Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221006224212.569555-1-gpiccoli@igalia.com> <20221006224212.569555-9-gpiccoli@igalia.com>
 <202210061614.8AA746094A@keescook> <CAMj1kXF4UyRMh2Y_KakeNBHvkHhTtavASTAxXinDO1rhPe_wYg@mail.gmail.com>
 <f857b97c-9fb5-8ef6-d1cb-3b8a02d0e655@igalia.com>
In-Reply-To: <f857b97c-9fb5-8ef6-d1cb-3b8a02d0e655@igalia.com>
From:   Ard Biesheuvel <ardb@kernel.org>
Date:   Fri, 7 Oct 2022 15:19:15 +0200
Message-ID: <CAMj1kXFy-2KddGu+dgebAdU9v2sindxVoiHLWuVhqYw+R=kqng@mail.gmail.com>
Subject: Re: [PATCH 8/8] efi: pstore: Add module parameter for setting the
 record size
To:     "Guilherme G. Piccoli" <gpiccoli@igalia.com>
Cc:     Kees Cook <keescook@chromium.org>, linux-kernel@vger.kernel.org,
        linux-fsdevel@vger.kernel.org, kernel-dev@igalia.com,
        kernel@gpiccoli.net, anton@enomsg.org, ccross@android.com,
        tony.luck@intel.com, linux-efi@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 7 Oct 2022 at 15:00, Guilherme G. Piccoli <gpiccoli@igalia.com> wrote:
>
> First of all, thanks Ard for the historical explanation!
>
>
> On 07/10/2022 06:11, Ard Biesheuvel wrote:
> > [...]
> >> I think it'd be great to make it configurable! Ard, do you have any
> >> sense of what the max/min, etc, should be here?
> >>
> >
> > Given that dbx on an arbitrary EFI system with secure boot enabled is
> > already almost 4k, that seems like a reasonable default. As for the
> > upper bound, there is no way to know what weird firmware bugs you
> > might tickle by choosing highly unusual values here.
> >
> > If you need to store lots of data, you might want to look at [0] for
> > some work done in the past on using capsule update for preserving data
> > across a reboot. In the general case, this is not as useful, as the
> > capsule is only delivered to the firmware after invoking the
> > ResetSystem() EFI runtime service (as opposed to SetVariable() calls
> > taking effect immediately). However, if you need to capture large
> > amounts of data, and can tolerate the uncertainty involved in the
> > capsule approach, it might be a reasonable option.
> >
> > [0] https://lore.kernel.org/all/20200312011335.70750-1-qiuxu.zhuo@intel.com/
>
> So, you mean 4K as the default? I can change, but I would try to not
> mess with the current users, is there a case you can imagine something
> like 4k would fail? Maybe 2K is safer?
>

Reducing the number of writes 4x on systems that can support this has
its own advantages, so I am willing to risk it.

> As for the maximum, I've tested with many values, and when it's larger
> than ~30000 for edk2/ovmf, it fails with EFI_OUT_OF_RESOURCES and
> doesn't collect the log; other than that, no issues observed.
>

OVMF has

OvmfPkg/OvmfPkgX64.dsc:
gEfiMdeModulePkgTokenSpaceGuid.PcdMaxVariableSize|0x2000
OvmfPkg/OvmfPkgX64.dsc:
gEfiMdeModulePkgTokenSpaceGuid.PcdMaxVariableSize|0x8400

where the first one is without secure boot and the second with secure boot.

Interestingly, the default is

gEfiMdeModulePkgTokenSpaceGuid.PcdMaxVariableSize|0x400

so this is probably where this 1k number comes from. So perhaps it is
better to leave it at 1k after all :-(

> When set to ~24000, the interesting is that we have fewer big logs in
> /sys/fs/pstore, so it's nice to see compared to the bunch of 1K files heheh
>
> Anyway, let's agree on the default and then I can resubmit that, I'm
> glad you both consider that it's a good idea =)
>
> Thanks,
>
>
> Guilherme