Merge patch series "ufs: Do not requeue while ungating the clock"

Bart Van Assche <bvanassche@acm.org> says:

In the traces we recorded while testing zoned storage we noticed that UFS
commands are requeued while the clock is being ungated. Command requeueing
makes it harder than necessary to preserve the command order. Hence this
patch series that modifies the SCSI core and also the UFS driver such that
clock ungating does not trigger command requeueing.

Link: https://lore.kernel.org/r/20230529202640.11883-1-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

8 files changed, 54 insertions, 76 deletions

diff --git a/drivers/scsi/hosts.c b/drivers/scsi/hosts.c
index f0bc8bbb3938..198edf03f929 100644
--- a/drivers/scsi/hosts.c
+++ b/drivers/scsi/hosts.c
@@ -441,6 +441,7 @@ struct Scsi_Host *scsi_host_alloc(const struct scsi_host_template *sht, int priv
 	shost->cmd_per_lun = sht->cmd_per_lun;
 	shost->no_write_same = sht->no_write_same;
 	shost->host_tagset = sht->host_tagset;
+	shost->queuecommand_may_block = sht->queuecommand_may_block;
 
 	if (shost_eh_deadline == -1 || !sht->eh_host_reset_handler)
 		shost->eh_deadline = -1;
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index d1a0b15d4514..496bdfc19c95 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -1984,6 +1984,8 @@ int scsi_mq_setup_tags(struct Scsi_Host *shost)
 	tag_set->flags = BLK_MQ_F_SHOULD_MERGE;
 	tag_set->flags |=
 		BLK_ALLOC_POLICY_TO_MQ_FLAG(shost->hostt->tag_alloc_policy);
+	if (shost->queuecommand_may_block)
+		tag_set->flags |= BLK_MQ_F_BLOCKING;
 	tag_set->driver_data = shost;
 	if (shost->host_tagset)
 		tag_set->flags |= BLK_MQ_F_TAG_HCTX_SHARED;
@@ -2943,11 +2945,20 @@ scsi_target_unblock(struct device *dev, enum scsi_device_state new_state)
 }
 EXPORT_SYMBOL_GPL(scsi_target_unblock);
 
+/**
+ * scsi_host_block - Try to transition all logical units to the SDEV_BLOCK state
+ * @shost: device to block
+ *
+ * Pause SCSI command processing for all logical units associated with the SCSI
+ * host and wait until pending scsi_queue_rq() calls have finished.
+ *
+ * Returns zero if successful or a negative error code upon failure.
+ */
 int
 scsi_host_block(struct Scsi_Host *shost)
 {
 	struct scsi_device *sdev;
-	int ret = 0;
+	int ret;
 
 	/*
 	 * Call scsi_internal_device_block_nowait so we can avoid
@@ -2959,20 +2970,14 @@ scsi_host_block(struct Scsi_Host *shost)
 		mutex_unlock(&sdev->state_mutex);
 		if (ret) {
 			scsi_device_put(sdev);
-			break;
+			return ret;
 		}
 	}
 
-	/*
-	 * SCSI never enables blk-mq's BLK_MQ_F_BLOCKING flag so
-	 * calling synchronize_rcu() once is enough.
-	 */
-	WARN_ON_ONCE(shost->tag_set.flags & BLK_MQ_F_BLOCKING);
+	/* Wait for ongoing scsi_queue_rq() calls to finish. */
+	blk_mq_wait_quiesce_done(&shost->tag_set);
 
