Re: [PATCH 04/22] arm64: dts: qcom: hamoa: Reserve low IOVA range for Iris
From: Vikash Garodia
Date: Wed Aug 12 2026 - 07:24:02 EST
On 8/12/2026 10:49 AM, Bryan O'Donoghue wrote:
On 11/08/2026 01:17, Dmitry Baryshkov wrote:
You are well aware of them, but still asking the same. Reasons,I think that we should have concentrated on it. From my point of view,
1. We have been attempting the subnodes for almost a year with multiple
pushbacks from multiple maintainers. We are closer and working on it to post
for iris, followed by venus, once review is acceptable for iris.
the subnodes solution is ready and the last obstacle was the
under-informing commit message. As you can see, Rob has (had?) questions
about this solution too.
Its taking too long.
I just suggested restricting concurrency but, that can't be right. The issue is IOVA - concurrency merely makes the bug more _likely_ to show up so in fact that won't work.
-rc1 hits soon, if sub-nodes isn't ready to go by that point there is no choice but to implement BROKEN or minItems, 100% up to qcom to decide which.
minItems is the way which could easily fix and patch all the kernels exposed with the known bug.
At the same time, we have identified a way to even reserve this in driver now (venus/iris) with [1] and the results are good with this as well. If not minItems, would prefer [1] in -rc1 to avoid further delays or to keep the crash exposed. Please review.
https://lore.kernel.org/linux-media/20260812-reserve_iova_in_driver-v1-0-ed62f801275c@xxxxxxxxxxxxxxxx/
Regards,
Vikash
---
bod