This version includes the following changes:
- Setting current base frequency as maximum for SST-BF with
kernel QOS changes
- Harmonize extended family decoded with the rest of the kernel
- Minor changes for error codes and messages
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When running intel-speed-select on unsupported CLX platforms, it prints
intel-speed-select: Invalid CPU model (85)
: Success
Because this is not a system error and errno is not set.
Replace err() with exit().
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When running an old version intel-speed-select tool on newer platforms,
even with "intel-speed-select -v", the tool only complains about
"Incompatible API version", without giving the current version info.
Print Version info whenever Incompatible API version is detected.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When running the "intel-speed-select -h" command, it returns
1. 0 when using a version that is API incompatible.
2. 1 when using a version that is API compatible.
And this is confusing.
Fix the program to return 0 for "-h" parameter, and return 1 whenever
"Incompatible API versions" is detected.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When decode and use CPU extended family ID in intel-speed-select, there
are several potential issues,
1. Mask with 0x0f to get CPU extended family ID is bogus because
CPU extended family ID takes 8 bits (bit 27:20).
2. Use CPU extended family ID fields without checking CPU family ID is
risky. Because Intel SDM says, "The Extended Family ID needs to be
examined only when the Family ID is 0FH."
3. Saving cpu family ID and cpu extended family ID separately doesn't
align with Linux kernel. And it may bring extra complexity when
making family specific changes in the future.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
SST-PP level change results in online/offline of CPUs with -o option.
The Linux intel-pstate driver internally stores the current HWP_REQ MSR
value during offline and restores them during online.
It is possible that during SST-PP level change, the new HWP_CAP limits
can be updated. So, when a CPU is online, the HWP_REQ MSR should be
updated to new values based on HWP_CAP values.
This is particularly problematic when either turbo is disabled or the
current HWP_REQ value (stored before online) is less than the base
frequency from the updated HWP_CAP MSR guaranteed value. If the HWP_REQ
MSR is not updated, then the performance will be limited to the value
before perf level change.
Hence the tool updates cpufreq scaling_max_freq to the newer
base_frequency value in this case. This step is not required when HWP
interrupts are enabled, as the perf level change should result in a new
interrupt with HWP_GUARANTEED_PERF_CHANGE_STATUS and the intel_pstate
driver will update to new limits.
But the tool needs to handle the case when HWP interrupts are not
enabled but there is no way for the tool to know that HWP interrupts are
enabled or not. So, it has to still update the scaling_max_freq.
With the QOS changes in the kernel, user space writes to scaling_max_freq
are treated as hard limits. So, when base frequency is increased with
SST-BF enabled, the cpufreq subsystem will still not allow setting to the
SST-BF high priority core frequency. So, the HWP_REQ MSR will still be
capped to the user-set scaling_max_freq after SST-PP level change.
To address this, instead of setting scaling_max_freq to the current HWP_CAP
highest frequency, set it to the maximum integer value to set the QOS limit
as unconstrained. In this case, the actual HWP_REQ maximum frequency will
still be capped to HWP_CAP highest performance by the intel-pstate driver.
So, it will not result in invalid HWP_REQ values.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version includes the following changes:
- Allow read only commands for non root users when permitted
- Fix file descriptor leak in isolate_cpus()
- Replace hardcoded libnl3 include path
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
The file descriptor opened in isolate_cpus() when (!level) is true was
not being closed before returning, causing a file descriptor leak in
both the error path and the success path.
When write() fails at line 950, the function returns at line 953 without
closing the file descriptor. Similarly, on success, the function returns
at line 956 without closing the file descriptor.
Add close(fd) calls before both return statements to fix the resource
leak. This follows the same pattern used elsewhere in the same function
where file descriptors are properly closed before returning (see lines
1005 and 1027).
Fixes: 997074df65 ("tools/power/x86/intel-speed-select: Use cgroup v2 isolation")
Signed-off-by: Malaya Kumar Rout <mrout@redhat.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Replace hardcoded libnl3 include path with pkg-config detection to
improve portability across different distributions and build environments.
The previous implementation used a fixed path constructed from the
compiler's sysroot, which could fail on systems with non-standard
library installations. Now the build system:
- Attempts to detect libnl-3.0 include paths using pkg-config
- Falls back to /usr/include/libnl3 if pkg-config is unavailable
- Maintains backward compatibility with existing build configurations
This ensures the tool builds correctly on a wider range of systems
while preserving existing behavior when pkg-config is not present.
