linux-kernel - Re: [PATCH] thunderbolt: Make iommu_dma

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: <YjSkAcxLDFmVdhEq@lahna>
Date:   Fri, 18 Mar 2022 17:23:45 +0200
From:   "mika.westerberg@...ux.intel.com" <mika.westerberg@...ux.intel.com>
To:     Robin Murphy <robin.murphy@....com>
Cc:     "Limonciello, Mario" <Mario.Limonciello@....com>,
        "andreas.noever@...il.com" <andreas.noever@...il.com>,
        "michael.jamet@...el.com" <michael.jamet@...el.com>,
        "YehezkelShB@...il.com" <YehezkelShB@...il.com>,
        "linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "iommu@...ts.linux-foundation.org" <iommu@...ts.linux-foundation.org>,
        "linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>
Subject: Re: [PATCH] thunderbolt: Make iommu_dma_protection more accurate

Hi Robin,

On Fri, Mar 18, 2022 at 03:15:19PM +0000, Robin Murphy wrote:
> > IMHO we should just trust the firmare provided information here
> > (otherwise we are screwed anyway as there is no way to tell if the
> > devices connected prior the OS can still do DMA), and use the external
> > facing port indicator to idenfity the ports that need DMA protection.
> 
> Indeed that's exactly what I want to do, but it begs the question of how we
> *find* the firmware-provided information in the first place!

Oh, right :) Its the combination of ACPI _DSD "ExternalFacingPort"
(which we already set, dev->external_facing, dev->untrusted for the
devices behind these ports IIRC) and the DMAR opt-in bit. All these are
already read by the kernel.

> I seem to have already started writing the dumb version that will walk the
> whole PCI segment and assume the presence of any external-facing port
> implies that we're good. Let me know if I should stop ;)

That sounds good to me, so don't stop just yet ;-)