[Arm-netbook] Arm processors

[Paul Boddie]

On Friday 3. March 2017 01.47.49 Siarhei Siamashka wrote:
> On Fri, Mar 3, 2017 at 12:48 AM, Paul Boddie <paul at boddie.org.uk> wrote:
> > 
> > Source:
> > https://olimex.wordpress.com/2017/02/07/fosdem-and-teres-i-update/
> > 
> > Some of that is specific to their laptop, but some of it seems relevant
> > to any A64 device. Maybe you could reconcile what the Olimex people are
> > saying with what you are claiming.
> 
> I guess, the emphasis was on *all* A64 features. And the mainline kernel
> clearly does not support *all* A64 features yet.

Sure, it wasn't good enough for a demonstration. But that doesn't need *all* 
features, either. Yes, I agree that it's not their focus and that they just 
want to demonstrate the hardware.

As people have noted before, it isn't always that helpful if such a hardware-
only focus continues through to retail because people can end up with hardware 
that they cannot support themselves, with software that violates the licensing 
terms.

> Also Olimex people are always saying that they don't do software and
> don't have software expertise in-house. The are not the best people
> to ask for this information.

Well, there was someone who appeared to be "in the know" replying to questions 
on their blog, so I was rather going off what they were saying. I don't expect 
the Olimex people to know the details, but I do expect them to know whether 
they can ship compliant software or not.

> > If you have any definitive information to the contrary, particularly
> > about the boot0 code that Luke appears to be referring to, please post
> > links to it.
> 
> Regarding Luke's claim stated in bold letters, here is the commit in the
> mainline U-Boot, which has added the A64 DRAM controller support:
> 
>    
> http://git.denx.de/?p=u-boot.git;a=commitdiff;h=1bc464be1fc559a3f6dc133429
> 7245d5b27b9b57

Thank you for posting this reference.

> But the reverse engineered A64 DRAM controller support code existed
> in experimental git branches many months before it finally landed
> upstream and anyone could try it.

I did see this on the A64 page:

"U-Boot 2017.03-rc1 saw the addition of the required DRAM init code, so SPL 
support is now enabled. However this version lacks support for loading the 
ATF, which limits the usability."

https://linux-sunxi.org/A64#Mainline_U-Boot

It was found under the empty boot0 section, so I guess one is supposed to read 
the page like a switch statement in the C programming language or something.

> > The linux-sunxi wiki was very vague on such matters last time I checked.
> > And yes, I have seen the "mainlining effort" page:
> > 
> > https://linux-sunxi.org/Linux_mainlining_effort#Status_Matrix
> 
> It's very good that you have found this page. You can clearly see
> many links to the work-in progress branches that are used for
> developing various drivers and test them.
> 
> If you don't understand something, you can always join the #linux-sunxi
> irc channel on freenode and ask around.

Well, I don't track Allwinner SoC support at all, really, but I thought that 
since we had the opportunity to get to the bottom of this matter, it might be 
best to ask someone who clearly has information about it. Publishing such 
information in obvious places is more helpful than having people "ask around", 
though, because it saves everyone time and reduces their confusion.

To that end, I added the following page to the Rhombus Tech wiki so that 
people have the ability to inform themselves more conveniently on the topic:

http://rhombus-tech.net/allwinner/a64/

Naturally, corrections and improvements are welcome.

Paul