linux-kernel - Re: [PATCH] arm64: dts: ti: k3-j722s-main: add legacy PCIe interrupts

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: <663079dc-54cf-4f29-a68b-fc9678c44af0@ti.com>
Date: Thu, 21 Aug 2025 10:39:15 +0530
From: s-vadapalli <s-vadapalli@...com>
To: Michael Walle <mwalle@...nel.org>
CC: s-vadapalli <s-vadapalli@...com>, Nishanth Menon <nm@...com>,
        Vignesh
 Raghavendra <vigneshr@...com>,
        Tero Kristo <kristo@...nel.org>, Rob Herring
	<robh@...nel.org>,
        Krzysztof Kozlowski <krzk+dt@...nel.org>,
        Conor Dooley
	<conor+dt@...nel.org>,
        <linux-arm-kernel@...ts.infradead.org>, <devicetree@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] arm64: dts: ti: k3-j722s-main: add legacy PCIe interrupts

On Wed, Aug 20, 2025 at 04:00:45PM +0200, Michael Walle wrote:
> Hi Siddharth,
> 
> On Wed Aug 20, 2025 at 3:52 PM CEST, s-vadapalli wrote:
> > On Tue, Aug 19, 2025 at 01:13:17PM +0200, Michael Walle wrote:
> >
> > Hello Michael,
> >
> >> The kernel will try to map the legacy interrupt pins, but the
> >> interrupt mapping is missing from the device tree and thus that
> >> fails with:
> >> 
> >>    pcieport 0000:00:00.0: of_irq_parse_pci: failed with rc=-22
> >> 
> >> Add the node for the legacy PCIe interrupts to fix that. This is just
> >> compile-time tested.
> >
> > INTx is not supported by the driver as explained at:
> > https://lore.kernel.org/r/be3e3c5f-0d48-41b0-87f4-2210f13b9460@ti.com/
> >
> > The patch to fix the error displayed in the logs was posted at:
> > https://lore.kernel.org/r/20240726135903.1255825-1-s-vadapalli@ti.com/
> > but wasn't accepted as-is. A different approach will be required to fix
> > of_irq_parse_pci() instead as pointed out at:
> > https://lore.kernel.org/r/20240729080006.GA8698@thinkpad/
> 
> Thanks for the pointers.
> 
> I've just checked the J722S errata sheet and there is no such
> erratum. So, is that sheet outdated or was it fixed?

According to the Errata documents, the errata only affects J721E SoC.
The procedure to enable the INTx support is the following:
1. Update DT bindings to indicate that INTx is valid only for SoCs other
   than J721E SoCs on the basis of the compatible.
2. Add support in the pci-j721e.c driver for INTx.
3. Finally, the current patch or patches for other SoCs to add INTx can
   be posted and merged.

The following series which was posted in 2021 corresponds to 1 and 2
above:
https://lore.kernel.org/r/20210811123846.31921-1-kishon@ti.com/

Please note that the above series was included in the downstream (TI)
Linux in the past. It was observed that despite the Errata not affecting
the AM64 SoC (contains a Cadence PCIe Controller and is programmed by the
same pci-j721e.c driver), legacy interrupts were missed and resulted in the
PCIe Endpoint device becoming non-functional. Resolving this issue
required retriggering the interrupt following which the device became
functional.

Therefore, it is encouraged that MSI/MSI-X are used instead. If INTx is a
necessity (unlike the current patch which only aims to prevent error logs
from showing up), then the aforementioned series can be used with the
disclaimer that it may still lead to interrupts being missed.

Regards,
Siddharth.