Re: [PATCH 1/7] efi: pstore: disentangle from deprecated efivars module

[Ard Biesheuvel]

On Thu, 24 Sep 2020 at 11:45, Ard Biesheuvel <ardb@xxxxxxxxxx> wrote:
>
> On Wed, 23 Sep 2020 at 23:02, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> >
> > On Wed, Sep 23, 2020 at 08:43:21PM +0200, Ard Biesheuvel wrote:
> > > On Wed, 23 Sep 2020 at 20:41, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> > > >
> > > > On Wed, Sep 23, 2020 at 06:13:58PM +0200, Ard Biesheuvel wrote:
> > > > > The EFI pstore implementation relies on the 'efivars' abstraction,
> > > > > which encapsulates the EFI variable store in a way that can be
> > > > > overridden by other backing stores, like the Google SMI one.
> > > > >
> > > > > On top of that, the EFI pstore implementation also relies on the
> > > > > efivars.ko module, which is a separate layer built on top of the
> > > > > 'efivars' abstraction that exposes the [deprecated] sysfs entries
> > > > > for each variable that exists in the backing store.
> > > > >
> > > > > Since the efivars.ko module is deprecated, and all users appear to
> > > > > have moved to the efivarfs file system instead, let's prepare for
> > > > > its removal, by removing EFI pstore's dependency on it.
> > > > >
> > > > > Signed-off-by: Ard Biesheuvel <ardb@xxxxxxxxxx>
> > > >
> > > > With this and the other pstore patch, do the pstore self-tests still
> > > > pass on an EFI system?
> > > >
> > > > If so, please consider both:
> > > >
> > > > Acked-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > > >
> > >
> > > Selftests? Excellent! Are they documented too?
> >
> > Not really, but they're pretty simple:
> >
> > cd tools/testing/selftests/pstore
> > *double-check "config" against running kernel config*
> > ./pstore_tests
> > ./pstore_crash_test
> > *wait for system to reboot*
> > cd tools/testing/selftests/pstore
> > ./pstore_post_reboot_tests
> >
> > (though please test before/after, just to make sure other deltas haven't
> > broken things before your series -- I don't test EFI pstore with high
> > frequency)
> >
>
> I have done the 'before' test on three different EFI Linux systems
> (x86, arm64 and ARM), and they all give me something like the below
>
> === Pstore unit tests (pstore_tests) ===
> UUID=109d02e6-9395-4274-9554-2c078e87a662
> Checking pstore backend is registered ... ok
>   backend=efi
>   cmdline=BOOT_IMAGE=/vmlinuz-5.3.0-59-generic
> root=/dev/mapper/crypt-root ro quiet splash vt.handoff=1
> Checking pstore console is registered ... FAIL
> Checking /dev/pmsg0 exists ... FAIL
> Writing unique string to /dev/pmsg0 ... FAIL
>
> So I'm not sure if there is any point to doing the 'after' test if
> this is the baseline.

In any case, I confirmed that  the new efi-pstore module
- exposes existing pstore dmesg entries correctly
- captures oops and panic messages as dmesg-efi-xxxx entries as before