raspberry pi 4

Sat Jul 13 05:50:19 UTC 2019

 > RPi and Beaglebone.  These are "legacy" boards that were the first ones
 > to run freebsd when we started adding armv6/7 support.
 > The other arm32 things basically range from "supported, but not much
 > ongoing activity" to "nobody has touched it for years, hard to call it
 > supported".

One interesting thing about Beaglebone - seems like it provides
functionality missing in other SBC including RPis:
https://www.embeddedrelated.com/showarticle/586.php
(I don't know does FreeBSD support RPUs in BB image or not)

'Zooming out' this discussion and trying to generalize
various interests of people playing/working with SCBs
it seems for me that there are 3 categories of boards:

1. A board interacting with user mostly via screen/GUI,
(laptop, POS terminal, kiosk) so the GPU support is essential.
All other peripheral devices are more or less standard.

2. High I/O throughput and headless (router, NAS).
In this case obviously things like data pipes:
SATA/PCIe/USB3.0/1G/10G Ethernet are essential
but the rest of peripheral devices are standard.

3. SBC designed to be connected to non-standard hardware
(sensors, actuators etc.) and require (close to)hard real
time support.

Regards,
Denis.

On 12/07/2019 12:58 am, Ian Lepore wrote:
> On Wed, 2019-07-10 at 12:19 -0700, Johannes Lundberg wrote:
>> On 7/10/19 11:10 AM, Ian Lepore wrote:
>>> On Wed, 2019-07-10 at 10:30 -0700, Johannes Lundberg wrote:
>>>> On 7/9/19 8:17 PM, Mark Linimon wrote:
>>>>> On Wed, Jul 10, 2019 at 09:52:43AM +0900, Denis Polygalov wrote:
>>>>>> but please let's enhance support of the good OS (FreeBSD)
>>>>>> on a *good* boards.
>>>>>
>>>>> Despite any technical advantages or disadvantages, RPI has the most
>>>>> mindshare, and we would be foolish to avoid it.
>>>>
>>>> Indeed. SBCs come and go. They are EOL before we even have a boot
>>>> prompt. Personally I would like to see a joint effort focused on one
>>>> board and make that work really well. Maybe an incentive would be the
>>>> foundation throwing money at it in the form of rewards for well defined
>>>> sub projects. The one most likely to survive longest is RPI but there
>>>> might be other valid alternatives as well. Thanks to Emmanuel's efforts
>>>> maybe Pine64 is a good alternative? I'm happy to help with graphics if
>>>> we would do such focused effort but as long as we're all over the place
>>>> I don't see much point in contributing with the limited time I have...
>>>>
>>>> Please note, this is not criticism in any way and I'm not trying to
>>>> diminish the work developers do on these boards. Everyone is free to
>>>> work on what they want. Question is, do we want a single board computer
>>>> that's actually usable for something or only as tinker toys? Without
>>>> direction, I'm afraid they will always be half working tinker toys due
>>>> to the limited amount of developers we have.
>>>>
>>>> If anyone disagrees, I welcome your point of view.
>>>>
>>>
>>> What you call a "half working tinker toy" is what we use to build and
>>> ship a dozen different products at $work.
>>
>> My apologies if I offended anyone. I didn't know that we had such good
>> support that you could actually ship products based on it. Maybe I ask
>> what board that is?
>>
>>
> 
> We use Freescale/Nxp imx6 SOMs from Boundary Devices, SolidRun, and
> Technexion (Wandboard's upstream vendor).  Usually we design our own
> carrier/motherboards and mount the vendor SOMs on them.  Sometimes
> we're able to just directly use the vendor's carrier boards when we
> don't need an fpga or other highly customized stuff on the board.  The
> products we build are all related in one way or another to precision
> timekeeping.
> 
> I think I'd sum up the state of freebsd arm32 support like this...
> 
> RPi and Beaglebone.  These are "legacy" boards that were the first ones
> to run freebsd when we started adding armv6/7 support.  There weren't
> that many boards available back then and often the developer/eval board
> cost hundreds of dollars, whereas these were cheap and easily
> available.  All the current arm32 devs hate working on the crappy old
> code for these boards and pretty much only apply fixes, reluctantly, as
> needed or requested by users.
> 
> imx6.  These are pretty well supported because I get paid to support
> them.  Audio and video support is weak because we don't use those at
> $work (and I don't have a strong personal interest in those areas).
> The most important thing that's missing is pcie support.
> 
> Allwinner.  Originally these boards were barely supported, because docs
> were hard to get.  Something changed and the docs became available, and
> several developers adopted the family out of personal interest, so it's
> currently the most-complete and best-supported arm32 family on freebsd.
> 
> Marvell.  Supported by the folks at Semihalf, mostly because people pay
> them to do so.  The 32-bit marvell world isn't very active these days.
> 
> The other arm32 things basically range from "supported, but not much
> ongoing activity" to "nobody has touched it for years, hard to call it
> supported".
> 
> -- Ian
> 
> 
> _______________________________________________
> freebsd-arm at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-arm
> To unsubscribe, send any mail to "freebsd-arm-unsubscribe at freebsd.org"
> 