Skip to contentSkip to conversation

cpufreq: Remove KERN_ERR output if nForce2 chipset not found

PATCH In review Roland Dreier 2 participants 4 messages Mar 6, 2009 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 cpufreq: Remove KERN_ERR output if nForce2 chipset not found

Distribution kernels are building cpufreq drivers into the kernel to get faster boot; if the nForce2 driver is built in, then on systems that are not supported by the driver (ie nearly all current systems), the message

cpufreq: No nForce2 chipset.

is printed at the KERN_ERR level, which means it goes to the console even if quiet boot is turned on.

The best way to handle this is just to delete the message, since the likelihood of it ever being of any use to anyone is very low, and other cpufreq drivers don't print anything if no matching hardware is found.

---
 arch/x86/kernel/cpu/cpufreq/cpufreq-nforce2.c |    4 +---
 1 files changed, 1 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/cpufreq/cpufreq-nforce2.c +1 -3
index 965ea52..055fca3 100644
@@ -419,10 +419,8 @@ static int __init nforce2_init(void)
 	/* TODO: do we need to detect the processor? */
 
 	/* detect chipset */
-	if (nforce2_detect_chipset()) {
-		printk(KERN_ERR "cpufreq: No nForce2 chipset.\n");
+	if (nforce2_detect_chipset())
 		return -ENODEV;
-	}
 
 	return cpufreq_register_driver(&nforce2_driver);
 }
#1 lore raw
Quoted 6 lines · On Friday 06 March 2009 15:55:56 Roland Dreier wrote:

Distribution kernels are building cpufreq drivers into the kernel to get faster boot; if the nForce2 driver is built in, then on systems that are not supported by the driver (ie nearly all current systems), the message

cpufreq: No nForce2 chipset.

Quoted 8 lines

is printed at the KERN_ERR level, which means it goes to the console even if quiet boot is turned on.

The best way to handle this is just to delete the message, since the likelihood of it ever being of any use to anyone is very low, and other cpufreq drivers don't print anything if no matching hardware is found.

Nb: Matthew Garrett submitted a similar-but-different patch a week or two ago on the cpufreq list that demoted the message to KERN_INFO.

Signature · 2 lines
Jarod Wilson
[email protected]
#2 lore raw
Quoted 2 lines

Nb: Matthew Garrett submitted a similar-but-different patch a week or two ago on the cpufreq list that demoted the message to KERN_INFO.

Sorry, missed that. I considered the demotion to KERN_INFO but a) no other cpufreq drivers seem to print anything if the device isn't found and b) I can't imagine a realistic case of when the message would *ever* be useful.

- R.

#3 lore raw
Quoted 7 lines · On Friday 06 March 2009 16:51:49 Roland Dreier wrote:
Quoted 2 lines

Nb: Matthew Garrett submitted a similar-but-different patch a week or two ago on the cpufreq list that demoted the message to KERN_INFO.

Sorry, missed that. I considered the demotion to KERN_INFO but a) no other cpufreq drivers seem to print anything if the device isn't found and b) I can't imagine a realistic case of when the message would *ever* be useful.

Good points. I wasn't favoring one or the other 'til now, but simply returning -ENODEV does indeed seem to be more in line with what other freq scaling drivers do.

Signature · 2 lines
Jarod Wilson
[email protected]