[fedora-arm] Who's using Kirkwood?

[Brendan Conoboy]

On 10/10/2012 10:47 AM, Derek Atkins wrote:
> Sure, but we're a decade later.  Kirkwood devices were just released
> what?  3 years ago?  I certainly got mine more recently than that.  I
> admit I've been running F12 on it, but that's only because there hadn't
> been another fedora release until F17.

The comparison to i686 isn't really very apt.  Kirkwood is more like 
i386, but even that's stretching the simile.  There several problems 
with armv5tel support over the long term.

1. It's not self hosting.  We have to use armv7 hosts to build most of 
the armv5 packages because only they have enough RAM, enough CPU time, 
fast enough swap.  Building UP packages on SMP systems causes issues for 
a number of multithreaded packages.  Transient failures, "bugs" that 
aren't really bugs, just packages written in the belief that armv5 code 
will be built and run on armv5 hosts.  This problem gets worse with 
every release.

2. The different ABI requires as much as 2 times the number of build 
hosts to support both hard and soft float ABIs.

3. Certain features such as atomic operations aren't available on armv5, 
reducing the number of packages that can be built for ARM in total: If 
it fails on armv5 but works on armv7, we still don't get it for armv7.

4. The contributors who do most of the Fedora ARM work are focused 
specifically on armv7, so the energy spent fixing armv5 specific build 
problems is time taken away from their interest.

5. On the whole, it's not a popular Fedora ARM target.  Raspberry pi, 
OMAP, highbank, this is where most (not all) of our known users have 
hardware and interest.  There are some Kirkwood users, clearly, but 
there are a lot more users of everything else.  We should get some 
updated download stats on this to demonstrate, but last I saw kirkwood 
was maybe 3% of usage.

Where does this leave us?  Dropping armv5tel anytime soon isn't being 
proposed- we'll certainly do F18.  Probably F19, too (We're already 
building for it).  But when we do logistics for moving koji services to 
PHX, most of the interested parties are thinking of just moving armv7hl. 
  The armv5tel builds can continue as they are, assuming Seneca wants to 
continue hosting them.  We're all volunteers here, so if you want to 
volunteer some time to keep armv5tel viable please do!  Nothing is 
written in stone or decided, but now is definitely the time to have the 
conversation.

-- 
Brendan Conoboy / Red Hat, Inc. / blc at redhat.com