Re: [PATCH v2 0/2] DMM DT adaptation

[Tomi Valkeinen <tomi.valkeinen-l0cyMroinI0@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 1487 bytes --]

On 16/09/13 12:30, Archit Taneja wrote:
> The DMM/Tiler block can used by omapdrm to allocate frame buffers. With the
> removal of address and irq data from the omap4 hwmods, the probe of DMM driver
> fails and omapdrm isn't able to utilize the DMM hardware.
> 
> Add DMM bindings for omap4 and omap5.
> 
> Changes in v2:
> - No changes, split out into a separate series containing only DT related parts.
> 
> Archit Taneja (2):
>   arm: dts: omap4+: Add DMM bindings
>   drm: omap: Enable DT support for DMM

This is not an objection as such, but I just want to point out that DMM
is part of the memory subsystem, and Tiler is part of DMM. So it's not
linked to the display subsystem in any way. And with a quick glance to
the TRM, DMM manages _all_ accesses to system memory.

In that sense, having a "DMM" driver as part of omapdrm is quite wrong.
If it was just a "Tiler" driver, it'd be better, but still wrong.

That said, if I remember right I brought this point up long ago, and it
was pointed out that only the Tiler part of DMM is configured at
runtime, and more or less all the registers of DMM are about Tiler. And
only omapdrm uses the Tiler. Thus it was deemed ok to have the DMM
driver in omapdrm.

I'd still rather have tiler as a separate driver, but that's a separate
issue from this series.

So, after fixing the issues I pointed in the other mail:

Acked-by: Tomi Valkeinen <tomi.valkeinen-l0cyMroinI0@public.gmane.org>

 Tomi



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 901 bytes --]