Skip to contentSkip to conversation

fpga: dfl: Log and clear errors on driver init

PATCH In review Russ Weight 4 participants 10 messages Oct 19, 2021 – Oct 21, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover fpga: dfl: Log and clear errors on driver init

These patches address a request to log and clear any prexisting errors on FPGA cards when the drivers load. Any existing errors will result in print statements to the kernel error log before the errors are cleared. These changes specifically affect the fme and port error registers.

Russ Weight (2):
fpga: dfl: afu: Clear port errors in afu init
fpga: dfl: fme: Clear fme global errors at driver init

 drivers/fpga/dfl-afu-error.c |  26 ++++---
 drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
 2 files changed, 100 insertions(+), 54 deletions(-)
Signature · 1 line
2.25.1
#1 lore raw

PATCH 1/2 fpga: dfl: afu: Clear port errors in afu init

When the AFU driver initializes, log any pre-existing errors and clear them.

---
 drivers/fpga/dfl-afu-error.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
drivers/fpga/dfl-afu-error.c +16 -10
index ab7be6217368..0dc60bf49902 100644
@@ -47,13 +47,13 @@ static void afu_port_err_mask(struct device *dev, bool mask)
 }
 
 /* clear port errors. */
-static int afu_port_err_clear(struct device *dev, u64 err)
+static int afu_port_err_clear(struct device *dev, u64 err, bool clear_all)
 {
 	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
 	struct platform_device *pdev = to_platform_device(dev);
+	u64 v, port_error, port_first_error;
 	void __iomem *base_err, *base_hdr;
 	int enable_ret = 0, ret = -EBUSY;
-	u64 v;
 
 	base_err = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_ERROR);
 	base_hdr = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_HEADER);
@@ -88,16 +88,21 @@ static int afu_port_err_clear(struct device *dev, u64 err)
 	__afu_port_err_mask(dev, true);
 
 	/* Clear errors if err input matches with current port errors.*/
-	v = readq(base_err + PORT_ERROR);
+	port_error = readq(base_err + PORT_ERROR);
 
-	if (v == err) {
-		writeq(v, base_err + PORT_ERROR);
+	if (clear_all || port_error == err) {
+		port_first_error = readq(base_err + PORT_FIRST_ERROR);
 
-		v = readq(base_err + PORT_FIRST_ERROR);
-		writeq(v, base_err + PORT_FIRST_ERROR);
+		if (clear_all && (port_error || port_first_error))
+			dev_warn(dev,
+				 "Port Error: 0x%llx, First Error 0x%llx\n",
+				 port_error, port_first_error);
+
+		writeq(port_error, base_err + PORT_ERROR);
+		writeq(port_first_error, base_err + PORT_FIRST_ERROR);
 	} else {
 		dev_warn(dev, "%s: received 0x%llx, expected 0x%llx\n",
-			 __func__, v, err);
+			 __func__, port_error, err);
 		ret = -EINVAL;
 	}
 
@@ -137,7 +142,7 @@ static ssize_t errors_store(struct device *dev, struct device_attribute *attr,
 	if (kstrtou64(buff, 0, &value))
 		return -EINVAL;
 
-	ret = afu_port_err_clear(dev, value);
+	ret = afu_port_err_clear(dev, value, false);
 
 	return ret ? ret : count;
 }
@@ -211,7 +216,8 @@ const struct attribute_group port_err_group = {
 static int port_err_init(struct platform_device *pdev,
 			 struct dfl_feature *feature)
 {
-	afu_port_err_mask(&pdev->dev, false);
+	if (afu_port_err_clear(&pdev->dev, 0, true))
+		afu_port_err_mask(&pdev->dev, false);
 
 	return 0;
 }
Signature · 1 line
2.25.1
maintainer
#2 lore raw

Re: [PATCH v1 1/2] fpga: dfl: afu: Clear port errors in afu init

Quoted 1 line · Russ Weight

When the AFU driver initializes, log any pre-existing errors and clear them.

Please wrapper the commit message to pass checkpatch.pl

Quoted 11 lines
---
 drivers/fpga/dfl-afu-error.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)
drivers/fpga/dfl-afu-error.c +0 -0
index ab7be6217368..0dc60bf49902 100644

drivers/fpga/dfl-afu-error.c static void afu_port_err_mask(struct device *dev, bool mask) · line 50

Quoted 5 lines
drivers/fpga/dfl-afu-error.c +1 -1
@@ -47,13 +47,13 @@ static void afu_port_err_mask(struct device *dev, bool mask)
 }
 
 /* clear port errors. */
-static int afu_port_err_clear(struct device *dev, u64 err)
+static int afu_port_err_clear(struct device *dev, u64 err, bool clear_all)

how about "clear_all_on_init"?

