linux-kernel - Re: r8169: disable ASPM during NAPI poll locking changes --> BUG: Invalid wait context --> PREEMPT

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: <1bea6e585a1be8ab89b9a3db2fb5c58431378f50.camel@gmx.de>
Date:   Mon, 22 May 2023 19:02:48 +0200
From:   Mike Galbraith <efault@....de>
To:     Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Cc:     Heiner Kallweit <hkallweit1@...il.com>,
        lkml <linux-kernel@...r.kernel.org>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: r8169: disable ASPM during NAPI poll locking changes --> BUG:
 Invalid wait context --> PREEMPT_RT pain

On Mon, 2023-05-22 at 15:43 +0200, Sebastian Andrzej Siewior wrote:
> On 2023-05-22 12:41:21 [+0200], Mike Galbraith wrote:
> > > I've been looking at the r8169 the other day and it seemed all good
> > > since it was only scheduling workqueues and NAPI. But now I see this
> > > mac_ocp_lock + config25_lock which are acquire in hardirq context but
> > > can't. This needs to be taken care.
> >
> > I did a patch converting the locks, but didn't post it hoping maybe
> > those locks weren't really really needed.  It doesn't matter for my
> > box, as it doesn't do anything but give RT chances to trip over bugs.
>
> I missed that part. Do you still see those 160us if you keep the locks
> as-it as drop the no-threads flag? I'm trying to figure out if this is
> due to hwirq context or just blocked PCI-bus due ASPM.

That 160us was with the nouveau left on by mistake, the worst ftrace
recorded with me beating up box was 77us (and it inflates).  Without
ftrace and keeping box quiet except for youtube and netperf with lappy,
it looks like rtmutex v spinlock would eventually end in a near tie,
both being noticeable, but for a box without an RT mission, hohum.

Me preferring to not notice them, <poof> all better.

no locks
policy: fifo: loadavg: 13.22 9.21 4.16 12/970 5737

T: 0 ( 4385) P:99 I:1000 C: 329625 Min:      1 Act:    4 Avg:    3 Max:      31
T: 1 ( 4386) P:99 I:1500 C: 219749 Min:      1 Act:    5 Avg:    3 Max:      24
T: 2 ( 4387) P:99 I:2000 C: 164811 Min:      1 Act:    4 Avg:    3 Max:      24
T: 3 ( 4388) P:99 I:2500 C: 131849 Min:      1 Act:    6 Avg:    4 Max:      25
T: 4 ( 4389) P:99 I:3000 C: 109874 Min:      1 Act:    4 Avg:    3 Max:      24
T: 5 ( 4390) P:99 I:3500 C:  94177 Min:      1 Act:    5 Avg:    4 Max:      25
T: 6 ( 4391) P:99 I:4000 C:  82405 Min:      1 Act:    5 Avg:    4 Max:      23
T: 7 ( 4392) P:99 I:4500 C:  73248 Min:      1 Act:    5 Avg:    4 Max:      27

spinlock_t IRQF_SHARED
policy: fifo: loadavg: 13.99 10.01 4.63 11/925 5459

T: 0 ( 5188) P:99 I:1000 C: 330449 Min:      1 Act:    4 Avg:    3 Max:      41
T: 1 ( 5189) P:99 I:1500 C: 220298 Min:      1 Act:    6 Avg:    4 Max:      25
T: 2 ( 5190) P:99 I:2000 C: 165223 Min:      1 Act:   12 Avg:    4 Max:      58
T: 3 ( 5191) P:99 I:2500 C: 132178 Min:      1 Act:    8 Avg:    4 Max:      44
T: 4 ( 5192) P:99 I:3000 C: 110148 Min:      1 Act:    8 Avg:    4 Max:      43
T: 5 ( 5193) P:99 I:3500 C:  94412 Min:      1 Act:    6 Avg:    4 Max:      23
T: 6 ( 5194) P:99 I:4000 C:  82611 Min:      1 Act:    6 Avg:    5 Max:      24
T: 7 ( 5195) P:99 I:4500 C:  73431 Min:      1 Act:   10 Avg:    5 Max:      38

raw_spinlock_t IRQF_NO_THREAD
policy: fifo: loadavg: 14.84 10.35 4.81 4/1031 5744

T: 0 ( 4680) P:99 I:1000 C: 347816 Min:      1 Act:    6 Avg:    4 Max:      29
T: 1 ( 4681) P:99 I:1500 C: 231877 Min:      1 Act:    6 Avg:    4 Max:      49
T: 2 ( 4682) P:99 I:2000 C: 173907 Min:      1 Act:    7 Avg:    5 Max:      39
T: 3 ( 4683) P:99 I:2500 C: 139125 Min:      1 Act:    6 Avg:    5 Max:      46
T: 4 ( 4684) P:99 I:3000 C: 115937 Min:      1 Act:    4 Avg:    4 Max:      34
T: 5 ( 4685) P:99 I:3500 C:  99375 Min:      1 Act:    7 Avg:    5 Max:      32
T: 6 ( 4686) P:99 I:4000 C:  86953 Min:      1 Act:    9 Avg:    5 Max:      28
T: 7 ( 4687) P:99 I:4500 C:  77291 Min:      1 Act:    6 Avg:    5 Max:      37