* [PATCH v1 1/2] scsi: ufs: Abort tasks before clear them from doorbell
[not found] <1598261952-29209-1-git-send-email-cang@codeaurora.org>
@ 2020-08-24 9:39 ` Can Guo
2020-08-24 9:39 ` [PATCH v1 2/2] scsi: ufs: Handle LINERESET indication in err handler Can Guo
1 sibling, 0 replies; 3+ messages in thread
From: Can Guo @ 2020-08-24 9:39 UTC (permalink / raw)
To: asutoshd, nguyenb, hongwus, rnayak, linux-scsi, kernel-team,
saravanak, salyzyn, cang
Cc: Alim Akhtar, Avri Altman, James E.J. Bottomley,
Martin K. Petersen, Stanley Chu, Bean Huo, Bart Van Assche,
open list
To recovery non-fatal errors, no full reset is required, err_handler only
clears those pending TRs/TMRs so that scsi layer can re-issue them. In
current err_handler, TRs are directly cleared from UFS host's doorbell but
not aborted from device side. However, according to the UFSHCI JEDEC spec,
the host software shall use UTP Transfer Request List CLear Register to
clear a task from UFS host's doorbell only when a UTP Transfer Request is
expected to not be completed, e.g. when the host software receives a
“FUNCTION COMPLETE” Task Management response which means a Transfer Request
was aborted. To follow the UFSHCI JEDEC spec, in err_handler, aborts one TR
before clearing it from doorbell.
Signed-off-by: Can Guo <cang@codeaurora.org>
---
drivers/scsi/ufs/ufshcd.c | 143 ++++++++++++++++++++++++++--------------------
1 file changed, 81 insertions(+), 62 deletions(-)
diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index b8441ad..000895f 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -235,6 +235,7 @@ static int ufshcd_setup_hba_vreg(struct ufs_hba *hba, bool on);
static int ufshcd_setup_vreg(struct ufs_hba *hba, bool on);
static inline int ufshcd_config_vreg_hpm(struct ufs_hba *hba,
struct ufs_vreg *vreg);
+static int ufshcd_try_to_abort_task(struct ufs_hba *hba, int tag);
static int ufshcd_wb_buf_flush_enable(struct ufs_hba *hba);
static int ufshcd_wb_buf_flush_disable(struct ufs_hba *hba);
static int ufshcd_wb_ctrl(struct ufs_hba *hba, bool enable);
@@ -5657,8 +5658,8 @@ static void ufshcd_err_handler(struct work_struct *work)
{
struct ufs_hba *hba;
unsigned long flags;
- u32 err_xfer = 0;
- u32 err_tm = 0;
+ bool err_xfer = false;
+ bool err_tm = false;
int err = 0;
int tag;
bool needs_reset = false;
@@ -5734,7 +5735,7 @@ static void ufshcd_err_handler(struct work_struct *work)
spin_unlock_irqrestore(hba->host->host_lock, flags);
/* Clear pending transfer requests */
for_each_set_bit(tag, &hba->outstanding_reqs, hba->nutrs) {
- if (ufshcd_clear_cmd(hba, tag)) {
+ if (ufshcd_try_to_abort_task(hba, tag)) {
err_xfer = true;
goto lock_skip_pending_xfer_clear;
}
@@ -6486,7 +6487,7 @@ static void ufshcd_set_req_abort_skip(struct ufs_hba *hba, unsigned long bitmap)
}
/**
- * ufshcd_abort - abort a specific command
+ * ufshcd_try_to_abort_task - abort a specific task
* @cmd: SCSI command pointer
*
* Abort the pending command in device by sending UFS_ABORT_TASK task management
@@ -6495,6 +6496,80 @@ static void ufshcd_set_req_abort_skip(struct ufs_hba *hba, unsigned long bitmap)
* issued. To avoid that, first issue UFS_QUERY_TASK to check if the command is
* really issued and then try to abort it.
