Bug 215641 - Dell G15 5510 s2idle fails in 5.16.11 works in 5.16.10
Summary: Dell G15 5510 s2idle fails in 5.16.11 works in 5.16.10
Status: RESOLVED UNREPRODUCIBLE
Alias: None
Product: Power Management
Classification: Unclassified
Component: Hibernation/Suspend (show other bugs)
Hardware: Intel Linux
: P1 normal
Assignee: Rafael J. Wysocki
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-02-25 18:04 UTC by Max Fedotov
Modified: 2022-03-19 17:15 UTC (History)
2 users (show)

See Also:
Kernel Version: 5.16.11
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
dmesg log (143.75 KB, text/plain)
2022-02-25 18:04 UTC, Max Fedotov
Details
dmesg suspend crash report (132.19 KB, text/plain)
2022-03-02 15:07 UTC, Max Fedotov
Details
system freeze report (112.24 KB, text/plain)
2022-03-09 07:24 UTC, Max Fedotov
Details

Description Max Fedotov 2022-02-25 18:04:10 UTC
Created attachment 300509 [details]
dmesg log

Fedora 35, after update to 5.16.11 s2idle sleep stopped working, works fine in 5.16.10.
Comment 1 Max Fedotov 2022-03-01 18:08:25 UTC
Bisected:

>35cd1e78f8ac045cf557cbd017c8397f082e309d is the first bad commit
>commit 35cd1e78f8ac045cf557cbd017c8397f082e309d
>Author: Al Cooper <alcooperx@gmail.com>
>Date:   Wed Dec 1 13:06:51 2021 -0500
>
>    phy: usb: Leave some clocks running during suspend
>    
>    [ Upstream commit 42fed57046fc74586d7058bd51a1c10ac9c690cb ]
>    
>    The PHY client driver does a phy_exit() call on suspend or rmmod and
>    the PHY driver needs to know the difference because some clocks need
>    to be kept running for suspend but can be shutdown on unbind/rmmod
>    (or if there are no PHY clients at all).
>    
>    The fix is to use a PM notifier so the driver can tell if a PHY
>    client is calling exit() because of a system suspend or a driver
>    unbind/rmmod.
>    
>    Signed-off-by: Al Cooper <alcooperx@gmail.com>
>    Acked-by: Florian Fainelli <f.fainelli@gmail.com>
>    Link: https://lore.kernel.org/r/20211201180653.35097-2-alcooperx@gmail.com
>    Signed-off-by: Vinod Koul <vkoul@kernel.org>
>    Signed-off-by: Sasha Levin <sashal@kernel.org>
>
> drivers/phy/broadcom/phy-brcm-usb.c | 38
> +++++++++++++++++++++++++++++++++++++
> 1 file changed, 38 insertions(+)
Comment 2 Hans de Goede 2022-03-01 19:09:28 UTC
I think you may have made an error with the bisect. The only file touched by the commit your bisect points to is not used (*) in Fedora x86_64 builds.

*) it is not even compiled.
Comment 3 Max Fedotov 2022-03-02 07:08:43 UTC
(In reply to Hans de Goede from comment #2)
> I think you may have made an error with the bisect. The only file touched by
> the commit your bisect points to is not used (*) in Fedora x86_64 builds.
> 
> *) it is not even compiled.

Are you sure? Comment on redhat bugzilla says otherwise.
https://bugzilla.redhat.com/show_bug.cgi?id=2059688
Comment 4 Hans de Goede 2022-03-02 09:09:49 UTC
(In reply to Max Fedotov from comment #3)
> (In reply to Hans de Goede from comment #2)
> > I think you may have made an error with the bisect. The only file touched
> by
> > the commit your bisect points to is not used (*) in Fedora x86_64 builds.
> > 
> > *) it is not even compiled.
> 
> Are you sure? Comment on redhat bugzilla says otherwise.
> https://bugzilla.redhat.com/show_bug.cgi?id=2059688

Yes I'm pretty sure, if you look at:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=42fed57046fc74586d7058bd51a1c10ac9c690cb

it only touches: drivers/phy/broadcom/phy-brcm-usb.c and if you look at drivers/phy/broadcom/Makefile that only gets build if when CONFIG_PHY_BRCM_USB is set and in e.g. /boot/config-5.17.0-0.rc5.102.fc36.x86_64 that is not set.

So I think that Justin did not look at the commit you are pointing to, but just assumed you were pointing to a specific patch which was recently merged to fix a set of suspend/resume issues. I've asked Justin which patch he meant in the RH bugzilla.

In the mean time, since I'm pretty sure that the commit you found is not the cause it would be helpful if you can redo the bisect.
Comment 5 JJ 2022-03-02 10:57:06 UTC
Hi, I think this might be the same issue reported here: https://bbs.archlinux.org/viewtopic.php?id=274292

If so, it appears in both linux and linux-lts and was probably introduced in 5.16.10 and backported to the 5.15 line in either 5.15.23 or 5.15.24.
Comment 6 Max Fedotov 2022-03-02 15:06:57 UTC
I redid the bisect and could not reproduce the problem. I then reinstalled Fedora 5.16.11 kernel and could not reproduce again. However, during the tests I encountered some oopses (sometimes during boot, sometimes during suspend attempt) which either made system freeze later when I tried to reboot or froze the system right away.

One of those is reported here: https://bugzilla.redhat.com/show_bug.cgi?id=2058504
It happens rarely during boot and causes random system problems like missing sound or problems during later reboot attempt.

Another crash happens randomly (rare) during suspend attempt and makes system unresponsive. dmesg output is attached.
Comment 7 Max Fedotov 2022-03-02 15:07:24 UTC
Created attachment 300518 [details]
dmesg suspend crash report
Comment 8 Hans de Goede 2022-03-08 19:52:11 UTC
(In reply to Max Fedotov from comment #7)
> Created attachment 300518 [details]
> dmesg suspend crash report

I took a quick look and that seems fw-loader related, assuming the backtrace is always the same ?

But I did not see any recent fw-loader changes, so not sure what to make of this.
Comment 9 Max Fedotov 2022-03-08 19:55:05 UTC
I'll check dmesg next time I encounter the problem to see if backtrace is the same. It seems random, I couldn't find reliable way to reproduce it.
Comment 10 Max Fedotov 2022-03-09 07:24:32 UTC
Created attachment 300547 [details]
system freeze report

I just got system freeze while watching video with dragon player. There's no meaningful backtrace, some page fault. Not sure if it's the same issue as before.
Comment 11 Hans de Goede 2022-03-09 16:39:25 UTC
(In reply to Max Fedotov from comment #10)
> Created attachment 300547 [details]
> system freeze report
> 
> I just got system freeze while watching video with dragon player. There's no
> meaningful backtrace, some page fault. Not sure if it's the same issue as
> before.

This one seems to indicate an issue with the nvidia binary driver. Have you tried to reproduce your issues without the nvidia binary driver being loaded?
Comment 12 Hans de Goede 2022-03-19 13:58:54 UTC
Ping? Can you please see if you are still seeing issues when not using the nvidia binary driver?
Comment 13 Max Fedotov 2022-03-19 15:07:39 UTC
I'm not having any issues with 5.16.14 or 5.16.15 kernels, not even ext4 crash, whatever it was it seems to be fixed.

Note You need to log in before you can comment on or make changes to this bug.