Skip to contentSkip to conversation

scsi: ufs: Let devices remain runtime suspended during system suspend

PATCH Superseded Adrian Hunter 2 participants 9 messages Sep 16, 2021 – Sep 20, 2021

Subsystems (2)

v4 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

v4 cover scsi: ufs: Let devices remain runtime suspended during system suspend

Hi

UFS devices can remain runtime suspended at system suspend time, if the conditions are right. Add support for that, first fixing the impediments.

Changes in V4:

scsi: ufs: Fix error handler clear ua deadlock

Do request-sense directly

Changes in V3:

scsi: ufs: Fix error handler clear ua deadlock

Correct commit message.
Amend stable tags to add dependent cherry picks

Changes in V2:

scsi: ufs: Let devices remain runtime suspended during system suspend

The ufs-hisi driver uses different RPM and SPM, but it is made explicit by a new parameter to suspend prepare.

Adrian Hunter (3):
scsi: ufs: Fix error handler clear ua deadlock
scsi: ufs: Fix runtime PM dependencies getting broken
scsi: ufs: Let devices remain runtime suspended during system suspend

 drivers/scsi/scsi_pm.c      |  16 ++-
 drivers/scsi/ufs/ufs-hisi.c |   8 +-
 drivers/scsi/ufs/ufs.h      |   3 +-
 drivers/scsi/ufs/ufshcd.c   | 255 ++++++++++++++++++++++++++++++++------------
 drivers/scsi/ufs/ufshcd.h   |  21 +++-
 include/scsi/scsi_device.h  |   1 +
 include/trace/events/ufs.h  |   5 +-
 7 files changed, 231 insertions(+), 78 deletions(-)

Regards
Adrian

#1 lore raw

PATCH v4 1/3 scsi: ufs: Fix error handler clear ua deadlock

There is no guarantee to be able to enter the queue if requests are blocked. That is because freezing the queue will block entry to the queue, but freezing also waits for outstanding requests which can make no progress while the queue is blocked.

That situation can happen when the error handler issues requests to clear unit attention condition. Requests are blocked by scsi_schedule_eh() and not cleared until the error handler has run. Requests can also be blocked if the ufshcd_state is UFSHCD_STATE_EH_SCHEDULED_FATAL, which can happen as a result of prior error handler activity. Requests cannot make progress when ufshcd_state is UFSHCD_STATE_EH_SCHEDULED_FATAL, and only the error handler can change that, so if the error handler is waiting to enter the queue and blk_mq_freeze_queue() is waiting for outstanding requests, they will deadlock.

Fix by issuing REQUEST_SENSE directly avoiding the SCSI queues, in a similar fashion as other device commands.

Fixes b294ff3e34490f ("scsi: ufs: core: Enable power management for wlun")
---
 drivers/scsi/ufs/ufs.h     |   3 +-
 drivers/scsi/ufs/ufshcd.c  | 201 ++++++++++++++++++++++++++-----------
 drivers/scsi/ufs/ufshcd.h  |   9 ++
 include/trace/events/ufs.h |   5 +-
 4 files changed, 160 insertions(+), 58 deletions(-)
drivers/scsi/ufs/ufs.h +2 -1
index 8c6b38b1b142..c60c6d7c62c1 100644
@@ -612,7 +612,8 @@ struct ufs_dev_info {
 enum ufs_trace_str_t {
 	UFS_CMD_SEND, UFS_CMD_COMP, UFS_DEV_COMP,
 	UFS_QUERY_SEND, UFS_QUERY_COMP, UFS_QUERY_ERR,
-	UFS_TM_SEND, UFS_TM_COMP, UFS_TM_ERR
+	UFS_TM_SEND, UFS_TM_COMP, UFS_TM_ERR,
+	UFS_SENSE_SEND, UFS_SENSE_COMP, UFS_SENSE_ERR,
 };
 
