Skip to contentSkip to conversation

Allow platform drivers to update UIC command timeout

PATCH Superseded Bao D. Nguyen 2 participants 6 messages Jul 9, 2024 – Jul 12, 2024

Subsystems (2)

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v3 cover Allow platform drivers to update UIC command timeout

The UIC command timeout default value remains as 500ms.

Allow platform drivers to change the UIC command timeout as they wish. During product development where a lot of debug logging/printing can occur, the uart may print from different modules with interrupt disabled for more than 500ms, causing interrupt starvation and UIC command timeout as a result. The UIC command timeout may eventually cause a watchdog timeout unnecessarily. With this change, the platform drivers can set a different UIC command timeout as desired. The supported values range from 500ms to 2 seconds.

v2 -> v3: - Addressed Bart's comments. Renamed UIC_CMD_TIMEOUT to
UIC_CMD_TIMEOUT_DEFAULT. Changed "Default.." to "Defaults..". Removed an extra line in the module description.
v1 -> v2: - Created kernel module parameter namely uic_cmd_timeout
as recommended by Bart. Addressed some other comments.
- Un-do the change in the include/ufs/ufshcd.h file which added the uic_cmd_timeout field to the hba struct.
- Removed the patch 2 in the series where the UIC command timeout value was overridden by the platform driver.

Bao D. Nguyen (1):
scsi: ufs: core: Support Updating UIC Command Timeout

 drivers/ufs/core/ufshcd.c | 34 +++++++++++++++++++++++++++++-----
 1 file changed, 29 insertions(+), 5 deletions(-)
Signature · 1 line
2.7.4
#1 lore raw

PATCH v3 1/1 scsi: ufs: core: Support Updating UIC Command Timeout

The default UIC command timeout still remains 500ms. Allow platform drivers to override the UIC command timeout if desired.

In a real product, the 500ms timeout value is probably good enough. However, during the product development where there are a lot of logging and debug messages being printed to the uart console, interrupt starvations happen occasionally because the uart may print long debug messages from different modules in the system. While printing, the uart may have interrupts disabled for more than 500ms, causing UIC command timeout.
The UIC command timeout would trigger more printing from the UFS driver, and eventually a watchdog timeout may occur unnecessarily.

Add support for overriding the UIC command timeout value with the newly created uic_cmd_timeout kernel module parameter. Default value is 500ms. Supported values range from 500ms to 2 seconds.

---
 drivers/ufs/core/ufshcd.c | 34 +++++++++++++++++++++++++++++-----
 1 file changed, 29 insertions(+), 5 deletions(-)
drivers/ufs/core/ufshcd.c +29 -5
index 21429ee..421e295 100644
@@ -51,8 +51,10 @@
 
 
 /* UIC command timeout, unit: ms */
-#define UIC_CMD_TIMEOUT	500
-
+enum {
+	UIC_CMD_TIMEOUT_DEFAULT	= 500,
+	UIC_CMD_TIMEOUT_MAX	= 2000,
+};
 /* NOP OUT retries waiting for NOP IN response */
 #define NOP_OUT_RETRIES    10
 /* Timeout after 50 msecs if NOP OUT hangs without response */
@@ -113,6 +115,28 @@ static bool is_mcq_supported(struct ufs_hba *hba)
 module_param(use_mcq_mode, bool, 0644);
 MODULE_PARM_DESC(use_mcq_mode, "Control MCQ mode for controllers starting from UFSHCI 4.0. 1 - enable MCQ, 0 - disable MCQ. MCQ is enabled by default");
 
+static int uic_cmd_timeout_set(const char *val, const struct kernel_param *kp)
+{
+	unsigned int n;
+	int ret;
+
+	ret = kstrtou32(val, 0, &n);
+	if (ret != 0 || n < UIC_CMD_TIMEOUT_DEFAULT || n > UIC_CMD_TIMEOUT_MAX)
+		return -EINVAL;
+
+	return param_set_int(val, kp);
+}
+
+static const struct kernel_param_ops uic_cmd_timeout_ops = {
+	.set = uic_cmd_timeout_set,
+	.get = param_get_uint,
+};
+
+static unsigned int uic_cmd_timeout = UIC_CMD_TIMEOUT_DEFAULT;
+module_param_cb(uic_cmd_timeout, &uic_cmd_timeout_ops, &uic_cmd_timeout, 0644);
+MODULE_PARM_DESC(uic_cmd_timeout,
+		"UFS UIC command timeout in milliseconds. Defaults to 500ms. Supported values range from 500ms to 2 seconds inclusively");
+
 #define ufshcd_toggle_vreg(_dev, _vreg, _on)				\
 	({                                                              \
 		int _ret;                                               \
@@ -2460,7 +2484,7 @@ static inline bool ufshcd_ready_for_uic_cmd(struct ufs_hba *hba)
 {
 	u32 val;
 	int ret = read_poll_timeout(ufshcd_readl, val, val & UIC_COMMAND_READY,
-				    500, UIC_CMD_TIMEOUT * 1000, false, hba,
+				    500, uic_cmd_timeout * 1000, false, hba,
 				    REG_CONTROLLER_STATUS);
 	return ret == 0;
 }
@@ -2520,7 +2544,7 @@ ufshcd_wait_for_uic_cmd(struct ufs_hba *hba, struct uic_command *uic_cmd)
 	lockdep_assert_held(&hba->uic_cmd_mutex);
 