*
+ * Returns zero on success, non-zero on failure
+ */
+static int ufshcd_try_to_abort_task(struct ufs_hba *hba, int tag)
+{
+ struct ufshcd_lrb *lrbp = &hba->lrb[tag];
+ int err = 0;
+ int poll_cnt;
+ u8 resp = 0xF;
+ u32 reg;
+
+ for (poll_cnt = 100; poll_cnt; poll_cnt--) {
+ err = ufshcd_issue_tm_cmd(hba, lrbp->lun, lrbp->task_tag,
+ UFS_QUERY_TASK, &resp);
+ if (!err && resp == UPIU_TASK_MANAGEMENT_FUNC_SUCCEEDED) {
+ /* cmd pending in the device */
+ dev_err(hba->dev, "%s: cmd pending in the device. tag = %d\n",
+ __func__, tag);
+ break;
+ } else if (!err && resp == UPIU_TASK_MANAGEMENT_FUNC_COMPL) {
+ /*
+ * cmd not pending in the device, check if it is
+ * in transition.
+ */
+ dev_err(hba->dev, "%s: cmd at tag %d not pending in the device.\n",
+ __func__, tag);
+ reg = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
+ if (reg & (1 << tag)) {
+ /* sleep for max. 200us to stabilize */
+ usleep_range(100, 200);
+ continue;
+ }
+ /* command completed already */
+ dev_err(hba->dev, "%s: cmd at tag %d successfully cleared from DB.\n",
+ __func__, tag);
+ goto out;
+ } else {
+ dev_err(hba->dev,
+ "%s: no response from device. tag = %d, err %d\n",
+ __func__, tag, err);
+ if (!err)
+ err = resp; /* service response error */
+ goto out;
+ }
+ }
+
+ if (!poll_cnt) {
+ err = -EBUSY;
+ goto out;
+ }
+
+ err = ufshcd_issue_tm_cmd(hba, lrbp->lun, lrbp->task_tag,
+ UFS_ABORT_TASK, &resp);
+ if (err || resp != UPIU_TASK_MANAGEMENT_FUNC_COMPL) {
+ if (!err) {
+ err = resp; /* service response error */
+ dev_err(hba->dev, "%s: issued. tag = %d, err %d\n",
+ __func__, tag, err);
+ }
+ goto out;
+ }
+
+ err = ufshcd_clear_cmd(hba, tag);
+ if (err)
+ dev_err(hba->dev, "%s: Failed clearing cmd at tag %d, err %d\n",
+ __func__, tag, err);
+
+out:
+ return err;
+}
+
+/**
+ * ufshcd_abort - scsi host template eh_abort_handler callback
+ * @cmd: SCSI command pointer
+ *
* Returns SUCCESS/FAILED
*/
static int ufshcd_abort(struct scsi_cmnd *cmd)
@@ -6504,8 +6579,6 @@ static int ufshcd_abort(struct scsi_cmnd *cmd)
unsigned long flags;
unsigned int tag;
int err = 0;
- int poll_cnt;
- u8 resp = 0xF;
struct ufshcd_lrb *lrbp;
u32 reg;
@@ -6574,63 +6647,9 @@ static int ufshcd_abort(struct scsi_cmnd *cmd)
goto out;
}
- for (poll_cnt = 100; poll_cnt; poll_cnt--) {
- err = ufshcd_issue_tm_cmd(hba, lrbp->lun, lrbp->task_tag,
- UFS_QUERY_TASK, &resp);
- if (!err && resp == UPIU_TASK_MANAGEMENT_FUNC_SUCCEEDED) {
- /* cmd pending in the device */
- dev_err(hba->dev, "%s: cmd pending in the device. tag = %d\n",
- __func__, tag);
- break;
- } else if (!err && resp == UPIU_TASK_MANAGEMENT_FUNC_COMPL) {
- /*
- * cmd not pending in the device, check if it is
- * in transition.