 /*
drivers/scsi/ufs/ufshcd.c +145 -56

271 lines not on this page · raw message

drivers/scsi/ufs/ufshcd.h +9 -0
index 4723f27a55d1..82c0f689cbdb 100644
@@ -56,6 +56,8 @@ struct ufs_hba;
 enum dev_cmd_type {
 	DEV_CMD_TYPE_NOP		= 0x0,
 	DEV_CMD_TYPE_QUERY		= 0x1,
+	DEV_CMD_TYPE_SENSE		= 0x2,
+	DEV_CMD_TYPE_INVALID		= 0xff,
 };
 
 enum ufs_event_type {
@@ -236,17 +238,24 @@ struct ufs_query {
 	struct ufs_query_res response;
 };
 
+/* UFS_DEV_CMD_BUF_SZ must be a power of 2 and bigger than UFS_SENSE_SIZE */
+#define UFS_DEV_CMD_BUF_SZ 32
+
 /**
  * struct ufs_dev_cmd - all assosiated fields with device management commands
  * @type: device management command type - Query, NOP OUT
  * @lock: lock to allow one command at a time
  * @complete: internal commands completion
+ * @buf: buffer for data
+ * @buf_dma_addr: DMA address of @buf
  */
 struct ufs_dev_cmd {
 	enum dev_cmd_type type;
 	struct mutex lock;
 	struct completion *complete;
 	struct ufs_query query;
+	void *buf;
+	dma_addr_t buf_dma_addr;
 };
 
 /**
include/trace/events/ufs.h +4 -1
index 599739ee7b20..efa2af4a3a5b 100644
@@ -46,7 +46,10 @@
 	EM(UFS_QUERY_ERR,	"query_complete_err")		\
 	EM(UFS_TM_SEND,		"tm_send")			\
 	EM(UFS_TM_COMP,		"tm_complete")			\
-	EMe(UFS_TM_ERR,		"tm_complete_err")
+	EM(UFS_TM_ERR,		"tm_complete_err")		\
+	EM(UFS_SENSE_SEND,	"sense_send")			\
+	EM(UFS_SENSE_COMP,	"sense_complete")		\
+	EMe(UFS_SENSE_ERR,	"sense_complete_err")
 
 #define UFS_CMD_TRACE_TSF_TYPES					\
 	EM(UFS_TSF_CDB,		"CDB")		                \
Signature · 1 line
2.17.1
#2 lore raw

Re: [PATCH V4 1/3] scsi: ufs: Fix error handler clear ua deadlock

Quoted 3 lines · Adrian Hunter
drivers/scsi/ufs/ufshcd.c +1 -1
-static void ufshcd_request_sense_done(struct request *rq, blk_status_t error)
+static int ufshcd_request_sense_direct(struct ufs_hba *hba, u8 wlun)
  {

[ ... ]

drivers/scsi/ufs/ufshcd.c static int ufshcd_add_lus(struct ufs_hba *hba) · line 7993

Quoted 2 lines
drivers/scsi/ufs/ufshcd.c +2 -0
+	/* The command queue cannot be frozen */
+	req = blk_get_request(q, REQ_OP_DRV_OUT, 0);

hba->cmd_queue shares a tag set with the request queues associated with SCSI LUNs. Since this blk_get_request() call happens from the context of the error handler before SCSI requests are unblocked, it will hang if all tags are in use for SCSI requests before the error handler starts.

Bart.

#3 lore raw

Re: [PATCH V4 1/3] scsi: ufs: Fix error handler clear ua deadlock

Quoted 5 lines · Bart Van Assche
Quoted 3 lines · Adrian Hunter
drivers/scsi/ufs/ufshcd.c +1 -1
-static void ufshcd_request_sense_done(struct request *rq, blk_status_t error)
+static int ufshcd_request_sense_direct(struct ufs_hba *hba, u8 wlun)

  {

[ ... ]

Quoted 7 lines
Quoted 2 lines

+    /* The command queue cannot be frozen */
+    req = blk_get_request(q, REQ_OP_DRV_OUT, 0);

hba->cmd_queue shares a tag set with the request queues associated with SCSI LUNs. Since this blk_get_request() call happens from the context of the error handler before SCSI requests are unblocked, it will hang if all tags are in use for SCSI requests before the error handler starts.

All the commands sent by ufshcd_probe_hba() take the same approach, so no difference there.

