Re: [PATCH RFC 06/12] rust: serial: Add minimal serial-core polling console support

[Shuai Liu <847807867@qq.com>]

Hi Kwanghoon,

On Fri, Oct 30, 2026 at 07:58:38AM -0400, Kwanghoon Son <asdfkwang@gmail.com> wrote:
> I don't understand for now, COMMON_CLK not possible on x86?
> or you mean CI not supported?

You are right to question this -- my comment was incorrect, sorry for
the confusion.

COMMON_CLK is not restricted away on x86. It only depends on
!HAVE_LEGACY_CLK, which x86 does not select, and several x86 platform
options (e.g. X86_INTEL_LPSS, X86_AMD_PLATFORM_DEVICE) select it
themselves. I verified that "make ARCH=x86 allmodconfig" results in
CONFIG_OF=y and CONFIG_COMMON_CLK=y, so with Rust enabled, as in the
Rust CI, all of the dependencies of SERIAL_BFLB_RUST (RUST, OF,
COMMON_CLK, HAS_IOMEM) are satisfiable on x86 and the driver is
eligible for build coverage from allmodconfig there.

To back this up, I also build-tested the series on x86 (LLVM=1,
allmodconfig-based configuration with Rust enabled): the serial-core
abstractions, the C helper glue and SERIAL_BFLB_RUST all compile
cleanly, and a full x86-64 vmlinux links with the driver built in.

What happened is that I conflated COMMON_CLK with arch-scoped
dependencies I had been looking at in other drivers (e.g. RISCV- or
OF-scoped Kconfig entries) and stated it without verifying it first.
It also means my suggestion of "depends on RISCV || COMPILE_TEST" in
that mail was unnecessary; the Kconfig as it stands is fine.

Thanks for catching this and for the question.

Best regards,
Shuai Liu