Implement hwmon API for reading/setting slow and fast PPT limit.
APU power is managed to system-level requirements through the PPT
(package power tracking) feature. PPT is intended to limit power to the
requirements of the power source and could be dynamically updated to
maximize APU performance within the system power budget.
Here FAST_PPT_LIMIT manages the ~10 ms moving average of APU power,
while SLOW_PPT_LIMIT manages the configurable, thermally significant
moving average of APU power (default ~5000 ms).
User could read slow/fast ppt limit using command "cat power*_cap" or
"sensors" in the hwmon device directory. User could adjust values of
slow/fast ppt limit as needed depending on workloads through command
"echo ## > power*_cap".
Example:
$ echo 15000000 > power1_cap
$ echo 18000000 > power2_cap
$ sensors
amdgpu-pci-0300
Adapter: PCI adapter
slowPPT: 9.04W (cap = 15.00 W)
fastPPT: 9.04W (cap = 18.00 W)
v2: align with existing interfaces for the getting/setting of PPT
limits. Encode the upper 8 bits of limit value to distinguish
slow and fast power limit type.
Signed-off-by: Xiaomeng Hou <Xiaomeng.Hou@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The original smu_get_power_limit callback accepts the power limit level
parameter as bool which limits to max and current. For possible needs to
retrieve other level like min, extend the parameter type using enum.
Signed-off-by: Xiaomeng Hou <Xiaomeng.Hou@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
If the user passes a "level" value which is higher than 31 then that
leads to shift wrapping. The undefined behavior will lead to a
syzkaller stack dump.
Fixes: 5632708f44 ("drm/amd/powerplay: add dpm force multiple levels on cz/tonga/fiji/polaris (v2)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Supplement of previous fix.
Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Currently, the the input to the pp_od_clk_voltage sysfs interface has
two adjacent separator characters, i.e. "\n\0", then we try to parse an
argument out of the empty string, whereas we really should just ignore
this case, and treat any number of adjacent separators as one separator
for arguments.
We do this here by simply skipping the argument parsing for all
empy-string arguments.
An example test case would be `echo 's 1 900' > pp_od_clk_voltage`
(note the two spaces after the 's'.
This also solves the issue where the written string ends with both a
newline, and a nul-terminator, '\n\0', as is the case for `echo 's 1
900' > pp_od_clk_voltage` since the recent rebase.
Got lost during code rebase/merge. No need to port this to other
branches.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 2456c290a7. Got lost
during code rebase/merge. No need to port this to other branches.
Signed-off-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to fix below build error while we are using the kconfig
without x86.
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c: In function
'vangogh_get_smu_metrics_data':
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c:300:10:
error: 'boot_cpu_data' undeclared (first use in this function); did you
mean 'boot_cpuid'?
300 | boot_cpu_data.x86_max_cores * sizeof(uint16_t));
| ^~~~~~~~~~~~~
| boot_cpuid
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c: In function
'vangogh_read_sensor':
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c:1320:11:
error: 'boot_cpu_data' undeclared (first use in this function); did you
mean 'boot_cpuid'?
1320 | *size = boot_cpu_data.x86_max_cores * sizeof(uint16_t);
| ^~~~~~~~~~~~~
| boot_cpuid
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c: In function
'vangogh_od_edit_dpm_table':
drivers/gpu/drm/amd/amdgpu/../pm/swsmu/smu11/vangogh_ppt.c:1460:19:
error: 'boot_cpu_data' undeclared (first use in this function); did you
mean 'boot_cpuid'?
1460 | if (input[0] >= boot_cpu_data.x86_max_cores) {
| ^~~~~~~~~~~~~
| boot_cpuid
v2: fix #ifdef and add comment for APU only
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to modify the fine grain tuning function for Renoir.
The fine grain tuning function uses the sysfs node -- pp_od_clk_voltage
to config gfxclk. Meanwhile, another sysfs
node -- power_dpm_force_perfomance_level also affects the gfx clk.
It will cause confusion when these two sysfs nodes works
together.
It is risky to add two new flags to common smu struct, so
this patch uses the existing flag to make these two sysfs nodes works
separately. Only when power_dpm_force_perfomance_level is changed
to "manual" mode, the fine grain function will be started.
In other profile modes, including "auto", "high", "low", "profile_peak",
"profile_standard", "profile_min_sclk", "profile_min_mclk",
the fine grain tuning function will be shut down and the frequency range
of gfx will be restored the default value.
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to implement the processor fine grain feature for vangogh.
It's similar with gfx clock, the only difference is below:
echo "p core_id level value" > pp_od_clk_voltage
1. "p" - set the cclk (processor) frequency
2. "core_id" - 0/1/2/3, represents which cpu core you want to select
2. "level" - 0 or 1, "0" represents the min value, "1" represents the
max value
3. "value" - the target value of cclk frequency, it should be limited in
the safe range
v2: fix some missing changes as Evan's suggestion.
v3: add version check and fix the restore.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The core processor clocks will be stored in smu metric table, then we
add this runtime information into amdgpu_pm_info interface.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to improve the fine grain tuning function for Renoir.
The fine grain tuning function uses the sysfs node -- pp_od_clk_voltage
to config gfxclk. Meanwhile, another sysfs
node -- power_dpm_force_perfomance_level also affects the gfx clk.