-	if (!ret)
-		synchronize_rcu();
-
-	return ret;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(scsi_host_block);
 
diff --git a/drivers/ufs/core/ufs-sysfs.c b/drivers/ufs/core/ufs-sysfs.c
index 883f0e44b54e..cdf3d5f2b77b 100644
--- a/drivers/ufs/core/ufs-sysfs.c
+++ b/drivers/ufs/core/ufs-sysfs.c
@@ -168,7 +168,7 @@ static ssize_t auto_hibern8_show(struct device *dev,
 	}
 
 	pm_runtime_get_sync(hba->dev);
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	ahit = ufshcd_readl(hba, REG_AUTO_HIBERNATE_IDLE_TIMER);
 	ufshcd_release(hba);
 	pm_runtime_put_sync(hba->dev);
diff --git a/drivers/ufs/core/ufshcd-crypto.c b/drivers/ufs/core/ufshcd-crypto.c
index 198360fe5e8e..f2c4422cab86 100644
--- a/drivers/ufs/core/ufshcd-crypto.c
+++ b/drivers/ufs/core/ufshcd-crypto.c
@@ -24,7 +24,7 @@ static int ufshcd_program_key(struct ufs_hba *hba,
 	u32 slot_offset = hba->crypto_cfg_register + slot * sizeof(*cfg);
 	int err = 0;
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 	if (hba->vops && hba->vops->program_key) {
 		err = hba->vops->program_key(hba, cfg, slot);
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h
index d53b93c21a0c..8f58c2169398 100644
--- a/drivers/ufs/core/ufshcd-priv.h
+++ b/drivers/ufs/core/ufshcd-priv.h
@@ -84,9 +84,6 @@ unsigned long ufshcd_mcq_poll_cqe_lock(struct ufs_hba *hba,
 int ufshcd_read_string_desc(struct ufs_hba *hba, u8 desc_index,
 			    u8 **buf, bool ascii);
 
-int ufshcd_hold(struct ufs_hba *hba, bool async);
-void ufshcd_release(struct ufs_hba *hba);
-
 int ufshcd_send_uic_cmd(struct ufs_hba *hba, struct uic_command *uic_cmd);
 
 int ufshcd_exec_raw_upiu_cmd(struct ufs_hba *hba,
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index fdf5073c7c6c..941e613548da 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -1189,7 +1189,7 @@ static int ufshcd_wait_for_doorbell_clr(struct ufs_hba *hba,
 	bool timeout = false, do_last_check = false;
 	ktime_t start;
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	spin_lock_irqsave(hba->host->host_lock, flags);
 	/*
 	 * Wait for all the outstanding tasks/transfer requests.
@@ -1310,7 +1310,7 @@ static int ufshcd_clock_scaling_prepare(struct ufs_hba *hba, u64 timeout_us)
 	}
 
 	/* let's not get into low power until clock scaling is completed */
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 out:
 	return ret;
@@ -1640,7 +1640,7 @@ static ssize_t ufshcd_clkscale_enable_store(struct device *dev,
 		goto out;
 
 	ufshcd_rpm_get_sync(hba);
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 	hba->clk_scaling.is_enabled = value;
 
@@ -1723,7 +1723,7 @@ static void ufshcd_ungate_work(struct work_struct *work)
 	spin_lock_irqsave(hba->host->host_lock, flags);
 	if (hba->clk_gating.state == CLKS_ON) {
 		spin_unlock_irqrestore(hba->host->host_lock, flags);
-		goto unblock_reqs;
+		return;
 	}
 
 	spin_unlock_irqrestore(hba->host->host_lock, flags);
@@ -1746,25 +1746,21 @@ static void ufshcd_ungate_work(struct work_struct *work)
 		}
 		hba->clk_gating.is_suspended = false;
 	}
-unblock_reqs:
-	ufshcd_scsi_unblock_requests(hba);
 }
 
 /**
  * ufshcd_hold - Enable clocks that were gated earlier due to ufshcd_release.
  * Also, exit from hibern8 mode and set the link as active.
  * @hba: per adapter instance
- * @async: This indicates whether caller should ungate clocks asynchronously.
  */
-int ufshcd_hold(struct ufs_hba *hba, bool async)
+void ufshcd_hold(struct ufs_hba *hba)
 {
-	int rc = 0;
 	bool flush_result;
 	unsigned long flags;
 
 	if (!ufshcd_is_clkgating_allowed(hba) ||
 	    !hba->clk_gating.is_initialized)
-		goto out;
+		return;
 	spin_lock_irqsave(hba->host->host_lock, flags);
 	hba->clk_gating.active_reqs++;
 
