Re: New test kernel - second attempt
Hi,
On Sun, 2025-09-21 at 11:18 -0400, Dennis Clarke wrote:
> On 9/21/25 01:35, John Paul Adrian Glaubitz wrote:
> > Hello Tony,
> >
> > On Sat, 2025-09-20 at 22:20 -0700, Tony Rodriguez wrote:
> > > On another note, unsure if Oracle fixed any S7-2 related issues with the latest CBE yet.
> > > I will download and attempt to install the latest CBE again. I did so several months ago.
> >
> > They have not made any new CBE release since 11.4.81, so it's naturally not fixed.
> >
>
> I only cared about this because it was a good way to test the machine.
> It boots and runs something. For one day. Not even that long. However
> that single little SAS disk will be pulled out, labeled, stuck into an
> envelope and then tossed into the back of a cardboard box somewhere.
> Maybe a written word or two on the envelope that says "open in case of
> busted S7-2". That is the end of that.
I honestly don't understand the hate. As we discussed before, the issue with
Solaris 11.4 CBE on the SPARC S7 is known and is already fixed on later releases
as far as I know.
It's unfortunate that Oracle hasn't a fix for Solaris 11.4.81 CBE yet, but that
doesn't question the whole quality of Solaris. It's merely a clear indicator that
Oracle means it when they say the CBE releases are unsupported.
I will try to reach out to some folks at Oracle though and see if we can get them
to publish a fix for Oracle 11.4.81 CBE to help users like you.
> > > **Since this appears to be Solaris (not Debian) and custom software related, we should
> > > probably discuss privately via email.
> >
> > No, please don't. A lot of people use Solaris as their hypervisor and there
> > is not point in keeping this discussion private.
>
> Well, my years and decades with Solaris are long behind me. I ran out
> and bought this machine exclusively to get Linux running on it. So did
> a friend of mine. He has the 512G memory machine! So there are at least
> two more people who are happy to toss money at this problem as well as
> plenty of time. I am not just going to "rage quit" ha ha.
Sure. But please keep in mind that Linux on the SPARC S7 is still a bit
of work-in-progress. It should be better with Linux 6.18 which contains
the fixes for the memory corruption bugs [1] and even better when we got
the CPU detection code in order though [2].
> Mostly I am too stubborn for thatand I want a top of the line modern
> sparc machine running Linux because Adrian says it is possible and also
> Tony claims he has it running. So lets do that. :)
My few tests on Tony's SPARC S7 were actually quite promising. It builds
Debian's kernel package which itself builds two fully fledged kernel
images (uni-processor and SMP) in just about 40 minutes which even beats
Debian's ppc64el [3] and s390x [4] buildds which build comparable kernel
packages.
Adrian
> [1] https://github.com/sparclinux/issues/issues/1
> [2] https://github.com/sparclinux/issues/issues/4
> [3] https://buildd.debian.org/status/logs.php?pkg=linux&arch=ppc64el
> [4] https://buildd.debian.org/status/logs.php?pkg=linux&arch=s390x
--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer
`. `' Physicist
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Reply to: