06.03.2020 16:37, Ulf Hansson пишет:
...

quoted

Technically, it should be possible to chain-load some custom secondary
bootloader instead of a kernel image, but this is not very practical
because now:

1. There is a need to make a custom bootloader and it is quite a lot of
work.

2. You'll have to tell everybody that a custom booloader may need to be
used in order to get a working eMMC.

Yeah, I get the point. It's not an optimal situation, but I assume
it's about informing developers. They can cope with this, no?

Perhaps no, it's not only about the informing. The need for a custom
bootloader creates other inconveniences because:

1. It won't be possible to boot a vanilla upstream kernel using
Android's "fastboot boot ..." without applying extra patches to kernel
for the partition table support. Advanced users usually tend to use
fastboot and it's also very useful for a regular development purposes as
well.

2. Somebody (a developer / advanced user) will have to create a custom
bootloader for each device in the first place. This is not what an
average person will be able to do and there are not that many developers
who would want to dedicate theirs time to this.

3. The entry barrier for upstreaming Android devices support to the
kernel is already quite enormous. Adding extra hurdles isn't a step into
the right direction, IMO.

quoted

3. NVIDIA's bootloader already passes a command line parameter to kernel
for locating GPT entry, but this hack is not acceptable for the upstream
kernel.

Well, I am just worried that we will end up with one partition format
per vendor/product, that wouldn't scale very well.

In any case, from mmc point of view I am less concerned, we can find a
way to support the needed bits. I just need to review the series more
carefully and provide some comments. :-)

However, before I do that, I would like to hear Jens opinion about
adding a new partition format, so I don't waste my time here.

Sure, no problems :) Let's wait for the comments from Jens.