RE: [PATCH] net: stmmac: protect statistics updates with a spinlock

[David Laight <David.Laight@ACULAB.COM>]

From: Andrew Lunn
> Sent: 05 January 2024 17:37
> 
> > This only leaves an atomic_t in hard irq context. I have tried to find
> > something that could relax the requirement, but AFAICS at least some
> > setups use several interrupts that can be delivered to different CPUs
> > simultaneously, and all of them will walk over all channels. So we're
> > left with an atomic_t here.
> 
> You might want to consider per CPU statistics. Since each CPU has its
> own structure of statistics, you don't need atomic.
> 
> The code actually using the statistics then needs to sum up the per
> CPU statistics, and using syncp should be sufficient for that.

Doesn't that consume rather a lot of memory on systems with
'silly numbers' of cpu?

Updating an atomic_t is (pretty much) the same as taking a lock.
unlock() is also likely to also contain an atomic operation.
So if you update more than two atomic_t it is likely that a lock
will be faster.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)