linux-kernel - Re: [RFC -tip] x86, apic: Merge x2apic code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 11 Nov 2010 23:59:55 +0300
From:	Cyrill Gorcunov <gorcunov@...il.com>
To:	Suresh Siddha <suresh.b.siddha@...el.com>
Cc:	Thomas Gleixner <tglx@...utronix.de>,
	Yinghai Lu <yinghai@...nel.org>, Ingo Molnar <mingo@...e.hu>,
	"H. Peter Anvin" <hpa@...or.com>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [RFC -tip] x86, apic: Merge x2apic code

On Thu, Nov 11, 2010 at 12:44:17PM -0800, Suresh Siddha wrote:
> On Wed, 2010-11-10 at 11:20 -0800, Cyrill Gorcunov wrote:
> > Hi, while being at x2apic_x.c code I noted that files look similar to
> > each other so I guessed may be we could merge them. As result --
> > this patch.
> 
> BTW, as you noticed, x2apic cluster mode allows IPI's to be sent to
> multiple cpu's (in the same cluster) at once. I was hoping sometime I
> will do this to see if it helps in anything.
> 
> So once we do that IPI code will look different.

yes, indeed, so enlight me a bit -- we have to map every cpu number from
cpumask to apic-id, then figure out if it belongs to same cluster, collect
such apic-ids and then send one ipi with one cluster and multiple apic-id
bits as a destination, or I miss something and there _can't_ ever be the
situation when first 16 cpus from cpumask belong to different clusters?

> 
> >  So comments/complains are appreciated and what is more important -- if we need
> > this patch at all.
> 
> May be we can do the merge for most of them but leave IPI code as it is
> (as that is also in the hot path). And may be while we are at this, we
> should probably look at the x2apic cluster IPI bits now aswell.
> 

yes, probably. Though I think first we need to resolve situation with
per-cluster IPIs.

> thanks,
> suresh
> 

 Cyrill
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/