linux-kernel - Re: [PATCH 2/2 v6] mm/mempolicy: Don't create weight sysfs for memoryless nodes

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: <9caca3a8-280a-45bd-a081-cf4a28f05f21@sk.com>
Date: Mon, 10 Mar 2025 21:26:48 +0900
From: Honggyu Kim <honggyu.kim@...com>
To: Gregory Price <gourry@...rry.net>
Cc: kernel_team@...ynix.com, Joshua Hahn <joshua.hahnjy@...il.com>,
 harry.yoo@...cle.com, ying.huang@...ux.alibaba.com,
 gregkh@...uxfoundation.org, rakie.kim@...com, akpm@...ux-foundation.org,
 rafael@...nel.org, lenb@...nel.org, dan.j.williams@...el.com,
 Jonathan.Cameron@...wei.com, dave.jiang@...el.com, horen.chuang@...ux.dev,
 hannes@...xchg.org, linux-kernel@...r.kernel.org,
 linux-acpi@...r.kernel.org, linux-mm@...ck.org, kernel-team@...a.com,
 yunjeong.mun@...com
Subject: Re: [PATCH 2/2 v6] mm/mempolicy: Don't create weight sysfs for
 memoryless nodes

Hi Gregory,

On 3/8/2025 2:51 AM, Gregory Price wrote:
> On Fri, Mar 07, 2025 at 08:46:46PM +0900, Honggyu Kim wrote:
>> You can see more info below.
>>
>>    $ cd /sys/devices/system/node
>>
>>    $ ls -d node*
>>    node0  node1  node2  node3
>>
>>    $ cat possible
>>    0-11
> 
> We're split across two threads now, but i'll add this context
> 
> I'm basically asking whether there should be 12 nodes possible. It seems
> like there should only be 4 nodes possible - 2 for sockets, 2 for host
> bridges.

Ack. If the N_POSSIBLE itself becomes 4, then I agree this problem can simply be 
resolved.

> 
> Unless I'm misunderstanding, it should be the case that a given physical
> address can only be hosted by 1 numa node (proximity domain).

Yeah, the proximity domain detects the node correctly as follows in dmesg.

  [  0.009915] ACPI: SRAT: Node 0 PXM 0 [mem 0x00000000-0x7fffffff]
  [  0.009917] ACPI: SRAT: Node 0 PXM 0 [mem 0x100000000-0x207fffffff]
  [  0.009919] ACPI: SRAT: Node 1 PXM 1 [mem 0x60f80000000-0x64f7fffffff]
  [  0.009924] ACPI: SRAT: Node 2 PXM 6 [mem 0x2080000000-0x807fffffff] hotplug
  [  0.009925] ACPI: SRAT: Node 3 PXM 7 [mem 0x64f80000000-0x6cf7fffffff] hotplug

It is printed even before CXL detection.

> 
> So it *should* be the case that you either have 4 nodes possible or 10
> nodes possible, not 12.  But I could be missing a piece of context.
> 
>> Which command do we need for this info specifically?  My output doesn't
>> provide some useful info for that.
>>
>>    $ acpidump -b
>>    $ iasl -d *
>>    $ cat cedt.dsl
>>        ...
>>    **** Unknown ACPI table signature [CEDT]
>>
> 
> You probably have an old version of acpidump here, you might need to get
> a newer version that knows about the CEDT.

I just used the newest acpica and was able to dump CEDT properly.  But its
output is also very long so it'd be helpful if you could tell us which specific
info you need.

> 
> You'll also want to get all the Memory Affinity entries from srat.dsl
> 
>> Not sure about it.  This must be fixed ASAP because current kernel is
>> broken on this issue and the fix should go into hotfix tree first.
>>
> 
> I agree something is broken, I'm not convinced what is broken.

Yeah, we should fix the broken status hopefully before v6.14 release.

Thanks,
Honggyu

> 
>> If you can think this is just a bandaid, but leaving it bleeding as is
>> not the right approach.
>>
> 
> This affects userland, we shouldn't thrash on this. Lets get it right.
> 
> ~Gregory