Quoted 18 lines
drivers/fpga/dfl-afu-error.c +2 -3
 {
 	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
 	struct platform_device *pdev = to_platform_device(dev);
+	u64 v, port_error, port_first_error;
 	void __iomem *base_err, *base_hdr;
 	int enable_ret = 0, ret = -EBUSY;
-	u64 v;
 
 	base_err = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_ERROR);
 	base_hdr = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_HEADER);
@@ -88,16 +88,21 @@ static int afu_port_err_clear(struct device *dev, u64 err)
 	__afu_port_err_mask(dev, true);
 
 	/* Clear errors if err input matches with current port errors.*/
-	v = readq(base_err + PORT_ERROR);
+	port_error = readq(base_err + PORT_ERROR);
 
-	if (v == err) {

drivers/fpga/dfl-afu-error.c static int afu_port_err_clear(struct device *dev, u64 err) · line 99

Quoted 10 lines
drivers/fpga/dfl-afu-error.c +6 -3
-		writeq(v, base_err + PORT_ERROR);
+	if (clear_all || port_error == err) {
+		port_first_error = readq(base_err + PORT_FIRST_ERROR);
 
-		v = readq(base_err + PORT_FIRST_ERROR);
-		writeq(v, base_err + PORT_FIRST_ERROR);
+		if (clear_all && (port_error || port_first_error))
+			dev_warn(dev,
+				 "Port Error: 0x%llx, First Error 0x%llx\n",
+				 port_error, port_first_error);

The log could be more specific that we are reporting errors on device initialization.

Thanks,
Yilun

Quoted 31 lines
drivers/fpga/dfl-afu-error.c +7 -3
+
+		writeq(port_error, base_err + PORT_ERROR);
+		writeq(port_first_error, base_err + PORT_FIRST_ERROR);
 	} else {
 		dev_warn(dev, "%s: received 0x%llx, expected 0x%llx\n",
-			 __func__, v, err);
+			 __func__, port_error, err);
 		ret = -EINVAL;
 	}
 
@@ -137,7 +142,7 @@ static ssize_t errors_store(struct device *dev, struct device_attribute *attr,
 	if (kstrtou64(buff, 0, &value))
 		return -EINVAL;
 
-	ret = afu_port_err_clear(dev, value);
+	ret = afu_port_err_clear(dev, value, false);
 
 	return ret ? ret : count;
 }
@@ -211,7 +216,8 @@ const struct attribute_group port_err_group = {
 static int port_err_init(struct platform_device *pdev,
 			 struct dfl_feature *feature)
 {
-	afu_port_err_mask(&pdev->dev, false);
+	if (afu_port_err_clear(&pdev->dev, 0, true))
+		afu_port_err_mask(&pdev->dev, false);
 
 	return 0;
 }
Signature · 1 line
2.25.1
reviewer
#3 lore raw

Re: [PATCH v1 1/2] fpga: dfl: afu: Clear port errors in afu init

In reply to Russ Weight
Quoted 12 lines · Russ Weight

When the AFU driver initializes, log any pre-existing errors and clear them.

---
  drivers/fpga/dfl-afu-error.c | 26 ++++++++++++++++----------
  1 file changed, 16 insertions(+), 10 deletions(-)
drivers/fpga/dfl-afu-error.c +0 -0
index ab7be6217368..0dc60bf49902 100644

drivers/fpga/dfl-afu-error.c static void afu_port_err_mask(struct device *dev, bool mask) · line 54

Quoted 9 lines
drivers/fpga/dfl-afu-error.c +2 -1
@@ -47,13 +47,13 @@ static void afu_port_err_mask(struct device *dev, bool mask)
  }
 
  /* clear port errors. */
-static int afu_port_err_clear(struct device *dev, u64 err)
+static int afu_port_err_clear(struct device *dev, u64 err, bool clear_all)
  {
  	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
  	struct platform_device *pdev = to_platform_device(dev);
+	u64 v, port_error, port_first_error;

v is only used now by the read of PORT_HDR_STS, could v be changed to a more descriptive variable like hdr_sts ?