 	if (wait_for_completion_timeout(&uic_cmd->done,
-					msecs_to_jiffies(UIC_CMD_TIMEOUT))) {
+					msecs_to_jiffies(uic_cmd_timeout))) {
 		ret = uic_cmd->argument2 & MASK_UIC_COMMAND_RESULT;
 	} else {
 		ret = -ETIMEDOUT;
@@ -4298,7 +4322,7 @@ static int ufshcd_uic_pwr_ctrl(struct ufs_hba *hba, struct uic_command *cmd)
 	}
 
 	if (!wait_for_completion_timeout(hba->uic_async_done,
-					 msecs_to_jiffies(UIC_CMD_TIMEOUT))) {
+					 msecs_to_jiffies(uic_cmd_timeout))) {
 		dev_err(hba->dev,
 			"pwr ctrl cmd 0x%x with mode 0x%x completion timeout\n",
 			cmd->command, cmd->argument3);
Signature · 1 line
2.7.4
#2 lore raw

Re: [PATCH v3 1/1] scsi: ufs: core: Support Updating UIC Command Timeout

drivers/ufs/core/ufshcd.c static bool is_mcq_supported(struct ufs_hba *hba) · line 128

Quoted 11 lines · Bao D. Nguyen
drivers/ufs/core/ufshcd.c +11 -0
+static int uic_cmd_timeout_set(const char *val, const struct kernel_param *kp)
+{
+	unsigned int n;
+	int ret;
+
+	ret = kstrtou32(val, 0, &n);
+	if (ret != 0 || n < UIC_CMD_TIMEOUT_DEFAULT || n > UIC_CMD_TIMEOUT_MAX)
+		return -EINVAL;
+
+	return param_set_int(val, kp);
+}

The above code converts 'val' twice to an integer: a first time by calling kstrtou32() and a second time by calling param_set_int(). Please remove one of the two string-to-integer conversions, e.g. by changing "param_set_int(val, kp)" into "uic_cmd_timeout = n" or *(unsigned int *)kp->arg = n".

Thanks,

Bart.

#3 lore raw

Re: [PATCH v3 1/1] scsi: ufs: core: Support Updating UIC Command Timeout

Quoted 12 lines · Bart Van Assche
Quoted 11 lines · Bao D. Nguyen

+static int uic_cmd_timeout_set(const char *val, const struct kernel_param *kp)

drivers/ufs/core/ufshcd.c +6 -0
+{
+    unsigned int n;
+    int ret;
+
+    ret = kstrtou32(val, 0, &n);
+    if (ret != 0 || n < UIC_CMD_TIMEOUT_DEFAULT || n > 

UIC_CMD_TIMEOUT_MAX)
+        return -EINVAL;
+

Quoted 8 lines
Quoted 2 lines

+    return param_set_int(val, kp);
+}

The above code converts 'val' twice to an integer: a first time by calling kstrtou32() and a second time by calling param_set_int(). Please remove one of the two string-to-integer conversions, e.g. by changing "param_set_int(val, kp)" into "uic_cmd_timeout = n" or *(unsigned int *)kp->arg = n".

Hi Bart,
My understanding is that in the kstrtou32() function, the the result of the conversion is written to the third parameter only which is '&n' in this case. 'val' does not get updated (and it cannot be updated because of its 'const' type). Please correct me if I am wrong.

Thanks, Bao

Quoted 4 lines

Thanks,

Bart.

#4 lore raw

Re: [PATCH v3 1/1] scsi: ufs: core: Support Updating UIC Command Timeout

Quoted 21 lines · Bao D. Nguyen
Quoted 12 lines · Bart Van Assche
Quoted 11 lines · Bao D. Nguyen

+static int uic_cmd_timeout_set(const char *val, const struct kernel_param *kp)

drivers/ufs/core/ufshcd.c +6 -0
+{
+    unsigned int n;
+    int ret;
+
+    ret = kstrtou32(val, 0, &n);
+    if (ret != 0 || n < UIC_CMD_TIMEOUT_DEFAULT || n > 

UIC_CMD_TIMEOUT_MAX)
+        return -EINVAL;
+

Quoted 8 lines
Quoted 2 lines

+    return param_set_int(val, kp);
+}

The above code converts 'val' twice to an integer: a first time by calling kstrtou32() and a second time by calling param_set_int(). Please remove one of the two string-to-integer conversions, e.g. by changing "param_set_int(val, kp)" into "uic_cmd_timeout = n" or *(unsigned int *)kp->arg = n".

Quoted 2 lines

Hi Bart,
My understanding is that in the kstrtou32() function, the  the result of the conversion is written to the third parameter only which is '&n' in this case. 'val' does not get updated (and it cannot be updated because of its 'const' type). Please correct me if I am wrong.

Hi Bart,
Maybe I misunderstood your comment. You probably was concerned about the execution time of param_set_int(val, kp) vs a single assignment instruction 'uic_cmd_timeout = n;', right? If that's the case, I'll update the patch.

Thanks, Bao

Quoted 8 lines

Thanks, Bao

Quoted 4 lines

Thanks,

Bart.

#5 lore raw

Re: [PATCH v3 1/1] scsi: ufs: core: Support Updating UIC Command Timeout

Quoted 1 line · Bao D. Nguyen

Maybe I misunderstood your comment. You probably was concerned about the execution time of param_set_int(val, kp) vs a single assignment instruction 'uic_cmd_timeout = n;', right? If that's the case, I'll update the patch.

I'm concerned about the two conversions yielding different results, e.g. because of different handling of base prefixes (e.g. 0x) or different handling of overflows. This patch would be easier to review if the conversion from string to int only happens once.

Thanks,

Bart.