Re: [RFC, Announce] Unified x86 architecture, arch/x86


On Sat, 21 Jul 2007, Andi Kleen wrote:

On Saturday 21 July 2007 00:32, Thomas Gleixner wrote:
We are pleased to announce a project we've been working on for some
time: the unified x86 architecture tree, or "arch/x86" - and we'd like
to solicit feedback about it.

Well you know my position on this. I think it's a bad idea because
it means we can never get rid of any old junk. IMNSHO arch/x86_64
is significantly cleaner and simpler in many ways than arch/i386 and I would
like to preserve that. Also in general arch/x86_64 is much easier to hack
than arch/i386 because it's easier to regression test and in general
has to care about much less junk. And I don't
know of any way to ever fix that for i386 besides splitting the old
stuff off completely.

I have to say honestly that it is much easier to work in the i386 arch
directories than the x86_64. But that may be my own feelings. You seem to
have a nice style that you like and think that it is cleaner. But it
doesn't really seem much cleaner to me. Somethings I like better with the
x86_64 code, and there's somethings I like better with the i386 code. But
from a comfort level, I have to go with worknig with the i386 code.



Besides radical file movements like this are bad anyways. They cause
a big break in patchkits and forward/backwards porting that doesn't
really help anybody.

I think it helps a lot of people. Especially those that are trying to add
things to _both_ i386 and x86_64.


This causes double maintenance
even for functionality that is conceptually the same for the 32-bit and
the 64-bit tree. (such as support for standard PC platform architecture
devices)

Not sure what you mean here? I would think that we have this "double
maintence" anyway. Fixes that are done in x86_64 probably should also be
done in i386. Why have it in two places?


It's not really the same platform: one is PC hardware going back forever
with zillions of bugs, the other is modern PC platforms which much less
bugs and quirks

hehe, I'm seeing a bunch of bugs and quirks appear. It's just that
x86_64 isn't as old as i386 to have as many of them. But give it time.


To see it otherwise it's more a junkification of arch/x86_64 than
a cleanup of arch/i386 -- in fact you didn't really clean up arch/i386
at all.

That was not the point of this patch. This patch was to unify the two so
that we can get started on the unification.


How did we do it?
-----------------

As an initial matter, we made it painstakingly sure that the resulting
.o files in a 32-bit build are bit for bit equal.

You got not a single line less code duplication then, so i don't really
see the point of this.


Did you read what tglx wrote? The point of this patch was to keep
everything the _same_. The fact that not a single line less code
duplication is a feature. A great starting point where we can easily
trace things back to the current arch separation, as well as move forward
in merging the two.

-- Steve

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • Re: [PATCH] ne.c fix for hibernate and rmmod oops fix
    ... Notable changes since the last patch: ... code doesn't even need the resources passed by the platform device. ... The ne.c driver already registers the resources it needs and there ... Note the user space ISA PnP doesn't work for suspend to disk, ...
    (Linux-Kernel)
  • [UNIX] XInetD 2.3.0 Code Audit Completed
    ... XInetD 2.3.0 Code Audit Completed ... There were, however, certain issues with patch merging, and the ... but the code remains far from clean and certain bugs are there by ...
    (Securiteam)
  • Re: Authentication failed suddenly
    ... >> off on such a platform, ... the developers don't consider it a fundamental part of the distribution. ... >The bugs are platform compatibility bugs. ... bugs and report them - rather than rambling in public fora ...
    (comp.security.ssh)
  • U5 Lazarus patch 1.1 out
    ... - Shadowlords will not display their names until you've "learned" them. ... - All reported game-crashing bugs in plot-critical NPCs should be fixed. ... - The game new begins with access to the mini-map function. ... included in this patch will not take effect until you begin a new character. ...
    (comp.sys.ibm.pc.games.rpg)
  • [PATCH] IRQ handling race and spurious IIR read in serial/8250.c
    ... 2nd patch below); I think this will fix the symptoms for you. ... in serial8250_start_tx delete the read from the IIR and the ... The bugs in detail (this discussion applies to 2.6.20 and also to ... the generated interrupt races with the straight-line code in ...
    (Linux-Kernel)