Re: ALIGN via ilog2 without gccisms (Re: [PATCH] Fix get_order())

[Oleg Verych]

On Wed, Mar 07, 2007 at 08:38:27AM -0800, Linus Torvalds wrote:
> 
> 
> On Wed, 7 Mar 2007, Oleg Verych wrote:
> >
> > Probably it can be used to get rid of gccisms and "type fluff" due to
> > bitwise arithmetics in ALIGN?
> 
> Hell no.
> 
> The typeof is there to make sure we have the right type, and it's simple.
> 
> The current ALIGN() macro is efficient as hell (generating just a simple 
> mask+add). Turning it into some kind of horrible thing that uses ilog2() 
> would be a total mistake.

GCC's assembler version of this macro is optimized as needed.

But i wanted to address Al's statement about using typeof():

,-*- [ Message-ID: <20061127044138.GA3078@xxxxxxxxxxxxxxxx> ]
|IOW, gcc allows type to leak out of scope it's been defined in (and
|typeof adds even more fun to the picture).  It not only goes against
|a _lot_ in C, it's actually not thought through by gcc folks.  Just
|try to mix that with variable-length arrays and watch it blow up
|in interesting ways...
`-*-

As for me, this is example of assembler's need, that very hard to
implement in C. Also, i doubt, C's shift doing any "type fluff" on its
only argument.

> Also, your ALIGN() macro was broken. That's not how ALIGN() is supposed to 
> work.

Yes, maybe so.
____

-
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/