New: REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
Subsystems (2)
https://bugzilla.kernel.org/show_bug.cgi?id=221505
Bug ID: 221505
Summary: REGRESSION][amd_pstate] "EPP cannot be set under
performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
Product: Power Management
Version: 2.5
Hardware: AMD
OS: Linux
Status: NEW
Severity: normal
Priority: P3
Component: cpufreq
Assignee: [email protected]
Reporter: [email protected]
Regression: No
Created attachment 310089
--> https://bugzilla.kernel.org/attachment.cgi?id=310089&action=edit
dmesg
Hardware: ASUS TUF Gaming A14 (2026) FA401EA
CPU: AMD Ryzen AI Max+ 932 (Strix Halo platform)
Kernel: 7.1.0-mainline (including patches for Bugzilla #221310) Power Daemon: power-profiles-daemon 0.30
Description:
On kernel 7.1, power-profiles-daemon fails to activate the "performance" profile. The D-Bus error returned is Code 26 (Device or resource busy).
After enabling dynamic debug for the amd_pstate module (echo 'module amd_pstate +p' > /sys/kernel/debug/dynamic_debug/control), I captured the following reason for the failure in dmesg:
[ 299.585428] amd_pstate: EPP cannot be set under performance policy
Steps to Reproduce:
Use a system with amd_pstate in active mode on kernel 7.1.
Attempt to switch profile: powerprofilesctl set performance.
Failed to communicate with power-profiles-daemon: g-io-error-quark:
GDBus.Error:org.gtk.GDBus.UnmappedGError.Quark._g_2dio_2derror_2dquark.Code26:
Failed to activate CPU driver 'amd_pstate': Error writing '/sys/devices/system/cpu/cpufreq/policy11/energy_performance_preference':
Device or resource busy (26)
The command fails, and dmesg logs: amd_pstate: EPP cannot be set under performance policy.
Expected Result:
The driver should allow setting the EPP value even if the policy is set to performance, or handle the transition gracefully without returning EBUSY to user-space daemons.
Additional Context:
This issue was observed on the new Strix Halo (Ryzen AI Max+) platform. The conflict between PPD's profile orchestration and the driver's internal state checks makes the "Performance" profile unusable via standard desktop tools (KDE/GNOME power settings).
I can build and test custom patches on this hardware
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #1 from Artem S. Tashkinov ([email protected]) --- *** Bug 221506 has been marked as a duplicate of this bug. ***
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected], | |[email protected]
--- Comment #2 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- Adding Mario – and also wondering if this is related to
https://lore.kernel.org/linux-pm/[email protected]/
and https://bugzilla.kernel.org/show_bug.cgi?id=221473 , but it sounds a bit
like this is something different.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #3 from Mario Limonciello (AMD) ([email protected]) --- Did you set CONFIG_X86_AMD_PSTATE_DYNAMIC_EPP
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #4 from Sergey ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #3)
Quoted 1 line
Did you set
CONFIG_X86_AMD_PSTATE_DYNAMIC_EPP
After enabling CONFIG_X86_AMD_PSTATE_DYNAMIC_EPP=y, power-profiles-daemon (PPD) loses the ability to switch to the power-saver profile. Attempting to set it via powerprofilesctl set power-saver results in the system defaulting back to balanced.
Technical details:
Governor limitation:
cat /sys/devices/system/cpu/cpufreq/policy0/scaling_available_governors returns only [performance, powersave].
PPD Failure:
Since only two governors are visible, PPD fails to map the "power-saver" profile correctly, even though the underlying hardware supports it.
Manual Override:
I discovered that the only way to successfully force the "power-saver" state is by writing directly to the platform profile:
echo "quiet" | sudo tee /sys/devices/platform/asus-nb-wmi/platform-profile/platform-profile-25/profile (Note: The path may vary, but the "quiet" hint works where PPD fails).
Steps to Reproduce:
Enable CONFIG_X86_AMD_PSTATE_DYNAMIC_EPP=y.
Try to set power-saver: powerprofilesctl set power-saver.
Check status: powerprofilesctl get returns balanced.
