Re: [PATCH v2] serial: sc16is7xx: Fix clock leak on invalid clock frequency
From: Philip Li
Date: Mon Nov 16 2026 - 00:11:55 EST
On Mon, Nov 16, 2026 at 11:10:42AM +0800, Xiangwu Lai wrote:
> Hi,
>
> thanks for the reports. The build error is an artifact of the test
> setup, not a problem in the patch itself.
Thanks for detail info, and sorry about the false reports, i will further
check to fix the bot issue.
Thanks
>
> Both reports build v2 on top of the linux-review branch created
> for the original 2026-10-30 posting, i.e. on top of the superseded
> v1. v1 redirected the invalid-frequency -EINVAL path to a "goto
> out_clk", making use of the error label that exists in mainline.
> v2 is a full rework that replaces v1: it switches to
> devm_clk_get_optional_enabled(), so the -EINVAL path keeps its
> original plain return, exactly like mainline, and the mainline
> out_clk label is removed. Since the -EINVAL path is identical in
> mainline and in v2, v2's diff does not touch that line at all.
> Stacking v2 on top of the v1 baseline therefore leaves v1's
> "goto out_clk" in the invalid-frequency path while the label is
> removed, which produces exactly the reported "use of undeclared
> label 'out_clk'" at sc16is7xx.c:1622.
>
> I verified this locally: v2 applies cleanly on top of current
> mainline and the result contains no reference to out_clk at all,
> while applying v2 on top of v1 reproduces the dangling goto
> reported here.
>
> So there is nothing to fix in v2. It was posted as a standalone
> replacement for v1 (as a new top-level thread), and the build
> error only shows up when it is stacked on top of the v1 baseline.
>
> Best regards,
> Xiangwu Lai
>
>
>