Re: [PATCH v2 0/3] efi: add support for non-standard capsule headers

From: Bryan O'Donoghue
Date: Tue Apr 25 2017 - 11:07:35 EST


On 25/04/17 16:01, Ard Biesheuvel wrote:
> On 25 April 2017 at 16:00, Matt Fleming <matt@xxxxxxxxxxxxxxxxxxx> wrote:
>> On Wed, 19 Apr, at 08:32:59PM, Jan Kiszka wrote:
>>> This picks up the patches Ard send before in [1], including the
>>> "left-over" patches 6..8.
>>>
>>> As Ard suggested, I've taken updated patches 6 and 7 of him from [2]
>>> which address reviewer comments. Furthermore, I've changed patch 8 to
>>> factor out the Quark quirk logic from the overloaded
>>> efi_capsule_setup_info as requested by Matt and also applied Andy's
>>> suggestion to have a quirk dispatcher table with callbacks.
>>>
>>> Tested successfully on the IOT2040 - still without a working Galileo
>>> board.
>>>
>>> Jan
>>>
>>> [1] http://www.spinics.net/lists/linux-efi/msg11194.html
>>> [2] https://git.kernel.org/pub/scm/linux/kernel/git/ardb/linux.git/log/?h=quark-capsule
>>>
>>> Cc: Matt Fleming <matt@xxxxxxxxxxxxxxxxxxx>
>>>
>>> Ard Biesheuvel (2):
>>> efi/capsule-loader: Redirect calls to efi_capsule_setup_info via weak
>>> alias
>>> efi/capsule-loader: Use page addresses rather than struct page
>>> pointers
>>>
>>> Jan Kiszka (1):
>>> efi/capsule: Add support for Quark security header
>>>
>>> arch/x86/platform/efi/quirks.c | 137 ++++++++++++++++++++++++++++++++++
>>> drivers/firmware/efi/Kconfig | 9 +++
>>> drivers/firmware/efi/capsule-loader.c | 66 ++++++++--------
>>> drivers/firmware/efi/capsule.c | 7 +-
>>> include/linux/efi.h | 14 +++-
>>> 5 files changed, 197 insertions(+), 36 deletions(-)
>>
>> OK, this looks like it's in good shape to me.
>>
>> Ard, are we waiting for anything else before we queue this up for
>> v4.13?
>
> Ideally, we'd get someone to test it with compliant capsule headers,
> but I'm happy to just queue it for v4.13, and get it in -next asap
> (i.e., right after the merge window closes)
>

I'm OOO until Thursday however, I could test this out on a Galileo for
you with the CSH stripped out. As Jan has pointed out Galileo aka Quark
x1000 non-secure (not to be called insecure) will just skips past the
CSH anyway - so - we can give it a conformant capsule by stripping the
CSH junk off the top.

How about you queue it up and I'll let you know if there's a problem
Thursday?

--
bod