Expected Result:
PPD should recognize that in active/dynamic_epp mode, "power-saver" should map to the powersave governor with an EPP hint of power (or quiet at the platform level)
Actual Result:
PPD is stuck between two profiles (balanced/performance)
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #5 from Mario Limonciello (AMD) ([email protected]) --- Yes - this is expected behavior with the series right now. I am proposing to drop the kconfig and make the dynamic epp mode opt in at runtiem or kernel command line.
https://lore.kernel.org/linux-pm/[email protected]/
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #6 from Sergey ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #5)
Quoted 6 lines
Yes - this is expected behavior with the series right now. I am proposing to drop the kconfig and make the dynamic epp mode opt in at runtiem or kernel command line.
https://lore.kernel.org/linux-pm/20260512221947.1652988-1-mario.
[email protected]/
I’ve tested the proposed patch on my Ryzen AI Max+ (Strix Halo) system (ASUS FA401EA), and here are the results:
With amd_dynamic_epp=enable:
The issue described in the Comment 4.
With amd_dynamic_epp disabled (or no parameter):
The issue described in the Description
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #7 from Mario Limonciello (AMD) ([email protected]) --- Can you please pair that patch with the patches on this branch?
https://git.kernel.org/pub/scm/linux/kernel/git/superm1/linux.git/log/?h=amd-pstate-fixes
There are multiple bugs at play here.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #8 from Sergey ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #7)
Quoted 5 lines
Can you please pair that patch with the patches on this branch?
https://git.kernel.org/pub/scm/linux/kernel/git/superm1/linux.git/log/?h=amd-
pstate-fixesThere are multiple bugs at play here.
I’ve tested the kernel from your amd-pstate-fixes branch (commit f9f16835d4dc), but unfortunately, the issues persist, and I’ve discovered a new conflict with the ASUS platform driver.
Current observations:
With amd_dynamic_epp=disable (or no parameter):
The behavior remains identical to the initial report. powerprofilesctl still fails with the error described in the original report (Description).
With amd_dynamic_epp=enable:
It breaks the integration with asus-nb-wmi.
Specifically, the /sys/devices/platform/asus-nb-wmi/throttle_thermal_policy (which controls ASUS thermal profiles) stops switching to value 2 (Silent/Power-save mode).
When I set "Power-saver" via PPD, it only toggles between 0 and 1.
The CPU TDP is not limited to its expected ~55W threshold unless I manually write 2 to the throttle_thermal_policy file.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #9 from Mario Limonciello (AMD) ([email protected]) ---
Quoted 1 line
With amd_dynamic_epp=enable:
I'm not sure why this is happening. What platform profiles does this other driver support?
Quoted 1 line
With amd_dynamic_epp=disable (or no parameter):
So you're still seeing this?
[ 299.585428] amd_pstate: EPP cannot be set under performance policy
Can I see the kernel log with dynamic debug for amd-pstate turned on? And also the output from the amd-pstate triage tool in amd-debug-tools.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #10 from Sergey ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #9)
Quoted 5 lines
Quoted 1 line
With amd_dynamic_epp=enable:
I'm not sure why this is happening. What platform profiles does this other driver support?
Quoted 8 lines
Quoted 1 line
With amd_dynamic_epp=disable (or no parameter):
So you're still seeing this?
[ 299.585428] amd_pstate: EPP cannot be set under performance policyCan I see the kernel log with dynamic debug for amd-pstate turned on? And also the output from the amd-pstate triage tool in amd-debug-tools.
