linux-stable/drivers/cpufreq/qcom-cpufreq-nvmem.c

648 lines
16 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2018, The Linux Foundation. All rights reserved.
*/
/*
* In Certain QCOM SoCs like apq8096 and msm8996 that have KRYO processors,
* the CPU frequency subset and voltage value of each OPP varies
* based on the silicon variant in use. Qualcomm Process Voltage Scaling Tables
* defines the voltage and frequency value based on the msm-id in SMEM
* and speedbin blown in the efuse combination.
* The qcom-cpufreq-nvmem driver reads the msm-id and efuse value from the SoC
* to provide the OPP framework with required information.
* This is used to determine the voltage and frequency value for each OPP of
* operating-points-v2 table when it is parsed by the OPP framework.
*/
#include <linux/cpu.h>
#include <linux/err.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/nvmem-consumer.h>
#include <linux/of.h>
#include <linux/platform_device.h>
cpufreq: qcom-nvmem: Preserve PM domain votes in system suspend >From the Linux point of view, the power domains used by the CPU must stay always-on. This is because we still need the CPU to keep running until the last instruction, which will typically be a firmware call that shuts down the CPU cleanly. At the moment the power domain votes (enable + performance state) are dropped during system suspend, which means the CPU could potentially malfunction while entering suspend. We need to distinguish between two different setups used with qcom-cpufreq-nvmem: 1. CPR power domain: The backing regulator used by CPR should stay always-on in Linux; it is typically disabled automatically by hardware when the CPU enters a deep idle state. However, we should pause the CPR state machine during system suspend. 2. RPMPD: The power domains used by the CPU should stay always-on in Linux (also across system suspend). The CPU typically only uses the *_AO ("active-only") variants of the power domains in RPMPD. For those, the RPM firmware will automatically drop the votes internally when the CPU enters a deep idle state. Make this work correctly by calling device_set_awake_path() on the virtual genpd devices, so that the votes are maintained across system suspend. The power domain drivers need to set GENPD_FLAG_ACTIVE_WAKEUP to opt into staying on during system suspend. For now we only set this for the RPMPD case. For CPR, not setting it will ensure the state machine is still paused during system suspend, while the backing regulator will stay on with "regulator-always-on". Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-11-14 10:07:44 +00:00
#include <linux/pm.h>
#include <linux/pm_domain.h>
#include <linux/pm_opp.h>
#include <linux/pm_runtime.h>
#include <linux/slab.h>
#include <linux/soc/qcom/smem.h>
#include <dt-bindings/arm/qcom,ids.h>
enum ipq806x_versions {
IPQ8062_VERSION = 0,
IPQ8064_VERSION,
IPQ8065_VERSION,
};
#define IPQ6000_VERSION BIT(2)
enum ipq8074_versions {
IPQ8074_HAWKEYE_VERSION = 0,
IPQ8074_ACORN_VERSION,
};
struct qcom_cpufreq_drv;
struct qcom_cpufreq_match_data {
int (*get_version)(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv);
const char **pd_names;
unsigned int num_pd_names;
};
struct qcom_cpufreq_drv_cpu {
int opp_token;
struct dev_pm_domain_list *pd_list;
};
struct qcom_cpufreq_drv {
u32 versions;
const struct qcom_cpufreq_match_data *data;
struct qcom_cpufreq_drv_cpu cpus[];
};
static struct platform_device *cpufreq_dt_pdev, *cpufreq_pdev;