- */
- dev_err(hba->dev, "%s: cmd at tag %d not pending in the device.\n",
- __func__, tag);
- reg = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
- if (reg & (1 << tag)) {
- /* sleep for max. 200us to stabilize */
- usleep_range(100, 200);
- continue;
- }
- /* command completed already */
- dev_err(hba->dev, "%s: cmd at tag %d successfully cleared from DB.\n",
- __func__, tag);
- goto out;
- } else {
- dev_err(hba->dev,
- "%s: no response from device. tag = %d, err %d\n",
- __func__, tag, err);
- if (!err)
- err = resp; /* service response error */
- goto out;
- }
- }
-
- if (!poll_cnt) {
- err = -EBUSY;
- goto out;
- }
-
- err = ufshcd_issue_tm_cmd(hba, lrbp->lun, lrbp->task_tag,
- UFS_ABORT_TASK, &resp);
- if (err || resp != UPIU_TASK_MANAGEMENT_FUNC_COMPL) {
- if (!err) {
- err = resp; /* service response error */
- dev_err(hba->dev, "%s: issued. tag = %d, err %d\n",
- __func__, tag, err);
- }
- goto out;
- }
-
- err = ufshcd_clear_cmd(hba, tag);
- if (err) {
- dev_err(hba->dev, "%s: Failed clearing cmd at tag %d, err %d\n",
- __func__, tag, err);
+ err = ufshcd_try_to_abort_task(hba, tag);
+ if (err)
goto out;
- }
spin_lock_irqsave(host->host_lock, flags);
__ufshcd_transfer_req_compl(hba, (1UL << tag));
--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project.
^ permalink raw reply related [flat|nested] 3+ messages in thread* [PATCH v1 2/2] scsi: ufs: Handle LINERESET indication in err handler
[not found] <1598261952-29209-1-git-send-email-cang@codeaurora.org>
2020-08-24 9:39 ` [PATCH v1 1/2] scsi: ufs: Abort tasks before clear them from doorbell Can Guo
@ 2020-08-24 9:39 ` Can Guo
1 sibling, 0 replies; 3+ messages in thread
From: Can Guo @ 2020-08-24 9:39 UTC (permalink / raw)
To: asutoshd, nguyenb, hongwus, rnayak, linux-scsi, kernel-team,
saravanak, salyzyn, cang
Cc: Alim Akhtar, Avri Altman, James E.J. Bottomley,
Martin K. Petersen, Stanley Chu, Bean Huo, Bart Van Assche,
Satya Tangirala, open list
PA Layer issues a LINERESET to the PHY at the recovery step in the Power
Mode change operation. If it happens during auto or mannual hibern8 enter,
even if hibern8 enter succeeds, UFS power mode shall be set to PWM-G1 mode
and kept in that mode after exit from hibern8, leading to bad performance.
Handle the LINERESET in the eh_work by restoring power mode to HS mode
after all pending reqs and tasks are cleared from doorbell.
Signed-off-by: Can Guo <cang@codeaurora.org>
---
drivers/scsi/ufs/ufshcd.c | 111 ++++++++++++++++++++++++++++++++++++++++------
drivers/scsi/ufs/ufshcd.h | 2 +
drivers/scsi/ufs/unipro.h | 3 ++
3 files changed, 102 insertions(+), 14 deletions(-)
diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 000895f..8cc127d 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -143,6 +143,7 @@ enum {
UFSHCD_UIC_NL_ERROR = (1 << 3), /* Network layer error */
UFSHCD_UIC_TL_ERROR = (1 << 4), /* Transport Layer error */
UFSHCD_UIC_DME_ERROR = (1 << 5), /* DME error */
+ UFSHCD_UIC_PA_GENERIC_ERROR = (1 << 6), /* Generic PA error */
};
#define ufshcd_set_eh_in_progress(h) \
@@ -4066,7 +4067,8 @@ static int ufshcd_change_power_mode(struct ufs_hba *hba,
int ret;
/* if already configured to the requested pwr_mode */
- if (pwr_mode->gear_rx == hba->pwr_info.gear_rx &&
+ if (!hba->force_pmc &&
+ pwr_mode->gear_rx == hba->pwr_info.gear_rx &&
