linux-kernel - Re: [PATCH v2 7/7] x86/resctrl: Determine if Sub-NUMA Cluster is enabled and initialize.

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: <CALPaoCh=j+5_dA2Vgj1DtXgFWGCjJDXt=MwSVtMmPuuouWYokg@mail.gmail.com>
Date:   Mon, 26 Jun 2023 14:18:01 +0200
From:   Peter Newman <peternewman@...gle.com>
To:     Tony Luck <tony.luck@...el.com>
Cc:     "Yu, Fenghua" <fenghua.yu@...el.com>,
        "Chatre, Reinette" <reinette.chatre@...el.com>,
        Jonathan Corbet <corbet@....net>,
        "x86@...nel.org" <x86@...nel.org>,
        Shaopeng Tan <tan.shaopeng@...itsu.com>,
        James Morse <james.morse@....com>,
        Jamie Iles <quic_jiles@...cinc.com>,
        Babu Moger <babu.moger@....com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-doc@...r.kernel.org" <linux-doc@...r.kernel.org>,
        "patches@...ts.linux.dev" <patches@...ts.linux.dev>,
        "Eranian, Stephane" <eranian@...gle.com>
Subject: Re: [PATCH v2 7/7] x86/resctrl: Determine if Sub-NUMA Cluster is
 enabled and initialize.

Hi Tony,

On Fri, Jun 23, 2023 at 10:20 PM Tony Luck <tony.luck@...el.com> wrote:
> On Fri, Jun 23, 2023 at 05:19:37PM +0200, Peter Newman wrote:
> > On Thu, Jun 22, 2023 at 6:02 PM Luck, Tony <tony.luck@...el.com> wrote:
> > >
> > > > Unfortunately I'm not getting as good of results with the new series.
> > > > The main difference seems to be updating the 0xca0 MSR instead of
> > > > applying the offset to PQR_ASSOC.
> > >
> > > I think I may have reversed the actions to update the MSR in one of
> > > my refactor/rebase. The comment here is correct, but that's not
> > > what the code is doing :-(
> > >
> > > Can you swap the bodies of these two functions and retest?
> >
> > It's a small improvement, but still not great. Still only node 0
> > giving believable results, but at least no more empty results from the
> > second package.
> >
> > I poked around in /proc/kcore and noticed that my snc_ways is still 1, though.

It turns out I just forgot that I had KASLR on. snc_ways was in fact 2.

The real problem was my test program was assuming that the absence of
the snc_ways file meant no SNC, so it was using cache IDs instead of
node IDs when choosing a mon_data subdirectory to read results from.

With that fixed, the results look good:

cpu: 95 (3), rmid: 17 (g16): 0 -> 32313974784 (32313974784)
cpu: 198 (3), rmid: 103 (g102): 0 -> 26078961664 (26078961664)
cpu: 117 (0), rmid: 59 (g58): 0 -> 26692599808 (26692599808)
cpu: 152 (1), rmid: 113 (g112): 0 -> 33368244224 (33368244224)
cpu: 33 (1), rmid: 77 (g76): 0 -> 26902077440 (26902077440)
cpu: 63 (2), rmid: 76 (g75): 0 -> 32478494720 (32478494720)
cpu: 90 (3), rmid: 94 (g93): 0 -> 31206088704 (31206088704)
cpu: 136 (0), rmid: 13 (g12): 0 -> 28095463424 (28095463424)
cpu: 37 (1), rmid: 177 (g175): 0 -> 31255060480 (31255060480)
cpu: 124 (0), rmid: 6 (g5): 0 -> 31128502272 (31128502272)
cpu: 102 (3), rmid: 68 (g67): 0 -> 28848963584 (28848963584)
cpu: 103 (3), rmid: 62 (g61): 0 -> 26091233280 (26091233280)
cpu: 71 (2), rmid: 123 (g122): 0 -> 29157933056 (29157933056)
cpu: 94 (3), rmid: 69 (g68): 0 -> 27776458752 (27776458752)
cpu: 102 (3), rmid: 174 (g172): 0 -> 26349281280 (26349281280)
cpu: 155 (1), rmid: 3 (g2): 0 -> 33489125376 (33489125376)
cpu: 40 (1), rmid: 16 (g15): 0 -> 27142750208 (27142750208)
cpu: 69 (2), rmid: 156 (g154): 0 -> 29294411776 (29294411776)
cpu: 121 (0), rmid: 63 (g62): 0 -> 30636777472 (30636777472)
cpu: 171 (2), rmid: 93 (g92): 0 -> 26103046144 (26103046144)

(and it turns out I never needed to manually look up the node IDs,
because the test output would have already told me, had the test been
working correctly)

Sorry for the extra trouble. The series works well for me.

Tested-By: Peter Newman <peternewman@...gle.com>

Thanks!
-Peter