It will cause confusion when these two sysfs nodes works
together.
And the flag "od_enabled" is used to control the overdrive function for
dGPU, like navi10, navi14 and navi21.
APU like Renior or Vangogh uses this "od_enabled" to configure
the frequency range of gfx clock, but the max value of frequency
range will not be higher than the safe limit, it is not "overdrive".
So this patch adds two new flags -- "fine_grain_enabled" and
"fine_grain_started" to avoid this confusion, the flag will
make these two sysfs nodes work separately.
The flag "fine_grain_enabled" is set as "enabled" by default,
so the fine grain tuning function will be enabled by default.
But the flag "fine_grain_started" is set as "false" by default,
so the fine grain function will not take effect until it is set as
"true".
Only when power_dpm_force_perfomance_level is changed to
"manual" mode, the flag "fine_grain_started" will be set as "true",
and the fine grain tuning function will be started.
In other profile modes, including "auto", "high", "low", "profile_peak",
"profile_standard", "profile_min_sclk", "profile_min_mclk",
the flag "fine_grain_started" will be set as "false", and the od range of
fine grain tuning function will be restored default value.
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to add two new sysfs nodes for vangogh:
pp_dpm_dclk and pp_dpm_vclk.
The two sysfs nodes are similar to pp_dpm_fclk/memclk/socclk.
pp_dpm_dclk represents the DPM frequency of dcn unit.
pp_dpm_vclk represents the DPM frequency of vcn unit.
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Acked-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This is supported by Sienna Cichlid, Navy Flounder and Dimgrey
Cavefish. For these ASICs, the target voltage calculation can be
illustrated by "voltage = voltage calculated from v/f curve +
overdrive vddgfx offset".
V2: limit the smu_version check for Sienna Cichlid only
Here are some sample usages about this new OD setting:
1. Check current vddgfx offset setting by
cat /sys/class/drm/card0/device/pp_od_clk_voltage
...
...
OD_VDDGFX_OFFSET:
0mV
...
...
2. Set new vddgfx offset by
echo "vo 10" > /sys/class/drm/card0/device/pp_od_clk_voltage
cat /sys/class/drm/card0/device/pp_od_clk_voltage
...
...
OD_VDDGFX_OFFSET:
10mV
...
...
3. Commit the new setting by
echo "c" > /sys/class/drm/card0/device/pp_od_clk_voltage
Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to set the initial value of pm info to zero.
The "value64" is ported to the hwmon and debugfs node, it is a uint64 type.
When it is used for NV10/VEGA10/VEGA20, its word size is appropriate,
because NV10/VEGA10/VEGA20 has a 64bit smu feature mask, which is separated to high 32bit and low 32bit.
But some asic has only 32bit smu feature mask,and this 32bit mask will fill the low 32bit of "value64".
So if this "value64" is not initialized to zero, the high 32bit will be
filled by a meaningless value, when the whole "value64" is ported to the
"SMC Feature Mask" in the "amdgpu_pm_info" on some specific asic, it
will be a wrong value.
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Populate current link speed, width and clock domain frequency.
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
These were leftover from the initial implementation, but
never used. Drop them.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Noticed-by: Ryan Taylor <ryan.taylor@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch is to add one sysfs file -- "pp_od_clk_voltage" for
Raven/Raven2/Picasso APU, which is only used by dGPU like VEGA10.
This sysfs file supports the feature to modify gfx engine clock(Mhz units), it can
be used to configure the min value and the max value for gfx clock limited in the
safe range.
Command guide:
echo "s level clock" > pp_od_clk_voltage
s - adjust teh sclk level
level - 0 or 1, "0" represents the min value, "1" represents the max value
clock - the clock value(Mhz units), like 400, 800 or 1200, the value must be within the
OD_RANGE limits.
Example:
$ cat pp_od_clk_voltage
OD_SCLK:
0: 200Mhz
1: 1400Mhz
OD_RANGE:
SCLK: 200MHz 1400MHz
$ echo "s 0 600" > pp_od_clk_voltage
$ echo "s 1 1000" > pp_od_clk_voltage
$ cat pp_od_clk_voltage
OD_SCLK:
0: 600Mhz
1: 1000Mhz
OD_RANGE:
SCLK: 200MHz 1400MHz
Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Add a static inline adev_to_drm() to obtain
the DRM device pointer from an amdgpu_device pointer.
Signed-off-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Get the amdgpu_device from the DRM device by use
of an inline function, drm_to_adev(). The inline
function resolves a pointer to struct drm_device
to a pointer to struct amdgpu_device.
v2: Use a typed visible static inline function
instead of an invisible macro.
Signed-off-by: Luben Tuikov <luben.tuikov@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
if other threads have holden the reset lock, recovery will
fail to try_lock. Therefore we introduce atomic hive->in_reset
and adev->in_gpu_reset, to avoid reentering GPU recovery.
v2:
drop "? true : false" in the definition of amdgpu_in_reset
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Dennis Li <Dennis.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We can get this on RENOIR and newer via the SMU metrics
table.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
FAMILY_KV is APUs and we already check for APUs.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The target is to provide a clear entry point(for power routines).
Also this can help to maintain a clear view about the frameworks
used on different ASICs. Hopefully all these can make power part
more friendly to play with.
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
