Re: [patch 07/13] ptp: Split out PTP_SYS_OFFSET ioctl code

[Vadim Fedorenko <vadim.fedorenko@linux.dev>]

On 21/06/2025 21:42, Thomas Gleixner wrote:
> On Sat, Jun 21 2025 at 21:14, Vadim Fedorenko wrote:
>> On 20/06/2025 14:24, Thomas Gleixner wrote:
>>> +	pct = &sysoff->ts[0];
>>> +	for (unsigned int i = 0; i < sysoff->n_samples; i++) {
>>> +		struct ptp_clock_info *ops = ptp->info;
>>
>> Looks like *ops initialization can be moved outside of the loop.
> 
> Well it can, but does it matter? No, because this is only a coding
> artifact. The compiler can evaluate ptp->info inside of the loop at his
> own peril even on both usage sites.
> 
> Though what's more important is that from a context point of view, ops
> belongs into the loop, because that's where it is used and not outside,
> no?

Well, in the original code it was always outside of any loops. And the
scope a bit bigger, all these functions work with ptp_clock object, so
the context is really subjective. It just looks a bit weird to de-
reference something, which is known not to change, every iteration. Of
course the compiler is smart enough to optimize it, but still..


> Thanks,
> 
>          tglx