Re: [PATCH v2 00/53] MIPS: SGI-IP27: add support for the IP35 (Bedrock hub) machines

On Tue, 20 Oct 2026 22:42:13 +0200 Imre Kaloz wrote:

On Tue, 20 Oct 2026 11:06:19 -0700 Jakub Kicinski wrote:

quoted

Why do we need support for machines discontinued 20+ years ago?  

For netdev the question is smaller than the series: patches 25 and 38,
44 lines in ioc3-eth, a driver that is already in tree for the same
part on IP27 and IP30. Patch 25 is a no-op on both of those (the bit it
tests is always set behind a classic Bridge); patch 38 is a MAC-address
fallback. I should have said in the cover that the series is meant to go
through the MIPS tree and that I'm only after an Ack for those two --
sorry for the noise on your queue.

As for why at all: same reason IP27, IP30 and IP32 are in tree. The
hardware is in use, Thomas maintains the platform, and people run and
test it; "discontinued" has never been the removal criterion, lack of
maintainers has. IP35 is IP27's direct successor on the same
Hub/XTalk/Bridge architecture with the same IOC3, plus its successor
IOC4, so most of the series is teaching existing code about one more
hub generation. And these boxes still earn their keep as a stress case
for arch and core code -- e.g. the kmem_cache_destroy() race that just
turned up on my Ultra 45:
https://lore.kernel.org/all/atDCYwNfKW3m6YXh@thinkstation/ (local)

I don't think you answered my question, or at least I'm too daft 
to catch what you mean. IP35 was not supported so nobody is using
it with Linux today. You're adding the support, and since it's 
a platform discontinued 20 years ago there's 0 chance of real users 
of any of this code.

What am I missing?