cpufreq: qcom-nvmem: Add MSM8909 When the MSM8909 SoC is used together with the PM8909 PMIC the primary power supply for the CPU (VDD_APC) is shared with other components to the SoC, namely the VDD_CX power domain typically supplied by the PM8909 S1 regulator. This means that all votes for necessary performance states go via the RPM firmware which collects the requirements from all the processors in the SoC. The RPM firmware then chooses the actual voltage based on the performance states ("corners"), depending on calibration values in the NVMEM and other factors. The MSM8909 SoC is also sometimes used with the PM8916 or PM660 PMIC. In that case there is a dedicated regulator connected to VDD_APC and Linux is responsible to do adaptive voltage scaling using CPR (similar to the existing code for QCS404). This difference can be described in the device tree, by either assigning the CPU a power domain from RPMPD or from the CPR driver. Describe this using "perf" as generic power domain name, which is also used already for SCMI based platforms. Also add a simple function that reads the speedbin from a NVMEM cell and sets it as-is for opp-supported-hw. The actual bit position can be described in the device tree without additional driver changes. Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Acked-by: Konrad Dybcio <konrad.dybcio@linaro.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> [ Viresh: Fixed rebase conflict. ] Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-10-18 08:06:04 +00:00
static int qcom_cpufreq_simple_get_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
u8 *speedbin;
*pvs_name = NULL;
speedbin = nvmem_cell_read(speedbin_nvmem, NULL);
if (IS_ERR(speedbin))
return PTR_ERR(speedbin);
dev_dbg(cpu_dev, "speedbin: %d\n", *speedbin);
drv->versions = 1 << *speedbin;
kfree(speedbin);
return 0;
}
static void get_krait_bin_format_a(struct device *cpu_dev,
int *speed, int *pvs,
u8 *buf)
{
u32 pte_efuse;
pte_efuse = *((u32 *)buf);
*speed = pte_efuse & 0xf;
if (*speed == 0xf)
*speed = (pte_efuse >> 4) & 0xf;
if (*speed == 0xf) {
*speed = 0;
dev_warn(cpu_dev, "Speed bin: Defaulting to %d\n", *speed);
} else {
dev_dbg(cpu_dev, "Speed bin: %d\n", *speed);
}
*pvs = (pte_efuse >> 10) & 0x7;
if (*pvs == 0x7)
*pvs = (pte_efuse >> 13) & 0x7;
if (*pvs == 0x7) {
*pvs = 0;
dev_warn(cpu_dev, "PVS bin: Defaulting to %d\n", *pvs);
} else {
dev_dbg(cpu_dev, "PVS bin: %d\n", *pvs);
}
}
static void get_krait_bin_format_b(struct device *cpu_dev,
int *speed, int *pvs, int *pvs_ver,
u8 *buf)
{
u32 pte_efuse, redundant_sel;
pte_efuse = *((u32 *)buf);
redundant_sel = (pte_efuse >> 24) & 0x7;
*pvs_ver = (pte_efuse >> 4) & 0x3;
switch (redundant_sel) {
case 1:
*pvs = ((pte_efuse >> 28) & 0x8) | ((pte_efuse >> 6) & 0x7);
*speed = (pte_efuse >> 27) & 0xf;
break;
case 2:
*pvs = (pte_efuse >> 27) & 0xf;
*speed = pte_efuse & 0x7;
break;
default:
/* 4 bits of PVS are in efuse register bits 31, 8-6. */
*pvs = ((pte_efuse >> 28) & 0x8) | ((pte_efuse >> 6) & 0x7);
*speed = pte_efuse & 0x7;
}
/* Check SPEED_BIN_BLOW_STATUS */
if (pte_efuse & BIT(3)) {
dev_dbg(cpu_dev, "Speed bin: %d\n", *speed);
} else {
dev_warn(cpu_dev, "Speed bin not set. Defaulting to 0!\n");
*speed = 0;
}
/* Check PVS_BLOW_STATUS */
pte_efuse = *(((u32 *)buf) + 1);
pte_efuse &= BIT(21);
if (pte_efuse) {
dev_dbg(cpu_dev, "PVS bin: %d\n", *pvs);
} else {
dev_warn(cpu_dev, "PVS bin not set. Defaulting to 0!\n");
