netdev - Re: RTL8402 stops working after hibernate/resume

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: <e1c7a37f-d8d0-a773-925c-987b92f12694@gmail.com>
Date:   Sat, 18 Jul 2020 14:07:50 +0200
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Petr Tesarik <ptesarik@...e.cz>
Cc:     Realtek linux nic maintainers <nic_swsd@...ltek.com>,
        netdev@...r.kernel.org
Subject: Re: RTL8402 stops working after hibernate/resume

On 16.07.2020 10:58, Petr Tesarik wrote:
> Hi Heiner,
> 
> first, thank you for looking into this!
> 
> On Wed, 15 Jul 2020 17:22:35 +0200
> Heiner Kallweit <hkallweit1@...il.com> wrote:
> 
>> On 15.07.2020 10:28, Petr Tesarik wrote:
>>> Hi all,
>>>
>>> I've encountered some issues on an Asus laptop. The RTL8402 receive
>>> queue behaves strangely after suspend to RAM and resume - many incoming
>>> packets are truncated, but not all and not always to the same length
>>> (most commonly 60 bytes, but I've also seen 150 bytes and other
>>> lengths).
>>>
>>> Reloading the driver can fix the problem, so I believe we must be
>>> missing some initialization on resume. I've already done some
>>> debugging, and the interface is not running when rtl8169_resume() is
>>> called, so __rtl8169_resume() is skipped, which means that almost
>>> nothing is done on resume.
>>>   
>> The dmesg log part in the opensuse bug report indicates that a userspace
>> tool (e.g. NetworkManager) brings down the interface on suspend.
>> On resume the interface is brought up again, and PHY driver is loaded.
>> Therefore it's ok that rtl8169_resume() is a no-op.
>>
>> The bug report mentions that the link was down before suspending.
>> Does the issue also happen if the link is up when suspending?
> 
> I have tried, and it makes no difference.
> 
>> Interesting would also be a test w/o a network manager.
>> Means the interface stays up during suspend/resume cycle.
> 
> I have stopped NetworkManager and configured a static IP address for
> the interface. Still the same result.
> 
> I have verified that the firmware is loaded, both before suspend and
> after resume:
> 
> zabulon:~ # ethtool -i eth0 
> driver: r8169
> version: 5.7.7-1-default
> firmware-version: rtl8402-1_0.0.1 10/26/11
> expansion-rom-version: 
> bus-info: 0000:03:00.2
> supports-statistics: yes
> supports-test: no
> supports-eeprom-access: no
> supports-register-dump: yes
> supports-priv-flags: no
> 
>> Unfortunately it's not known whether it's a regression, and I have no
>> test hw with this chip version.
>>
>> Also you could test whether the same happens with the r8101 vendor driver.
> 
> I was not aware of this alternative driver... Anyway, I have built
> r8101 from git (v1.035.03) for kernel 5.7.7. When loaded, it hangs the
> machine hard. I mean like not even SysRq+B works...
> 
> Petr T
> 
>>> Some more information can be found in this openSUSE bug report:
>>>
>>> https://bugzilla.opensuse.org/show_bug.cgi?id=1174098
>>>
>>> The laptop is not (yet) in production, so I can do further debugging if
>>> needed.
>>>
>>> Petr T
>>>   
>> Heiner
> 

Maybe the following gives us an idea:
Please do "ethtool -d <if>" after boot and after resume from suspend,
and check for differences.