linux-kernel - Re: [PATCH] arm64: dts: ti: k3-j721s2: fix overlapping GICD memory region

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]

Date:   Thu, 16 Jun 2022 13:19:08 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     Marc Zyngier <maz@...nel.org>
Cc:     Matt Ranostay <mranostay@...com>, Nishanth Menon <nm@...com>,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] arm64: dts: ti: k3-j721s2: fix overlapping GICD memory
 region

On 2022-06-16 13:13, Marc Zyngier wrote:
> On Thu, 16 Jun 2022 13:07:23 +0100,
> Robin Murphy <robin.murphy@....com> wrote:
>>
>> On 2022-06-16 11:51, Matt Ranostay wrote:
>>> GICD region was overlapping with GICR causing the latter to not map
>>> successfully, and in turn the gic-v3 driver would fail to initialize.
>>>
>>> This issue was hidden till commit 2b2cd74a06c3 ("irqchip/gic-v3: Claim iomem resources")
>>> replaced of_iomap() calls with of_io_request_and_map() that internally
>>> called request_mem_region().
>>>
>>> Respective console output before this patchset:
>>>
>>> [    0.000000] GICv3: /bus@...000/interrupt-controller@...0000: couldn't map region 0
>>
>> Oh, it's nice that this finds bugs, but it seems I hadn't fully
>> considered that making the simple easy change in the DT paths results
>> in different behaviour from ACPI.
>>
>> Marc, would you like a fix for this to remain non-fatal even in the
>> face of a dodgy DT, or are you happy with being a bit stricter now?
> 
> I'd rather we work around it. I shout at people for breaking existing
> DTs, so this should apply to the GIC as well. A nice WARN_ON_ONCE()
> should do, if you don't mind writing the patch.

Indeed, I think that would be my default preference as well - apologies 
for the oversight! I'll spin the patch shortly (and make the warning 
consistent for the ACPI side too).

Cheers,
Robin.