Hi Greg, Am Mittwoch, 29. Januar 2025, 14:36:07 CET schrieb Greg Kroah-Hartman: > On Wed, Jan 29, 2025 at 02:27:05PM +0100, Jan Čermák wrote: > > Hi Greg, everyone, > > > > unfortunately, this patch introduced a regression on rk356x boards, as the > > current DTS is missing the reset names. This was pointed out in 6.12 series > > by Chukun Pan [1], it applies here as well. Real world examples of breakages > > are M.2 NVMe on ODROID-M1S [2] and USB 3 ports on ODROID-M1 [3]. This patch > > shouldn't have been applied without the device tree change or extra fallback > > code, as suggested in the discussion for Chukun's original commits [4]. > > Version 6.6.74 is still affected by the bug. > > > > Regards, > > Jan > > > > [1] > > https://lore.kernel.org/stable/20241231021010.17792-1-amadeus@xxxxxxxxxx/ > > [2] https://github.com/home-assistant/operating-system/issues/3837 > > [3] https://github.com/home-assistant/operating-system/issues/3841 > > [4] https://lore.kernel.org/all/20250103033016.79544-1-amadeus@xxxxxxxxxx/ > > > > #regzbot introduced: v6.6.68..v6.6.69 > > So where should it be reverted from, 6.6.y and 6.12.y? Or should a > different specific commit be backported instead? Alternatively, ways forward would be: - 6.13 contains the devicetree change, the failing phy-patch requires: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8b9c12757f919157752646faf3821abf2b7d2a64 - there is a pending fix to the phy driver that acts as a fallback for old DTs, waiting for phy maintainers coming back from winter-break https://lore.kernel.org/all/20250106100001.1344418-2-amadeus@xxxxxxxxxx/ > And this isn't an issue on 6.13, right? correct, 6.13 got the dt change that was part of the original series. Of course the fallback thing from above should've been part of the original submission in the first place. Heiko
