Re: [PATCH v3 1/7] seqnum_ops: Introduce Sequence Number Ops

[Shuah Khan]

On 2/5/21 2:58 AM, Greg KH wrote:
> On Wed, Feb 03, 2021 at 11:11:57AM -0700, Shuah Khan wrote:
> > +static inline u32 seqnum32_inc(struct seqnum32 *seq)
> > +{
> > +	atomic_t val = ATOMIC_INIT(seq->seqnum);
> > +
> > +	seq->seqnum = (u32) atomic_inc_return(&val);
> > +	if (seq->seqnum >= UINT_MAX)
> > +		pr_info("Sequence Number overflow %u detected\n",
> > +			seq->seqnum);
> > +	return seq->seqnum;
>
> As Peter points out, this is doing doing what you think it is doing :(
>
> Why do you not just have seq->seqnum be a real atomic variable?  Trying
> to switch to/from one like this does not work as there is no
> "atomic-ness" happening here at all.

Yes. This is sloppy on my part. As Peter and Rafael also pointed. I have
to start paying more attention to my inner voice.

thanks,
-- Shuah