Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1586468669-21456-1-git-send-email-victor.erminpour@oracle.com>
 <CAMj1kXGgYYx=9rq-Ficw3vS6FX_0nb-hRStXijh7H4zdD=+Gaw@mail.gmail.com> <aee6c7cc-24c7-2822-47d5-1e05413a8024@oracle.com>
In-Reply-To: <aee6c7cc-24c7-2822-47d5-1e05413a8024@oracle.com>
From:   Ard Biesheuvel <ardb@kernel.org>
Date:   Sat, 11 Apr 2020 08:39:00 +0200
Message-ID: <CAMj1kXFF3gR+LyE1VbLuwnRA3CzQtF90mv7rpw9x2RcyugO1JA@mail.gmail.com>
Subject: Re: [PATCH] efi/libstub/arm64: Enable __efistub_global define in
 .data section
To:     Victor Erminpour <victor.erminpour@oracle.com>,
        Arvind Sankar <nivedita@alum.mit.edu>
Cc:     linux-efi <linux-efi@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, 11 Apr 2020 at 00:12, Victor Erminpour
<victor.erminpour@oracle.com> wrote:
>
>
>
> On 4/10/20 1:09 AM, Ard Biesheuvel wrote:
> > On Thu, 9 Apr 2020 at 23:44, Victor Erminpour
> > <victor.erminpour@oracle.com> wrote:
> >>
> >> Enable the __efistub_global define to place variables in the
> >> .data section for both CONFIG_ARM and CONFIG_ARM64.
> >>
> >> This places the EFIstub sys_table variable and other EFIstub
> >> static variables in the .data section for both CONFIG_ARM and
> >> CONFIG_ARM64.
> >>
> >
> > What does that achieve?
>
> Hi Ard,
>
> Without placing these global variables in .data, I get the
> following errors when booting an ARM64 EFI system:
>
> EFI stub: ERROR: Exit boot services failed.
> EFI stub: ERROR: Failed to update FDT and exit boot services
>

Which boot loader are you using? Does this involve shim?

Also, does it help if you add 'efi=no_disable_early_pci_dma'?


>
> I know that the ARM64 linker script is supposed to put the
> .init.bss into the .init.data section, but I don't think this
> is happening for all systems.
>
> Having it explicitly enabled for CONFIG_ARM64 worked for me.
>

OK, thanks for the report. However, we will be removing
__efistub_global entirely during the next cycle, so this is not the
right fix.