On Tue, Jun 16, 2020 at 03:07:17AM +0000, Mark Tomlinson wrote:
> On Mon, 2020-06-15 at 15:32 +0100, Mark Brown wrote:

> > Again was this done for a reason and if so do we understand why doing
> > this from interrupt context is safe - how long can the interrupts be
> > when stuffing the FIFO from interrupt context?

> As I'm porting a Broadcom patch, I'm hoping someone else can add
> something to this. From the history it appears there was a hard limit

If you didn't write this code then it should have at least one signoff
from the original source, I can't do anything with this without signoffs
- please see Documentation/process/submitting-patches.rst for what this
is and why it's important.

> (no small chunks), and this was changed to doing it in chunks with
> patch 345309fa7c0c92, apparently to improve performance. I believe this
> change further improves performance, but as the patch arrived without
> any documentation, I'm not certain.

Have you tested the impact on performance?