On 01.09.2019 08:58, Adam Zerella wrote:
> This resolves a type conversion from 'char *' to 'unsigned short'.
Could you explain this? There's no ...
> --- a/arch/x86/xen/efi.c
> +++ b/arch/x86/xen/efi.c
> @@ -118,8 +118,8 @@ static enum efi_secureboot_mode
> xen_efi_get_secureboot(void)
> unsigned long size;
>
> size = sizeof(secboot);
> - status = efi.get_variable(L"SecureBoot", &efi_variable_guid,
> - NULL, &size, &secboot);
> + status = efi.get_variable((efi_char16_t *)L"SecureBoot",
> + &efi_variable_guid, NULL, &size, &secboot);
... "char *" resulting as type for L"" type strings, hence there
should be no need for a cast: In fact I consider such casts
dangerous, as they may hide actual problems. To me this looks
more like something that wants fixing in sparse; the compilers,
after all, have no issue with such wide character string literals.
> @@ -158,7 +158,7 @@ static enum efi_secureboot_mode
> xen_efi_get_secureboot(void)
> return efi_secureboot_mode_unknown;
> }
>
> -void __init xen_efi_init(struct boot_params *boot_params)
> +static void __init xen_efi_init(struct boot_params *boot_params)
> {
> efi_system_table_t *efi_systab_xen;
If I was a maintainer of this code, I'd request this not be part
of a patch with a title being entirely unrelated to the change.
Jan
