linux-kernel - Re: [PATCH] PCI/DOE: Expose the DOE protocols via sysfs

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

Message-ID: <20230731175216.GA19517@wunner.de>
Date:   Mon, 31 Jul 2023 19:52:16 +0200
From:   Lukas Wunner <lukas@...ner.de>
To:     Alistair Francis <alistair23@...il.com>
Cc:     Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        bhelgaas@...gle.com, linux-pci@...r.kernel.org,
        linux-kernel@...r.kernel.org,
        Alistair Francis <alistair.francis@....com>
Subject: Re: [PATCH] PCI/DOE: Expose the DOE protocols via sysfs

On Mon, Jul 31, 2023 at 11:30:29AM -0400, Alistair Francis wrote:
> Yep, a big benefit is lspci or other userspace tools to be able to see
> what DOE protocols are supported.

Fair enough, but please make that motivation explicit in the commit
message.

> I also have plans to expose DOE mailboxes to userspace. That way we
> can run the SPDM requester code (using libspdm) in userspace to
> communicate with devices using SPDM. That will allow device
> authentication for example.

That duplicates our effort to bring up authentication in the kernel:
https://github.com/l1k/linux/commits/doe

We had a lengthy debate on the pros and cons of doing SPDM in-kernel
versus in user space.  We realized that when resuming a device from
D3cold or recovering it after reset, the user space component performing
SPDM authentication might reside on the device (disk, network) being
resumed or reset, preventing its execution.

When resuming from system sleep, devices need to be reauthenticated
during the ->resume_noirq phase, i.e. with device interrupts disabled,
as drivers are allowed access to devices already in that phase.
So authentication (and encryption establishment) needs to happen this
early, when user space isn't available yet.

See the commit message of "PCI/CMA: Reauthenticate devices on reset
and resume" on the above-linked branch for details.

Thanks,

Lukas