Closes:https://bugzilla.kernel.org/show_bug.cgi?id=220819
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When permitted by the file /dev/isst_interface, allow to issue commands
for non root users.
When user id is non root, check if "/dev/isst_interface" can still be
opened. If this file can be opened, allow all read only commands.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version includes the following changes:
- Check feature status to check if the feature enablement was successful
- Reset SST-TF bucket structure to display valid bucket info
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
With SST-TF version 2 only 3 buckets are present. The information in
others buckets can be junk. So initialize the info structure of type
isst_turbo_freq_info, before issing ioctl to get bucket information.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
After change of enable/disable status of SST-CP, SST-TF and SST-BF
check if the hardware status change was successful. If not successful
even after retries, return failure.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version includes the following changes:
- Displays SST-PP2 revision fields.
- Skips updating uncore frequency limits on newer generations of CPUs.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
On SST PP level switch, skip adjusting the uncore frequency limit and
allow the hardware to handle this on newer platforms.
As newer generations of CPUs have changed the extended family identifier,
use this identifier to exclude the update.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version has fix for:
- Display of die ID and optimize array size for multi package
systems.
- Fix build warning with cross compiler
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This helps when using a cross-compiler for building intel-speed-select,
currently, its hardcoded to pick libnl3 headers from build host which may
not be same as build target when cross compiling.
cc -print-sysroot will print nothing if compiler is configured without
a sysroot and result in same string as it is now.
Fixes errors with gcc configured with host include poisoning e.g.
cc1: error: include location "/usr/include/libnl3" is unsafe for
cross-compilation [-Werror=poison-system-directories]
Signed-off-by: Khem Raj <raj.khem@gmail.com>
[ srinivas: Changelog edits for checkpatch warning ]
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Even when there is no die exported by CPUID leaf 0x1F, the kernel version
after 6.9 will show non zero die_id in the sysfs. In that case maximum
die_id can still match maximum power domain ID. So the condition to check
if the power domain ID is same a die_id to prevent duplicate display
doesn't hold true.
The better condition is to check if the maximum die_id is more than the
maximum package_id. If the die_id is exposed by CPUID leaf 0x1F, the
maximum die_id will be more than maximum package_id.
With this change tracking of max_punit_id is not used, so remove storing
max_punit_id.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
In the function for_each_online_power_domain_in_set() to pick one CPU
from each power domain a three-dimensional array is used, which assumes
that a package contains multiple dies, that means the die_id from
/sys/devices/system/cpu/cpu0/topology/die_id is only local to package.
If it is not unique, still there will be no functional issues in the
current generation of products, but the MAX_DIE_PER_PACKAGE will need to
be increased for future products with many packages.
After kernel version 6.9 die ID is unique system wide not per package.
Even if the CPU topology has no dies, the ID will still increment across
package. In this case the die_id in package 0 will be 0 and die_id in
package 1 will be 1 in a 2-package system.
Since the die count must be same for packages, just count the number of
dies in package 0 and calculate die index from
/sys/devices/system/cpu/cpu0/topology/die_id which is only unique within
a package.
In this way the array size
"int cpus[MAX_PACKAGE_COUNT][MAX_DIE_PER_PACKAGE][MAX_PUNIT_PER_DIE]"
doesn't have to increase with increasing package count.
No functional change is expected.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When SST-TF is disabled, the TRL (Turbo Ratio Limit) of config level 0
is getting restored. But the TRL of current level should be restored
which may not be config level 0.
This is caused by a bug in treating config level as TRL level. So
arguments needs to be swapped.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version addresses one issue:
- Fix updating TRL MSR after SST-TF is disabled in auto mode.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When SST-TF is disabled in auto mode, the performance is getting
limited.
This is caused by wrong programming of Turbo Ratio Limit (TRL) MSR.
This MSR always accepts the frequency ratio in 100 MHz unit. When the
TPMI is sending TRL in 1 MHz unit, change to 100 MHz, before updating
TRL MSR.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version addresses issues with:
- Support of SST BF/TF support per level
- Increase number of CPUs displayed
- Present all TRL levels for turbo-freq
- Fix display for unsupported levels
- Support multiple dies
- Increase die count
- Change CPU display for non compute domain
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
When there is no CPU in a power domain, display "None" instead of -1.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
SST BF and TF can be enabled/disabled per level. So check the current
level support from the mask of supported levels.