With amd_dynamic_epp=enable:
cat /sys/devices/platform/asus-nb-wmi/platform-profile/platform-profile-25/name asus-wmi
cat /sys/devices/platform/asus-nb-wmi/platform-profile/platform-profile-25/profile balanced
cat /sys/devices/platform/asus-nb-wmi/platform-profile/platform-profile-25/choices quiet balanced performance
With amd_dynamic_epp=disable (or no parameter):
[ 128.393895] amd_pstate: EPP cannot be set under performance policy
🐧 Arch Linux
🐧 Kernel: 7.1.0-rc2-1-mainline-gf9f16835d4dc-dirty
○ 'status': active
○ 'prefcore': enabled
🐧 ITMT: Y
💻 CPU: AMD RYZEN AI MAX+ 392 w/ Radeon 8060S
🔋 Per-CPU sysfs files
+---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------+
| CPU # | CPU Min Freq | CPU Nonlinear Freq | CPU Max Freq | Scaling
Min Freq | Scaling Max Freq | Energy Performance Preference | Prefcore |
Boost |
|---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------|
| 0 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 206 |
0 |
| 1 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 216 |
0 |
| 2 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 201 |
0 |
| 3 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 216 |
0 |
| 4 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 196 |
0 |
| 5 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 211 |
0 |
| 6 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 176 |
0 |
| 7 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 166 |
0 |
| 8 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 186 |
0 |
| 9 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 181 |
0 |
| 10 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 171 |
0 |
| 11 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 191 |
0 |
| 12 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 206 |
0 |
| 13 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 216 |
0 |
| 14 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 201 |
0 |
| 15 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 216 |
0 |
| 16 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 196 |
0 |
| 17 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 211 |
0 |
| 18 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 176 |
0 |
| 19 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 166 |
0 |
| 20 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 186 |
0 |
| 21 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 181 |
0 |
| 22 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 171 |
0 |
| 23 | 609334 | 1980334 | 3199000 |
609334 | 3199000 | power | 191 |
0 |
+---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------+
🔋 CPPC MSRs
+---------+----------+----------+--------------------+---------+------------+
| CPU # | Enable | Status | Cap 1 | Cap 2 | Request |
|---------+----------+----------+--------------------+---------+------------|
| 0 | 1 | 0 | 0xce694114ce694114 | 0x0 | 0xff001569 |
| 1 | 1 | 0 | 0xd8694114d8694114 | 0x0 | 0xff001569 |
| 2 | 1 | 0 | 0xc9694114c9694114 | 0x0 | 0xff001569 |
| 3 | 1 | 0 | 0xd8694114d8694114 | 0x0 | 0xff001569 |
| 4 | 1 | 0 | 0xc4694114c4694114 | 0x0 | 0xff001569 |
| 5 | 1 | 0 | 0xd3694114d3694114 | 0x0 | 0xff001569 |
| 6 | 1 | 0 | 0xb0694114b0694114 | 0x0 | 0xff001569 |
| 7 | 1 | 0 | 0xa6694114a6694114 | 0x0 | 0xff001569 |
| 8 | 1 | 0 | 0xba694114ba694114 | 0x0 | 0xff001569 |
| 9 | 1 | 0 | 0xb5694114b5694114 | 0x0 | 0xff001569 |
| 10 | 1 | 0 | 0xab694114ab694114 | 0x0 | 0xff001569 |
| 11 | 1 | 0 | 0xbf694114bf694114 | 0x0 | 0xff001569 |
