Re: NetBSD 6.0 and earmv7hf

["Ernst J. Oud" <ernstoud%euronet.nl@localhost>]

"Kulesa, Craig A - (ckulesa)" <ckulesa%email.arizona.edu@localhost> wrote in message 
news:6CC6AD7F-F05A-41F5-900B-DEBF9DCE7132%email.arizona.edu@localhost...

>> Harold Gutch wrote:

>> Does "build.sh -m evbarm -a arm" (or perhaps "-m evbarm-el") perhaps
>> do the job?

>I think the main problem in this thread is discussing earmv7hf in the 
>context of NetBSD 6.

>NetBSD 6's arm releases are all OABI.  That is, only MACHINE_ARCH's of 
>'arm' and 'armeb'
> are supported in release builds of NetBSD 6.x.  That's why there are no 
> package binary
> releases for anything but plain "arm" for NetBSD 6.

> EABI support (earmv4, earmv6hf, earmv7hf, etc) started in earnest with the 
> NetBSD 7.0 release.

But the SOC in the latest TC (BCM53019) is earmv7hf according to all info I 
can find (same SOC as in Raspberry Pi2).

> The question to confirm is: "what is the ABI on the late-generation Time 
> Capsules"?

It runs NetBSD 6.0 (according to uname -a) and it has an earmv7hf SOC... 
therefore it must be EABI?

> If it's still OABI, the NetBSD 6 pkgsrc packages would normally run on it 
> (unless Apple built it as hard-float).

They don't run (ENOEXEC, SH says "Cannot run ELF binary". More evidence that 
it is EABI?

>If it's EABI (because Apple backported EABI to NetBSD 6, or built from a 
>6.99.x snapshot using the EABI),
> then we may be looking at a hybrid that doesn't represent *anything* 
> NetBSD has supported in a release.

> -Craig

I don't rule out that Broadcom supplied some form of toolchain to Apple, not 
using GCC and creating some hybrid...

Ernst