On Mon, Aug 21, 2023 at 04:19:54PM +0800, Herbert Xu wrote:
> On Sat, Aug 19, 2023 at 09:33:18AM +0200, Ard Biesheuvel wrote:
> >
> > No, that otx2_write128() routine looks buggy, actually, The ! at the
> > end means writeback, and so the register holding addr will be
> > modified, which is not reflect in the asm constraints. It also lacks a
> > barrier.
>
> OK. But at least having a helper called write128 looks a lot
> cleaner than just having unexplained assembly in the code.
I guess we want something similar to how writeq() is handled on 32-bit
architectures (see include/linux/io-64-nonatomic-{hi-lo,lo-hi}.h.
It's then CPU-dependent on whether you get atomicity.
Will
