Re: [PATCH v5 OLK-5.10 5/8] soc: hisilicon: hisi_hbmdev: Add power domain control methods
在 2023/3/3 14:19, Kefeng Wang 写道:
On 2023/3/3 11:49, Zhang Zekun wrote:
Offering: HULK hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I67QNJ CVE: NA
------------------------------------------------------------------
Platform devices which supports power control are often required to be power off/on together with the devices in the same power domain. However, there isn't a generic driver that support the power control logic of these devices.
ACPI container seems to be a good place to hold these control logic. Add platform devices in the same power domain in a ACPI container, we can easily get the locality information about these devices and can moniter the power of these devices in the same power domain together.
This patch provide three userspace control interface to control the power of devices together in the container: - state: Echo online to state to power up the devices in the container and then online these devices which will be triggered by BIOS. Echo offline to the state to offline and eject the child devices in the container which are ejectable. - pxms: show the pxms of devices which are present in the container.
In our scenario, we need to control the power of HBM memory devices which can be power consuming and will only be used in some specialized scenarios, such as HPC. HBM memory devices in a socket are in the same power domain, and should be power off/on together. We have come up with an idea that put these power control logic in a specialized driver, but ACPI container seems to be a more generic place to hold these control logic.
Signed-off-by: Zhang Zekun <zhangzekun11@huawei.com> --- v6: - get container dev form container_subsys
v5: - check containers have PNP0C80 before create attibute files
v4: - prettify the code
v3: - move the common code to hisi_internal.h
drivers/base/container.c | 1 + drivers/soc/Kconfig | 1 + drivers/soc/Makefile | 1 + drivers/soc/hisilicon/Kconfig | 19 +++ drivers/soc/hisilicon/Makefile | 3 + drivers/soc/hisilicon/hisi_hbmdev.c | 196 ++++++++++++++++++++++++++ drivers/soc/hisilicon/hisi_internal.h | 31 ++++ 7 files changed, 252 insertions(+) create mode 100644 drivers/soc/hisilicon/Kconfig create mode 100644 drivers/soc/hisilicon/Makefile create mode 100644 drivers/soc/hisilicon/hisi_hbmdev.c create mode 100644 drivers/soc/hisilicon/hisi_internal.h
diff --git a/drivers/base/container.c b/drivers/base/container.c index 1ba42d2d3532..12e572d0c69b 100644 --- a/drivers/base/container.c +++ b/drivers/base/container.c @@ -30,6 +30,7 @@ struct bus_type container_subsys = { .online = trivial_online, .offline = container_offline, }; +EXPORT_SYMBOL_GPL(container_subsys); void __init container_dev_init(void) { diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig index 425ab6f7e375..f7c59b063321 100644 --- a/drivers/soc/Kconfig +++ b/drivers/soc/Kconfig @@ -23,5 +23,6 @@ source "drivers/soc/versatile/Kconfig" source "drivers/soc/xilinx/Kconfig" source "drivers/soc/zte/Kconfig" source "drivers/soc/kendryte/Kconfig" +source "drivers/soc/hisilicon/Kconfig" endmenu diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile index 36452bed86ef..68f186e00e44 100644 --- a/drivers/soc/Makefile +++ b/drivers/soc/Makefile @@ -29,3 +29,4 @@ obj-$(CONFIG_PLAT_VERSATILE) += versatile/ obj-y += xilinx/ obj-$(CONFIG_ARCH_ZX) += zte/ obj-$(CONFIG_SOC_KENDRYTE) += kendryte/ +obj-y += hisilicon/ diff --git a/drivers/soc/hisilicon/Kconfig b/drivers/soc/hisilicon/Kconfig new file mode 100644 index 000000000000..497787af004e --- /dev/null +++ b/drivers/soc/hisilicon/Kconfig @@ -0,0 +1,19 @@ +# SPDX-License-Identifier: GPL-2.0 +# +# Hisilicon SoC drivers +# +menu "Hisilicon SoC driver support" + +config HISI_HBMDEV + tristate "add extra support for hbm memory device" + depends on ACPI_HOTPLUG_MEMORY + select ACPI_CONTAINER + help + This driver add extra supports for memory devices. The driver + provides methods for userpace to control the power of memory + devices in a container. + + To compile this driver as a module, choose M here: + the module will be called hisi_hbmdev. + +endmenu diff --git a/drivers/soc/hisilicon/Makefile b/drivers/soc/hisilicon/Makefile new file mode 100644 index 000000000000..22e87acb1ab3 --- /dev/null +++ b/drivers/soc/hisilicon/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0 + +obj-$(CONFIG_HISI_HBMDEV) += hisi_hbmdev.o diff --git a/drivers/soc/hisilicon/hisi_hbmdev.c b/drivers/soc/hisilicon/hisi_hbmdev.c new file mode 100644 index 000000000000..8485efc00684 --- /dev/null +++ b/drivers/soc/hisilicon/hisi_hbmdev.c @@ -0,0 +1,196 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) Huawei Technologies Co., Ltd. 2023. All rights reserved. + */ + +#include <linux/kobject.h> +#include <linux/module.h> +#include <linux/nodemask.h> +#include <linux/acpi.h> +#include <linux/container.h> + +#include "hisi_internal.h" + +#define ACPI_MEMORY_DEVICE_HID "PNP0C80" +#define ACPI_GENERIC_CONTAINER_DEVICE_HID "PNP0A06" + +struct memory_dev { + struct kobject *memdev_kobj; +}; + +static struct memory_dev *mdev; + +static int hbmdev_find(struct acpi_device *adev, void *arg) +{ + const char *hid = acpi_device_hid(adev); + bool *found = arg; + + if (!strcmp(hid, ACPI_MEMORY_DEVICE_HID)) { + *found = true; + return -1; + } + + return 0; +} + +static bool has_hbmdev(struct device *dev) +{ + struct acpi_device *adev = ACPI_COMPANION(dev); + const char *hid = acpi_device_hid(adev); + bool found = false; + + if (strcmp(hid, ACPI_GENERIC_CONTAINER_DEVICE_HID)) + return found; + + acpi_dev_for_each_child(adev, hbmdev_find, &found); + + return found; +} + +static int get_pxm(struct acpi_device *acpi_device, void *arg) +{ + acpi_handle handle = acpi_device->handle; + nodemask_t *mask = arg; + unsigned long long sta; + acpi_status status; + int nid; + + status = acpi_evaluate_integer(handle, "_STA", NULL, &sta); + if (ACPI_SUCCESS(status) && (sta & ACPI_STA_DEVICE_ENABLED)) { + nid = acpi_get_node(handle); + if (nid >= 0) + node_set(nid, *mask); + } + + return 0; +} + +static ssize_t pxms_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct acpi_device *adev = ACPI_COMPANION(dev); + nodemask_t mask; + + nodes_clear(mask); + acpi_dev_for_each_child(adev, get_pxm, &mask); + + return sysfs_emit(buf, "%*pbl\n", + nodemask_pr_args(&mask)); +} +static DEVICE_ATTR_RO(pxms); + +static int memdev_power_on(struct acpi_device *adev) +{ + acpi_handle handle = adev->handle; + acpi_status status; + + status = acpi_evaluate_object(handle, "_ON", NULL, NULL); + if (ACPI_FAILURE(status)) { + acpi_handle_warn(handle, "Power on failed (0x%x)\n", status); + return -ENODEV; + } + + return 0; +} + +static int eject_device(struct acpi_device *acpi_device, void *not_used) +{ + acpi_object_type unused; + acpi_status status; + + status = acpi_get_type(acpi_device->handle, &unused); + if (ACPI_FAILURE(status) || !acpi_device->flags.ejectable) + return -ENODEV; + + get_device(&acpi_device->dev); + status = acpi_hotplug_schedule(acpi_device, ACPI_OST_EC_OSPM_EJECT); + if (ACPI_SUCCESS(status)) + return 0; + + put_device(&acpi_device->dev); + acpi_evaluate_ost(acpi_device->handle, ACPI_OST_EC_OSPM_EJECT, + ACPI_OST_SC_NON_SPECIFIC_FAILURE, NULL); + + return status == AE_NO_MEMORY ? -ENOMEM : -EAGAIN; +} + +static int memdev_power_off(struct acpi_device *adev) +{ + return acpi_dev_for_each_child(adev, eject_device, NULL); +} + +static ssize_t state_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct acpi_device *adev = ACPI_COMPANION(dev); + const int type = online_type_from_str(buf); + int ret = -EINVAL; + + switch (type) { + case STATE_ONLINE: + ret = memdev_power_on(adev); + break; + case STATE_OFFLINE: + ret = memdev_power_off(adev); + break; + default: + break; + } + + if (ret) + return ret; + + return count; +} +static DEVICE_ATTR_WO(state); + +static int container_add(struct device *dev, void *data) +{ + if (!has_hbmdev(dev)) + return 0; + + device_create_file(dev, &dev_attr_state); + device_create_file(dev, &dev_attr_pxms); + + return 0; +} + +static int container_remove(struct device *dev, void *data) +{ + if (!has_hbmdev(dev)) + return 0; + + device_remove_file(dev, &dev_attr_state); + device_remove_file(dev, &dev_attr_pxms); + + return 0; +} + +static int __init mdev_init(void) +{ + mdev = kzalloc(sizeof(struct memory_dev), GFP_KERNEL); + if (!mdev) + return -ENOMEM; + + mdev->memdev_kobj = kobject_create_and_add("hbm_memory", kernel_kobj); + if (!mdev->memdev_kobj) { + kfree(mdev); + return -ENOMEM; + } + + bus_for_each_dev(&container_subsys, NULL, NULL, container_add); + return 0; +}
这个还是没有解决是否加载的问题； 1） 通过遍历，你已经拿到了所有的container device， 后面移除的时候就不需要再次遍历了；这里可以搞个链表； 2）另外判断是否有hbm，搞个全局变量，如果有了，才继续执行后面的初始化，否则就直接返回了；
按照这个思路这里可以改成这样： module_init 1. 首先判断所有容器中是否有hbm device， 构建链表 2. 如果链表不为空 那么执行模块的初始化，创建hbm_memory目录 如果链表为空 直接返回，释放链表 module_exit 释放链表中的资源，释放链表
+module_init(mdev_init); + +static void __exit mdev_exit(void) +{ + bus_for_each_dev(&container_subsys, NULL, NULL, container_remove); + kobject_put(mdev->memdev_kobj); + kfree(mdev); +} +module_exit(mdev_exit); + +MODULE_LICENSE("GPL v2"); +MODULE_AUTHOR("Zhang Zekun <zhangzekun11@huawei.com>"); diff --git a/drivers/soc/hisilicon/hisi_internal.h b/drivers/soc/hisilicon/hisi_internal.h new file mode 100644 index 000000000000..5345174f6b84 --- /dev/null +++ b/drivers/soc/hisilicon/hisi_internal.h @@ -0,0 +1,31 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Copyright (C) Huawei Technologies Co., Ltd. 2023. All rights reserved. + */ + +#ifndef _HISI_INTERNAL_H +#define _HISI_INTERNAL_H + +enum { + STATE_ONLINE, + STATE_OFFLINE, +}; + +static const char *const online_type_to_str[] = { + [STATE_ONLINE] = "online", + [STATE_OFFLINE] = "offline", +}; + +static inline int online_type_from_str(const char *str) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(online_type_to_str); i++) { + if (sysfs_streq(str, online_type_to_str[i])) + return i; + } + + return -EINVAL; +} + +#endif
participants (1)
-
zhangzekun (A)