Quoted 11 lines
drivers/fpga/dfl-afu-error.c +0 -2
  	void __iomem *base_err, *base_hdr;
  	int enable_ret = 0, ret = -EBUSY;
-	u64 v;
 
  	base_err = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_ERROR);
  	base_hdr = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_HEADER);
@@ -88,16 +88,21 @@ static int afu_port_err_clear(struct device *dev, u64 err)
  	__afu_port_err_mask(dev, true);
 
  	/* Clear errors if err input matches with current port errors.*/
-	v = readq(base_err + PORT_ERROR);

drivers/fpga/dfl-afu-error.c static int afu_port_err_clear(struct device *dev, u64 err) · line 96

Quoted 10 lines
drivers/fpga/dfl-afu-error.c +4 -4
+	port_error = readq(base_err + PORT_ERROR);
 
-	if (v == err) {
-		writeq(v, base_err + PORT_ERROR);
+	if (clear_all || port_error == err) {
+		port_first_error = readq(base_err + PORT_FIRST_ERROR);
 
-		v = readq(base_err + PORT_FIRST_ERROR);
-		writeq(v, base_err + PORT_FIRST_ERROR);
+		if (clear_all && (port_error || port_first_error))

likely with clear_all that this dev_warn will over report.

how about removing clear_all && from if-check ?

Quoted 23 lines
drivers/fpga/dfl-afu-error.c +8 -2
+			dev_warn(dev,
+				 "Port Error: 0x%llx, First Error 0x%llx\n",
+				 port_error, port_first_error);
+
+		writeq(port_error, base_err + PORT_ERROR);
+		writeq(port_first_error, base_err + PORT_FIRST_ERROR);
  	} else {
  		dev_warn(dev, "%s: received 0x%llx, expected 0x%llx\n",
-			 __func__, v, err);
+			 __func__, port_error, err);
  		ret = -EINVAL;
  	}
 
@@ -137,7 +142,7 @@ static ssize_t errors_store(struct device *dev, struct device_attribute *attr,
  	if (kstrtou64(buff, 0, &value))
  		return -EINVAL;
 
-	ret = afu_port_err_clear(dev, value);
+	ret = afu_port_err_clear(dev, value, false);
 
  	return ret ? ret : count;
  }

drivers/fpga/dfl-afu-error.c const struct attribute_group port_err_group = { · line 220

Quoted 6 lines
drivers/fpga/dfl-afu-error.c +2 -1
@@ -211,7 +216,8 @@ const struct attribute_group port_err_group = {
  static int port_err_init(struct platform_device *pdev,
  			 struct dfl_feature *feature)
  {
-	afu_port_err_mask(&pdev->dev, false);
+	if (afu_port_err_clear(&pdev->dev, 0, true))
+		afu_port_err_mask(&pdev->dev, false);

There is a __afu_port_err_mask at the end of afu_port_err_clear so this call isn't needed.

Tom

Quoted 3 lines
drivers/fpga/dfl-afu-error.c +0 -0
  	return 0;
  }
#4 lore raw

Re: [PATCH v1 1/2] fpga: dfl: afu: Clear port errors in afu init

Quoted 43 lines · Tom Rix
Quoted 12 lines · Russ Weight

When the AFU driver initializes, log any pre-existing errors and clear them.

---
  drivers/fpga/dfl-afu-error.c | 26 ++++++++++++++++----------

  1 file changed, 16 insertions(+), 10 deletions(-)
drivers/fpga/dfl-afu-error.c +0 -0
index ab7be6217368..0dc60bf49902 100644

drivers/fpga/dfl-afu-error.c static void afu_port_err_mask(struct device *dev, bool mask) · line 53

Quoted 8 lines
drivers/fpga/dfl-afu-error.c +2 -1
@@ -47,13 +47,13 @@ static void afu_port_err_mask(struct device *dev, bool mask)
   }
     /* clear port errors. */
-static int afu_port_err_clear(struct device *dev, u64 err)
+static int afu_port_err_clear(struct device *dev, u64 err, bool clear_all)
   {
       struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
       struct platform_device *pdev = to_platform_device(dev);
+    u64 v, port_error, port_first_error;

v is only used now by the read of PORT_HDR_STS, could v be changed to a more descriptive variable like hdr_sts ?

