On Sat, Feb 12, 2000 at 12:14:40PM +0100, Philipp Rumpf wrote:
> > Philipp's point is still valid though, u32 is only 2 byte aligned on
>
> Alan's point. My point was you could rely on gcc not padding structures
> when you always explicitly padded to natural alignment boundaries. Eg
> when you do struct { u16 s; u32 l; } you deserve to get burned, but when
> you do struct { u16 s; u16 unused; u32 l; } you can expect gcc not to
> be smart on you.
Alas, you can't. If accessing quantities of type short is exceptionally
slow, then it may well be that the ABI decress Thou Shalt Pad. gcc has
__attribute__(packed) to allow you to work around this.
Just do as the plan9'ers do and represent all externally visible data
as an array of unsigned chars. It's difficult to get that wrong.
-
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/
This archive was generated by hypermail 2b29 : Sat Feb 12 2000 - 18:27:17 EST