*pvs = 0;
}
dev_dbg(cpu_dev, "PVS version: %d\n", *pvs_ver);
}
static int qcom_cpufreq_kryo_name_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
size_t len;
u32 msm_id;
u8 *speedbin;
int ret;
*pvs_name = NULL;
ret = qcom_smem_get_soc_id(&msm_id);
if (ret)
return ret;
speedbin = nvmem_cell_read(speedbin_nvmem, &len);
if (IS_ERR(speedbin))
return PTR_ERR(speedbin);
switch (msm_id) {
case QCOM_ID_MSM8996:
case QCOM_ID_APQ8096:
case QCOM_ID_IPQ5332:
case QCOM_ID_IPQ5322:
case QCOM_ID_IPQ5312:
case QCOM_ID_IPQ5302:
case QCOM_ID_IPQ5300:
case QCOM_ID_IPQ5321:
case QCOM_ID_IPQ9514:
case QCOM_ID_IPQ9550:
case QCOM_ID_IPQ9554:
case QCOM_ID_IPQ9570:
case QCOM_ID_IPQ9574:
drv->versions = 1 << (unsigned int)(*speedbin);
break;
case QCOM_ID_MSM8996SG:
case QCOM_ID_APQ8096SG:
drv->versions = 1 << ((unsigned int)(*speedbin) + 4);
break;
default:
BUG();
break;
}
kfree(speedbin);
return 0;
}
static int qcom_cpufreq_krait_name_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
int speed = 0, pvs = 0, pvs_ver = 0;
u8 *speedbin;
size_t len;
int ret = 0;
speedbin = nvmem_cell_read(speedbin_nvmem, &len);
if (IS_ERR(speedbin))
return PTR_ERR(speedbin);
switch (len) {
case 4:
get_krait_bin_format_a(cpu_dev, &speed, &pvs, speedbin);
break;
case 8:
get_krait_bin_format_b(cpu_dev, &speed, &pvs, &pvs_ver,
speedbin);
break;
default:
dev_err(cpu_dev, "Unable to read nvmem data. Defaulting to 0!\n");
ret = -ENODEV;
goto len_error;
}
snprintf(*pvs_name, sizeof("speedXX-pvsXX-vXX"), "speed%d-pvs%d-v%d",
speed, pvs, pvs_ver);
drv->versions = (1 << speed);
len_error:
kfree(speedbin);
return ret;
}
static int qcom_cpufreq_ipq8064_name_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
int speed = 0, pvs = 0;
int msm_id, ret = 0;
u8 *speedbin;
size_t len;
speedbin = nvmem_cell_read(speedbin_nvmem, &len);
if (IS_ERR(speedbin))
return PTR_ERR(speedbin);
if (len != 4) {
dev_err(cpu_dev, "Unable to read nvmem data. Defaulting to 0!\n");
ret = -ENODEV;
goto exit;
}
get_krait_bin_format_a(cpu_dev, &speed, &pvs, speedbin);
ret = qcom_smem_get_soc_id(&msm_id);
if (ret)
goto exit;
switch (msm_id) {
case QCOM_ID_IPQ8062:
drv->versions = BIT(IPQ8062_VERSION);
break;
case QCOM_ID_IPQ8064:
case QCOM_ID_IPQ8066:
case QCOM_ID_IPQ8068:
drv->versions = BIT(IPQ8064_VERSION);
break;
case QCOM_ID_IPQ8065:
case QCOM_ID_IPQ8069:
drv->versions = BIT(IPQ8065_VERSION);
break;
default:
dev_err(cpu_dev,
"SoC ID %u is not part of IPQ8064 family, limiting to 1.0GHz!\n",
msm_id);
drv->versions = BIT(IPQ8062_VERSION);
break;
}
/* IPQ8064 speed is never fused. Only pvs values are fused. */
snprintf(*pvs_name, sizeof("speed0-pvsXX"), "speed0-pvs%d", pvs);
exit:
kfree(speedbin);
return ret;
}
static int qcom_cpufreq_ipq6018_name_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
u32 msm_id;
int ret;
u8 *speedbin;
*pvs_name = NULL;
ret = qcom_smem_get_soc_id(&msm_id);
if (ret)
return ret;
speedbin = nvmem_cell_read(speedbin_nvmem, NULL);
if (IS_ERR(speedbin))
return PTR_ERR(speedbin);
switch (msm_id) {
case QCOM_ID_IPQ6005:
case QCOM_ID_IPQ6010:
case QCOM_ID_IPQ6018:
case QCOM_ID_IPQ6028:
/* Fuse Value Freq BIT to set
* ---------------------------------
* 2b0 No Limit BIT(0)
* 2b1 1.5 GHz BIT(1)
*/
drv->versions = 1 << (unsigned int)(*speedbin);
break;
case QCOM_ID_IPQ6000:
/*
* IPQ6018 family only has one bit to advertise the CPU
* speed-bin, but that is not enough for IPQ6000 which
* is only rated up to 1.2GHz.