pwr_mode->gear_tx == hba->pwr_info.gear_tx &&
pwr_mode->lane_rx == hba->pwr_info.lane_rx &&
pwr_mode->lane_tx == hba->pwr_info.lane_tx &&
@@ -4494,6 +4496,8 @@ static int ufshcd_link_startup(struct ufs_hba *hba)
if (ret)
goto out;
+ /* Clear UECPA once due to LINERESET has happened during LINK_STARTUP */
+ ufshcd_readl(hba, REG_UIC_ERROR_CODE_PHY_ADAPTER_LAYER);
ret = ufshcd_make_hba_operational(hba);
out:
if (ret) {
@@ -5650,6 +5654,22 @@ static inline void ufshcd_recover_pm_error(struct ufs_hba *hba)
}
#endif
+static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
+{
+ struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
+ u32 mode;
+
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
+
+ if (pwr_info->pwr_rx != ((mode >> PWRMODE_RX_OFFSET) & PWRMODE_MASK))
+ return true;
+
+ if (pwr_info->pwr_tx != (mode & PWRMODE_MASK))
+ return true;
+
+ return false;
+}
+
/**
* ufshcd_err_handler - handle UFS errors that require s/w attention
* @work: pointer to work structure
@@ -5660,9 +5680,9 @@ static void ufshcd_err_handler(struct work_struct *work)
unsigned long flags;
bool err_xfer = false;
bool err_tm = false;
- int err = 0;
+ int err = 0, pmc_err;
int tag;
- bool needs_reset = false;
+ bool needs_reset = false, needs_restore = false;
hba = container_of(work, struct ufs_hba, eh_work);
@@ -5710,8 +5730,9 @@ static void ufshcd_err_handler(struct work_struct *work)
UFSHCD_UIC_DL_TCx_REPLAY_ERROR))))
needs_reset = true;
- if (hba->saved_err & (INT_FATAL_ERRORS | UIC_ERROR |
- UFSHCD_UIC_HIBERN8_MASK)) {
+ if ((hba->saved_err & (INT_FATAL_ERRORS | UFSHCD_UIC_HIBERN8_MASK)) ||
+ (hba->saved_uic_err &&
+ (hba->saved_uic_err != UFSHCD_UIC_PA_GENERIC_ERROR))) {
bool pr_prdt = !!(hba->saved_err & SYSTEM_BUS_FATAL_ERROR);
spin_unlock_irqrestore(hba->host->host_lock, flags);
@@ -5729,8 +5750,25 @@ static void ufshcd_err_handler(struct work_struct *work)
* host reset and restore
*/
if (needs_reset)
- goto skip_pending_xfer_clear;
+ goto do_reset;
+ /*
+ * If LINERESET was caught, UFS might have been put to PWM mode,
+ * check if power mode restore is needed.
+ */
+ if (hba->saved_uic_err & UFSHCD_UIC_PA_GENERIC_ERROR) {
+ hba->saved_uic_err &= ~UFSHCD_UIC_PA_GENERIC_ERROR;
+ if (!hba->saved_uic_err)
+ hba->saved_err &= ~UIC_ERROR;
+ spin_unlock_irqrestore(hba->host->host_lock, flags);
+ if (ufshcd_is_pwr_mode_restore_needed(hba))
+ needs_restore = true;
+ spin_lock_irqsave(hba->host->host_lock, flags);
+ if (!hba->saved_err && !needs_restore)
+ goto skip_err_handling;
+ }
+
+ hba->silence_err_logs = true;
/* release lock as clear command might sleep */
spin_unlock_irqrestore(hba->host->host_lock, flags);
/* Clear pending transfer requests */
@@ -5754,11 +5792,38 @@ static void ufshcd_err_handler(struct work_struct *work)
/* Complete the requests that are cleared by s/w */
ufshcd_complete_requests(hba);
+ hba->silence_err_logs = false;
- if (err_xfer || err_tm)
+ if (err_xfer || err_tm) {
needs_reset = true;
+ goto do_reset;
+ }
-skip_pending_xfer_clear:
+ /*
+ * After all reqs and tasks are cleared from doorbell,
+ * now it is safe to retore power mode.
+ */
+ if (needs_restore) {
+ spin_unlock_irqrestore(hba->host->host_lock, flags);
+ /*
+ * Hold the scaling lock just in case dev cmds
+ * are sent via bsg and/or sysfs.