#4 lore raw

Re: [PATCH V4 1/3] scsi: ufs: Fix error handler clear ua deadlock

Quoted 9 lines · Adrian Hunter
Quoted 8 lines · Bart Van Assche
Quoted 3 lines · Adrian Hunter
drivers/scsi/ufs/ufshcd.c +1 -1
-static void ufshcd_request_sense_done(struct request *rq, blk_status_t error)
+static int ufshcd_request_sense_direct(struct ufs_hba *hba, u8 wlun)
   {

[ ... ]

Quoted 2 lines

+    /* The command queue cannot be frozen */
+    req = blk_get_request(q, REQ_OP_DRV_OUT, 0);

Quoted 7 lines
Quoted 4 lines

hba->cmd_queue shares a tag set with the request queues associated with SCSI LUNs. Since this blk_get_request() call happens from the context of the error handler before SCSI requests are unblocked, it will hang if all tags are in use for SCSI requests before the error handler starts.

All the commands sent by ufshcd_probe_hba() take the same approach, so no difference there.

The same comment applies to the commands sent by ufshcd_probe_hba() I think. If you would like to address this issue, how about passing the flag BLK_MQ_REQ_RESERVED to the blk_get_request() call in ufshcd_exec_dev_cmd()? The following additional changes are required to make this work:
* Add a 'reserved_tags' member to struct scsi_host_template, e.g. close to tag_alloc_policy.
* Modify scsi_mq_setup_tags() such that it copies the reserved_tags value from the host template into tag_set->reserved_tags.
* Set 'reserved_tags' in the UFS driver SCSI host template.

Thanks,

Bart.

#5 lore raw

Re: [PATCH V4 1/3] scsi: ufs: Fix error handler clear ua deadlock

Quoted 18 lines · Bart Van Assche
Quoted 9 lines · Adrian Hunter
Quoted 8 lines · Bart Van Assche
Quoted 3 lines · Adrian Hunter
drivers/scsi/ufs/ufshcd.c +1 -1
-static void ufshcd_request_sense_done(struct request *rq, blk_status_t error)
+static int ufshcd_request_sense_direct(struct ufs_hba *hba, u8 wlun)

   {

[ ... ]

Quoted 2 lines

+    /* The command queue cannot be frozen */
+    req = blk_get_request(q, REQ_OP_DRV_OUT, 0);

Quoted 7 lines
Quoted 4 lines

hba->cmd_queue shares a tag set with the request queues associated with SCSI LUNs. Since this blk_get_request() call happens from the context of the error handler before SCSI requests are unblocked, it will hang if all tags are in use for SCSI requests before the error handler starts.

All the commands sent by ufshcd_probe_hba() take the same approach, so no difference there.

Quoted 9 lines

The same comment applies to the commands sent by ufshcd_probe_hba() I think. If you would like to address this issue, how about passing the flag BLK_MQ_REQ_RESERVED to the blk_get_request() call in ufshcd_exec_dev_cmd()? The following additional changes are required to make this work:
* Add a 'reserved_tags' member to struct scsi_host_template, e.g. close to tag_alloc_policy.
* Modify scsi_mq_setup_tags() such that it copies the reserved_tags value from the host template into tag_set->reserved_tags.
* Set 'reserved_tags' in the UFS driver SCSI host template.

I think these things only happen on the reset path, so all the slots should be free, even if all the tags are allocated. With some care, it might be possible simply to grab a free slot.

#6 lore raw

Re: [PATCH V4 1/3] scsi: ufs: Fix error handler clear ua deadlock

Quoted 11 lines · Adrian Hunter
Quoted 9 lines · Bart Van Assche

The same comment applies to the commands sent by ufshcd_probe_hba() I think. If you would like to address this issue, how about passing the flag BLK_MQ_REQ_RESERVED to the blk_get_request() call in ufshcd_exec_dev_cmd()? The following additional changes are required to make this work:
* Add a 'reserved_tags' member to struct scsi_host_template, e.g. close to tag_alloc_policy.
* Modify scsi_mq_setup_tags() such that it copies the reserved_tags value from the host template into tag_set->reserved_tags.
* Set 'reserved_tags' in the UFS driver SCSI host template.

