Bug#995927: linux-image-amd64: since kernel 5.9 module e100 causes system-wide problems after suspend
Control: tags -1 + moreinfo
Hi,
On Fri, Oct 08, 2021 at 12:03:01PM +0200, hikaru wrote:
> Package: linux-image-amd64
> Version: 5.10.46-5
> Severity: normal
> Tags: upstream
> X-Debbugs-Cc: hikaru.debian@web.de
>
> Dear Maintainers,
>
> I own an old notebook with the following wired ethernet adapter:
>
> 07:08.0 Ethernet controller: Intel Corporation PRO/100 VE Network Connection (rev 02)
> Subsystem: Fujitsu Technology Solutions PRO/100 VE Network Connection
> Flags: bus master, medium devsel, latency 64, IRQ 20
> Memory at dc000000 (32-bit, non-prefetchable) [size=4K]
> I/O ports at 5000 [size=64]
> Capabilities: [dc] Power Management version 2
> Kernel driver in use: e100
> Kernel modules: e100
>
> Under Bullseye (kernel 5.10), after a suspend2RAM/resume cycle, e100 causes
> problems which seem to be related to "e100: use generic power management" [1],
> which wass introduced with kernel 5.9. In the comment it says:
> "Compile-tested only."
>
>
> The symptoms are not always consistent, but usually they are as follows:
>
> 1. After resume, there is no network at all (not only via the wired adapter,
> but also via the intel wifi adapter), at least for several (5?) minutes.
> Network may or may not return after that time.
>
> 2. During that time, certain interactions with the system just stall -
> especially those connected to network tasks (e.g. network-manager-gnome, ip).
> In case "ip a" is successful it will usually show that there is a
> network connection (ip adress obtained via dhcp), but no network activity is
> possible (e.g. ping).
> (I ruled out network manager as the culprit, by uninstalling it. The problems
> remained the same.)
>
> 3. When performing a shutdown or reboot after such a suspend/resume cycle, the
> operating system and the HDD will shut down, but the computer will not
> actually power off/reboot. Instead it will stay on indefinitely while messages
> like these remain on the screen:
>
> Okt 06 18:32:19 amilo systemd[1]: Reached target Reboot.
> Okt 06 18:32:19 amilo systemd[1]: Shutting down.
> Okt 06 18:32:19 amilo systemd[1]: Using hardware watchdog 'iTCO_wdt', version 0, device /dev/watchdog
> Okt 06 18:32:19 amilo systemd[1]: Set hardware watchdog to 10min.
> Okt 06 18:32:19 amilo kernel: watchdog: watchdog0: watchdog did not stop!
> Okt 06 18:32:19 amilo systemd-shutdown[1]: Syncing filesystems and block devices.
> Okt 06 18:32:19 amilo systemd-shutdown[1]: Sending SIGTERM to remaining processes...
> Okt 06 18:32:19 amilo systemd-journald[213]: Journal stopped
>
> 4. During that shutdown the kernel may or may not throw exceptions, of
> which I have a photo, but no easily accessible log.
>
>
> Buster (kernel 4.19, up to bpo kernel 5.8) does not exhibit this behavior.
> However, it does with bpo kernels 5.9 and 5.10.
>
> The problems disappear (with kernels >= 5.9) when disabling the
> ethernet adapter in the BIOS or when blacklisting e100.
>
> For sake of transparency, I opened a thread in the unofficial German
> debianforum.de, which led me to file this bug report. [2]
>
>
> [1] https://github.com/torvalds/linux/commit/69a74aef8a18eef20fb0044b5e164af41b84db21
> [2] https://debianforum.de/forum/viewtopic.php?f=26&t=182239
Can you report the upstream directly to upstream people (and ideally
keep us in the loop)?
Regards,
Salvatore
Reply to: