linux-kernel - Re: [PATCH V2] arch_topology: support parsing cluster

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]

Message-ID: <20220513084654.woxpjvtmtxx7cpik@bogus>
Date:   Fri, 13 May 2022 09:46:54 +0100
From:   Sudeep Holla <sudeep.holla@....com>
To:     王擎 <wangqing@...o.com>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Rafael J. Wysocki" <rafael@...nel.org>,
        Sudeep Holla <sudeep.holla@....com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "dietmar.eggemann@....com" <dietmar.eggemann@....com>
Subject: Re: [PATCH V2] arch_topology: support parsing cluster_id from DT

On Fri, May 13, 2022 at 08:30:10AM +0000, 王擎 wrote:
> 
> >> From: Wang Qing <wangqing@...o.com>
> >> 
> >> Use nested cluster structures in DT to support describing multi-level
> >> cluster topologies and increase the parsing of nested cluster.
> >> 
> >> Notice: the clusters describing in DT currently are not physical
> >> boundaries, since changing "cluster" to "socket" is too involved and error
> >> prone, this patch will not have any effect on one-level cluster topo, but
> >> can support the mutil-level cluster topo to support CLUSTER_SCHED.
> >
> >Sorry the socket/package_id is broken. If we are playing with cluster_id
> >which is now wrongly presented as package_id, you are forced to fix that
> >too. We don't want to break that in a different way or leave that as is
> >since the cluster_id and package ids now show up as same now. Earlier the
> >cluster_id was -1.
> >
> >I had a look when I started reviewing your patch. Assuming we don't need
> >nested cluster support yet, I have some patches(not built or tested
> >unfortunately yet). Let me know your thoughts. If you think you still
> >need support for some kind of nested cluster, build that on top of this.
> >Also I haven't bothered about sched domains as this purely relates to
> >topology and how this is mapped to sched domain is orthogonal.
> >
> >If anything is broken, that needs to be fixed separately there. I see the
> >idea here is correct and would like to push the patches once I build/test
> >and get some review/more testing.
> >
> >Regards,
> >Sudeep
>
> You have to modify all DTS(rename "cluster" to "socket"), otherwise, 
> new package_id = -1 and new cluster_id = old package_id.
>

Nope. I am handling absence of socket nodes and that is a must for backward
compatibility with existing DT.

> This will affect MC and CLS useage, do you have any plans about this?
>

I don't have much knowledge on scheduler domains and I will defer that to
the experts. I am just trying to get the topology read from DT correct and
to align with PPTT. Though LLC is not yet considered but that is not part
of cpu-map. I am trying to get only the /cpu-map part of topology correct
at first. We(you, me or anyone interested) can get the LLC part updated
after that.

--
Regards,
Sudeep