Quoted 3 lines

I think these things only happen on the reset path, so all the slots should be free, even if all the tags are allocated. With some care, it might be possible simply to grab a free slot.

Hmm ... my understanding is that ufshcd_try_to_abort_task() may fail and hence that it is not guaranteed that there will be free slot. Anyway, I'm fine with the patch in its current shape and implementing the mechanism explained in my previous email at a later time.

Bart.

#7 lore raw

PATCH v4 2/3 scsi: ufs: Fix runtime PM dependencies getting broken

In reply to Adrian Hunter

UFS SCSI devices make use of device links to establish PM dependencies. However, SCSI PM will force devices' runtime PM state to be active during system resume. That can break runtime PM dependencies for UFS devices. Fix by adding a flag 'preserve_rpm' to let UFS SCSI devices opt-out of the unwanted behaviour.

Fixes b294ff3e34490f ("scsi: ufs: core: Enable power management for wlun")
---
 drivers/scsi/scsi_pm.c     | 16 +++++++++++-----
 drivers/scsi/ufs/ufshcd.c  |  1 +
 include/scsi/scsi_device.h |  1 +
 3 files changed, 13 insertions(+), 5 deletions(-)
drivers/scsi/scsi_pm.c +11 -5
index 3717eea37ecb..0557c1ad304d 100644
@@ -73,13 +73,22 @@ static int scsi_dev_type_resume(struct device *dev,
 		int (*cb)(struct device *, const struct dev_pm_ops *))
 {
 	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
+	struct scsi_device *sdev = NULL;
+	bool preserve_rpm = false;
 	int err = 0;
 
+	if (scsi_is_sdev_device(dev)) {
+		sdev = to_scsi_device(dev);
+		preserve_rpm = sdev->preserve_rpm;
+		if (preserve_rpm && pm_runtime_suspended(dev))
+			return 0;
+	}
+
 	err = cb(dev, pm);
 	scsi_device_resume(to_scsi_device(dev));
 	dev_dbg(dev, "scsi resume: %d\n", err);
 
-	if (err == 0) {
+	if (err == 0 && !preserve_rpm) {
 		pm_runtime_disable(dev);
 		err = pm_runtime_set_active(dev);
 		pm_runtime_enable(dev);
@@ -91,11 +100,8 @@ static int scsi_dev_type_resume(struct device *dev,
 		 *
 		 * The resume hook will correct runtime PM status of the disk.
 		 */
-		if (!err && scsi_is_sdev_device(dev)) {
-			struct scsi_device *sdev = to_scsi_device(dev);
-
+		if (!err && sdev)
 			blk_set_runtime_active(sdev->request_queue);
-		}
 	}
 
 	return err;
drivers/scsi/ufs/ufshcd.c +1 -0
index 2802083e8c3f..9c9df2ad2a34 100644
@@ -5053,6 +5053,7 @@ static int ufshcd_slave_configure(struct scsi_device *sdev)
 		pm_runtime_get_noresume(&sdev->sdev_gendev);
 	else if (ufshcd_is_rpm_autosuspend_allowed(hba))
 		sdev->rpm_autosuspend = 1;
+	sdev->preserve_rpm = 1;
 
 	ufshcd_crypto_setup_rq_keyslot_manager(hba, q);
 
include/scsi/scsi_device.h +1 -0
index 09a17f6e93a7..47eb30a6b7b2 100644
@@ -197,6 +197,7 @@ struct scsi_device {
 	unsigned no_read_disc_info:1;	/* Avoid READ_DISC_INFO cmds */
 	unsigned no_read_capacity_16:1; /* Avoid READ_CAPACITY_16 cmds */
 	unsigned try_rc_10_first:1;	/* Try READ_CAPACACITY_10 first */
+	unsigned preserve_rpm:1;	/* Preserve runtime PM */
 	unsigned security_supported:1;	/* Supports Security Protocols */
 	unsigned is_visible:1;	/* is the device visible in sysfs */
 	unsigned wce_default_on:1;	/* Cache is ON by default */
Signature · 1 line
2.17.1
#8 lore raw

PATCH v4 3/3 scsi: ufs: Let devices remain runtime suspended during system suspend

In reply to Adrian Hunter

If the UFS Device WLUN is runtime suspended and is in the same power mode, link state and b_rpm_dev_flush_capable (BKOP or WB buffer flush etc) state, then it can remain runtime suspended instead of being runtime resumed and then system suspended.