* So for IPQ6000 manually set BIT(2) based on SMEM ID.
*/
drv->versions = IPQ6000_VERSION;
break;
default:
dev_err(cpu_dev,
"SoC ID %u is not part of IPQ6018 family, limiting to 1.2GHz!\n",
msm_id);
drv->versions = IPQ6000_VERSION;
break;
}
kfree(speedbin);
return 0;
}
static int qcom_cpufreq_ipq8074_name_version(struct device *cpu_dev,
struct nvmem_cell *speedbin_nvmem,
char **pvs_name,
struct qcom_cpufreq_drv *drv)
{
u32 msm_id;
int ret;
*pvs_name = NULL;
ret = qcom_smem_get_soc_id(&msm_id);
if (ret)
return ret;
switch (msm_id) {
case QCOM_ID_IPQ8070A:
case QCOM_ID_IPQ8071A:
case QCOM_ID_IPQ8172:
case QCOM_ID_IPQ8173:
case QCOM_ID_IPQ8174:
drv->versions = BIT(IPQ8074_ACORN_VERSION);
break;
case QCOM_ID_IPQ8072A:
case QCOM_ID_IPQ8074A:
case QCOM_ID_IPQ8076A:
case QCOM_ID_IPQ8078A:
drv->versions = BIT(IPQ8074_HAWKEYE_VERSION);
break;
default:
dev_err(cpu_dev,
"SoC ID %u is not part of IPQ8074 family, limiting to 1.4GHz!\n",
msm_id);
drv->versions = BIT(IPQ8074_ACORN_VERSION);
break;
}
return 0;
}
static const struct qcom_cpufreq_match_data match_data_kryo = {
.get_version = qcom_cpufreq_kryo_name_version,
};
static const struct qcom_cpufreq_match_data match_data_krait = {
.get_version = qcom_cpufreq_krait_name_version,
};
cpufreq: qcom-nvmem: Add MSM8909 When the MSM8909 SoC is used together with the PM8909 PMIC the primary power supply for the CPU (VDD_APC) is shared with other components to the SoC, namely the VDD_CX power domain typically supplied by the PM8909 S1 regulator. This means that all votes for necessary performance states go via the RPM firmware which collects the requirements from all the processors in the SoC. The RPM firmware then chooses the actual voltage based on the performance states ("corners"), depending on calibration values in the NVMEM and other factors. The MSM8909 SoC is also sometimes used with the PM8916 or PM660 PMIC. In that case there is a dedicated regulator connected to VDD_APC and Linux is responsible to do adaptive voltage scaling using CPR (similar to the existing code for QCS404). This difference can be described in the device tree, by either assigning the CPU a power domain from RPMPD or from the CPR driver. Describe this using "perf" as generic power domain name, which is also used already for SCMI based platforms. Also add a simple function that reads the speedbin from a NVMEM cell and sets it as-is for opp-supported-hw. The actual bit position can be described in the device tree without additional driver changes. Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Acked-by: Konrad Dybcio <konrad.dybcio@linaro.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> [ Viresh: Fixed rebase conflict. ] Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-10-18 08:06:04 +00:00
static const struct qcom_cpufreq_match_data match_data_msm8909 = {
.get_version = qcom_cpufreq_simple_get_version,
.pd_names = (const char *[]) { "perf" },
.num_pd_names = 1,
cpufreq: qcom-nvmem: Add MSM8909 When the MSM8909 SoC is used together with the PM8909 PMIC the primary power supply for the CPU (VDD_APC) is shared with other components to the SoC, namely the VDD_CX power domain typically supplied by the PM8909 S1 regulator. This means that all votes for necessary performance states go via the RPM firmware which collects the requirements from all the processors in the SoC. The RPM firmware then chooses the actual voltage based on the performance states ("corners"), depending on calibration values in the NVMEM and other factors. The MSM8909 SoC is also sometimes used with the PM8916 or PM660 PMIC. In that case there is a dedicated regulator connected to VDD_APC and Linux is responsible to do adaptive voltage scaling using CPR (similar to the existing code for QCS404). This difference can be described in the device tree, by either assigning the CPU a power domain from RPMPD or from the CPR driver. Describe this using "perf" as generic power domain name, which is also used already for SCMI based platforms. Also add a simple function that reads the speedbin from a NVMEM cell and sets it as-is for opp-supported-hw. The actual bit position can be described in the device tree without additional driver changes. Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Acked-by: Konrad Dybcio <konrad.dybcio@linaro.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> [ Viresh: Fixed rebase conflict. ] Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-10-18 08:06:04 +00:00
};
static const struct qcom_cpufreq_match_data match_data_qcs404 = {
.pd_names = (const char *[]) { "cpr" },
.num_pd_names = 1,
};
static const struct qcom_cpufreq_match_data match_data_ipq6018 = {
.get_version = qcom_cpufreq_ipq6018_name_version,
};
static const struct qcom_cpufreq_match_data match_data_ipq8064 = {
.get_version = qcom_cpufreq_ipq8064_name_version,
};
static const struct qcom_cpufreq_match_data match_data_ipq8074 = {
.get_version = qcom_cpufreq_ipq8074_name_version,
};
static void qcom_cpufreq_suspend_pd_devs(struct qcom_cpufreq_drv *drv, unsigned int cpu)
{
struct dev_pm_domain_list *pd_list = drv->cpus[cpu].pd_list;
int i;
if (!pd_list)
return;
for (i = 0; i < pd_list->num_pds; i++)
device_set_awake_path(pd_list->pd_devs[i]);
}
static int qcom_cpufreq_probe(struct platform_device *pdev)
{
struct qcom_cpufreq_drv *drv;
struct nvmem_cell *speedbin_nvmem;
struct device *cpu_dev;
cpufreq: qcom: fix writes in read-only memory region This commit fixes a kernel oops because of a write in some read-only memory: [ 9.068287] Unable to handle kernel write to read-only memory at virtual address ffff800009240ad8 ..snip.. [ 9.138790] Internal error: Oops: 9600004f [#1] PREEMPT SMP ..snip.. [ 9.269161] Call trace: [ 9.276271] __memcpy+0x5c/0x230 [ 9.278531] snprintf+0x58/0x80 [ 9.282002] qcom_cpufreq_msm8939_name_version+0xb4/0x190 [ 9.284869] qcom_cpufreq_probe+0xc8/0x39c ..snip.. The following line defines a pointer that point to a char buffer stored in read-only memory: char *pvs_name = "speedXX-pvsXX-vXX"; This pointer is meant to hold a template "speedXX-pvsXX-vXX" where the XX values get overridden by the qcom_cpufreq_krait_name_version function. Since the template is actually stored in read-only memory, when the function executes the following call we get an oops: snprintf(*pvs_name, sizeof("speedXX-pvsXX-vXX"), "speed%d-pvs%d-v%d", speed, pvs, pvs_ver); To fix this issue, we instead store the template name onto the stack by using the following syntax: char pvs_name_buffer[] = "speedXX-pvsXX-vXX"; Because the `pvs_name` needs to be able to be assigned to NULL, the template buffer is stored in the pvs_name_buffer and not under the pvs_name variable. Cc: v5.7+ <stable@vger.kernel.org> # v5.7+ Fixes: a8811ec764f9 ("cpufreq: qcom: Add support for krait based socs") Signed-off-by: Fabien Parent <fabien.parent@linaro.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2022-10-15 13:04:23 +00:00
char pvs_name_buffer[] = "speedXX-pvsXX-vXX";
char *pvs_name = pvs_name_buffer;
unsigned cpu;
const struct of_device_id *match;
int ret;
cpu_dev = get_cpu_device(0);
if (!cpu_dev)
return -ENODEV;
struct device_node *np __free(device_node) =
dev_pm_opp_of_get_opp_desc_node(cpu_dev);
if (!np)
return -ENOENT;
ret = of_device_is_compatible(np, "operating-points-v2-kryo-cpu") ||
of_device_is_compatible(np, "operating-points-v2-krait-cpu");
if (!ret)
return -ENOENT;
drv = devm_kzalloc(&pdev->dev, struct_size(drv, cpus, num_possible_cpus()),
GFP_KERNEL);
if (!drv)
return -ENOMEM;
match = pdev->dev.platform_data;
drv->data = match->data;
if (!drv->data)
return -ENODEV;
if (drv->data->get_version) {
speedbin_nvmem = of_nvmem_cell_get(np, NULL);
if (IS_ERR(speedbin_nvmem))
return dev_err_probe(cpu_dev, PTR_ERR(speedbin_nvmem),
"Could not get nvmem cell\n");
ret = drv->data->get_version(cpu_dev,
speedbin_nvmem, &pvs_name, drv);
if (ret) {
nvmem_cell_put(speedbin_nvmem);
return ret;
}
nvmem_cell_put(speedbin_nvmem);
}
for_each_possible_cpu(cpu) {
struct dev_pm_opp_config config = {
.supported_hw = NULL,
};
cpu_dev = get_cpu_device(cpu);
if (NULL == cpu_dev) {
ret = -ENODEV;
goto free_opp;
}
if (drv->data->get_version) {
config.supported_hw = &drv->versions;
config.supported_hw_count = 1;
if (pvs_name)
config.prop_name = pvs_name;
}
if (config.supported_hw) {
drv->cpus[cpu].opp_token = dev_pm_opp_set_config(cpu_dev, &config);
if (drv->cpus[cpu].opp_token < 0) {
ret = drv->cpus[cpu].opp_token;
dev_err(cpu_dev, "Failed to set OPP config\n");
goto free_opp;
}
}
if (drv->data->pd_names) {
struct dev_pm_domain_attach_data attach_data = {
.pd_names = drv->data->pd_names,
.num_pd_names = drv->data->num_pd_names,
.pd_flags = PD_FLAG_DEV_LINK_ON |
PD_FLAG_REQUIRED_OPP,
};
ret = dev_pm_domain_attach_list(cpu_dev, &attach_data,
&drv->cpus[cpu].pd_list);
if (ret < 0)
goto free_opp;
}
}
cpufreq_dt_pdev = platform_device_register_simple("cpufreq-dt", -1,
NULL, 0);
if (!IS_ERR(cpufreq_dt_pdev)) {
platform_set_drvdata(pdev, drv);
return 0;
}
ret = PTR_ERR(cpufreq_dt_pdev);
dev_err(cpu_dev, "Failed to register platform device\n");
free_opp:
for_each_possible_cpu(cpu) {
dev_pm_domain_detach_list(drv->cpus[cpu].pd_list);
dev_pm_opp_clear_config(drv->cpus[cpu].opp_token);
}
return ret;
}
static void qcom_cpufreq_remove(struct platform_device *pdev)
{
struct qcom_cpufreq_drv *drv = platform_get_drvdata(pdev);
unsigned int cpu;
platform_device_unregister(cpufreq_dt_pdev);
for_each_possible_cpu(cpu) {
dev_pm_domain_detach_list(drv->cpus[cpu].pd_list);
dev_pm_opp_clear_config(drv->cpus[cpu].opp_token);
}
}
cpufreq: qcom-nvmem: Preserve PM domain votes in system suspend >From the Linux point of view, the power domains used by the CPU must stay always-on. This is because we still need the CPU to keep running until the last instruction, which will typically be a firmware call that shuts down the CPU cleanly. At the moment the power domain votes (enable + performance state) are dropped during system suspend, which means the CPU could potentially malfunction while entering suspend. We need to distinguish between two different setups used with qcom-cpufreq-nvmem: 1. CPR power domain: The backing regulator used by CPR should stay always-on in Linux; it is typically disabled automatically by hardware when the CPU enters a deep idle state. However, we should pause the CPR state machine during system suspend. 2. RPMPD: The power domains used by the CPU should stay always-on in Linux (also across system suspend). The CPU typically only uses the *_AO ("active-only") variants of the power domains in RPMPD. For those, the RPM firmware will automatically drop the votes internally when the CPU enters a deep idle state. Make this work correctly by calling device_set_awake_path() on the virtual genpd devices, so that the votes are maintained across system suspend. The power domain drivers need to set GENPD_FLAG_ACTIVE_WAKEUP to opt into staying on during system suspend. For now we only set this for the RPMPD case. For CPR, not setting it will ensure the state machine is still paused during system suspend, while the backing regulator will stay on with "regulator-always-on". Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-11-14 10:07:44 +00:00
