linux-kernel - Re: linux acpi (thunderbolt? bug)

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:   Fri, 16 Feb 2018 18:23:13 +0300
From:   Yuriy Vostrikov <delamonpansie@...il.com>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     linux-kernel@...r.kernel.org, x86@...nel.org
Subject: Re: linux acpi (thunderbolt? bug)

On 15 February 2018 at 11:52, Thomas Gleixner <tglx@...utronix.de> wrote:
> On Wed, 14 Feb 2018, Thomas Gleixner wrote:
>> On Wed, 14 Feb 2018, Yuriy Vostrikov wrote:
>> > after sleep 1 time
>> > name:   VECTOR
>> >  size:   0
>> >  mapped: 35
>> >  flags:  0x00000041
>> > Online bitmaps:        2
>> > Global available:    385
>> > Global reserved:      12
>> > Total allocated:      32
>> > System: 41: 0-19,32,50,128,238-255
>> >  | CPU | avl | man | act | vectors
>> >      0   185     1    18  33-43,46-49,51-53
>> >      1   200     1     3  33-37
>>
>> The accounting is already screwed. CPU1 claims to have 3 allocated vectors,
>> but the allocation bitmap has 5 bits set !?!
>>
>> I have no idea yet how that can happen. Lemme stare into the code some more
>> and I came back to you.
>
> Still confused.
>
> Does this happen if you boot w/o the thunderbolt thingy and then do
> suspend/resume cycles?
>
> Can you please take snapshots from:
>
>     /proc/interrupts
>     /sys/kernel/debug/irq/*
>
> right after boot, after the unplug, before suspend and after resume?
>

Apparently, timing is important: problem manifests if the laptop goes
to sleep shortly after unplug.
If there is some delay between unplugging and sleeping, then there is
no problem.
I'm attaching tar.gz with two runs: run-1 with the problem and run-2
without. Dumps include output
of dmesg in time of making a snapshot.

Hope this clarifies the situation a bit.

Thank you,
Yuriy.

Download attachment "dump.tar.gz" of type "application/x-gzip" (181264 bytes)