Skip to contentSkip to conversation

cpufreq/intel_pstate: Do not issue the not supported message on !Intel

PATCH In review Borislav Petkov 4 participants 7 messages Mar 30, 2019 – Apr 2, 2019

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

PATCH 1/1 cpufreq/intel_pstate: Do not issue the not supported message on !Intel

From: Borislav Petkov <[email protected]>

Issue the CPU-not-supported message only on Intel machines as this driver is Intel-only. Which means, the print statement can remain KERN_INFO for ease of debugging (no need to enable it first in dynamic debug).

While at it, correct it to say CPU "model" which is what that test does.

Fixes 076b862c7e44 ("cpufreq: intel_pstate: Add reasons for failure and debug messages")
Cc Erwan Velu <[email protected]>
Cc Len Brown <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
CC Srinivas Pandruvada <[email protected]>
Cc Viresh Kumar <[email protected]>
---
 drivers/cpufreq/intel_pstate.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/cpufreq/intel_pstate.c +3 -1
index ea62e3f02d56..19854f01e2fa 100644
@@ -2608,7 +2608,9 @@ static int __init intel_pstate_init(void)
 	} else {
 		id = x86_match_cpu(intel_pstate_cpu_ids);
 		if (!id) {
-			pr_info("CPU ID not supported\n");
+			if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL)
+				pr_info("CPU model not supported\n");
+
 			return -ENODEV;
 		}
 
Signature · 1 line
2.21.0
#1 lore raw
Quoted 5 lines

index ea62e3f02d56..19854f01e2fa 100644

drivers/cpufreq/intel_pstate.c static int __init intel_pstate_init(void) · line 2615

Quoted 9 lines
drivers/cpufreq/intel_pstate.c +3 -1
@@ -2608,7 +2608,9 @@ static int __init intel_pstate_init(void)
  	} else {
  		id = x86_match_cpu(intel_pstate_cpu_ids);
  		if (!id) {
-			pr_info("CPU ID not supported\n");
+			if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL)
+				pr_info("CPU model not supported\n");
+
  			return -ENODEV;
  		}

That's a good catch but I was wondering why not putting this vendor condition at the initial "if (noload)" statement.

I mean, if we don't run an intel CPU there is no need of making the x86_match_cpu().

This commit is also killing the case of reporting an unsupported intel processor.

I'd suggest something like this and keep the 'CPUID not supported' part untouched.

    if (no_load) || (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)

        return -ENODEV

#2 lore raw
Quoted 4 lines · Erwan Velu

I'd suggest something like this and keep the 'CPUID not supported' part untouched.

    if (no_load) || (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)

Yeah, that's the usual thing we do in such cases and the better idea, I'll do that in v2.

Thx.

Signature · 4 lines
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.
#3 lore raw

PATCH 1/1 cpufreq/intel_pstate: Load only on Intel hardware

From: Borislav Petkov <[email protected]>

This driver is Intel-only so loading on anything which is not Intel is pointless. Prevent it from doing so.

While at it, correct the "not supported" print statement to say CPU "model" which is what that test does.

Cc Len Brown <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
CC Srinivas Pandruvada <[email protected]>
Cc Viresh Kumar <[email protected]>
---
 drivers/cpufreq/intel_pstate.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/cpufreq/intel_pstate.c +4 -1
index b599c7318aab..2986119dd31f 100644
@@ -2596,6 +2596,9 @@ static int __init intel_pstate_init(void)
 	const struct x86_cpu_id *id;
 	int rc;
 
+	if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
+		return -ENODEV;
+
 	if (no_load)
 		return -ENODEV;
 
@@ -2611,7 +2614,7 @@ static int __init intel_pstate_init(void)
 	} else {
 		id = x86_match_cpu(intel_pstate_cpu_ids);
 		if (!id) {
-			pr_info("CPU ID not supported\n");
+			pr_info("CPU model not supported\n");
 			return -ENODEV;
 		}
 
Signature · 8 lines
2.21.0


-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.
#4 lore raw

Re: [PATCH] cpufreq/intel_pstate: Load only on Intel hardware

Le 01/04/2019 à 17:03, Borislav Petkov a écrit :

Quoted 8 lines

From: Borislav Petkov <[email protected]>

This driver is Intel-only so loading on anything which is not Intel is pointless. Prevent it from doing so.

While at it, correct the "not supported" print statement to say CPU "model" which is what that test does.

Looks good to me !

Thanks !

Erwan,

Reviewed-by on v1 1/1
#5 lore raw

Re: [PATCH] cpufreq/intel_pstate: Load only on Intel hardware

In reply to Borislav Petkov
Quoted 4 lines · CEST Borislav Petkov

From: Borislav Petkov <[email protected]>

This driver is Intel-only so loading on anything which is not Intel is pointless. Prevent it from doing so.

Nice.
I wondered whether there are more of these to find by review, instead of waiting for the next message to show up.

I ended up in the "not so straight forward" IOMMU init macros... and continued with daily work again.

Anyway there are a lot files showing up when grepping the kernel for intel files/drivers, maybe someone who is involved in the one or other comes up with something similar...

Thomas

FWIW:

Quoted 25 lines

While at it, correct the "not supported" print statement to say CPU "model" which is what that test does.

Cc Len Brown <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
CC Srinivas Pandruvada <[email protected]>
Cc Viresh Kumar <[email protected]>
---
 drivers/cpufreq/intel_pstate.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/cpufreq/intel_pstate.c +3 -0
index b599c7318aab..2986119dd31f 100644
@@ -2596,6 +2596,9 @@ static int __init intel_pstate_init(void)
 	const struct x86_cpu_id *id;
 	int rc;
 
+	if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
+		return -ENODEV;
+

...

#6 lore raw

Re: [PATCH] cpufreq/intel_pstate: Load only on Intel hardware

In reply to Borislav Petkov
Quoted 35 lines · CEST Borislav Petkov

From: Borislav Petkov <[email protected]>

This driver is Intel-only so loading on anything which is not Intel is pointless. Prevent it from doing so.

While at it, correct the "not supported" print statement to say CPU "model" which is what that test does.

Cc Len Brown <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
CC Srinivas Pandruvada <[email protected]>
Cc Viresh Kumar <[email protected]>
---
 drivers/cpufreq/intel_pstate.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/cpufreq/intel_pstate.c +3 -0
index b599c7318aab..2986119dd31f 100644
@@ -2596,6 +2596,9 @@ static int __init intel_pstate_init(void)
 	const struct x86_cpu_id *id;
 	int rc;
 
+	if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
+		return -ENODEV;
+
 	if (no_load)
 		return -ENODEV;
 

drivers/cpufreq/intel_pstate.c static int __init intel_pstate_init(void) · line 2620

Quoted 8 lines
drivers/cpufreq/intel_pstate.c +1 -1
@@ -2611,7 +2614,7 @@ static int __init intel_pstate_init(void)
 	} else {
 		id = x86_match_cpu(intel_pstate_cpu_ids);
 		if (!id) {
-			pr_info("CPU ID not supported\n");
+			pr_info("CPU model not supported\n");
 			return -ENODEV;
 		}
 

Applied, thanks!