linux-kernel - Re: [PATCH v2 0/7] arm64: Make EFI calls preemptible

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <CAMj1kXEonDpry5ZJoApGfqP95qFByAWGxcCFtfFycspT6-V8Rw@mail.gmail.com>
Date: Mon, 15 Sep 2025 11:05:11 +0200
From: Ard Biesheuvel <ardb@...nel.org>
To: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Cc: Ard Biesheuvel <ardb+git@...gle.com>, linux-efi@...r.kernel.org, 
	linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org, 
	Will Deacon <will@...nel.org>, Mark Rutland <mark.rutland@....com>, 
	Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH v2 0/7] arm64: Make EFI calls preemptible

Hi,

Thanks for taking a look.

On Mon, 15 Sept 2025 at 10:52, Sebastian Andrzej Siewior
<bigeasy@...utronix.de> wrote:
>
> On 2025-09-05 15:30:36 [+0200], Ard Biesheuvel wrote:
> > From: Ard Biesheuvel <ardb@...nel.org>
> >
> …
> > Note that this is only a partial solution in terms of RT guarantees,
> > given that the runtime services execute at the same privilege level as
> > the kernel, and can therefore disable interrupts (and therefore
> > preemption) directly. But it should prevent scheduling latency spikes
> > for EFI calls that simply take a long time to run to completion.
>
> That sounds nice. There is no feature flag that could tell if a specific
> EFI-call (or any) will disable interrupts, right?

Sadly, no. At runtime, the EFI APIs that manage this at a higher level
of abstraction are no longer available, and so the only available
option for firmware to ensure that code runs uninterrupted is to mask
interrupts at the CPU side. Everything else (timers, interrupt
controllers) is owned by the OS at this point, so runtime firmware
cannot touch it (even if it wanted to - it has no idea where memory
mapped peripherals live in the OS's memory map that it runs under)

It would be nice if we could sandbox this in a VM but that is not
straight-forward.

> But if the source code
> is available, you could check.
>

Even though much of the code is based on the public reference
implementation, the tweaks that require playing with interrupt
masking/unmasking are often part of the downstream, closed source
forks.