The following patches have cleared the way for that to happen:
scsi: ufs: Fix runtime PM dependencies getting broken
scsi: ufs: Fix error handler clear ua deadlock

So amend the logic accordingly.

Note, the ufs-hisi driver uses different RPM and SPM, but it is made explicit by a new parameter to suspend prepare.

---
 drivers/scsi/ufs/ufs-hisi.c |  8 +++++-
 drivers/scsi/ufs/ufshcd.c   | 53 ++++++++++++++++++++++++++++---------
 drivers/scsi/ufs/ufshcd.h   | 12 ++++++++-
 3 files changed, 58 insertions(+), 15 deletions(-)
drivers/scsi/ufs/ufs-hisi.c +7 -1
index 6b706de8354b..4a08fb35642c 100644
@@ -396,6 +396,12 @@ static int ufs_hisi_pwr_change_notify(struct ufs_hba *hba,
 	return ret;
 }
 
+static int ufs_hisi_suspend_prepare(struct device *dev)
+{
+	/* RPM and SPM are different. Refer ufs_hisi_suspend() */
+	return __ufshcd_suspend_prepare(dev, false);
+}
+
 static int ufs_hisi_suspend(struct ufs_hba *hba, enum ufs_pm_op pm_op)
 {
 	struct ufs_hisi_host *host = ufshcd_get_variant(hba);
@@ -574,7 +580,7 @@ static int ufs_hisi_remove(struct platform_device *pdev)
 static const struct dev_pm_ops ufs_hisi_pm_ops = {
 	SET_SYSTEM_SLEEP_PM_OPS(ufshcd_system_suspend, ufshcd_system_resume)
 	SET_RUNTIME_PM_OPS(ufshcd_runtime_suspend, ufshcd_runtime_resume, NULL)
-	.prepare	 = ufshcd_suspend_prepare,
+	.prepare	 = ufs_hisi_suspend_prepare,
 	.complete	 = ufshcd_resume_complete,
 };
 
drivers/scsi/ufs/ufshcd.c +40 -13

73 lines not on this page · raw message

drivers/scsi/ufs/ufshcd.h +11 -1
index 82c0f689cbdb..49e8dd70eb5e 100644
@@ -924,7 +924,6 @@ struct ufs_hba {
 #endif
 	u32 luns_avail;
 	bool complete_put;
-	bool rpmb_complete_put;
 };
 
 /* Returns true if clocks can be gated. Otherwise false */
@@ -1184,6 +1183,7 @@ int ufshcd_exec_raw_upiu_cmd(struct ufs_hba *hba,
 
 int ufshcd_wb_toggle(struct ufs_hba *hba, bool enable);
 int ufshcd_suspend_prepare(struct device *dev);
+int __ufshcd_suspend_prepare(struct device *dev, bool rpm_ok_for_spm);
 void ufshcd_resume_complete(struct device *dev);
 
 /* Wrapper functions for safely calling variant operations */
@@ -1392,6 +1392,16 @@ static inline int ufshcd_rpm_put_sync(struct ufs_hba *hba)
 	return pm_runtime_put_sync(&hba->sdev_ufs_device->sdev_gendev);
 }
 
+static inline void ufshcd_rpm_get_noresume(struct ufs_hba *hba)
+{
+	pm_runtime_get_noresume(&hba->sdev_ufs_device->sdev_gendev);
+}
+
+static inline int ufshcd_rpm_resume(struct ufs_hba *hba)
+{
+	return pm_runtime_resume(&hba->sdev_ufs_device->sdev_gendev);
+}
+
 static inline int ufshcd_rpm_put(struct ufs_hba *hba)
 {
 	return pm_runtime_put(&hba->sdev_ufs_device->sdev_gendev);
Signature · 1 line
2.17.1