@@ -1781,15 +1777,10 @@ start:
 		 */
 		if (ufshcd_can_hibern8_during_gating(hba) &&
 		    ufshcd_is_link_hibern8(hba)) {
-			if (async) {
-				rc = -EAGAIN;
-				hba->clk_gating.active_reqs--;
-				break;
-			}
 			spin_unlock_irqrestore(hba->host->host_lock, flags);
 			flush_result = flush_work(&hba->clk_gating.ungate_work);
 			if (hba->clk_gating.is_suspended && !flush_result)
-				goto out;
+				return;
 			spin_lock_irqsave(hba->host->host_lock, flags);
 			goto start;
 		}
@@ -1811,21 +1802,14 @@ start:
 		hba->clk_gating.state = REQ_CLKS_ON;
 		trace_ufshcd_clk_gating(dev_name(hba->dev),
 					hba->clk_gating.state);
-		if (queue_work(hba->clk_gating.clk_gating_workq,
-			       &hba->clk_gating.ungate_work))
-			ufshcd_scsi_block_requests(hba);
+		queue_work(hba->clk_gating.clk_gating_workq,
+			   &hba->clk_gating.ungate_work);
 		/*
 		 * fall through to check if we should wait for this
 		 * work to be done or not.
 		 */
 		fallthrough;
 	case REQ_CLKS_ON:
-		if (async) {
-			rc = -EAGAIN;
-			hba->clk_gating.active_reqs--;
-			break;
-		}
-
 		spin_unlock_irqrestore(hba->host->host_lock, flags);
 		flush_work(&hba->clk_gating.ungate_work);
 		/* Make sure state is CLKS_ON before returning */
@@ -1837,8 +1821,6 @@ start:
 		break;
 	}
 	spin_unlock_irqrestore(hba->host->host_lock, flags);
-out:
-	return rc;
 }
 EXPORT_SYMBOL_GPL(ufshcd_hold);
 
@@ -2070,7 +2052,7 @@ static void ufshcd_exit_clk_gating(struct ufs_hba *hba)
 	ufshcd_remove_clk_gating_sysfs(hba);
 
 	/* Ungate the clock if necessary. */
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	hba->clk_gating.is_initialized = false;
 	ufshcd_release(hba);
 
@@ -2468,7 +2450,7 @@ int ufshcd_send_uic_cmd(struct ufs_hba *hba, struct uic_command *uic_cmd)
 	if (hba->quirks & UFSHCD_QUIRK_BROKEN_UIC_CMD)
 		return 0;
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	mutex_lock(&hba->uic_cmd_mutex);
 	ufshcd_add_delay_before_dme_cmd(hba);
 
@@ -2871,12 +2853,6 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 
 	WARN_ONCE(tag < 0 || tag >= hba->nutrs, "Invalid tag %d\n", tag);
 
-	/*
-	 * Allows the UFS error handler to wait for prior ufshcd_queuecommand()
-	 * calls.
-	 */
-	rcu_read_lock();
-
 	switch (hba->ufshcd_state) {
 	case UFSHCD_STATE_OPERATIONAL:
 		break;
@@ -2922,13 +2898,7 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 
 	hba->req_abort_count = 0;
 
-	err = ufshcd_hold(hba, true);
-	if (err) {
-		err = SCSI_MLQUEUE_HOST_BUSY;
-		goto out;
-	}
-	WARN_ON(ufshcd_is_clkgating_allowed(hba) &&
-		(hba->clk_gating.state != CLKS_ON));
+	ufshcd_hold(hba);
 
 	lrbp = &hba->lrb[tag];
 	WARN_ON(lrbp->cmd);
@@ -2958,8 +2928,6 @@ static int ufshcd_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *cmd)
 	ufshcd_send_command(hba, tag, hwq);
 
 out:
-	rcu_read_unlock();
-
 	if (ufs_trigger_eh()) {
 		unsigned long flags;
 
