Skip to contentSkip to conversation

hwmon: (ltc2945) Don't crash the kernel unnecessarily

PATCH In review Guenter Roeck 2 participants 3 messages Apr 21, 2014 – Apr 23, 2014 Hardware monitoring

lore
Share & tools
lore

Guenter Roeck

author maintainer
#0 lore raw

PATCH 1/1 hwmon: (ltc2945) Don't crash the kernel unnecessarily

An implementation error should not crash the kernel if it is avoidable. Replace BUG() with WARN_ONCE().

---
 drivers/hwmon/ltc2945.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/hwmon/ltc2945.c +3 -3
index c104cc3..c9cddf5 100644
@@ -1,4 +1,4 @@
-/*
+    /*
  * Driver for Linear Technology LTC2945 I2C Power Monitor
  *
  * Copyright (c) 2014 Guenter Roeck
@@ -314,8 +314,8 @@ static ssize_t ltc2945_reset_history(struct device *dev,
 		reg = LTC2945_MAX_ADIN_H;
 		break;
 	default:
-		BUG();
-		break;
+		WARN_ONCE(1, "Bad register: 0x%x\n", reg);
+		return -EINVAL;
 	}
 	/* Reset maximum */
 	ret = regmap_bulk_write(regmap, reg, buf_max, num_regs);
Signature · 7 lines
1.7.9.7


_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
Reviewed-by on v1 1/1
#1 lore raw

Hi Guenter,

Quoted 19 lines · Guenter Roeck

An implementation error should not crash the kernel if it is avoidable. Replace BUG() with WARN_ONCE().

---
 drivers/hwmon/ltc2945.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/hwmon/ltc2945.c +1 -1
index c104cc3..c9cddf5 100644
@@ -1,4 +1,4 @@
-/*
+    /*
  * Driver for Linear Technology LTC2945 I2C Power Monitor
  *
  * Copyright (c) 2014 Guenter Roeck

drivers/hwmon/ltc2945.c static ssize_t ltc2945_reset_history(struct device *dev, · line 321

Quoted 10 lines
drivers/hwmon/ltc2945.c +2 -2
@@ -314,8 +314,8 @@ static ssize_t ltc2945_reset_history(struct device *dev,
 		reg = LTC2945_MAX_ADIN_H;
 		break;
 	default:
-		BUG();
-		break;
+		WARN_ONCE(1, "Bad register: 0x%x\n", reg);
+		return -EINVAL;
 	}
 	/* Reset maximum */
 	ret = regmap_bulk_write(regmap, reg, buf_max, num_regs);

Agreed. -EINVAL may be confusing though, as this normally means the user input was invalid (which isn't the case here.) Maybe we can come up with a better error code. Maybe -EFAULT? Or be original and use -EBADSLT? ;)

It's purely theoretical anyway, as we both know that this piece of code is never going to be executed. TBH if it was me I would just drop this test altogether. But anyway:

Signature · 7 lines
Jean Delvare
SUSE L3 Support

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
author maintainer
#2 lore raw
In reply to Guenter Roeck
Quoted 38 lines · Jean Delvare

Hi Guenter,

Quoted 19 lines · Guenter Roeck

An implementation error should not crash the kernel if it is avoidable. Replace BUG() with WARN_ONCE().

---
 drivers/hwmon/ltc2945.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/hwmon/ltc2945.c +1 -1
index c104cc3..c9cddf5 100644
@@ -1,4 +1,4 @@
-/*
+    /*
  * Driver for Linear Technology LTC2945 I2C Power Monitor
  *
  * Copyright (c) 2014 Guenter Roeck

drivers/hwmon/ltc2945.c static ssize_t ltc2945_reset_history(struct device *dev, · line 321

Quoted 10 lines
drivers/hwmon/ltc2945.c +2 -2
@@ -314,8 +314,8 @@ static ssize_t ltc2945_reset_history(struct device *dev,
 		reg = LTC2945_MAX_ADIN_H;
 		break;
 	default:
-		BUG();
-		break;
+		WARN_ONCE(1, "Bad register: 0x%x\n", reg);
+		return -EINVAL;
 	}
 	/* Reset maximum */
 	ret = regmap_bulk_write(regmap, reg, buf_max, num_regs);

Agreed. -EINVAL may be confusing though, as this normally means the user input was invalid (which isn't the case here.) Maybe we can come up with a better error code. Maybe -EFAULT? Or be original and use -EBADSLT? ;)

Quoted 4 lines

It's purely theoretical anyway, as we both know that this piece of code is never going to be executed. TBH if it was me I would just drop this test altogether. But anyway:

Guess I'll leave it as is ... as you said, it is theoretic anyway, at least for now. I don't want to drop the case entirely; someone might add code later on and screw it up. Also I can't just drop the default:
case without getting a gcc complaint. Sure, I could silently set the variables to some default, but that doesn't seem to be right either.

It would be nice to have something like -EIMPLEMENTATION_ERROR for such cases ...

Thanks a lot for the reviews!

Guenter

_______________________________________________
lm-sensors mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors