Disabling Intel turbo on non IDA featured processor generally correct?

Disabling Intel turbo on non IDA featured processor generally correct?

[Thomas Renninger]

[-- Attachment #1: Type: text/plain, Size: 422 bytes --]

Hello Srinivas,

Since patch:
ac4e04d9e378f5aa826c2406ad7871ae1b6a6fb9

there is a user reporting turbo not working, while it worked before and
he claims it works on Windows 11 as well.
No related BIOS option avail.

The patch mentions Skylake-X systems, while the user seem to have another CPU:
https://bugzilla.opensuse.org/show_bug.cgi?id=1252385

Can someone from Intel please double check.

Thanks,

           Thomas

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: Disabling Intel turbo on non IDA featured processor generally correct?

[srinivas pandruvada]

Hi Thomas,

The BIOS shipped with system disabled turbo on boot and user had to
manually force via the sysfs. With the new change that is not an
option.

We are trying to contact the manufacturer, but no solution yet.


Thanks,
Srinivas




On Tue, 2025-10-28 at 13:47 +0100, Thomas Renninger wrote:
> Hello Srinivas,
> 
> Since patch:
> ac4e04d9e378f5aa826c2406ad7871ae1b6a6fb9
> 
> there is a user reporting turbo not working, while it worked before
> and
> he claims it works on Windows 11 as well.
> No related BIOS option avail.
> 
> The patch mentions Skylake-X systems, while the user seem to have
> another CPU:
> https://bugzilla.opensuse.org/show_bug.cgi?id=1252385
> 
> Can someone from Intel please double check.
> 
> Thanks,
> 
>            Thomas

Re: Disabling Intel turbo on non IDA featured processor generally correct?

[Thomas Renninger]

[-- Attachment #1: Type: text/plain, Size: 984 bytes --]

On Mittwoch, 5. November 2025 14:06:42 Mitteleuropäische Normalzeit srinivas 
pandruvada wrote:
> Hi Thomas,
> 
> The BIOS shipped with system disabled turbo on boot and user had to
> manually force via the sysfs. With the new change that is not an
> option.

User (adding EZ, sorry for missing you out on initial post)
claims that cpuid shows:
So I ran the cpuid commend and it returned:
Thermal and Power Management Features (6):
      digital thermometer                     = true
      Intel Turbo Boost Technology            = true
 ...

Not sure whether this really is the same bit and whether this is prove that 
IDA feature bit has been switched by Linux OS (by SMI/firmware?) after 
initializing?

IDA cpufeature bit is not set in /proc/cpuinfo

EZ probably can do the one or other msr read/write if you need someone tests.

Good luck,

         Thomas
> 
> We are trying to contact the manufacturer, but no solution yet.

That is appreciated!

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 525 bytes --]

Re: Disabling Intel turbo on non IDA featured processor generally correct?

[srinivas pandruvada]

[-- Attachment #1: Type: text/plain, Size: 1480 bytes --]

On Wed, 2025-11-05 at 15:55 +0100, Thomas Renninger wrote:
> On Mittwoch, 5. November 2025 14:06:42 Mitteleuropäische Normalzeit
> srinivas 
> pandruvada wrote:
> > Hi Thomas,
> > 
> > The BIOS shipped with system disabled turbo on boot and user had to
> > manually force via the sysfs. With the new change that is not an
> > option.
> 
> User (adding EZ, sorry for missing you out on initial post)
> claims that cpuid shows:
> So I ran the cpuid commend and it returned:
> Thermal and Power Management Features (6):
>       digital thermometer                     = true
>       Intel Turbo Boost Technology            = true
>  ...
> 
> Not sure whether this really is the same bit and whether this is
> prove that 
> IDA feature bit has been switched by Linux OS (by
> SMI/firmware?)Fixes: 0fb5be7fea98 ("cpufreq: intel_pstate: Unchecked
> MSR aceess in legacy mode") after 
> initializing?
> 
> IDA cpufeature bit is not set in /proc/cpuinfo
> 
> EZ probably can do the one or other msr read/write if you need
> someone tests.
> 
If the BIOS does it correctly there is a way to know as the limits can
be dynamic.

Here probably we need to reevaluate again. Try the attached change and
check if this brings back the old hack.

Thanks,
Srinivas


> Good luck,
> 
>          Thomas
> > 
> > We are trying to contact the manufacturer, but no solution yet.
> 
> That is appreciated!

[-- Attachment #2: 0001-cpufreq-intel_pstate-Reevaluate-IDA-presence-on-no_t.patch --]
[-- Type: text/x-patch, Size: 1494 bytes --]

From 405d27e871f7bc85a786f84877a35da54c813b39 Mon Sep 17 00:00:00 2001
From: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Date: Wed, 5 Nov 2025 09:57:03 -0800
Subject: [PATCH] cpufreq: intel_pstate: Reevaluate IDA presence on no_turbo
 attribute change

If hardware disabled IDA (Intel Dynamic Acceleration technology) feature
before OS boot, turbo mode support will be disabled permanently. In this
case CPUID.06H: EAX[1] reports 0 and attribute
"/sys/devices/system/cpu/intel_pstate/no_turbo" will show "1" and status
can't be changed to "0".

When no_turbo is written with 0, in this case evaluate CPUID.06H: EAX[1]
again. If the feature status is changed to 1 post OS boot then allow to
enable turbo mode.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
---
 drivers/cpufreq/intel_pstate.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c
index 43e847e9f741..0ec45a610b45 100644
--- a/drivers/cpufreq/intel_pstate.c
+++ b/drivers/cpufreq/intel_pstate.c
@@ -596,10 +596,15 @@ static void intel_pstate_hybrid_hwp_adjust(struct cpudata *cpu)
 
 static bool turbo_is_disabled(void)
 {
+	unsigned int eax, ebx, ecx, edx;
 	u64 misc_en;
 
-	if (!cpu_feature_enabled(X86_FEATURE_IDA))
+	eax = 0;
+	cpuid(6, &eax, &ebx, &ecx, &edx);
+	if (!(eax & BIT(1))) {
+		pr_info("Turbo is disabled\n");
 		return true;
+	}
 
 	rdmsrl(MSR_IA32_MISC_ENABLE, misc_en);
 
-- 
2.43.0