Quoted 7 lines
drivers/fpga/dfl-afu-error.c +0 -1
       void __iomem *base_err, *base_hdr;
       int enable_ret = 0, ret = -EBUSY;
-    u64 v;
         base_err = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_ERROR);
       base_hdr = dfl_get_feature_ioaddr_by_id(dev, PORT_FEATURE_ID_HEADER);
@@ -88,16 +88,21 @@ static int afu_port_err_clear(struct device *dev, u64 err)
       __afu_port_err_mask(dev, true);

drivers/fpga/dfl-afu-error.c static int afu_port_err_clear(struct device *dev, u64 err) · line 95

Quoted 10 lines
drivers/fpga/dfl-afu-error.c +4 -3
         /* Clear errors if err input matches with current port errors.*/
-    v = readq(base_err + PORT_ERROR);
+    port_error = readq(base_err + PORT_ERROR);
   -    if (v == err) {
-        writeq(v, base_err + PORT_ERROR);
+    if (clear_all || port_error == err) {
+        port_first_error = readq(base_err + PORT_FIRST_ERROR);
   -        v = readq(base_err + PORT_FIRST_ERROR);
-        writeq(v, base_err + PORT_FIRST_ERROR);
+        if (clear_all && (port_error || port_first_error))

likely with clear_all that this dev_warn will over report.

Quoted 1 line

how about removing clear_all && from if-check ?

I think that would make it report more often? clear_all is only set at the time that the driver initializes. With the current condition, errors will only be logged when the driver loads, if and only if there are actually errors to report.

Quoted 21 lines
Quoted 21 lines
drivers/fpga/dfl-afu-error.c +7 -1
+            dev_warn(dev,
+                 "Port Error: 0x%llx, First Error 0x%llx\n",
+                 port_error, port_first_error);
+
+        writeq(port_error, base_err + PORT_ERROR);
+        writeq(port_first_error, base_err + PORT_FIRST_ERROR);
       } else {
           dev_warn(dev, "%s: received 0x%llx, expected 0x%llx\n",
-             __func__, v, err);
+             __func__, port_error, err);
           ret = -EINVAL;
       }
   @@ -137,7 +142,7 @@ static ssize_t errors_store(struct device *dev, struct device_attribute *attr,
       if (kstrtou64(buff, 0, &value))

          return -EINVAL;
  -    ret = afu_port_err_clear(dev, value);
+    ret = afu_port_err_clear(dev, value, false);
        return ret ? ret : count;
  }

drivers/fpga/dfl-afu-error.c +0 -0
@@ -211,7 +216,8 @@ const struct attribute_group port_err_group = {
Quoted 8 lines

drivers/fpga/dfl-afu-error.c const struct attribute_group port_err_group = { · line 220

Quoted 6 lines
drivers/fpga/dfl-afu-error.c +2 -1
   static int port_err_init(struct platform_device *pdev,
                struct dfl_feature *feature)
   {
-    afu_port_err_mask(&pdev->dev, false);
+    if (afu_port_err_clear(&pdev->dev, 0, true))
+        afu_port_err_mask(&pdev->dev, false);

There is a __afu_port_err_mask at the end of afu_port_err_clear so this call isn't needed.

The condition statements says to only clear the mask explicitly if afu_port_err_clear() fails. If it succeeds, then the explicit call will not happen, because as you have pointed out, afu_port_err_clear() clear the mask.

Thanks,
- Russ

Quoted 6 lines

Tom

Quoted 2 lines
drivers/fpga/dfl-afu-error.c +0 -0
         return 0;
   }
#5 lore raw

PATCH 2/2 fpga: dfl: fme: Clear fme global errors at driver init

In reply to Russ Weight

When the fme-error driver initializes, log any pre-existing errors and clear them. To avoid code replication, common code is gathered into a common fme_err_clear() function and a structure (err_reg) is created to describe each of the error registers, the corresponding mask registers, and the default mask for each register.

---
 drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
 1 file changed, 84 insertions(+), 44 deletions(-)
drivers/fpga/dfl-fme-error.c +84 -44
index 51c2892ec06d..44da7b30c469 100644
@@ -39,6 +39,27 @@
 
 #define ERROR_MASK		GENMASK_ULL(63, 0)
 
+struct err_reg {
+	char *name;
+	u32 err_offset;
+	u32 mask_offset;
+	u32 mask_value;
+};
+
+static struct err_reg pcie0_reg = {
+	.name = "PCIE0",
+	.err_offset = PCIE0_ERROR,
+	.mask_offset = PCIE0_ERROR_MASK,
+	.mask_value = 0ULL
+};
+
+static struct err_reg pcie1_reg = {
+	.name = "PCIE1",
+	.err_offset = PCIE1_ERROR,
+	.mask_offset = PCIE1_ERROR_MASK,
+	.mask_value = 0ULL
+};
+
 static ssize_t pcie0_errors_show(struct device *dev,
 				 struct device_attribute *attr, char *buf)
 {
@@ -55,31 +76,48 @@ static ssize_t pcie0_errors_show(struct device *dev,
 	return sprintf(buf, "0x%llx\n", (unsigned long long)value);
 }
 
