lkml.org 
[lkml]   [2016]   [Nov]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] x86/efi: Fix EFI memmap pointer size warning
On Sun, 06 Nov, at 02:02:54PM, Borislav Petkov wrote:
> Hi Matt,
>
> please doublecheck me on this but I think we're fine using an unsigned
> long.
>
> Thanks.
>
> ---
> From: Borislav Petkov <bp@suse.de>
> Date: Sun, 6 Nov 2016 13:49:10 +0100
> Subject: [PATCH] x86/efi: Fix EFI memmap pointer size warning
> MIME-Version: 1.0
> Content-Type: text/plain; charset=UTF-8
> Content-Transfer-Encoding: 8bit
>
> Fix this when building on 32-bit:
>
> arch/x86/platform/efi/efi.c: In function ‘__efi_enter_virtual_mode’:
> arch/x86/platform/efi/efi.c:911:5: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
> (efi_memory_desc_t *)pa);
> ^
> arch/x86/platform/efi/efi.c:918:5: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
> (efi_memory_desc_t *)pa);
> ^
>
> The @pa local variable is declared as phys_addr_t and that is a u64 when
> CONFIG_PHYS_ADDR_T_64BIT=y. (The last is enabled on 32-bit on a PAE
> build.)
>
> However, its value comes from __pa() which is basically doing pointer
> arithmetic and checking, and returns unsigned long as it is the native
> pointer width.
>
> So let's use an unsigned long too. It should be fine to do so because
> the later users cast it to a pointer too.
>
> Signed-off-by: Borislav Petkov <bp@suse.de>
> Cc: Matt Fleming <matt@codeblueprint.co.uk>
> ---
> arch/x86/platform/efi/efi.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/platform/efi/efi.c b/arch/x86/platform/efi/efi.c
> index bf99aa7005eb..936a488d6cf6 100644
> --- a/arch/x86/platform/efi/efi.c
> +++ b/arch/x86/platform/efi/efi.c
> @@ -861,7 +861,7 @@ static void __init __efi_enter_virtual_mode(void)
> int count = 0, pg_shift = 0;
> void *new_memmap = NULL;
> efi_status_t status;
> - phys_addr_t pa;
> + unsigned long pa;
>
> efi.systab = NULL;
>

Right, this does look correct given that __pa() uses 'unsigned long'.

I think the reason this is safe on PAE is that __get_free_pages() is
guaranteed to return a 32-bit address, and will not return HIGHMEM
pages. Which makes __pa() legal and 'unsigned long' the correct type.

Want me to take this through the EFI tree?

\
 
 \ /
  Last update: 2016-11-07 11:00    [W:0.043 / U:0.052 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site