On 16 June 2017 at 09:45, Christoph Hellwig [off-list ref] wrote:

On Thu, Jun 15, 2017 at 02:12:58PM +0200, Linus Walleij wrote:

quoted

Currently the RPMB partition spawns a separate block device
named /dev/mmcblkNrpmb for each device with an RPMB partition,
including the creation of a block queue with its own kernel
thread and all overhead associated with this. On the Ux500
HREFv60 platform, for example, the two eMMCs means that two
block queues with separate threads are created for no use
whatsoever.

Yikes!  What an amazingly stupid design decision.

Unfortunate, there is more. :-)

We are actually registering at least three more block devices per eMMC
card (two boot partitions, and one general purpose partition). Except
for the main partition of course.

The difference compared to rpmb from the above, is that those are
actually general read/write partitions.

So all these partitions are on the same eMMC card, but being I/O
scheduled separately because there are separate block devices. Yeah,
starvation, latency, etc - all bad things comes with it. :-)

My point is, this is only the first step in re-working and fixing this
- and we really appreciate your review!

[...]

Kind regards
Uffe