Re: questionable use of qemu-m68k for building
On 12/06/2015 05:13 PM, Thorsten Glaser wrote:
> Ähm, I was *specifically* commenting on the kswapd bug. I wonder
> what words you lie into my mouth.
I'm not laying words into your mouth, but I know you and I read
the subject of your mail and I read the IRC messages that you
were quoting. Those are pretty condescending which is really annoying.
> Even Andreas (whose message I read too late) said it was a known
> kernel bug (can it be fixed, pretty please?).
Yes, a bug in the m68k kernel. Which means something that needs
to be fixed, irrelevant whether you're hosting this on a 32-bit
or a 64-bit machine.
>> for example, tells me that I am right and you are wrong:
> I did not wish to make this a matter of right or wrong, but I
> cannot help but throw an argument into this particular pot:
> Would you choose speed over consistency?
What do you mean by consistency, you are using the wrong word.
There might be an issue with miscompiled code, but this would be
rather an issue with gcc than with the emulating environment.
Yes, qemu-m68k has some bugs. But those can be ironed out and
the best way to do it is to start actually using it and obeserve
the results that it produces. I have used the same approach to
get gcc-4.9 and gcc-5 fixed on sh4, with great success. And
Laurent is doing an awesome job fixing the bugs and I am glad
I am able to help him with some minor patches even though I
am new to the qemu codebase.
Once the bugs are ironed out, we have a massive speed advantage
which will eventually mean that we will be finally be able to
catch up, even with release architectures. And that is really
worth all the hassle and the bugs.
> I think that’s one of the major cultural differences between
> Linux and BSD people (and, incidentally, MySQL users and
> database users).
See, again, you're being condescending. This sentence alone
is complete non-sense. And, no, I am not going to use this
as an initiative to start a flamewar. The fact that Linux
is massively more successful than any flavour of BSD is enough
justification we need.
> So even the right vs. wrong is subjective here.
I don't think it's subjective. It's clearly the right approach
to improve things and not remain standing at the same point
in development forever. Time moves on and even Jordan Hubbard,
the *co-founder* of FreeBSD agrees with this stance:
> But, by all means: you’re doing the work, and I have been
> trying to reduce my involvement in Debian/m68k (as I’ve
> brought it to a state which solved my initial problems,
> and I’ve got other things to do, and I never meant to
> stay infinitely), so, from the doöcracy PoV, you’re right
> (merely by doing things, mind you).
I have done the same for *both* sh4 *and* sparc64. So?
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer - firstname.lastname@example.org
`. `' Freie Universitaet Berlin - email@example.com
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913