static int qcom_cpufreq_suspend(struct device *dev)
{
struct qcom_cpufreq_drv *drv = dev_get_drvdata(dev);
unsigned int cpu;
for_each_possible_cpu(cpu)
qcom_cpufreq_suspend_pd_devs(drv, cpu);
cpufreq: qcom-nvmem: Preserve PM domain votes in system suspend >From the Linux point of view, the power domains used by the CPU must stay always-on. This is because we still need the CPU to keep running until the last instruction, which will typically be a firmware call that shuts down the CPU cleanly. At the moment the power domain votes (enable + performance state) are dropped during system suspend, which means the CPU could potentially malfunction while entering suspend. We need to distinguish between two different setups used with qcom-cpufreq-nvmem: 1. CPR power domain: The backing regulator used by CPR should stay always-on in Linux; it is typically disabled automatically by hardware when the CPU enters a deep idle state. However, we should pause the CPR state machine during system suspend. 2. RPMPD: The power domains used by the CPU should stay always-on in Linux (also across system suspend). The CPU typically only uses the *_AO ("active-only") variants of the power domains in RPMPD. For those, the RPM firmware will automatically drop the votes internally when the CPU enters a deep idle state. Make this work correctly by calling device_set_awake_path() on the virtual genpd devices, so that the votes are maintained across system suspend. The power domain drivers need to set GENPD_FLAG_ACTIVE_WAKEUP to opt into staying on during system suspend. For now we only set this for the RPMPD case. For CPR, not setting it will ensure the state machine is still paused during system suspend, while the backing regulator will stay on with "regulator-always-on". Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-11-14 10:07:44 +00:00
return 0;
}
static DEFINE_SIMPLE_DEV_PM_OPS(qcom_cpufreq_pm_ops, qcom_cpufreq_suspend, NULL);
static struct platform_driver qcom_cpufreq_driver = {
.probe = qcom_cpufreq_probe,
.remove = qcom_cpufreq_remove,
.driver = {
.name = "qcom-cpufreq-nvmem",
cpufreq: qcom-nvmem: Preserve PM domain votes in system suspend >From the Linux point of view, the power domains used by the CPU must stay always-on. This is because we still need the CPU to keep running until the last instruction, which will typically be a firmware call that shuts down the CPU cleanly. At the moment the power domain votes (enable + performance state) are dropped during system suspend, which means the CPU could potentially malfunction while entering suspend. We need to distinguish between two different setups used with qcom-cpufreq-nvmem: 1. CPR power domain: The backing regulator used by CPR should stay always-on in Linux; it is typically disabled automatically by hardware when the CPU enters a deep idle state. However, we should pause the CPR state machine during system suspend. 2. RPMPD: The power domains used by the CPU should stay always-on in Linux (also across system suspend). The CPU typically only uses the *_AO ("active-only") variants of the power domains in RPMPD. For those, the RPM firmware will automatically drop the votes internally when the CPU enters a deep idle state. Make this work correctly by calling device_set_awake_path() on the virtual genpd devices, so that the votes are maintained across system suspend. The power domain drivers need to set GENPD_FLAG_ACTIVE_WAKEUP to opt into staying on during system suspend. For now we only set this for the RPMPD case. For CPR, not setting it will ensure the state machine is still paused during system suspend, while the backing regulator will stay on with "regulator-always-on". Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-11-14 10:07:44 +00:00
