Re: [PATCH RFC 01/14] x86/asm: add iosubmit_cmds512() based on movdir64b CPU instruction

[Dan Williams]

On Wed, Nov 20, 2019 at 4:53 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
> On Wed, 20 Nov 2019, Dan Williams wrote:
> > On Wed, Nov 20, 2019 at 3:27 PM Borislav Petkov <bp@xxxxxxxxx> wrote:
> > >
> > > On Wed, Nov 20, 2019 at 03:19:23PM -0800, Luck, Tony wrote:
> > > > That's the underlying functionality of the MOVDIR64B instruction. A
> > > > posted write so no way to know if it succeeded.
> > >
> > > So how do you know whether any of the writes went through?
> >
> > It's identical to the writel() mmio-write to start a SATA command
> > transfer. The higher level device driver protocol validates that the
> > command went through, ultimately with a timeout. There's no return
> > value for iosubmit_cmds512() for the same reason there's no return
> > value for the other iowrite primitives.
>
> With the difference that other iowrite primitive have no dependencies on
> cpu feature bits and cannot fail on the software level.

True, but that would be a driver coding mistake flagged by the
WARN_ON_ONCE, and the failure is static. The driver must check for
static_cpu_has(X86_FEATURE_MOVDIR64B) once at init, but it need not
check again on every command submission.