Hi, On Mon, 23 Jul 2007, Andi Kleen wrote:
> On Monday 23 July 2007 18:06:03 Satyam Sharma wrote: > > From: Satyam Sharma <[EMAIL PROTECTED]> > > > > [7/8] i386: bitops: Kill needless usage of __asm__ __volatile__ > > > > Another oddity I noticed in this file. The semantics of __volatile__ > > when used to qualify inline __asm__ are that the compiler will not > > (1) elid, or, (2) reorder, or, (3) intersperse, our inline asm with > > the rest of the generated code. > > > > However, we do not want these guarantees in the unlocked variants of the > > bitops functions. > > I thought so too and did a similar transformation while moving > some string functions out of line. After that recent misadventure > I would be very very careful with this. Ah, ok, so this must be the case we'd stumbled upon recently on the other thread. I hadn't got your fix for this, so didn't know. > Overall I'm sorry to say, but the risk:gain ratio of this > patch is imho totally out of whack. The patch does look correct to me, we're only killing the use of __volatile__ from places that don't require it (the guarantee-less variants). Without losing it, I really don't see how the compiler can ever combine multiple bitops, which does sound beneficial when the caller has already implemented higher-level locking around the usage of these operations. Satyam - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
