Re: [patch] pagecache-2.3.9-H3, bmap & ext2fs cleanup patch

Linus Torvalds (torvalds@transmeta.com)
Sat, 26 Jun 1999 18:31:35 -0700 (PDT)


On Sat, 26 Jun 1999, Albert D. Cahalan wrote:
>
> If Linux will use 3 bits to represent a hole (and I hope not),
> please hide it in a macro. The kernel, all 1.5 million lines of
> it, is a bit hard to follow when you can't hack Linux full-time.

The Linux VFS layer simply does not care about holes. Never has, and
hopefully never will.

Not a _single_ place cares. It shouldn't.

And that's why I'm so opposed to adding a bit that says "hole". Because we
don't care about that bit anyway.

Basically, not a single place checks the three bits that I claim are
equivalent to a hole, because nobody NEEDS to. The semantics of "holes"
come from completely logical rules, without any need for any explicit bit.

> Coherency problems are bugs which can be tested for with debug code.
> #define CHECK_BH_BITS 1

Ehh..

I much prefer code that cannot be buggy, to code that has to have asserts
for bugs.

Eliminating a source of bugs is always a good idea. Remember: the only
code that is _guaranteed_ to be bug-free is the code that isn't even
written. Ergo, you should avoid trying to add unnecessary code at all
costs.

Asserts do not help. Not writing the buggy code in the first place _does_
help.

Linus

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