[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <a10e34df-bbd8-462a-8851-51eabda87775@oss.qualcomm.com>
Date: Mon, 9 Dec 2024 15:56:19 -0800
From: Jeff Johnson <jeff.johnson@....qualcomm.com>
To: Zhang Zekun <zhangzekun11@...wei.com>, xuwei5@...ilicon.com,
lihuisong@...wei.com, Jonathan.Cameron@...wei.com
Cc: linux-kernel@...r.kernel.org, liuyongqiang13@...wei.com
Subject: Re: [PATCH 1/2] soc: hisilicon: kunpeng_hbmdev: Add support for
controling the power of hbm memory
On 12/6/24 03:28, Zhang Zekun wrote:
> Add a driver for High Bandwidth Memory (HBM) devices, which will provide
> user space interfaces to power on/off the HBM devices. In Kunpeng servers,
> we need to control the power of HBM 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.
>
> HBM devices will be configured with ACPI device id "PNP0C80", and be used
> as a cpuless numa node. HBM devices in the same power domain will be put
> into the same container. ACPI function "_ON" and "_OFF" are reponsible
> for power on/off the HBM device, and notify the OS to fully online/offline
> the HBM memory.
>
> Signed-off-by: Zhang Zekun <zhangzekun11@...wei.com>
> ---
...
> diff --git a/drivers/soc/hisilicon/Kconfig b/drivers/soc/hisilicon/Kconfig
> index 6d7c244d2e78..b3ca7d6f5d01 100644
> --- a/drivers/soc/hisilicon/Kconfig
> +++ b/drivers/soc/hisilicon/Kconfig
> @@ -21,4 +21,16 @@ config KUNPENG_HCCS
> health status and port information of HCCS, or reducing system
> power consumption on Kunpeng SoC.
>
> +config KUNPENG_HBMDEV
> + bool "add extra support for hbm memory device"
> + depends on ACPI_HOTPLUG_MEMORY
> + select ACPI_CONTAINER
> + help
> + The driver provides methods for userpace to control the power
> + of HBM memory devices on Kunpeng soc, which can help to save
> + energy. The functionality of the driver would require dedicated
> + BIOS configuration.
> +
> + If not sure, say N.
> +
> endmenu
> diff --git a/drivers/soc/hisilicon/Makefile b/drivers/soc/hisilicon/Makefile
> index 226e747e70d6..08048d73586e 100644
> --- a/drivers/soc/hisilicon/Makefile
> +++ b/drivers/soc/hisilicon/Makefile
> @@ -1,2 +1,3 @@
> # SPDX-License-Identifier: GPL-2.0-only
> obj-$(CONFIG_KUNPENG_HCCS) += kunpeng_hccs.o
> +obj-$(CONFIG_KUNPENG_HBMDEV) += kunpeng_hbmdev.o
...
> diff --git a/drivers/soc/hisilicon/kunpeng_hbmdev.c b/drivers/soc/hisilicon/kunpeng_hbmdev.c
> new file mode 100644
> index 000000000000..1945676ff502
> --- /dev/null
> +++ b/drivers/soc/hisilicon/kunpeng_hbmdev.c
> @@ -0,0 +1,210 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2024 Huawei Technologies Co., Ltd
> + */
...
> +MODULE_LICENSE("GPL");
> +MODULE_AUTHOR("Zhang Zekun <zhangzekun11@...wei.com>");
Since commit 1fffe7a34c89 ("script: modpost: emit a warning when the
description is missing"), a module without a MODULE_DESCRIPTION() will
result in a warning with make W=1. My usual guidance is to add the
missing MODULE_DESCRIPTION() to avoid such warnings. But in this case,
due to how your Kconfig & Makefile are defined, this can NEVER be built
as a module, and hence I'd remove all of the MODULE_*() macros.
/jeff
Powered by blists - more mailing lists