| 12 | 1 | 0 | 0xce694114 | 0x0 | 0xff001569 |
| 13 | 1 | 0 | 0xd8694114 | 0x0 | 0xff001569 |
| 14 | 1 | 0 | 0xc9694114 | 0x0 | 0xff001569 |
| 15 | 1 | 0 | 0xd8694114 | 0x0 | 0xff001569 |
| 16 | 1 | 0 | 0xc4694114 | 0x0 | 0xff001569 |
| 17 | 1 | 0 | 0xd3694114 | 0x0 | 0xff001569 |
| 18 | 1 | 0 | 0xb0694114 | 0x0 | 0xff001569 |
| 19 | 1 | 0 | 0xa6694114 | 0x0 | 0xff001569 |
| 20 | 1 | 0 | 0xba694114 | 0x0 | 0xff001569 |
| 21 | 1 | 0 | 0xb5694114 | 0x0 | 0xff001569 |
| 22 | 1 | 0 | 0xab694114 | 0x0 | 0xff001569 |
| 23 | 1 | 0 | 0xbf694114 | 0x0 | 0xff001569 |
+---------+----------+----------+--------------------+---------+------------+
🔋 MSR_AMD_CPPC_CAP1 (decoded)
+---------+---------------+------------------+----------------+----------------+
| CPU # | Lowest Perf | Nonlinear Perf | Nominal Perf | Highest Perf
|
|---------+---------------+------------------+----------------+----------------|
| 0 | 20 | 65 | 105 | 206
|
| 1 | 20 | 65 | 105 | 216
|
| 2 | 20 | 65 | 105 | 201
|
| 3 | 20 | 65 | 105 | 216
|
| 4 | 20 | 65 | 105 | 196
|
| 5 | 20 | 65 | 105 | 211
|
| 6 | 20 | 65 | 105 | 176
|
| 7 | 20 | 65 | 105 | 166
|
| 8 | 20 | 65 | 105 | 186
|
| 9 | 20 | 65 | 105 | 181
|
| 10 | 20 | 65 | 105 | 171
|
| 11 | 20 | 65 | 105 | 191
|
| 12 | 20 | 65 | 105 | 206
|
| 13 | 20 | 65 | 105 | 216
|
| 14 | 20 | 65 | 105 | 201
|
| 15 | 20 | 65 | 105 | 216
|
| 16 | 20 | 65 | 105 | 196
|
| 17 | 20 | 65 | 105 | 211
|
| 18 | 20 | 65 | 105 | 176
|
| 19 | 20 | 65 | 105 | 166
|
| 20 | 20 | 65 | 105 | 186
|
| 21 | 20 | 65 | 105 | 181
|
| 22 | 20 | 65 | 105 | 171
|
| 23 | 20 | 65 | 105 | 191
|
+---------+---------------+------------------+----------------+----------------+
🔋 MSR_AMD_CPPC_REQ (decoded)
+---------+------------+------------+----------------+---------------------------+
| CPU # | Min Perf | Max Perf | Desired Perf | Energy Performance
Perf |
|---------+------------+------------+----------------+---------------------------|
| 0 | 21 | 105 | 0 |
255 |
| 1 | 21 | 105 | 0 |
255 |
| 2 | 21 | 105 | 0 |
255 |
| 3 | 21 | 105 | 0 |
255 |
| 4 | 21 | 105 | 0 |
255 |
| 5 | 21 | 105 | 0 |
255 |
| 6 | 21 | 105 | 0 |
255 |
| 7 | 21 | 105 | 0 |
255 |
| 8 | 21 | 105 | 0 |
255 |
| 9 | 21 | 105 | 0 |
255 |
| 10 | 21 | 105 | 0 |
255 |
| 11 | 21 | 105 | 0 |
255 |
| 12 | 21 | 105 | 0 |
255 |
| 13 | 21 | 105 | 0 |
255 |
| 14 | 21 | 105 | 0 |
255 |
| 15 | 21 | 105 | 0 |
255 |
| 16 | 21 | 105 | 0 |
255 |
| 17 | 21 | 105 | 0 |
255 |
| 18 | 21 | 105 | 0 |
255 |
| 19 | 21 | 105 | 0 |
255 |
| 20 | 21 | 105 | 0 |
255 |
| 21 | 21 | 105 | 0 |
255 |
| 22 | 21 | 105 | 0 |
255 |
| 23 | 21 | 105 | 0 |
255 |
+---------+------------+------------+----------------+---------------------------+Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
Rhys Edwards ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #11 from Rhys Edwards ([email protected]) --- Not sure if i'm adding much additional info here not already known but i also noticed this issue recently and believe it is also linked to the frequency limiting problem in https://bugzilla.kernel.org/show_bug.cgi?id=221473 as i have/had both issues
I've had cachyos installed using rc kernel on a desktop computer and last week or so noticed cpu temps/wattage/frequency was much lower than normal, frequency was specifically capping out at exactly 3Ghz regardless of workload.
I don't recall changing any pstate profiles at any point since i installed a few months ago so i assume it was on performance(?)