@@ -3253,7 +3221,7 @@ int ufshcd_query_flag(struct ufs_hba *hba, enum query_opcode opcode,
 
 	BUG_ON(!hba);
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	mutex_lock(&hba->dev_cmd.lock);
 	ufshcd_init_query(hba, &request, &response, opcode, idn, index,
 			selector);
@@ -3327,7 +3295,7 @@ int ufshcd_query_attr(struct ufs_hba *hba, enum query_opcode opcode,
 		return -EINVAL;
 	}
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 	mutex_lock(&hba->dev_cmd.lock);
 	ufshcd_init_query(hba, &request, &response, opcode, idn, index,
@@ -3423,7 +3391,7 @@ static int __ufshcd_query_descriptor(struct ufs_hba *hba,
 		return -EINVAL;
 	}
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 	mutex_lock(&hba->dev_cmd.lock);
 	ufshcd_init_query(hba, &request, &response, opcode, idn, index,
@@ -4241,7 +4209,7 @@ int ufshcd_uic_change_pwr_mode(struct ufs_hba *hba, u8 mode)
 	uic_cmd.command = UIC_CMD_DME_SET;
 	uic_cmd.argument1 = UIC_ARG_MIB(PA_PWRMODE);
 	uic_cmd.argument3 = mode;
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	ret = ufshcd_uic_pwr_ctrl(hba, &uic_cmd);
 	ufshcd_release(hba);
 
@@ -4348,7 +4316,7 @@ void ufshcd_auto_hibern8_update(struct ufs_hba *hba, u32 ahit)
 	if (update &&
 	    !pm_runtime_suspended(&hba->ufs_device_wlun->sdev_gendev)) {
 		ufshcd_rpm_get_sync(hba);
-		ufshcd_hold(hba, false);
+		ufshcd_hold(hba);
 		ufshcd_auto_hibern8_enable(hba);
 		ufshcd_release(hba);
 		ufshcd_rpm_put_sync(hba);
@@ -4941,7 +4909,7 @@ static int ufshcd_verify_dev_init(struct ufs_hba *hba)
 	int err = 0;
 	int retries;
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	mutex_lock(&hba->dev_cmd.lock);
 	for (retries = NOP_OUT_RETRIES; retries > 0; retries--) {
 		err = ufshcd_exec_dev_cmd(hba, DEV_CMD_TYPE_NOP,
@@ -6227,22 +6195,22 @@ static void ufshcd_err_handling_prepare(struct ufs_hba *hba)
 		ufshcd_setup_vreg(hba, true);
 		ufshcd_config_vreg_hpm(hba, hba->vreg_info.vccq);
 		ufshcd_config_vreg_hpm(hba, hba->vreg_info.vccq2);
-		ufshcd_hold(hba, false);
+		ufshcd_hold(hba);
 		if (!ufshcd_is_clkgating_allowed(hba))
 			ufshcd_setup_clocks(hba, true);
 		ufshcd_release(hba);
 		pm_op = hba->is_sys_suspended ? UFS_SYSTEM_PM : UFS_RUNTIME_PM;
 		ufshcd_vops_resume(hba, pm_op);
 	} else {
-		ufshcd_hold(hba, false);
+		ufshcd_hold(hba);
 		if (ufshcd_is_clkscaling_supported(hba) &&
 		    hba->clk_scaling.is_enabled)
 			ufshcd_suspend_clkscaling(hba);
 		ufshcd_clk_scaling_allow(hba, false);
 	}
 	ufshcd_scsi_block_requests(hba);
-	/* Drain ufshcd_queuecommand() */
-	synchronize_rcu();
+	/* Wait for ongoing ufshcd_queuecommand() calls to finish. */
+	blk_mq_wait_quiesce_done(&hba->host->tag_set);
 	cancel_work_sync(&hba->eeh_work);
 }
 
