linux-kernel - Re: [PATCH v3] mm: fix panic in __alloc

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, 9 Dec 2021 10:56:42 +0100
From:   Michal Hocko <mhocko@...e.com>
To:     Alexey Makhalov <amakhalov@...are.com>
Cc:     Dennis Zhou <dennis@...nel.org>,
        Eric Dumazet <eric.dumazet@...il.com>,
        "linux-mm@...ck.org" <linux-mm@...ck.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        David Hildenbrand <david@...hat.com>,
        Oscar Salvador <osalvador@...e.de>, Tejun Heo <tj@...nel.org>,
        Christoph Lameter <cl@...ux.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "stable@...r.kernel.org" <stable@...r.kernel.org>
Subject: Re: [PATCH v3] mm: fix panic in __alloc_pages

On Thu 09-12-21 09:28:55, Alexey Makhalov wrote:
> 
> 
> > On Dec 9, 2021, at 12:46 AM, Michal Hocko <mhocko@...e.com> wrote:
> > 
> > On Thu 09-12-21 02:16:17, Alexey Makhalov wrote:
> >> This patch calls alloc_percpu() from setup_arch() while percpu
> >> allocator is not yet initialized (before setup_per_cpu_areas()).
> > 
> > Yeah, I haven't realized the pcp is not available. I was not really sure
> > about that. Could you try with the alloc_percpu dropped?
> > 
> > Thanks for testing!
> > -- 
> > Michal Hocko
> > SUSE Labs
> 
> It boots now. dmesg has these new messages:
> 
> [    0.081777] Node 4 uninitialized by the platform. Please report with boot dmesg.
> [    0.081790] Initmem setup node 4 [mem 0x0000000000000000-0x0000000000000000]
> ...
> [    0.086441] Node 127 uninitialized by the platform. Please report with boot dmesg.
> [    0.086454] Initmem setup node 127 [mem 0x0000000000000000-0x0000000000000000]

Interesting that only those two didn't get a proper arch specific
initialization. Could you check why? I assume init_cpu_to_node
doesn't see any CPU pointing at this node. Wondering why that would be
the case but that can be a bug in the affinity tables.

> vCPU/node hot add works.
> Onlining works as well, but with warning. I do not think it is related to the patch:
> [   36.838838] CPU4 has been hot-added
> [   36.838987] acpi_processor_hotadd_init:205 cpu 4, node 4, online 0, ndata 00000000e9c7f79b
> [   48.480498] Built 4 zonelists, mobility grouping on.  Total pages: 961440
> [   48.480508] Policy zone: Normal
> [   48.508318] smpboot: Booting Node 4 Processor 4 APIC 0x8
> [   48.509255] Disabled fast string operations
> [   48.509807] smpboot: CPU 4 Converting physical 8 to logical package 4
> [   48.509825] smpboot: CPU 4 Converting physical 0 to logical die 4
> [   48.510040] WARNING: workqueue cpumask: online intersect > possible intersect

I will double check. There are changes required on the hotplug side. I
would like to see that this one doesn't blow up before diving there.

> [   48.510324] vmware: vmware-stealtime: cpu 4, pa 3e667000
> [   48.511311] Will online and init hotplugged CPU: 4
> 
> Hot remove does not quite work. It might be issue in ACPI/Firmware code or Hypervisor. Debugging…
> 
> Do you want me to perform any specific tests?

No, not really. AFAIU your issue has been reproducible during boot and
that seems to be fixed. I will work on the hotplug side of the things
and post something resembling a real patch soon. That would require also
memory hotplug testing.

Thanks for your help!
-- 
Michal Hocko
SUSE Labs