CPU: AMD Ryzen 7 5800x
Mobo: MSI x570 Tomahawk
Ram: 32GB DDR4 3600Mhz
GPU: AMD RX570 4GB
It seems mostly fixed when i switched back to balanced using powerprofilesctl, but now i also can't set back to performance getting the error mentioned here:
Failed to communicate with power-profiles-daemon: g-io-error-quark:
GDBus.Error:org.gtk.GDBus.UnmappedGError.Quark._g_2dio_2derror_2dquark.Code26:
Failed to activate CPU driver 'amd_pstate': Error writing '/sys/devices/system/cpu/cpufreq/policy11/energy_performance_preference':
Device or resource busy (26)
Here is my p-state triage if it's any use...
🐧 CachyOS
🐧 Kernel: 7.1.0-rc3-1-cachyos-rc
○ 'status': active
○ 'prefcore': enabled
🐧 ITMT: Y
💻 CPU: AMD Ryzen 7 5800X 8-Core Processor
🔋 Per-CPU sysfs files
+---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------+
| CPU # | CPU Min Freq | CPU Nonlinear Freq | CPU Max Freq | Scaling
Min Freq | Scaling Max Freq | Energy Performance Preference | Prefcore |
Boost |
|---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------|
| 0 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 176 |
1 |
| 1 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 196 |
1 |
| 2 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 186 |
1 |
| 3 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 181 |
1 |
| 4 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 171 |
1 |
| 5 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 191 |
1 |
| 6 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 166 |
1 |
| 7 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 196 |
1 |
| 8 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 176 |
1 |
| 9 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 196 |
1 |
| 10 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 186 |
1 |
| 11 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 181 |
1 |
| 12 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 171 |
1 |
| 13 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 191 |
1 |
| 14 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 166 |
1 |
| 15 | 555531 | 1754308 | 4853585 |
1754308 | 4853585 | balance_performance | 196 |
1 |
+---------+----------------+----------------------+----------------+--------------------+--------------------+---------------------------------+------------+---------+
🔋 CPPC MSRs
+---------+----------+----------+---------+---------+-----------+
| CPU # | Enable | Status | Cap 1 | Cap 2 | Request |
|---------+----------+----------+---------+---------+-----------|
| 0 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 1 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 2 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 3 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 4 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 5 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 6 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 7 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 8 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 9 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 10 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 11 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 12 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 13 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 14 | 0 | 0 | 0x0 | 0x0 | 0x0 |
| 15 | 0 | 0 | 0x0 | 0x0 | 0x0 |
+---------+----------+----------+---------+---------+-----------+
🔋 MSR_AMD_CPPC_CAP1 (decoded)
+---------+---------------+------------------+----------------+----------------+
| CPU # | Lowest Perf | Nonlinear Perf | Nominal Perf | Highest Perf
|
|---------+---------------+------------------+----------------+----------------|
| 0 | 0 | 0 | 0 | 0
|
| 1 | 0 | 0 | 0 | 0
|
| 2 | 0 | 0 | 0 | 0
|
| 3 | 0 | 0 | 0 | 0
|
| 4 | 0 | 0 | 0 | 0
|
| 5 | 0 | 0 | 0 | 0
|
| 6 | 0 | 0 | 0 | 0
|
| 7 | 0 | 0 | 0 | 0
|
| 8 | 0 | 0 | 0 | 0
|
| 9 | 0 | 0 | 0 | 0
|
| 10 | 0 | 0 | 0 | 0
|
| 11 | 0 | 0 | 0 | 0
|
| 12 | 0 | 0 | 0 | 0
|
| 13 | 0 | 0 | 0 | 0
|
| 14 | 0 | 0 | 0 | 0
|
| 15 | 0 | 0 | 0 | 0
|
+---------+---------------+------------------+----------------+----------------+
🔋 MSR_AMD_CPPC_REQ (decoded)
+---------+------------+------------+----------------+---------------------------+
| CPU # | Min Perf | Max Perf | Desired Perf | Energy Performance
Perf |
|---------+------------+------------+----------------+---------------------------|
| 0 | 0 | 0 | 0 |
0 |
| 1 | 0 | 0 | 0 |
0 |
| 2 | 0 | 0 | 0 |
0 |
| 3 | 0 | 0 | 0 |
0 |
| 4 | 0 | 0 | 0 |
0 |
| 5 | 0 | 0 | 0 |
0 |
| 6 | 0 | 0 | 0 |
0 |
| 7 | 0 | 0 | 0 |
0 |
| 8 | 0 | 0 | 0 |
0 |
| 9 | 0 | 0 | 0 |
0 |
| 10 | 0 | 0 | 0 |
0 |
| 11 | 0 | 0 | 0 |
0 |
| 12 | 0 | 0 | 0 |
0 |
| 13 | 0 | 0 | 0 |
0 |
| 14 | 0 | 0 | 0 |
0 |
| 15 | 0 | 0 | 0 |
0 |
+---------+------------+------------+----------------+---------------------------+Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #12 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- Is this still an issue with 7.1-rc5 after various fixes were mainlined in the
past two weeks?
