Re: [PATCH v6 0/9] dmaengine: add DMA channel bus and fsl-edma per-channel IOMMU support
From: Jason Gunthorpe
Date: Tue Nov 17 2026 - 08:00:31 EST
On Tue, Nov 17, 2026 at 04:30:15PM +0800, Peng Fan wrote:
> On Mon, Nov 16, 2026 at 03:30:21PM -0400, Jason Gunthorpe wrote:
> >On Thu, Nov 12, 2026 at 03:53:03PM +0800, Peng Fan (OSS) wrote:
> >> I used iommu-map for eDMA, but iommu experts rejected it:
> >> "Using iommu-map to describe synthetic dma_chan_dev devices that the
> >> kernel creates is the same kind of DT abuse from over here:"
> >> More info: https://lore.kernel.org/all/20260917144804.GI3196566@xxxxxxxx/
> >>
> >> So I followed here [2] as Krzysztof wrote for QCOM VPU:
> >> "Thanks Jason for context and detailed arguments. Robin did not chime in
> >> yet, but from what I understood the DT-child-node approach will be the
> >> way we should go here.
> >>
> >> I accept above point of view and I am fine with this, thus Vishnu and
> >> Vikash - please go ahead with DT-child-node solution we had some time ago."
> >> [2] https://lore.kernel.org/linux-iommu/c7b956a9-d3e8-4e18-b780-5d08f5cd2ca1@xxxxxxxxxx/
> >>
> >> This series adds per-channel DT child node support to fsl-edma and, as a
> >> prerequisite, introduces a proper bus for DMA channel devices.
> >
> >To be fair this is discouraged too, now that I've seen 3-4 people
> >trying to do exactly the same thing I'm not as positive about the
> >DT-child-node hack.
>
> I may not describe it clear in cover-letter.
>
> I understand the concern. Let me explain why I believe it is justified here
> rather than a hack.
>
> On i.MX eDMA each channel has its own interrupt, its own register region,
> and on i.MX8QM even its own power domain. Each channel may also have its
> own Stream ID, or share one when SIDs are scarce. These are genuine
> per-channel hardware resources, not software fiction.
>
> In dmaengine each channel is already modeled as a channel device (see
> dmaengine_chan_dev()), and this series puts those devices on a real bus
> (dmaengine_chan_bus_type) as a prerequisite. The DT child node simply
> describes that existing device, so its reg/interrupts/power-domains/iommus
> are attached per channel instead of as index-matched flat lists on the
> controller.
If they are real devices then why can't they come out as platform
devices?
I think you are overstating how independent they are. Many single
devices have per "channel" resources like MMIO and IRQs. That's
totally normal. It doesn't make the sub resources independent devices.
AFAICT what you have here is some HW designer used SID where they
should have used SSID and you are trying to hack DT to make that
usable because Linux does not support a SID as SSID model today.
Again, since this is the 3rd - 4th case like this, it seems like a
popular hardware choice and we should try to do better to make Linux
support if we can.
Jason