-static ssize_t pcie0_errors_store(struct device *dev,
-				  struct device_attribute *attr,
-				  const char *buf, size_t count)
+static int fme_err_clear(struct device *dev, struct err_reg *reg,
+			 u64 err, bool clear_all)
 {
 	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
 	void __iomem *base;
 	int ret = 0;
-	u64 v, val;
-
-	if (kstrtou64(buf, 0, &val))
-		return -EINVAL;
+	u64 v;
 
 	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
 
 	mutex_lock(&pdata->lock);
-	writeq(GENMASK_ULL(63, 0), base + PCIE0_ERROR_MASK);
+	writeq(GENMASK_ULL(63, 0), base + reg->mask_offset);
 
-	v = readq(base + PCIE0_ERROR);
-	if (val == v)
-		writeq(v, base + PCIE0_ERROR);
-	else
+	v = readq(base + reg->err_offset);
+	if (clear_all || err == v) {
+		if (clear_all && v)
+			dev_warn(dev, "%s: %s Errors: 0x%llx\n",
+				 __func__, reg->name, v);
+
+		writeq(v, base + reg->err_offset);
+	} else {
 		ret = -EINVAL;
+	}
 
-	writeq(0ULL, base + PCIE0_ERROR_MASK);
+	writeq(reg->mask_value, base + reg->mask_offset);
 	mutex_unlock(&pdata->lock);
+
+	return ret;
+}
+
+static ssize_t pcie0_errors_store(struct device *dev,
+				  struct device_attribute *attr,
+				  const char *buf, size_t count)
+{
+	u64 val;
+	int ret;
+
+	if (kstrtou64(buf, 0, &val))
+		return -EINVAL;
+
+	ret = fme_err_clear(dev, &pcie0_reg, val, false);
+
 	return ret ? ret : count;
 }
 static DEVICE_ATTR_RW(pcie0_errors);
@@ -104,27 +142,14 @@ static ssize_t pcie1_errors_store(struct device *dev,
 				  struct device_attribute *attr,
 				  const char *buf, size_t count)
 {
-	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
-	void __iomem *base;
-	int ret = 0;
-	u64 v, val;
+	u64 val;
+	int ret;
 
 	if (kstrtou64(buf, 0, &val))
 		return -EINVAL;
 
-	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
+	ret = fme_err_clear(dev, &pcie1_reg, val, false);
 
-	mutex_lock(&pdata->lock);
-	writeq(GENMASK_ULL(63, 0), base + PCIE1_ERROR_MASK);
-
-	v = readq(base + PCIE1_ERROR);
-	if (val == v)
-		writeq(v, base + PCIE1_ERROR);
-	else
-		ret = -EINVAL;
-
-	writeq(0ULL, base + PCIE1_ERROR_MASK);
-	mutex_unlock(&pdata->lock);
 	return ret ? ret : count;
 }
 static DEVICE_ATTR_RW(pcie1_errors);
@@ -218,29 +243,26 @@ static ssize_t fme_errors_store(struct device *dev,
 				struct device_attribute *attr,
 				const char *buf, size_t count)
 {
-	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
+	static struct err_reg fme_reg = {
+		.name = "FME",
+		.err_offset = FME_ERROR,
+		.mask_offset = FME_ERROR_MASK,
+		.mask_value = 0ULL
+	};
 	void __iomem *base;
-	u64 v, val;
-	int ret = 0;
+	u64 val;
+	int ret;
 
 	if (kstrtou64(buf, 0, &val))
 		return -EINVAL;
 
+	/* Workaround: disable MBP_ERROR if feature revision is 0 */
 	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
+	if (!dfl_feature_revision(base))
+		fme_reg.mask_value = MBP_ERROR;
 
-	mutex_lock(&pdata->lock);
-	writeq(GENMASK_ULL(63, 0), base + FME_ERROR_MASK);
+	ret = fme_err_clear(dev, &fme_reg, val, false);
 
