Re: FreeBSD 14: Poll armv6 deprecated or removed

From: Mark Millard via freebsd-arm <>
Date: Thu, 18 Nov 2021 19:32:25 UTC
On 2021-Nov-4, at 14:18, Mark Millard <> wrote:

> On 2021-Nov-4, at 13:39, tech-lists <> wrote:
>> On Thu, Nov 04, 2021 at 11:53:18AM -0700, Mark Millard via freebsd-arch wrote:
>>> Without one or more developers willing to keep ARM11 based RPi* FreeBSD
>>> working as FreeBSD updates, the code will break. Other architectures
>>> have been removed for such. Folks that do not want to work on such code
>>> do not want to have to work on it to keep FreeBSD building and operating
>>> for other architectures that have active developmers/maintainers.
>>> If there were active FreeBSD developers for ARM11 RPi*'s, the removal
>>> would have been unlikely to be proposed at all, even if the use was
>>> minor. FreeBSD is driven by the developer context directly, not the
>>> usage context directly. 
>> OK. I can understand that. No developers want to work on it so no
>> interest. That's straightforward, logical, bad for me but I can
>> understand it and work around it. But that was not mentioned by the OP.
>> On Thu, Oct 28, 2021 at 09:44:20AM -0600, Warner Losh wrote:
>>>> Given that the number of available and useful armv6 boards has fallen
>>>> to almost zero, the time has come to look hard at armv6.
>> I'm objecting to this because "available and useful" is impossible to measure. "Available" is going to be a very large number, because of
>> the number of sales and popularity of these boards, and that they are
>> durable. So stuff made years ago can logically be presumed to be still
>> in working order. Even if 0.1% of rpi1b users used freebsd on their
>> boards, it'll still be a big number. FreeBSD does not record anywhere the context in which it is used. And "useful" depends on who is using it for what and is an opinion.
>>> NetBSD supports a lot of systems that FreeBSD does not. That fact has
>>> never justified having support for those systems in FreeBSD. 
>> I'm not saying that. What I'm asking is the reasoning.
>> "we don't want to support it anymore" is a reason
>> "no devs are interested" is a reason
>> "the number of available and useful armv6 boards has fallen to almost
>> zero" is objectively false and so therefore is not a reason. And because
>> it is not a reason then justifications following it will also be
>> incorrect.
> I'll note that:
> indicates: ARMv6 (Raspberry Pi 1 only)
> so NetBSD does not have general armv6 support, just support for
> the RPi*'s that are ARM11 based. (Another page mentions RPi0 and
> RPi0w examples as "expected to work", although needing FDT files.
> See: and its
> earmv6hf material.)
> The lack of a variety of sources of armv6 or ARM11 that NetBSD
> supports is likely a kind of property being referenced: even for
> NetBSD no other ARM11's are targeted.
> Basically, even for NetBSD, one has to be interested in supporting
> (some) RPi*'s in order to be interested in supporting ARM11. There
> is not much of any other ARM11 market for NetBSD (or FreeBSD).
>> I'm interested to know what NetBSD's reasons are in having tier-1
>> support for armv6, but I'll ask that on their lists.

I'll also note that Fedora is at the proposal stage for
removal of armv7 (yes: 7) in fc37. From:

is (in part):

Detailed Description
The ARMv7 arm architecture was the second variant of the arm architecture that Fedora has supported, the first was ARMv5, the third is aarch64. The proposal is to retire ARMv7 as part of the Fedora 37 release. This will allow ARMv7/armhfp to be supported until the Fedora 36 end of life in around June 2023.

Overall arm32 is generally waning with generally few new ARMv7 devices added to Fedora in recent releases. To add to that a number of newer Fedora features designed to improve speed and security of the Fedora release are causing 32 bit architectures in general primarily due to the process memory limit when linking large applications. The ARMv7/armhfp is the last fully supported 32 bit architecture, we still currently build i686 packages, but it's not shipped as artefacts.

Benefit to Fedora
The primary benefit is to maintainers of the ARM architecture, the various toolchain teams and package maintainers in general.

Mark Millard
marklmi at
( went
away in early 2018-Mar)