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #13 from Alex Ungur ([email protected]) ---
I use 7.1-rc5 and I can switch just fine to balanced or power-saver but not to performance:
```
$ powerprofilesctl
performance:
CpuDriver: amd_pstate
PlatformDriver: platform_profile
Degraded: no
* balanced:
CpuDriver: amd_pstate
PlatformDriver: platform_profile
power-saver:
CpuDriver: amd_pstate
PlatformDriver: platform_profile
$ powerprofilesctl set power-saver # ok
$ powerprofilesctl set balanced # ok
$ powerprofilesctl set performance # not ok
Failed to communicate with power-profiles-daemon: g-io-error-quark:
GDBus.Error:org.gtk.GDBus.UnmappedGError.Quark._g_2dio_2derror_2dquark.Code26:
Failed to activate CPU driver 'amd_pstate': Error writing
'/sys/devices/system/cpu/cpufreq/policy11/energy_performance_preference':
Resource busy (26)
```It worked fine with rc4 though... I could switch to any profile, incl. performance.
My machine is:
```
$ inxi -CM
Machine:
Type: Laptop System: ASUSTeK product: ASUS Vivobook S 16 M5606WA_M5606WA
v: 1.0 serial: <superuser required>
Mobo: ASUSTeK model: M5606WA v: 1.0 serial: <superuser required>
UEFI: American Megatrends LLC. v: M5606WA.316 date: 01/05/2026
CPU:
Info: 12-core model: AMD Ryzen AI 9 HX 370 w/ Radeon 890M bits: 64
type: MT MCP cache: L2: 12 MiB
Speed (MHz): avg: 3585 min/max: 605/5158:3289 cores: 1: 3585 2: 3585
3: 3585 4: 3585 5: 3585 6: 3585 7: 3585 8: 3585 9: 3585 10: 3585 11: 3585
12: 3585 13: 3585 14: 3585 15: 3585 16: 3585 17: 3585 18: 3585 19: 3585
20: 3585 21: 3585 22: 3585 23: 3585 24: 3585
```Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #14 from Mario Limonciello (AMD) ([email protected]) --- This is with dynamic epp off you're hitting that on 7.1-rc5?
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #15 from Alex Ungur ([email protected]) --- I believe it's off:
``` $ cat /sys/devices/system/cpu/amd_pstate/status active $ cat /sys/devices/system/cpu/amd_pstate/dynamic_epp disabled $ cat /sys/devices/system/cpu/amd_pstate/prefcore enabled ```
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
Axel Rank ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #16 from Axel Rank ([email protected]) ---
Yes I can confirm this bug since 7.1-rc2 and still in 7.1-rc5 Additional there is a shutdown bug since then. System doesn't power off completely.
CPU: AMD Ryzen 9 7940HS
GPU: AMD Radeon 780M Graphics
Host: VivoBook_ASUSLaptop M1605XA
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #17 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- Mario, what's the status here? I know, you and your colleagues already did a
lot of work to improve things that came up here, but it looks like we still
have one important problem left with no solution in sight. Or did I miss
anything (which is pretty likely from my external point)?
Quoted 1 line
since 7.1-rc2 and still in 7.1-rc5
Given that the problems Alex' had to deal with changed over time I wonder (a) if you are really seeing the exact same thing (b) if this bug slowly gets too confusing and if it might be wise to start new, separate ones.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #18 from Mario Limonciello (AMD) ([email protected]) --- This is a resource availability issue (me).
