Re: [GIT PULL] RISC-V SpacemiT Devicetrees for v7.3

[Yixun Lan <dlan@kernel.org>]

hi Arnd,

On 09:22 Tue 11 Aug     , Arnd Bergmann wrote:
> On Tue, Aug 11, 2026, at 04:49, Troy Mitchell wrote:
> > On Tue Aug 11, 2026 at 12:45 AM +08, Arnd Bergmann wrote:
> >> On Sun, Aug 2, 2026, at 04:42, Yixun Lan wrote:
> >>
> >>> Troy Mitchell (1):
> >>>       riscv: dts: spacemit: k3: add i2s0-i2s5 nodes
> >>
> >> This commit caused a build failure, so I reverted it for the 7.3
> >> merge window:
> >>
> >>     Lexical error: ../arch/riscv/boot/dts/spacemit/k3.dtsi:1181.20-39 Unexpected 'CLK_MPMU_I2S_SYSCLK'
> >>
> >> Please make sure that each pull request you send is free of
> >> build failures and warnings and works on its own.
> >>
Ok, Learned

> >
> > The missing macro has been merged here [1].
> >
> > Link: 
> > https://lore.kernel.org/all/20260717-k3-clk-fix-i2s-v1-1-e95001a692ee@linux.spacemit.com/ 
> > [1]
> 
> Right, the problem was that your patch to use the macro was
> applied on a different branch that caused the build to
> fail. Yixun Lan should have caught this problem and rejected
> your patch before sending the pull request when doing
> a 'make dtbs_check'.
> 

Sorry, I didn't catch this problem due to test DT together with the 
clock patches.. I'll be more careful next time to send a PR without
unmerged cross-tree dependency, thanks for taking care of this..

> This is a common problem with unstructured SoC devices:
> If you can't just use the number from a datasheet, you
> have to wait until the binding has been agreed and merged.
> 
> Earlier in the cycle this could have been resolved by
> ensuring the binding change is used as the base for both
> the clk driver and the dts branch, but now you'll have
> to wait until after -rc1 is out to resend.
> 

-- 
Yixun Lan (dlan)