.pm = pm_sleep_ptr(&qcom_cpufreq_pm_ops),
},
};
static const struct of_device_id qcom_cpufreq_match_list[] __initconst __maybe_unused = {
{ .compatible = "qcom,apq8096", .data = &match_data_kryo },
cpufreq: qcom-nvmem: Add MSM8909 When the MSM8909 SoC is used together with the PM8909 PMIC the primary power supply for the CPU (VDD_APC) is shared with other components to the SoC, namely the VDD_CX power domain typically supplied by the PM8909 S1 regulator. This means that all votes for necessary performance states go via the RPM firmware which collects the requirements from all the processors in the SoC. The RPM firmware then chooses the actual voltage based on the performance states ("corners"), depending on calibration values in the NVMEM and other factors. The MSM8909 SoC is also sometimes used with the PM8916 or PM660 PMIC. In that case there is a dedicated regulator connected to VDD_APC and Linux is responsible to do adaptive voltage scaling using CPR (similar to the existing code for QCS404). This difference can be described in the device tree, by either assigning the CPU a power domain from RPMPD or from the CPR driver. Describe this using "perf" as generic power domain name, which is also used already for SCMI based platforms. Also add a simple function that reads the speedbin from a NVMEM cell and sets it as-is for opp-supported-hw. The actual bit position can be described in the device tree without additional driver changes. Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com> Acked-by: Konrad Dybcio <konrad.dybcio@linaro.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> [ Viresh: Fixed rebase conflict. ] Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
2023-10-18 08:06:04 +00:00
{ .compatible = "qcom,msm8909", .data = &match_data_msm8909 },
{ .compatible = "qcom,msm8996", .data = &match_data_kryo },
{ .compatible = "qcom,qcs404", .data = &match_data_qcs404 },
{ .compatible = "qcom,ipq5332", .data = &match_data_kryo },
{ .compatible = "qcom,ipq6018", .data = &match_data_ipq6018 },
{ .compatible = "qcom,ipq8064", .data = &match_data_ipq8064 },
{ .compatible = "qcom,ipq8074", .data = &match_data_ipq8074 },
{ .compatible = "qcom,apq8064", .data = &match_data_krait },
{ .compatible = "qcom,ipq9574", .data = &match_data_kryo },
{ .compatible = "qcom,msm8974", .data = &match_data_krait },
{ .compatible = "qcom,msm8960", .data = &match_data_krait },
{},
};
MODULE_DEVICE_TABLE(of, qcom_cpufreq_match_list);
/*
* Since the driver depends on smem and nvmem drivers, which may
* return EPROBE_DEFER, all the real activity is done in the probe,
* which may be defered as well. The init here is only registering
* the driver and the platform device.
*/
static int __init qcom_cpufreq_init(void)
{
struct device_node *np __free(device_node) = of_find_node_by_path("/");
const struct of_device_id *match;
int ret;
if (!np)
return -ENODEV;
match = of_match_node(qcom_cpufreq_match_list, np);
if (!match)
return -ENODEV;
ret = platform_driver_register(&qcom_cpufreq_driver);
if (unlikely(ret < 0))
return ret;
cpufreq_pdev = platform_device_register_data(NULL, "qcom-cpufreq-nvmem",
-1, match, sizeof(*match));
ret = PTR_ERR_OR_ZERO(cpufreq_pdev);
if (0 == ret)
return 0;
platform_driver_unregister(&qcom_cpufreq_driver);
return ret;
}
module_init(qcom_cpufreq_init);
static void __exit qcom_cpufreq_exit(void)
{
platform_device_unregister(cpufreq_pdev);
platform_driver_unregister(&qcom_cpufreq_driver);
}
module_exit(qcom_cpufreq_exit);
MODULE_DESCRIPTION("Qualcomm Technologies, Inc. CPUfreq driver");
MODULE_LICENSE("GPL v2");