I haven't had time to make any progress on this, so yes sorry no solution in sight yet for the performance mode issue right now.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #19 from Sergey ([email protected]) ---
(In reply to The Linux kernel's regression tracker (Thorsten Leemhuis) from comment #17)
Quoted 10 lines
Mario, what's the status here? I know, you and your colleagues already did a lot of work to improve things that came up here, but it looks like we still have one important problem left with no solution in sight. Or did I miss anything (which is pretty likely from my external point)?
Quoted 1 line
since 7.1-rc2 and still in 7.1-rc5
Given that the problems Alex' had to deal with changed over time I wonder
(a) if you are really seeing the exact same thing (b) if this bug slowly gets too confusing and if it might be wise to start new, separate ones.
I don't know if this will help, but I'm having this problem on kernel 7.1-rc1. I tried using git bisect to find the problematic commit, but it's 13 iterations long and sometimes the kernel won't start, so I gave up. If you have a list of suspicious commits, I could search for the problematic one this weekend if that helps.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #20 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- (In reply to Sergey from comment #19)
Quoted 2 lines
I don't know if this will help, but I'm having this problem on kernel 7.1-rc1.
Quite a few problems were fixed since then and a bisection likely a waste of time. Check if 7.1-rc5 works better and report back please.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #21 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #18)
Quoted 3 lines
This is a resource availability issue (me).
I haven't had time to make any progress on this, so yes sorry no solution in sight yet for the performance mode issue right now.
Understood, that's how it is sometimes. Thing is: 7.1 final is not far away any more, so allow me to ask (sorry, that's why I'm here for; I really don't want to make your life harder):
I assume "revert one or a handful changes now to resolve this for 7.1 [and reapply them later in improved form]" is not a option here, right?
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #22 from Mario Limonciello (AMD) ([email protected]) ---
Quoted 3 lines
Understood, that's how it is sometimes. Thing is: 7.1 final is not far away any more, so allow me to ask (sorry, that's why I'm here for; I really don't want to make your life harder):
I know; I really want to get to this. If I don't; I'll get something for 7.2 and we can backport it. It's certainly still on my radar.
Quoted 2 lines
I assume "revert one or a handful changes now to resolve this for 7.1 [and reapply them later in improved form]" is not a option here, right?
If I understood the root cause I would have a better answer for you.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #23 from Sergey ([email protected]) ---
(In reply to The Linux kernel's regression tracker (Thorsten Leemhuis) from comment #20)
Quoted 6 lines
(In reply to Sergey from comment #19)
Quoted 2 lines
I don't know if this will help, but I'm having this problem on kernel 7.1-rc1.
Quite a few problems were fixed since then and a bisection likely a waste of time. Check if 7.1-rc5 works better and report back please.
The problem on 7.1-rc5 is repeated
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #24 from The Linux kernel's regression tracker (Thorsten Leemhuis) ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #22)
Quoted 1 line
If I understood the root cause I would have a better answer for you.
One more question then, if I may: does a bisection make sense, or are things just to messy/complicated due to the other fixes in this area (or if, how to best approach one given those fixes? apply them at each step without committing them?)?
Side note: Alex, Sergey, you both seems to have Asus laptops, wondering if that's part of the problem here.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #25 from Sergey ([email protected]) ---
(In reply to The Linux kernel's regression tracker (Thorsten Leemhuis) from comment #24)
Quoted 10 lines
(In reply to Mario Limonciello (AMD) from comment #22)
Quoted 1 line
If I understood the root cause I would have a better answer for you.
One more question then, if I may: does a bisection make sense, or are things just to messy/complicated due to the other fixes in this area (or if, how to best approach one given those fixes? apply them at each step without committing them?)?
Side note: Alex, Sergey, you both seems to have Asus laptops, wondering if that's part of the problem here.
I can't say for sure, but I can test the 7.1-rc5 kernel on a mini PC with an AMD Ryzen 9 7945HX this weekend and report back here if there are any issues.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #26 from Alex Ungur ([email protected]) ---
If there's any additional information I can provide about my laptop, happy to share, just let me know.
