Re: [PATCH v3 2/2] platform/x86: thinklmi: improved DMI handling

From: Ilpo Järvinen
Date: Tue Jun 10 2025 - 03:38:43 EST


On Mon, 9 Jun 2025, Mark Pearson wrote:

> Fix issues reported by kernel test robot.
> - Require DMI for think-lmi.
> - Check return from getting serial string
>
> Reported-by: kernel test robot <lkp@xxxxxxxxx>
> Closes: https://lore.kernel.org/oe-kbuild-all/202506062319.F0IpDxF6-lkp@xxxxxxxxx/
>
> Signed-off-by: Mark Pearson <mpearson-lenovo@xxxxxxxxx>
> ---
> - New patch added to series.
>
> drivers/platform/x86/lenovo/Kconfig | 1 +
> drivers/platform/x86/lenovo/think-lmi.c | 8 +++++---
> 2 files changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/platform/x86/lenovo/Kconfig b/drivers/platform/x86/lenovo/Kconfig
> index a4b565283768..207dd7f88ed0 100644
> --- a/drivers/platform/x86/lenovo/Kconfig
> +++ b/drivers/platform/x86/lenovo/Kconfig
> @@ -197,6 +197,7 @@ config THINKPAD_ACPI_HOTKEY_POLL
> config THINKPAD_LMI
> tristate "Lenovo WMI-based systems management driver"
> depends on ACPI_WMI
> + depends on DMI
> select FW_ATTR_CLASS
> help
> This driver allows changing BIOS settings on Lenovo machines whose
> diff --git a/drivers/platform/x86/lenovo/think-lmi.c b/drivers/platform/x86/lenovo/think-lmi.c
> index 143d9fdedb65..8f70c60f791f 100644
> --- a/drivers/platform/x86/lenovo/think-lmi.c
> +++ b/drivers/platform/x86/lenovo/think-lmi.c
> @@ -772,6 +772,7 @@ static ssize_t certificate_store(struct kobject *kobj,
> struct tlmi_pwd_setting *setting = to_tlmi_pwd_setting(kobj);
> enum cert_install_mode install_mode = TLMI_CERT_INSTALL;
> char *auth_str, *new_cert;
> + const char *serial;
> char *signature;
> char *guid;
> int ret;
> @@ -789,9 +790,10 @@ static ssize_t certificate_store(struct kobject *kobj,
> return -EACCES;
>
> /* Format: 'serial#, signature' */
> - auth_str = cert_command(setting,
> - dmi_get_system_info(DMI_PRODUCT_SERIAL),
> - setting->signature);
> + serial = dmi_get_system_info(DMI_PRODUCT_SERIAL);
> + if (!serial)
> + return -EINVAL;

This should not return -EINVAL as it is not a problem with the input
parameters. Perhaps -ENODEV would make sense instead?

> + auth_str = cert_command(setting, serial, setting->signature);
> if (!auth_str)
> return -ENOMEM;
>
>

--
i.