/drivers/scsi/cxlflash/main.h
@@ -107,10 +107,15 @@ cxlflash_assign_ops(struct dev_dependent_vals *ddv)
 {
 	const struct cxlflash_backend_ops *ops = NULL;
 
+#ifdef CONFIG_OCXL
 	if (ddv->flags & CXLFLASH_OCXL_DEV)
 		ops = &cxlflash_ocxl_ops;
+#endif
+
+#ifdef CONFIG_CXL
 	if (!(ddv->flags & CXLFLASH_OCXL_DEV))
 		ops = &cxlflash_cxl_ops;
+#endif
 
 	return ops;
 }
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 5ba6e62..e489d89 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -14,8 +14,9 @@
 
 #include <linux/delay.h>
 #include <linux/file.h>
+#include <linux/interrupt.h>
+#include <linux/pci.h>
 #include <linux/syscalls.h>
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 
 #include <scsi/scsi.h>
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 5deef57..66e445a 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -12,8 +12,9 @@
  * 2 of the License, or (at your option) any later version.
  */
 
+#include <linux/interrupt.h>
+#include <linux/pci.h>
 #include <linux/syscalls.h>
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 #include <asm/bitsperlong.h>
 
-- 
2.1.0

[PATCH 7/7] cxlflash: Isolate external module dependencies

[Uma Krishnan]

Depending on the underlying transport, cxlflash has a dependency on either
the CXL or OCXL drivers, which are enabled via their Kconfig option.
Instead of having a module wide dependency on these config options, it is
better to isolate the object modules that are dependent on the CXL and OCXL
drivers and adjust the module dependencies accordingly.

This commit isolates the object files that are dependent on CXL and/or
OCXL. The cxl/ocxl fops used in the core driver are tucked under an ifdef
to avoid compilation errors.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
---
 drivers/scsi/cxlflash/Kconfig     | 2 +-
 drivers/scsi/cxlflash/Makefile    | 4 +++-
 drivers/scsi/cxlflash/lunmgt.c    | 4 +++-
 drivers/scsi/cxlflash/main.c      | 2 --
 drivers/scsi/cxlflash/main.h      | 5 +++++
 drivers/scsi/cxlflash/superpipe.c | 3 ++-
 drivers/scsi/cxlflash/vlun.c      | 3 ++-
 7 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/drivers/scsi/cxlflash/Kconfig b/drivers/scsi/cxlflash/Kconfig
index e2a3a1b..f1b17e3 100644
--- a/drivers/scsi/cxlflash/Kconfig
+++ b/drivers/scsi/cxlflash/Kconfig
@@ -4,7 +4,7 @@
 
 config CXLFLASH
 	tristate "Support for IBM CAPI Flash"
-	depends on PCI && SCSI && CXL && OCXL && EEH
+	depends on PCI && SCSI && (CXL || OCXL) && EEH
 	select IRQ_POLL
 	default m
 	help
diff --git a/drivers/scsi/cxlflash/Makefile b/drivers/scsi/cxlflash/Makefile
index 5124c68..283377d 100644
--- a/drivers/scsi/cxlflash/Makefile
+++ b/drivers/scsi/cxlflash/Makefile
@@ -1,2 +1,4 @@
 obj-$(CONFIG_CXLFLASH) += cxlflash.o
-cxlflash-y += main.o superpipe.o lunmgt.o vlun.o cxl_hw.o ocxl_hw.o
+cxlflash-y += main.o superpipe.o lunmgt.o vlun.o
+cxlflash-$(CONFIG_CXL) += cxl_hw.o
+cxlflash-$(CONFIG_OCXL) += ocxl_hw.o
diff --git a/drivers/scsi/cxlflash/lunmgt.c b/drivers/scsi/cxlflash/lunmgt.c
index 4d232e2..edea125 100644
--- a/drivers/scsi/cxlflash/lunmgt.c
+++ b/drivers/scsi/cxlflash/lunmgt.c
@@ -12,9 +12,11 @@
  * 2 of the License, or (at your option) any later version.
  */
 
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+
 #include <scsi/scsi_host.h>
 #include <uapi/scsi/cxlflash_ioctl.h>
 