@@ -6887,7 +6855,7 @@ static int __ufshcd_issue_tm_cmd(struct ufs_hba *hba,
 		return PTR_ERR(req);
 
 	req->end_io_data = &wait;
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 
 	spin_lock_irqsave(host->host_lock, flags);
 
@@ -7124,7 +7092,7 @@ int ufshcd_exec_raw_upiu_cmd(struct ufs_hba *hba,
 		cmd_type = DEV_CMD_TYPE_NOP;
 		fallthrough;
 	case UPIU_TRANSACTION_QUERY_REQ:
-		ufshcd_hold(hba, false);
+		ufshcd_hold(hba);
 		mutex_lock(&hba->dev_cmd.lock);
 		err = ufshcd_issue_devman_upiu_cmd(hba, req_upiu, rsp_upiu,
 						   desc_buff, buff_len,
@@ -7190,7 +7158,7 @@ int ufshcd_advanced_rpmb_req_handler(struct ufs_hba *hba, struct utp_upiu_req *r
 	u16 ehs_len;
 
 	/* Protects use of hba->reserved_slot. */
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	mutex_lock(&hba->dev_cmd.lock);
 	down_read(&hba->clk_scaling_lock);
 
@@ -7425,7 +7393,7 @@ static int ufshcd_abort(struct scsi_cmnd *cmd)
 
 	WARN_ONCE(tag < 0, "Invalid tag %d\n", tag);
 
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	reg = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
 	/* If command is already aborted/completed, return FAILED. */
 	if (!(test_bit(tag, &hba->outstanding_reqs))) {
@@ -9416,7 +9384,7 @@ static int __ufshcd_wl_suspend(struct ufs_hba *hba, enum ufs_pm_op pm_op)
 	 * If we can't transition into any of the low power modes
 	 * just gate the clocks.
 	 */
-	ufshcd_hold(hba, false);
+	ufshcd_hold(hba);
 	hba->clk_gating.is_suspended = true;
 
 	if (ufshcd_is_clkscaling_supported(hba))
@@ -10204,6 +10172,7 @@ int ufshcd_init(struct ufs_hba *hba, void __iomem *mmio_base, unsigned int irq)
 	host->max_channel = UFSHCD_MAX_CHANNEL;
 	host->unique_id = host->host_no;
 	host->max_cmd_len = UFS_CDB_SIZE;
+	host->queuecommand_may_block = !!(hba->caps & UFSHCD_CAP_CLK_GATING);
 
 	hba->max_pwr_info.is_valid = false;
 
diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h
index 0f29799efa02..70b7475dcf56 100644
--- a/include/scsi/scsi_host.h
+++ b/include/scsi/scsi_host.h
@@ -458,6 +458,9 @@ struct scsi_host_template {
 	/* True if the host uses host-wide tagspace */
 	unsigned host_tagset:1;
 
+	/* The queuecommand callback may block. See also BLK_MQ_F_BLOCKING. */
+	unsigned queuecommand_may_block:1;
+
 	/*
 	 * Countdown for host blocking with no commands outstanding.
 	 */
@@ -653,6 +656,9 @@ struct Scsi_Host {
 	/* True if the host uses host-wide tagspace */
 	unsigned host_tagset:1;
 
+	/* The queuecommand callback may block. See also BLK_MQ_F_BLOCKING. */
+	unsigned queuecommand_may_block:1;
+
 	/* Host responded with short (<36 bytes) INQUIRY result */
 	unsigned short_inquiry:1;
 
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index f7553293ba98..8039c2b72502 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -1358,7 +1358,7 @@ void ufshcd_fixup_dev_quirks(struct ufs_hba *hba,
 int ufshcd_read_string_desc(struct ufs_hba *hba, u8 desc_index,
 			    u8 **buf, bool ascii);
 
-int ufshcd_hold(struct ufs_hba *hba, bool async);
+void ufshcd_hold(struct ufs_hba *hba);
 void ufshcd_release(struct ufs_hba *hba);
 
 void ufshcd_clkgate_delay_set(struct device *dev, unsigned long value);