I can also try building and rc4 kernel over the weekend, as I 100% remember it working. If that indeed works, wouldn't it make the bisection simpler?
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
Rhys Edwards ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #27 from Rhys Edwards ([email protected]) --- Just updated to 7.1-rc5 as part of usual CachyOS update, can confirm issue is not fixed still exactly the same as before, pretty sure it started at rc1 or rc2 but unsure. If i swap back to the 7.0 kernel using the kernel manager performance mode works fine again, if i set my mode to performance in 7.0 and then boot back to 7.1-rc i get that same weird half broken frequency limiting thing where it won't go past 3.0Ghz until i go back to balanced, but then get the communication GDBus error trying to go back to performance again.
❯ cat /sys/devices/system/cpu/amd_pstate/status active
~
❯ cat /sys/devices/system/cpu/amd_pstate/dynamic_epp disabled
~
❯ cat /sys/devices/system/cpu/amd_pstate/prefcore enabled
This is a pretty standard desktop computer, no battery involved and nothing ASUS. Specs were listed in previous message. If there is anything else i can run to provide details let me know.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
Marco Scardovi ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #28 from Marco Scardovi ([email protected]) --- @Mario I took a look into it and made a quick patch. I hope that it can help mitigate the problem.
From 379f0641e4bd7b287f398e1798530b1b09c5ad0f Mon Sep 17 00:00:00 2001
From: Marco Scardovi <[email protected]>
Date: Sat, 30 May 2026 16:30:50 +0200
Subject: [PATCH 1/2] cpufreq/amd-pstate: Allow EPP=0 under performance policy
Under the cpufreq "performance" policy, the amd-pstate driver currently rejects all EPP writes with -EBUSY. However, writing the EPP value of AMD_CPPC_EPP_PERFORMANCE (0) is semantically consistent with the performance policy as EPP=0 corresponds to the most performance-oriented setting. Allowing EPP=0 writes under the performance policy prevents userspace power management utilities like power-profiles-daemon from receiving -EBUSY and failing when switching profiles.
Only reject non-zero EPP values when the policy is set to performance.
--- drivers/cpufreq/amd-pstate.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/cpufreq/amd-pstate.c +0 -0
@@ -1423,7 +1423,8 @@ ssize_t store_energy_performance_preference(struct
cpufreq_policy *policy,
epp = cpudata->epp_default_dc;
}- if (cpudata->policy == CPUFREQ_POLICY_PERFORMANCE) {
+ if (epp != AMD_CPPC_EPP_PERFORMANCE &&
+ cpudata->policy == CPUFREQ_POLICY_PERFORMANCE) {
pr_debug("EPP cannot be set under performance policy\n");
return -EBUSY;
}Signature · 7 lines
2.54.0 -- You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #29 from Alex Ungur ([email protected]) --- That patch works for me on 7.1-rc5 (AMD Ryzen AI 9 HX 370).
``` $ cat /sys/devices/system/cpu/amd_pstate/status active $ cat /sys/devices/system/cpu/amd_pstate/dynamic_epp disabled $ cat /sys/devices/system/cpu/amd_pstate/prefcore enabled ```
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #30 from Mario Limonciello (AMD) ([email protected]) --- Very similar to what I just posted earlier today.
https://lore.kernel.org/linux-pm/[email protected]/T/#u
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #31 from Marco Scardovi ([email protected]) --- (In reply to Mario Limonciello (AMD) from comment #30)
Quoted 4 lines
Very similar to what I just posted earlier today.
Oh sorry, I didn't notice it
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
--- Comment #32 from Mario Limonciello (AMD) ([email protected]) --- No problem, and thanks for debugging in parallel! Feel free to leave some tags for review and test on the submission.
Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
[Bug 221505] REGRESSION][amd_pstate] "EPP cannot be set under performance policy" (EBUSY) when switching to performance profile via PPD on 7.1
In reply to [email protected]https://bugzilla.kernel.org/show_bug.cgi?id=221505
Mario Limonciello (AMD) ([email protected]) changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
Assignee|[email protected] |[email protected]Signature · 4 lines
You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.