linux-kernel - Re: [PATCH] PCI: Add quirk for ASPEED AST1150 bridge to prevent false RID aliasing

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>] [day] [month] [year] [list]

Message-ID: <72c7fb61-a4b8-4443-86ff-84345fe1ced4@arm.com>
Date: Wed, 17 Dec 2025 13:48:34 +0000
From: Robin Murphy <robin.murphy@....com>
To: Nirmoy Das <nirmoyd@...dia.com>, Bjorn Helgaas <bhelgaas@...gle.com>
Cc: linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
 Jason Gunthorpe <jgg@...dia.com>, Will Deacon <will@...nel.org>,
 Joerg Roedel <joro@...tes.org>, iommu@...ts.linux.dev,
 jammy_huang@...eedtech.com
Subject: Re: [PATCH] PCI: Add quirk for ASPEED AST1150 bridge to prevent false
 RID aliasing

On 17/12/2025 1:32 pm, Nirmoy Das wrote:
> ASPEED BMC controllers have VGA and USB functions behind a PCIe-to-PCI
> bridge that causes them to share the same stream ID:
> 
>    [e0]---00.0-[e1-e2]----00.0-[e2]--+-00.0  ASPEED Graphics Family
>                                      \-02.0  ASPEED USB Controller
> 
> Both devices get stream ID 0x5e200 due to bridge aliasing, causing the
> USB controller to be rejected with 'Aliasing StreamID unsupported'.
> 
> Per ASPEED, the AST1150 operates in PCI mode where downstream devices
> generate their own distinct Requester IDs. Set
> PCI_DEV_FLAGS_BRIDGE_XLATE_ROOT to stop false alias detection.

I don't think that's really the right quirk - that one effectively means 
"RID aliasing upstream of this bridge is irrelevant", whereas what you 
want here is "RID aliasing *at* this bridge doesn't happen even though 
it looks like it could". With BRIDGE_XLATE_ROOT, I think if you then put 
this behind a contrived upstream chain of additional PCIe->PCI->PCIE 
bridges that *could* legitimately introduce further aliasing, it would 
give the wrong result.

Thanks,
Robin.

> Signed-off-by: Nirmoy Das <nirmoyd@...dia.com>
> ---
>   drivers/pci/quirks.c | 14 ++++++++++++++
>   1 file changed, 14 insertions(+)
> 
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index b9c252aa6fe0..a7a1bf4c4354 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -4453,6 +4453,20 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_BROADCOM, 0x9000,
>   DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_BROADCOM, 0x9084,
>   				quirk_bridge_cavm_thrx2_pcie_root);
>   
> +/*
> + * ASPEED AST1150 is a PCIe-to-PCI bridge that operates in PCI mode (not PCI-X).
> + * Although it reports as a PCIe-to-PCI bridge, the downstream PCI bus does not
> + * perform conventional Requester ID aliasing - each device behind the bridge
> + * generates its own distinct Requester ID. This quirk stops false alias
> + * detection at the bridge, fixing IOMMU StreamID conflicts that break DMA for
> + * devices like the USB controller behind this bridge.
> + */
> +static void quirk_aspeed_ast1150_bridge(struct pci_dev *pdev)
> +{
> +	pdev->dev_flags |= PCI_DEV_FLAGS_BRIDGE_XLATE_ROOT;
> +}
> +DECLARE_PCI_FIXUP_HEADER(0x1a03, 0x1150, quirk_aspeed_ast1150_bridge);
> +
>   /*
>    * Intersil/Techwell TW686[4589]-based video capture cards have an empty (zero)
>    * class code.  Fix it.