+ */
+ down_write(&hba->clk_scaling_lock);
+ hba->force_pmc = true;
+ pmc_err = ufshcd_config_pwr_mode(hba, &(hba->pwr_info));
+ if (pmc_err) {
+ needs_reset = true;
+ dev_err(hba->dev, "%s: Failed to restore power mode, err = %d\n",
+ __func__, pmc_err);
+ }
+ hba->force_pmc = false;
+ ufshcd_print_pwr_info(hba);
+ up_write(&hba->clk_scaling_lock);
+ spin_lock_irqsave(hba->host->host_lock, flags);
+ }
+
+do_reset:
/* Fatal errors need reset */
if (needs_reset) {
unsigned long max_doorbells = (1UL << hba->nutrs) - 1;
@@ -5814,17 +5879,33 @@ static irqreturn_t ufshcd_update_uic_error(struct ufs_hba *hba)
u32 reg;
irqreturn_t retval = IRQ_NONE;
- /* PHY layer lane error */
+ /* PHY layer error */
reg = ufshcd_readl(hba, REG_UIC_ERROR_CODE_PHY_ADAPTER_LAYER);
- /* Ignore LINERESET indication, as this is not an error */
if ((reg & UIC_PHY_ADAPTER_LAYER_ERROR) &&
- (reg & UIC_PHY_ADAPTER_LAYER_LANE_ERR_MASK)) {
+ (reg & UIC_PHY_ADAPTER_LAYER_ERROR_CODE_MASK)) {
+ ufshcd_update_reg_hist(&hba->ufs_stats.pa_err, reg);
/*
* To know whether this error is fatal or not, DB timeout
* must be checked but this error is handled separately.
*/
- dev_dbg(hba->dev, "%s: UIC Lane error reported\n", __func__);
- ufshcd_update_reg_hist(&hba->ufs_stats.pa_err, reg);
+ if (reg & UIC_PHY_ADAPTER_LAYER_LANE_ERR_MASK)
+ dev_dbg(hba->dev, "%s: UIC Lane error reported\n",
+ __func__);
+
+ /* Got a LINERESET indication. */
+ if (reg & UIC_PHY_ADAPTER_LAYER_GENERIC_ERROR) {
+ struct uic_command *cmd = NULL;
+
+ hba->uic_error |= UFSHCD_UIC_PA_GENERIC_ERROR;
+ if (hba->uic_async_done && hba->active_uic_cmd)
+ cmd = hba->active_uic_cmd;
+ /*
+ * Ignore the LINERESET during power mode change
+ * operation via DME_SET command.
+ */
+ if (cmd && (cmd->command == UIC_CMD_DME_SET))
+ hba->uic_error &= ~UFSHCD_UIC_PA_GENERIC_ERROR;
+ }
retval |= IRQ_HANDLED;
}
@@ -5941,7 +6022,9 @@ static irqreturn_t ufshcd_check_errors(struct ufs_hba *hba)
hba->saved_uic_err |= hba->uic_error;
/* dump controller state before resetting */
- if (hba->saved_err & (INT_FATAL_ERRORS | UIC_ERROR)) {
+ if ((hba->saved_err & (INT_FATAL_ERRORS)) ||
+ (hba->saved_uic_err &&
+ (hba->saved_uic_err != UFSHCD_UIC_PA_GENERIC_ERROR))) {
dev_err(hba->dev, "%s: saved_err 0x%x saved_uic_err 0x%x\n",
__func__, hba->saved_err,
hba->saved_uic_err);
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index 618b253..8817103 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -629,6 +629,7 @@ struct ufs_hba_variant_params {
* @saved_err: sticky error mask
* @saved_uic_err: sticky UIC error mask
* @force_reset: flag to force eh_work perform a full reset
+ * @force_pmc: flag to force a power mode change
* @silence_err_logs: flag to silence error logs
* @dev_cmd: ufs device management command information
* @last_dme_cmd_tstamp: time stamp of the last completed DME command
@@ -728,6 +729,7 @@ struct ufs_hba {
u32 saved_uic_err;
struct ufs_stats ufs_stats;
bool force_reset;
+ bool force_pmc;
bool silence_err_logs;
/* Device management request data */
diff --git a/drivers/scsi/ufs/unipro.h b/drivers/scsi/ufs/unipro.h
index 4ee6478..f6b52ce 100644
--- a/drivers/scsi/ufs/unipro.h
+++ b/drivers/scsi/ufs/unipro.h
@@ -205,6 +205,9 @@ enum {
UNCHANGED = 7,
};
+#define PWRMODE_MASK 0xF
+#define PWRMODE_RX_OFFSET 4
+
/* PA TX/RX Frequency Series */
enum {
PA_HS_MODE_A = 1,
--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project.
^ permalink raw reply related [flat|nested] 3+ messages in thread