-	v = readq(base + FME_ERROR);
-	if (val == v)
-		writeq(v, base + FME_ERROR);
-	else
-		ret = -EINVAL;
-
-	/* Workaround: disable MBP_ERROR if feature revision is 0 */
-	writeq(dfl_feature_revision(base) ? 0ULL : MBP_ERROR,
-	       base + FME_ERROR_MASK);
-	mutex_unlock(&pdata->lock);
 	return ret ? ret : count;
 }
 static DEVICE_ATTR_RW(fme_errors);
@@ -338,6 +360,24 @@ static void fme_err_mask(struct device *dev, bool mask)
 static int fme_global_err_init(struct platform_device *pdev,
 			       struct dfl_feature *feature)
 {
+	static struct err_reg fme_reg = {
+		.name = "FME",
+		.err_offset = FME_ERROR,
+		.mask_offset = FME_ERROR_MASK,
+		.mask_value = 0ULL
+	};
+	void __iomem *base;
+
+	/* Workaround: disable MBP_ERROR if feature revision is 0 */
+	base = dfl_get_feature_ioaddr_by_id(&pdev->dev,
+					    FME_FEATURE_ID_GLOBAL_ERR);
+	if (!dfl_feature_revision(base))
+		fme_reg.mask_value = MBP_ERROR;
+
+	(void)fme_err_clear(&pdev->dev, &pcie0_reg, 0ULL, true);
+	(void)fme_err_clear(&pdev->dev, &pcie1_reg, 0ULL, true);
+	(void)fme_err_clear(&pdev->dev, &fme_reg, 0ULL, true);
+
 	fme_err_mask(&pdev->dev, false);
 
 	return 0;
Signature · 1 line
2.25.1
reviewer
#6 lore raw

Re: [PATCH v1 2/2] fpga: dfl: fme: Clear fme global errors at driver init

Quoted 30 lines · Russ Weight

When the fme-error driver initializes, log any pre-existing errors and clear them. To avoid code replication, common code is gathered into a common fme_err_clear() function and a structure (err_reg) is created to describe each of the error registers, the corresponding mask registers, and the default mask for each register.

---
  drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
  1 file changed, 84 insertions(+), 44 deletions(-)
drivers/fpga/dfl-fme-error.c +11 -0
index 51c2892ec06d..44da7b30c469 100644
@@ -39,6 +39,27 @@
 
  #define ERROR_MASK		GENMASK_ULL(63, 0)
 
+struct err_reg {
+	char *name;
+	u32 err_offset;
+	u32 mask_offset;
+	u32 mask_value;
+};
+
+static struct err_reg pcie0_reg = {
+	.name = "PCIE0",
+	.err_offset = PCIE0_ERROR,
+	.mask_offset = PCIE0_ERROR_MASK,

drivers/fpga/dfl-fme-error.c line 62

Quoted 10 lines
drivers/fpga/dfl-fme-error.c +10 -0
+	.mask_value = 0ULL
+};
+
+static struct err_reg pcie1_reg = {
+	.name = "PCIE1",
+	.err_offset = PCIE1_ERROR,
+	.mask_offset = PCIE1_ERROR_MASK,
+	.mask_value = 0ULL
+};
+

Can these be stored in an array ?

Quoted 25 lines
drivers/fpga/dfl-fme-error.c +3 -7
  static ssize_t pcie0_errors_show(struct device *dev,
  				 struct device_attribute *attr, char *buf)
  {
@@ -55,31 +76,48 @@ static ssize_t pcie0_errors_show(struct device *dev,
  	return sprintf(buf, "0x%llx\n", (unsigned long long)value);
  }
 
-static ssize_t pcie0_errors_store(struct device *dev,
-				  struct device_attribute *attr,
-				  const char *buf, size_t count)
+static int fme_err_clear(struct device *dev, struct err_reg *reg,
+			 u64 err, bool clear_all)
  {
  	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
  	void __iomem *base;
  	int ret = 0;
-	u64 v, val;
-
-	if (kstrtou64(buf, 0, &val))
-		return -EINVAL;
+	u64 v;
 
  	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
 
  	mutex_lock(&pdata->lock);

drivers/fpga/dfl-fme-error.c static ssize_t pcie0_errors_show(struct device *dev, · line 94

Quoted 10 lines
drivers/fpga/dfl-fme-error.c +4 -5
-	writeq(GENMASK_ULL(63, 0), base + PCIE0_ERROR_MASK);
+	writeq(GENMASK_ULL(63, 0), base + reg->mask_offset);
 
-	v = readq(base + PCIE0_ERROR);
-	if (val == v)
-		writeq(v, base + PCIE0_ERROR);
-	else
+	v = readq(base + reg->err_offset);
+	if (clear_all || err == v) {
+		if (clear_all && v)

clear_all will over report.