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c
index cd7dcc5..6637116 100644
--- a/drivers/scsi/cxlflash/main.c
+++ b/drivers/scsi/cxlflash/main.c
@@ -19,8 +19,6 @@
 
 #include <asm/unaligned.h>
 
-#include <misc/cxl.h>
-
 #include <scsi/scsi_cmnd.h>
 #include <scsi/scsi_host.h>
 #include <uapi/scsi/cxlflash_ioctl.h>
diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h
index ed4908e..2a39778 100644
--- a/drivers/scsi/cxlflash/main.h
+++ b/drivers/scsi/cxlflash/main.h
@@ -107,10 +107,15 @@ cxlflash_assign_ops(struct dev_dependent_vals *ddv)
 {
 	const struct cxlflash_backend_ops *ops = NULL;
 
+#ifdef CONFIG_OCXL
 	if (ddv->flags & CXLFLASH_OCXL_DEV)
 		ops = &cxlflash_ocxl_ops;
+#endif
+
+#ifdef CONFIG_CXL
 	if (!(ddv->flags & CXLFLASH_OCXL_DEV))
 		ops = &cxlflash_cxl_ops;
+#endif
 
 	return ops;
 }
diff --git a/drivers/scsi/cxlflash/superpipe.c b/drivers/scsi/cxlflash/superpipe.c
index 5ba6e62..e489d89 100644
--- a/drivers/scsi/cxlflash/superpipe.c
+++ b/drivers/scsi/cxlflash/superpipe.c
@@ -14,8 +14,9 @@
 
 #include <linux/delay.h>
 #include <linux/file.h>
+#include <linux/interrupt.h>
+#include <linux/pci.h>
 #include <linux/syscalls.h>
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 
 #include <scsi/scsi.h>
diff --git a/drivers/scsi/cxlflash/vlun.c b/drivers/scsi/cxlflash/vlun.c
index 5deef57..66e445a 100644
--- a/drivers/scsi/cxlflash/vlun.c
+++ b/drivers/scsi/cxlflash/vlun.c
@@ -12,8 +12,9 @@
  * 2 of the License, or (at your option) any later version.
  */
 
+#include <linux/interrupt.h>
+#include <linux/pci.h>
 #include <linux/syscalls.h>
-#include <misc/cxl.h>
 #include <asm/unaligned.h>
 #include <asm/bitsperlong.h>
 
-- 
2.1.0

Re: [PATCH 7/7] cxlflash: Isolate external module dependencies

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:06:19PM -0500, Uma Krishnan wrote:
> Depending on the underlying transport, cxlflash has a dependency on either
> the CXL or OCXL drivers, which are enabled via their Kconfig option.
> Instead of having a module wide dependency on these config options, it is
> better to isolate the object modules that are dependent on the CXL and OCXL
> drivers and adjust the module dependencies accordingly.
> 
> This commit isolates the object files that are dependent on CXL and/or
> OCXL. The cxl/ocxl fops used in the core driver are tucked under an ifdef
> to avoid compilation errors.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 7/7] cxlflash: Isolate external module dependencies

[Matthew R. Ochs]

On Fri, May 11, 2018 at 02:06:19PM -0500, Uma Krishnan wrote:
> Depending on the underlying transport, cxlflash has a dependency on either
> the CXL or OCXL drivers, which are enabled via their Kconfig option.
> Instead of having a module wide dependency on these config options, it is
> better to isolate the object modules that are dependent on the CXL and OCXL
> drivers and adjust the module dependencies accordingly.
> 
> This commit isolates the object files that are dependent on CXL and/or
> OCXL. The cxl/ocxl fops used in the core driver are tucked under an ifdef
> to avoid compilation errors.
> 
> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>

Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>

Re: [PATCH 0/7] Miscellaneous patches and bug fixes

[Martin K. Petersen]

Uma,

> This patch series adds few improvements to the cxlflash driver and it
> also contains couple of bug fixes.

Applied to 4.18/scsi-queue, thank you!

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH 0/7] Miscellaneous patches and bug fixes

[Martin K. Petersen]

Uma,

> This patch series adds few improvements to the cxlflash driver and it
> also contains couple of bug fixes.

Applied to 4.18/scsi-queue, thank you!

-- 
Martin K. Petersen	Oracle Linux Engineering