[U-Boot] [PATCH v2 03/13] x86: Change __kernel_size_t conditionals to use compiler provided defines

[Simon Glass]

Hi Bin,

On 12 June 2018 at 09:36, Bin Meng <bmeng.cn at gmail.com> wrote:
> Since commit bb0bb91cf0aa ("efi_stub: Use efi_uintn_t"), EFI x86
> 64-bit payload does not work anymore. The call to GetMemoryMap()
> in efi_stub.c fails with return code EFI_INVALID_PARAMETER. Since
> the payload itself is still 32-bit U-Boot, efi_uintn_t gets wrongly
> interpreted as int, but it should actually be long in a 64-bit EFI
> environment.
>
> This changes the x86 __kernel_size_t conditionals to use compiler
> provided defines instead. That way we always adhere to the build
> environment we're in and the definitions adjust automatically.
>
> Fixes: bb0bb91cf0aa ("efi_stub: Use efi_uintn_t")
> Signed-off-by: Bin Meng <bmeng.cn at gmail.com>
>
> ---
>
> Changes in v2:
> - new patch to "change __kernel_size_t conditionals to use compiler
>   provided defines"
>
>  arch/x86/include/asm/posix_types.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/include/asm/posix_types.h b/arch/x86/include/asm/posix_types.h
> index 717f6cb..a19f1a0 100644
> --- a/arch/x86/include/asm/posix_types.h
> +++ b/arch/x86/include/asm/posix_types.h
> @@ -16,7 +16,7 @@ typedef int           __kernel_pid_t;
>  typedef unsigned short __kernel_ipc_pid_t;
>  typedef unsigned short __kernel_uid_t;
>  typedef unsigned short __kernel_gid_t;
> -#if CONFIG_IS_ENABLED(X86_64)
> +#if defined __x86_64__

Ick.

I worry that this is hiding an EFI peculiarity in generic code.

We have a lot of occurrences of '-#if CONFIG_IS_ENABLED(X86_64)' and
it is not obvious why this one is wrong and the others are correct. A
change to posix_types will presumably affect other things too.

I'm not sure of the best way to solve this,

Is it possible /sensible to introduce a new CONFIG which selects int
or long for these types? I am not even sure what it could be called,
but then it could normally be set according to the X86_64 setting (in
U-Boot or SPL) but changed by EFI.

Or is this not a config setting, but a modulation of a config setting?
If so, then perhaps we can have:

#if CONFIG_IS_ENABLED(X86_64)
# if <some EFI condition>
#define USE_LONG_FOR_64BIT
#else
#define USE_INT_FOR_64BIT
#else
#define USE_LONG_FOR_64BIT
#endif

But in any case I think we need something explicit for the particular
'64-bit int' needed by this particular build of EFI.

I suppose another possibility is that efi_stub.c should not use
efi_uintn_t and we should just revert that patch? At present the
64-bit stub is quite clearly differentiated from the rest of U-Boot.
Are we just going to end up with a spiderweb of problems here?

>  typedef unsigned long  __kernel_size_t;
>  typedef long           __kernel_ssize_t;
>  #else
> --
> 2.7.4
>

Regards,
Simon