Quoted 31 lines
drivers/fpga/dfl-fme-error.c +23 -1
+			dev_warn(dev, "%s: %s Errors: 0x%llx\n",
+				 __func__, reg->name, v);
+
+		writeq(v, base + reg->err_offset);
+	} else {
  		ret = -EINVAL;
+	}
 
-	writeq(0ULL, base + PCIE0_ERROR_MASK);
+	writeq(reg->mask_value, base + reg->mask_offset);
  	mutex_unlock(&pdata->lock);
+
+	return ret;
+}
+
+static ssize_t pcie0_errors_store(struct device *dev,
+				  struct device_attribute *attr,
+				  const char *buf, size_t count)
+{
+	u64 val;
+	int ret;
+
+	if (kstrtou64(buf, 0, &val))
+		return -EINVAL;
+
+	ret = fme_err_clear(dev, &pcie0_reg, val, false);
+
  	return ret ? ret : count;
  }
  static DEVICE_ATTR_RW(pcie0_errors);

drivers/fpga/dfl-fme-error.c static ssize_t pcie1_errors_store(struct device *dev, · line 144

Quoted 3 lines
drivers/fpga/dfl-fme-error.c +0 -0
@@ -104,27 +142,14 @@ static ssize_t pcie1_errors_store(struct device *dev,
  				  struct device_attribute *attr,
  				  const char *buf, size_t count)
  {

This looks like a copy of pcie0_errors_store can these functions be consolidated ?

Quoted 76 lines
drivers/fpga/dfl-fme-error.c +18 -31
-	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
-	void __iomem *base;
-	int ret = 0;
-	u64 v, val;
+	u64 val;
+	int ret;
 
  	if (kstrtou64(buf, 0, &val))
  		return -EINVAL;
 
-	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
+	ret = fme_err_clear(dev, &pcie1_reg, val, false);
 
-	mutex_lock(&pdata->lock);
-	writeq(GENMASK_ULL(63, 0), base + PCIE1_ERROR_MASK);
-
-	v = readq(base + PCIE1_ERROR);
-	if (val == v)
-		writeq(v, base + PCIE1_ERROR);
-	else
-		ret = -EINVAL;
-
-	writeq(0ULL, base + PCIE1_ERROR_MASK);
-	mutex_unlock(&pdata->lock);
  	return ret ? ret : count;
  }
  static DEVICE_ATTR_RW(pcie1_errors);
@@ -218,29 +243,26 @@ static ssize_t fme_errors_store(struct device *dev,
  				struct device_attribute *attr,
  				const char *buf, size_t count)
  {
-	struct dfl_feature_platform_data *pdata = dev_get_platdata(dev);
+	static struct err_reg fme_reg = {
+		.name = "FME",
+		.err_offset = FME_ERROR,
+		.mask_offset = FME_ERROR_MASK,
+		.mask_value = 0ULL
+	};
  	void __iomem *base;
-	u64 v, val;
-	int ret = 0;
+	u64 val;
+	int ret;
 
  	if (kstrtou64(buf, 0, &val))
  		return -EINVAL;
 
+	/* Workaround: disable MBP_ERROR if feature revision is 0 */
  	base = dfl_get_feature_ioaddr_by_id(dev, FME_FEATURE_ID_GLOBAL_ERR);
+	if (!dfl_feature_revision(base))
+		fme_reg.mask_value = MBP_ERROR;
 
-	mutex_lock(&pdata->lock);
-	writeq(GENMASK_ULL(63, 0), base + FME_ERROR_MASK);
+	ret = fme_err_clear(dev, &fme_reg, val, false);
 
-	v = readq(base + FME_ERROR);
-	if (val == v)
-		writeq(v, base + FME_ERROR);
-	else
-		ret = -EINVAL;
-
-	/* Workaround: disable MBP_ERROR if feature revision is 0 */
-	writeq(dfl_feature_revision(base) ? 0ULL : MBP_ERROR,
-	       base + FME_ERROR_MASK);
-	mutex_unlock(&pdata->lock);
  	return ret ? ret : count;
  }
  static DEVICE_ATTR_RW(fme_errors);
@@ -338,6 +360,24 @@ static void fme_err_mask(struct device *dev, bool mask)
  static int fme_global_err_init(struct platform_device *pdev,
  			       struct dfl_feature *feature)
  {
+	static struct err_reg fme_reg = {
+		.name = "FME",
+		.err_offset = FME_ERROR,

drivers/fpga/dfl-fme-error.c static void fme_err_mask(struct device *dev, bool mask) · line 375

Quoted 10 lines
drivers/fpga/dfl-fme-error.c +10 -0
+		.mask_offset = FME_ERROR_MASK,
+		.mask_value = 0ULL
+	};
+	void __iomem *base;
+
+	/* Workaround: disable MBP_ERROR if feature revision is 0 */
+	base = dfl_get_feature_ioaddr_by_id(&pdev->dev,
+					    FME_FEATURE_ID_GLOBAL_ERR);
+	if (!dfl_feature_revision(base))
+		fme_reg.mask_value = MBP_ERROR;

A similar block above to set fme_reg.mask_value.

