Re: [tip:efi/urgent] efi: Align 'efi_guid_t' to 64 bits

From: Ard Biesheuvel
Date: Wed Dec 19 2018 - 17:50:28 EST


On Tue, 18 Dec 2018 at 00:20, Heinrich Schuchardt <xypron.glpk@xxxxxx> wrote:
>
> On 12/17/18 11:42 PM, Ard Biesheuvel wrote:
> > On Mon, 17 Dec 2018 at 23:33, Heinrich Schuchardt <xypron.glpk@xxxxxx> wrote:
> >>
> >> On 12/17/18 7:16 PM, tip-bot for Heinrich Schuchardt wrote:
> >>> Commit-ID: 793423cf07e51e3185b8680167115813589c057d
> >>> Gitweb: https://git.kernel.org/tip/793423cf07e51e3185b8680167115813589c057d
> >>> Author: Heinrich Schuchardt <xypron.glpk@xxxxxx>
> >>> AuthorDate: Mon, 17 Dec 2018 19:02:14 +0100
> >>> Committer: Ingo Molnar <mingo@xxxxxxxxxx>
> >>> CommitDate: Mon, 17 Dec 2018 19:12:48 +0100
> >>>
> >>> efi: Align 'efi_guid_t' to 64 bits
> >>>
> >>> The UEFI Specification Version 2.7 Errata A defines:
> >>>
> >>> "EFI_GUID
> >>> 128-bit buffer containing a unique identifier value.
> >>> Unless otherwise specified, aligned on a 64-bit boundary."
> >>>
> >>> Before this patch efi_guid_t was only 8-bit aligned.
> >>>
> >>> Note that this could potentially trigger alignment faults during
> >>> EFI runtime services calls on 32-bit ARM, given that it does not
> >>> permit load/store double or load/store multiple instructions to
> >>> operate on memory addresses that are not 32-bit aligned.
> >>>
> >>> Signed-off-by: Heinrich Schuchardt <xypron.glpk@xxxxxx>
> >>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
> >>> Cc: <stable@xxxxxxxxxxxxxxx> # v4.9+, or earlier if possible
> >>> Cc: Andy Lutomirski <luto@xxxxxxxxxx>
> >>> Cc: Borislav Petkov <bp@xxxxxxxxx>
> >>> Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
> >>> Cc: H. Peter Anvin <hpa@xxxxxxxxx>
> >>> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> >>> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> >>> Cc: Qian Cai <cai@xxxxxx>
> >>> Cc: Rik van Riel <riel@xxxxxxxxxxx>
> >>> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> >>> Cc: linux-efi@xxxxxxxxxxxxxxx
> >>> Link: http://lkml.kernel.org/r/20181217180214.9436-3-ard.biesheuvel@xxxxxxxxxx
> >>> Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
> >>> ---
> >>> include/linux/efi.h | 2 +-
> >>> 1 file changed, 1 insertion(+), 1 deletion(-)
> >>>
> >>> diff --git a/include/linux/efi.h b/include/linux/efi.h
> >>> index 100ce4a4aff6..e6480c805932 100644
> >>> --- a/include/linux/efi.h
> >>> +++ b/include/linux/efi.h
> >>> @@ -48,7 +48,7 @@ typedef u16 efi_char16_t; /* UNICODE character */
> >>> typedef u64 efi_physical_addr_t;
> >>> typedef void *efi_handle_t;
> >>>
> >>> -typedef guid_t efi_guid_t;
> >>> +
> >>>
> >>> #define EFI_GUID(a,b,c,d0,d1,d2,d3,d4,d5,d6,d7) \
> >>> GUID_INIT(a, b, c, d0, d1, d2, d3, d4, d5, d6, d7)
> >>>
> >>
> >> Before rushing this patch in, we should carefully review its side
> >> effects, e.g. on 32bit system this changes the size of
> >> efi_config_table_32_t from 20 to 24, which is part of the interface to
> >> the UEFI firmware.
> >>
> >
> > grmbl.
> >
> > Thanks for spotting that.
> >
> > The UEFI spec defines a GUID struct as { UINT32; UINT16; UINT16;
> > UINT8[8]; } so its natural alignment is 32 bits not 64 bits. The
> > alignment issue on ARM would be solved by using __aligned(4) rather
> > than __aligned(8), while not affecting the size of the config table
> > struct (and potentially others) on 32-bit architectures.
> >
> > Ingo, apologies for the breakage. Do you prefer a replacement patch or
> > a followup patch?
> >
>
> The UEFI spec explicitly requires EFI_GUID to be 64-bit aligned. On the
> other hand neither EDK2 nor GRUB not U-Boot cared about this requirement
> up to now. So the array of efi_config_table_32_t had 20 byte long
> members at least on Linux, EDK2, GRUB, U-Boot, and possibly on other
> UEFI implementations though the UEFI spec does not mention packing here.
>
> As the 4.20 kernel release is imminent, please, keep this patch out
> before causing breakage.
>

Ingo,

Let's just drop Heinrich's patch for now, and I will come back to it
for the next release. There is very little EFI on 32-bit ARM in the
field, and this patch in its current form will break IA-32 as well, so
a quick followup fix is probably not the best approach here.