Bug#983818: linux-image-5.10.0-3-arm64: often fails to bring up eth0 / dwmac_rk module

[Salvatore Bonaccorso <carnil@debian.org>]

Control: tags -1 + moreinfo

On Mon, Mar 01, 2021 at 03:33:27PM -0800, Forest wrote:
> Package: src:linux
> Version: 5.10.13-1
> Severity: critical
> Justification: breaks unrelated software
> 
> Dear Maintainer,
> 
> When booting recent kernels on a RockPro64 board (rk3399), eth0 often fails
> to come up, leaving this headless box practically unusable without serial
> console intervention. Logging in on the console and using rmmod/modprobe to
> reload dwmac_rk revives the network interface and allows normal operation
> until the next reboot.
> 
> Logs are included below, but the last relevant dmesg errors seem to be:
> rk_gmac-dwmac fe300000.ethernet eth0: stmmac_hw_setup: DMA engine initialization failed
> rk_gmac-dwmac fe300000.ethernet eth0: stmmac_open: Hw setup failed
> 
> The problem doesn't occur on every boot. I haven't determined what conditions
> make it more likely. It's possible that scheduling an fsck makes it happen
> less frequently (perhaps there's a timing issue?) but it's hard to say; it
> still happens fairly often even when I fsck on every boot.
> 
> The problem appeared some time in the past two or three months. I keep up
> with unstable kernel updates, so maybe a semi-recent change in the unstable
> kernel caused it? Or maybe I was just lucky until recently.
> 
> Curiously, when dropbear ssh launches from initramfs, it never has trouble
> with eth0. The problem doesn't show up until after dropbear has run and I
> have unlocked my root filesystem over ssh and boot continues. I am using an
> initramfs-tools tweak to make dropbear work around #968519, so I suppose that
> bug and the present problem could be related but merely avoided by dropbear.
> However, #968519 was present long before the present problem appeared.
> 
> Here's someone else experiencing this problem:
> https://forum.pine64.org/showthread.php?tid=9351&pid=87304#pid87304

Could you try with the current kernel in unstable? We are at 5.14.6-2,
which had some rk3399 related changes. If you still can reproduce the
issue my best guess would be to make a report upstream, presumably
something around contacting 

Giuseppe Cavallaro <peppe.cavallaro@st.com> (supporter:STMMAC ETHERNET DRIVER)
Alexandre Torgue <alexandre.torgue@foss.st.com> (supporter:STMMAC ETHERNET DRIVER)
Jose Abreu <joabreu@synopsys.com> (supporter:STMMAC ETHERNET DRIVER)
"David S. Miller" <davem@davemloft.net> (maintainer:NETWORKING DRIVERS)
Jakub Kicinski <kuba@kernel.org> (maintainer:NETWORKING DRIVERS)
Maxime Coquelin <mcoquelin.stm32@gmail.com> (maintainer:ARM/STM32 ARCHITECTURE)
Philipp Zabel <p.zabel@pengutronix.de> (maintainer:RESET CONTROLLER FRAMEWORK)
Liam Girdwood <lgirdwood@gmail.com> (supporter:VOLTAGE AND CURRENT REGULATOR FRAMEWORK)
Mark Brown <broonie@kernel.org> (supporter:VOLTAGE AND CURRENT REGULATOR FRAMEWORK)
netdev@vger.kernel.org (open list:STMMAC ETHERNET DRIVER)
linux-stm32@st-md-mailman.stormreply.com (moderated list:ARM/STM32 ARCHITECTURE)
linux-arm-kernel@lists.infradead.org (moderated list:ARM/STM32 ARCHITECTURE)
linux-kernel@vger.kernel.org (open list)

(and keeping us downstream in the loop).

Regards,
Salvatore