linux-kernel - Re: [PATCH] EDAC: expose per-dimm error counts in sysfs

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: <2DF9D2DA-2584-4623-BC1A-69314AA80F2A@fb.com>
Date:   Thu, 27 Oct 2016 21:23:15 +0000
From:   Aaron Miller <aaronmiller@...com>
To:     Borislav Petkov <bp@...en8.de>
CC:     Mauro Carvalho Chehab <mchehab@...nel.org>,
        "linux-edac@...r.kernel.org" <linux-edac@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] EDAC: expose per-dimm error counts in sysfs

If your system is like the one I’m testing on, only the channel 0 DIMM slots are populated, and you injected an error for an unpopulated slot, for which no dimmX directory gets created.

In edac_mc_sysfs.c:

    for (i = 0; i < mci->tot_dimms; i++) {
        struct dimm_info *dimm = mci->dimms[i];
        /* Only expose populated DIMMs */
        if (!dimm->nr_pages)
            continue;


I can repro what you saw here: 

$ cd /sys/devices/system/edac/mc/mc0
$ grep . dimm*/*location
dimm0/dimm_location:channel 0 slot 0
dimm3/dimm_location:channel 1 slot 0
dimm6/dimm_location:channel 2 slot 0
dimm9/dimm_location:channel 3 slot 0

$ echo 1 > /sys/kernel/debug/edac/mc0/fake_inject_channel
$ echo 2 > /sys/kernel/debug/edac/mc0/fake_inject_slot
$ echo 3 > /sys/kernel/debug/edac/mc0/fake_inject_count
$ echo 1 > /sys/kernel/debug/edac/mc0/fake_inject
$ cat ce_count
3

$ grep . dimm*/*ce_count
dimm0/dimm_ce_count:0
dimm3/dimm_ce_count:0
dimm6/dimm_ce_count:0
dimm9/dimm_ce_count:0


And I get what I expect for a populated slot:

$ echo 0 > /sys/kernel/debug/edac/mc0/fake_inject_slot
$ echo 1 > /sys/kernel/debug/edac/mc0/fake_inject
$ cat ce_count
6

$ grep . dimm*/*ce_count
dimm0/dimm_ce_count:0
dimm3/dimm_ce_count:3
dimm6/dimm_ce_count:0
dimm9/dimm_ce_count:0



On 10/27/16, 11:07 AM, "Borislav Petkov" <bp@...en8.de> wrote:

    On Tue, Oct 25, 2016 at 04:25:51PM -0700, Aaron Miller wrote:
    
    <--- This patch needs a commit message.
    
    Especially as to *why* we need this.
    
    > Signed-off-by: Aaron Miller <aaronmiller@...com>
    > ---
    >  drivers/edac/edac_mc_sysfs.c | 38 ++++++++++++++++++++++++++++++++++++++
    >  1 file changed, 38 insertions(+)
    
    Regardless, something's still not right yet:
    
    $ echo 1 > /sys/kernel/debug/edac/mc0/fake_inject_channel
    $ echo 2 > /sys/kernel/debug/edac/mc0/fake_inject_slot
    $ echo 3 > /sys/kernel/debug/edac/mc0/fake_inject_count
           ^
    
    $ echo 1 > /sys/kernel/debug/edac/mc0/fake_inject
    
    $ grep . /sys/devices/system/edac/mc/mc0/*count
    /sys/devices/system/edac/mc/mc0/ce_count:3
    					 ^
    
    /sys/devices/system/edac/mc/mc0/ce_noinfo_count:0
    /sys/devices/system/edac/mc/mc0/ue_count:0
    /sys/devices/system/edac/mc/mc0/ue_noinfo_count:0
    
    $ grep -r . /sys/devices/system/edac/mc/mc0/dimm*/* 2>/dev/null | grep ce_count
    /sys/devices/system/edac/mc/mc0/dimm0/dimm_ce_count:0
    /sys/devices/system/edac/mc/mc0/dimm3/dimm_ce_count:0
    /sys/devices/system/edac/mc/mc0/dimm6/dimm_ce_count:0
    /sys/devices/system/edac/mc/mc0/dimm9/dimm_ce_count:0
    						    ^
    
    There should be 3 somewhere in the DIMM counters...
    
    -- 
    Regards/Gruss,
        Boris.
    
    ECO tip #101: Trim your mails when you reply.