linux-kernel - Re: [PATCH v6 6/6] arm64: Use SYSTEM_OFF2 PSCI call to power off for hibernate

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAMj1kXFma8-GqKuOs5-UAQY9asbq2p9EubSjjbywaURa4T4WnA@mail.gmail.com>
Date: Mon, 4 Nov 2024 14:54:12 +0100
From: Ard Biesheuvel <ardb@...nel.org>
To: Lorenzo Pieralisi <lpieralisi@...nel.org>
Cc: David Woodhouse <dwmw2@...radead.org>, sami.mujawar@....com, 
	Paolo Bonzini <pbonzini@...hat.com>, Jonathan Corbet <corbet@....net>, Marc Zyngier <maz@...nel.org>, 
	Oliver Upton <oliver.upton@...ux.dev>, James Morse <james.morse@....com>, 
	Suzuki K Poulose <suzuki.poulose@....com>, Zenghui Yu <yuzenghui@...wei.com>, 
	Catalin Marinas <catalin.marinas@....com>, Will Deacon <will@...nel.org>, 
	Mark Rutland <mark.rutland@....com>, "Rafael J. Wysocki" <rafael@...nel.org>, Pavel Machek <pavel@....cz>, 
	Len Brown <len.brown@...el.com>, Shuah Khan <shuah@...nel.org>, 
	David Woodhouse <dwmw@...zon.co.uk>, kvm@...r.kernel.org, linux-doc@...r.kernel.org, 
	linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org, 
	kvmarm@...ts.linux.dev, linux-pm@...r.kernel.org, 
	linux-kselftest@...r.kernel.org, 
	Francesco Lavra <francescolavra.fl@...il.com>, Miguel Luis <miguel.luis@...cle.com>
Subject: Re: [PATCH v6 6/6] arm64: Use SYSTEM_OFF2 PSCI call to power off for hibernate

On Fri, 1 Nov 2024 at 18:49, Lorenzo Pieralisi <lpieralisi@...nel.org> wrote:
>
> [+Ard, Sami, for EFI]
>
> On Thu, Oct 31, 2024 at 06:55:43PM +0100, Lorenzo Pieralisi wrote:
> > On Sat, Oct 19, 2024 at 06:15:47PM +0100, David Woodhouse wrote:
> >
> > [...]
> >
> > > +#ifdef CONFIG_HIBERNATION
> > > +static int psci_sys_hibernate(struct sys_off_data *data)
> > > +{
> > > +   /*
> > > +    * Zero is an acceptable alternative to PSCI_1_3_OFF_TYPE_HIBERNATE_OFF
> > > +    * and is supported by hypervisors implementing an earlier version
> > > +    * of the pSCI v1.3 spec.
> > > +    */
> >
> > It is obvious but with this patch applied a host kernel would start executing
> > SYSTEM_OFF2 too if supported in firmware to hibernate, it is not a hypervisor
> > only code path.
> >
> > Related to that: is it now always safe to override
> >
> > commit 60c0d45a7f7a ("efi/arm64: use UEFI for system reset and poweroff")
> >
> > for hibernation ? It is not very clear to me why overriding PSCI for
> > poweroff was the right thing to do - tried to follow that patch history but
> > the question remains (it is related to UpdateCapsule() but I don't know
> > how that applies to the hibernation use case).
>
> RFC: It is unclear to me what happens in current mainline if we try to
> hibernate with EFI runtime services enabled and a capsule update pending (we
> issue EFI ResetSystem(EFI_RESET_SHUTDOWN,..) which might not be compatible
> with the reset required by the pending capsule update request) what happens
> in this case I don't know but at least the choice is all contained in
> EFI firmware.
>
> Then if in the same scenario now we are switching to PSCI SYSTEM_OFF2 for the
> hibernate reset I suspect that what happens to the in-flight capsule
> update requests strictly depends on what "reset" PSCI SYSTEM_OFF2 will
> end up doing ?
>
> I think this is just a corner case and it is unlikely it has been ever
> tested (is it even possible ? Looking at EFI folks) - it would be good
> to clarify it at least to make sure we understand this code path.
>

I'm not aware of any OS that actually uses capsule update at runtime
(both Windows and Linux queue up the capsule and call the
UpdateCapsule() runtime service at boot time after a reboot).

So it is unlikely that this would break anything, and I'd actually be
inclined to disable capsule update at runtime altogether.

I will also note that hibernation with EFI is flaky in general, given
that EFI memory regions may move around