[Arm-netbook] A20 CPU cards going out; micro-engineering board being made up.

luke.leighton luke.leighton at gmail.com
Wed Jun 5 22:02:51 BST 2013


ok, the invoices for the 1st batch of A20 test samples are going out:
we're not charging VAT and they will be specifically marked (indelibly
and permanently in order to comply with customs instructions) as test
samples, as well as marking them clearly as "of negligable value" on
the customs forms when they are shipped.  if recipients *are* charged
VAT or Customs Duty, you *should* claim it back.  if you are asked to
fill in any customs forms, please be absolutely clear, according to
the laws in your country, as to how to deal with test / engineering
samples.  for those people in the UK, that's here:
http://customs.hmrc.gov.uk/channelsPortalWebApp/channelsPortalWebApp.portal?_nfpb=true&_pageLabel=pageLibrary_ShowContent&id=HMCE_CL_000061&propertyType=document

ok.  so.  henrik's pointed out that by the time these arrive, he
should have support for the A20 in sunxi u-boot.

i've uploaded a kernel tree here which includes a rather important
bug-fix (in some USB debugging messages ironically) and potentially
something irrelevant (a reset of the USB bus) but that's what's in the
A10 source code so i decided to add it and it works:
http://git.rhombus-tech.net/?p=linux.git;a=tree;h=refs/heads/lkcl-3.3-a20;hb=refs/heads/lkcl-3.3-a20

and the config i'm currently experimenting with is here:
http://rhombus-tech.net/allwinner/a20/boot/eoma68-a20-config

and the script.fex is here:
http://rhombus-tech.net/allwinner/a20/boot/script.fex

please note that to get a compile of the kernel, allwinner have *not*
used anything and i mean that truly they have used *nothing* that the
free software community has worked with and enhanced over the past 18
months, they have regressed *right* back and completely ignored
everything we've done.  it is therefore a bit of a shock to have to
run that age-old "build.sh" script and to have that really odd and
separate compile of the power-management assembly code _not_
integrated into the kernel build itself, because they used a different
gcc compiler for it.  YMMV here and you may even find that those
stupid, stupid console framebuffer crashes are back.  fixed in the
community 3.4 code, fscked in allwinner's own code.  awesome.

also one thing i find hilarious is that MALI is entirely missing from
the allwinner a20 3.3 release.  they've promised to get a 3.4 kernel
out by the end of june, however the question remains in my mind, "how
the hell can any of the people claiming to be selling A20 tablets
*actually* be selling them when the 3D engine isn't integrated into
the kernel source code yet??" :)

anyway there's probably a perfectly good answer to that.

now.

last thing: i mentioned this before, but we have had an offer of $100
donation to go towards a 1st batch board.  i'd like recommendations
from people on whom should best receive a board, or if there is anyone
who feels that they could significantly contribute in a strategic or
an actual concrete way to furthering the goals of the rhombus tech
project.

"by default" if i don't get any submissions then i have a person i'd
like to recommend: the decision however i will defer to those people
who volunteered quite some time ago to help out in rare instances like
this, busy as they are.  it's been so long they've probably forgotten
but i'll remind them shortly :)

ok that's all... oh wait!  yes, nearly forgot: we've now got a PCB
layout for a micro-engineering board:
http://rhombus-tech.net/community_ideas/micro_engineering_board/

these are designed not as being price-conscious in *any* way but as
convenience instead, and so are using parts that are DIL (through-hole
i.e. easily hand-solderable) and also available directly off of
digikey.  i think what we'll do is say make one prototype, see if it
works, and if it does then get some PCBs printed up and if anyone
wants to get a group buy together for some components we'll work
something out.

ok yep that's all :)

l.



More information about the arm-netbook mailing list