These should be consolidated.

drivers/fpga/dfl-fme-error.c static void fme_err_mask(struct device *dev, bool mask) · line 377

Quoted 2 lines
drivers/fpga/dfl-fme-error.c +2 -0
+
+	(void)fme_err_clear(&pdev->dev, &pcie0_reg, 0ULL, true);

remove casting return to (void)

Tom

Quoted 6 lines
drivers/fpga/dfl-fme-error.c +3 -0
+	(void)fme_err_clear(&pdev->dev, &pcie1_reg, 0ULL, true);
+	(void)fme_err_clear(&pdev->dev, &fme_reg, 0ULL, true);
+
  	fme_err_mask(&pdev->dev, false);
 
  	return 0;
#7 lore raw
In reply to Russ Weight
Quoted 6 lines

Subject: [PATCH v1 0/2] fpga: dfl: Log and clear errors on driver init

These patches address a request to log and clear any prexisting errors on FPGA cards when the drivers load. Any existing errors will result in print statements to the kernel error log before the errors are cleared. These changes specifically affect the fme and port error registers.

Could you please explain more about why we need this change?
As we have user interface to log and clear errors already, is it a better choice to let userspace log and clear them during AFU initialization?

Hao

Quoted 11 lines

Russ Weight (2):
fpga: dfl: afu: Clear port errors in afu init
fpga: dfl: fme: Clear fme global errors at driver init

 drivers/fpga/dfl-afu-error.c |  26 ++++---
 drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
 2 files changed, 100 insertions(+), 54 deletions(-)

--
2.25.1

#8 lore raw
Quoted 9 lines · Hao
Quoted 6 lines

Subject: [PATCH v1 0/2] fpga: dfl: Log and clear errors on driver init

These patches address a request to log and clear any prexisting errors on FPGA cards when the drivers load. Any existing errors will result in print statements to the kernel error log before the errors are cleared. These changes specifically affect the fme and port error registers.

Could you please explain more about why we need this change?
As we have user interface to log and clear errors already, is it a better choice to let userspace log and clear them during AFU initialization?

In the new architecture we are offering more flexibility to customers for adding functions. With these designs it becomes nearly impossible to design the AFU interface handler to recover from errors and resume operation afterwards. The proposed solution is to flag the source of the error and then capture it in sticky registers so that they can be read out from SW in the event of a crash/warm boot. To ensure that we capture these errors, the proposal is to log them in the kernel log and clear them at driver initialization.

- Russ

Quoted 12 lines

Hao

Quoted 10 lines

Russ Weight (2):
fpga: dfl: afu: Clear port errors in afu init
fpga: dfl: fme: Clear fme global errors at driver init

 drivers/fpga/dfl-afu-error.c |  26 ++++---
 drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
 2 files changed, 100 insertions(+), 54 deletions(-)

--
2.25.1

#9 lore raw
Quoted 10 lines · Hao
Quoted 9 lines
Quoted 6 lines

Subject: [PATCH v1 0/2] fpga: dfl: Log and clear errors on driver init

These patches address a request to log and clear any prexisting errors on FPGA cards when the drivers load. Any existing errors will result in print statements to the kernel error log before the errors are cleared. These changes specifically affect the fme and port error registers.

Could you please explain more about why we need this change?
As we have user interface to log and clear errors already, is it a better choice to let userspace log and clear them during AFU initialization?

Quoted 8 lines

In the new architecture we are offering more flexibility to customers for adding functions. With these designs it becomes nearly impossible to design the AFU interface handler to recover from errors and resume operation afterwards. The proposed solution is to flag the source of the error and then capture it in sticky registers so that they can be read out from SW in the event of a crash/warm boot. To ensure that we capture these errors, the proposal is to log them in the kernel log and clear them at driver initialization.

The error can be logged and cleared at user space driver initialization, as current usage model is let userspace handle everything including error.

Hao

Quoted 15 lines

- Russ

Quoted 12 lines

Hao

Quoted 10 lines

Russ Weight (2):
fpga: dfl: afu: Clear port errors in afu init
fpga: dfl: fme: Clear fme global errors at driver init

 drivers/fpga/dfl-afu-error.c |  26 ++++---
 drivers/fpga/dfl-fme-error.c | 128 +++++++++++++++++++++++------------
 2 files changed, 100 insertions(+), 54 deletions(-)

--
2.25.1