[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: OABI [was Re: COMPAT_NETBSD32 in GENERIC.common]
What’s the situation for the 32-bit Acorn machines?
Sent from my iPhone.
> On Nov 21, 2018, at 1:42 PM, Martin Husemann <martin%duskware.de@localhost> wrote:
>> On Wed, Nov 21, 2018 at 08:29:38AM -0500, Greg Troxel wrote:
>> it seems oabi is only sensible for a small number of older ARM
>> processors with a kind of floating point unit that is not found on
>> anything recent. Is that right?
> We have used oabi many of our arm ports upto NetBSD 8, and for all of
> them upto NetBSD 6.
> Removing oabi support from kernels means no binary compatibility for
> those machines when running our own older binaries. This is bad for
> testing new kernels with old userland.
> No oabi compat support in userland means you can not just drop in an old
> NetBSD/oabi binary into a -current machine and run it, instead you have
> to create a chroot or similar. This is (IMHO) not that bad.
>> Finally, I see some MACHINE_ARCH values in build.sh that I can't quite
>> map to the language others use.
>> earm means what processor architecture? armv1 v2? v3?
>> earmhf means what particular kind of floating point unit?
> earm means some arm with eabi. We currently default it to armv4.
> I think we only use hf for modern cpus, so VFP.
>> earmv4: does that mean "armv4t", with thumb, which as I understand it is
>> pretty much every armv4 cpu except strongarm, wich is armv4 without
> No. My shark (strongarm 1100, armv4, no thumb instructions) currently
> runs earmv4.
Main Index |
Thread Index |