This change from a single level to mask for info.sst_tf_support and
info.sst_tf_support is indicated by API version change. Use as mask for
API version above 2. In this way there is no change in behavior when
running on older kernel with API version 2.
Since the tool can support now API version 3, update the supported API
version.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Currently max 128 CPUs can be displayed in the enable CPU list. Double
the range. Since the size is big for stack allocation, change to static.
Here changing to static is fine as these functions are called in serial.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
For turbo-freq feature, only 3 levels of frequencies are displayed even
if platform support more. Present all levels based on the CPU model.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
During call to "intel-speed-select turbo-freq info" some junk values are
reported for unsupported levels. Initialize the structure fact_info with
0s, so that isst_fact_display_information() will skip "0" values in the
frequency.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
When the die id is same as punit compute die ID, treat them same. In this
case, when for_each_online_power_domain_in_set() is called, then don't
loop for each punit in a die. Just loop for all punits in a package.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
TPMI platform information supports up to 16 compute dies. So increase the
range.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
This version addresses issues with:
- When CPU 0 hotplug is not possible, try cgroup v2 isolation
without any user input
- Fix turbo mode enable/disable swapped
- Sanitize command line integer and hex arguments
- Add more error messages
- Increase CPU count in one request
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
From kernel version 6.5, CPU 0 hotplug capability is deprecated.
If some SST profile doesn't have CPU 0, then it is no longer possible to
offline CPU 0. This means that user space threads will still run on
CPU 0.
To workaround this issue, use cgroup v2 isolation feature. Whenever there
/sys/devices/system/cpu/cpu0/online file is absent or open fails, isolate
CPU 0 via CPU cgroup v2 isolation. Also add a command line option to
force even if the /sys/devices/system/cpu/cpu0/online is present.
The previous commit "01bcb56f059e ("tools/power/x86/intel-speed-select:
Prevent CPU 0 offline") was just warning about this issue based on the
kernel version 6.5 and above. With this new approach, instead of warning
take action to mitigate the issue.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
With the increase in the CPU count, this count needs to be updated.
Increase max CPU count to 512.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When core-power is getting enabled, if the feaure is not supported,
display error.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
The command for turbo-mode enable and disable is swapped. Fix that.
Previously turbo-mode enable was actually disabling and disable was
enabling.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
TRL (turbo ratio limit) argument is passed in hex string. Clarify that
in the help.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
If the command takes some integer arguments, make sure the command
contains only digits. Same for Hex arguments. Otherwise return error.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version addresses issues with:
- CPU count display for power domain != 0
- Support more than 8 sockets
- Error on max CPU count exceeds in one request
- Prevent trying CPU 0 hotplug for kernel version 6.5 or later
- Change mem-frequency display to max-mem-frequency
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
The mem-frequency displayed by each profile is not the actual memory
frequency of DIMMs, but the maximum the CPU can support.
Change the mem-frequency field to max-mem-frequency.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Kernel 6.5 version deprecated CPU 0 hotplug. This will cause all
requests to fail to offline CPU 0. Check version number of kernel
and ignore CPU 0 hotplug request with debug aid to use cgroup
isolation feature for CPU 0.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
There is a limit on number of CPUs in one request. This is set to 256.
Currently tool silently ignores request for count over 256. Give an
error message to indicate this.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
MAX_PACKAGE_COUNT limits the intel-speed-select to systems with 8 sockets or fewer.
On a system with more than 8 sockets intel-speed-select silently ignores everything
beyond the 8th socket, rendering the tool useless for those systems.
Increase MAX_PACKAGE_COUNT to support systems with up to 32 sockets.
Signed-off-by: Frank Ramsay <frank.ramsay@hpe.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Fix CPU count display for power domain != 0. In the function
punit_id is always 0, so it never incremented cpu count for power
domain id != 0.
Update punit_id after call to update_punit_cpu_info() to what is
actually received from the kernel.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
This version addresses issues with core power configuration for
non CPU dies. Also address issue with JSON formatting of output.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When core-power configuration or enabled is modified, this is only done
for compute dies. But the config must also be set to cores with no CPUs.
Without this the configuration is not affective.
On displaying config information, allow display for non compute dies
also.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
