Re: [RFC] mmc: meson-gx: use memcpy_to/fromio for dram-access-quirk

From: Robin Murphy
Date: Wed Jun 09 2021 - 11:01:25 EST


On 2021-06-09 15:55, Neil Armstrong wrote:
Hi Robin,

On 09/06/2021 16:45, Robin Murphy wrote:
On 2021-06-09 14:07, Neil Armstrong wrote:
Hi,

On 08/06/2021 17:50, Marek Szyprowski wrote:
Hi

On 08.06.2021 17:33, Neil Armstrong wrote:
It has been reported that usage of memcpy() to/from an iomem mapping is invalid,
and and recent arm64 memcpy update [1] triggers a memory abort when dram-access-quirk
is used on the G12A/G12B platforms.

This adds a local sg_copy_to_buffer which makes usage of io versions of memcpy
when dram-access-quirk is enabled.

Fixes: acdc8e71d9bb ("mmc: meson-gx: add dram-access-quirk")
Reported-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
Suggested-by: Mark Rutland <mark.rutland@xxxxxxx>
Signed-off-by: Neil Armstrong <narmstrong@xxxxxxxxxxxx>

[1] 285133040e6c ("arm64: Import latest memcpy()/memmove() implementation")

Signed-off-by: Neil Armstrong <narmstrong@xxxxxxxxxxxx>
---
Hi Ulf, Marek, Mark,

I haven't tested the patch yet, but should fix issue reported at [2].

Works fine here and fixed the issue.

Tested-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>

Thanks, I'll need to rework to pass an __iomem pointer to memcpy_to/fromio so sparse doesn't scream anymore.

Hmm, might it be worth factoring out general sg_copy_{to,from}_iomem() helpers? From a quick grep I found at least mv_cesa_sg_copy() already doing essentially the same thing as meson_mmc_copy_buffer().

It's definitely worth it, but since we need a quick fix, we should have meson_mmc_copy_buffer() as a fix then we should definitely move to sg_copy_{to,from}_iomem() helpers

Oh, that makes sense for sure - I was just doing some general thinking out loud before I forget about the whole thing :)

Cheers,
Robin.