Re: [PATCH 1/1] inside-secure irq balance

[Sven Auhagen <sven.auhagen@voleatech.de>]

On Thu, Jul 16, 2020 at 08:44:23AM +0000, Van Leeuwen, Pascal wrote:
> > -----Original Message-----
> > From: linux-crypto-owner@vger.kernel.org <linux-crypto-owner@vger.kernel.org> On Behalf Of Herbert Xu
> > Sent: Thursday, July 16, 2020 9:22 AM
> > To: Sven Auhagen <sven.auhagen@voleatech.de>
> > Cc: linux-crypto@vger.kernel.org
> > Subject: Re: [PATCH 1/1] inside-secure irq balance
> >
> > <<< External Email >>>
> > Sven Auhagen <sven.auhagen@voleatech.de> wrote:
> > >
> > > +       // Set affinity
> > > +       cpu = ring_id % num_online_cpus();
> > > +       irq_set_affinity_hint(irq, get_cpu_mask(cpu));
> > > +
> >
> > This doesn't look right.  There is no guarantee that the online
> > CPUs are the lowest bits in the bitmask.  Also, what are you going
> > to do when the CPUs go down (or up)?
> >

After some further reading this is only a hint.
If the CPU is not online a different one will be used.
If the CPU goes offline the cpu hotplug code makes sure to move the irq
to a different CPU or remove the hint completely.

This should be safe to use and btw other crypto drivers do it the same way.
For example cavium nitrox or cavium cpt.

Best
Sven

> 
> Ok, I was just about to test this patch with my hardware, but I suppose I can spare myself the
> trouble if it doesn't make sense. I already had a hunch it was too simplistic for general use.
> However, he does get a very significant speed boost out of this, which makes sense as having
> the interrupts properly distributed AND pinned to a fixed CPU ensures proper workload
> distribution and cache locality. In fact, this was the whole idea behind having multiple rings
> and interrupts.
> 
> So is there a better way to achieve the same goal from the driver? Or is this really something
> you cannot fix in the crypto driver itself?
> 
> > Cheers,
> > --
> > Email: Herbert Xu <herbert@gondor.apana.org.au>
> > Home Page: https://eur03.safelinks.protection.outlook.com/?url=http:%2F%2Fgondor.apana.org.au%2F~herbert%2F&amp;data=02%7C01%7Csven.auhagen%40voleatech.de%7C42783499b8fa4d11a9c608d8296474d2%7Cb82a99f679814a7295344d35298f847b%7C0%7C0%7C637304858734739951&amp;sdata=GNleSUVRQe56P%2BkG6OQ3JH7AkXzKve6UP6ai5dKpN0M%3D&amp;reserved=0
> > PGP Key: https://eur03.safelinks.protection.outlook.com/?url=http:%2F%2Fgondor.apana.org.au%2F~herbert%2Fpubkey.txt&amp;data=02%7C01%7Csven.auhagen%40voleatech.de%7C42783499b8fa4d11a9c608d8296474d2%7Cb82a99f679814a7295344d35298f847b%7C0%7C0%7C637304858734739951&amp;sdata=nqUVTBAMn1ifyR6lj9nyxBFQZNR9Au8r0aUJR44ziyc%3D&amp;reserved=0
> 
> Regards,
> Pascal van Leeuwen
> Silicon IP Architect Multi-Protocol Engines, Rambus Security
> Rambus ROTW Holding BV
> +31-73 6581953
> 
> Note: The Inside Secure/Verimatrix Silicon IP team was recently acquired by Rambus.
> Please be so kind to update your e-mail address book with my new e-mail address.
> 
> 
> ** This message and any attachments are for the sole use of the intended recipient(s). It may contain information that is confidential and privileged. If you are not the intended recipient of this message, you are prohibited from printing, copying, forwarding or saving it. Please delete the message and attachments and notify the sender immediately. **
> 
> Rambus Inc.<https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.rambus.com%2F&amp;data=02%7C01%7Csven.auhagen%40voleatech.de%7C42783499b8fa4d11a9c608d8296474d2%7Cb82a99f679814a7295344d35298f847b%7C0%7C0%7C637304858734739951&amp;sdata=gCBXI0rNikA%2FG2ME7RxWwwmkuUNl9wRlyQqDGbFoGHk%3D&amp;reserved=0>