Re: [PATCH v2 1/1] rust: add `rustcheck` make target for check-only builds

[Nicolas Schier <nsc@kernel.org>]

On Tue, Sep 23, 2025 at 06:47:20AM +0300, Onur Özkan wrote:
> On Mon, 22 Sep 2025 08:02:36 +0200
> Nicolas Schier <nicolas.schier@linux.dev> wrote:
> 
> > On Mon, Sep 15, 2025 at 02:53:11PM +0300, Onur Özkan wrote:
> > > Adds a new `rustcheck` make target to run a check-only build
> > > similar to `cargo check`. This allows us to verify that the Rust
> > > sources can build without building/linking final artifacts,
> > > which speeds up the iteration (a lot) during development.
> > > 
> > > The target also supports the CLIPPY flag (e.g., `make LLVM=1
> > > rustcheck CLIPPY=1) to run Clippy in a faster way.
> > > 
> > > Also, unlike `make LLVM=1`, it doesn't compile large amounts of C
> > > code (on a fresh checkout) when the goal is only to check that
> > > Rust builds are not broken after some changes.
> > > 
> > > Suggested-by: Benno Losin <lossin@kernel.org>
> > > Link:
> > > https://rust-for-linux.zulipchat.com/#narrow/channel/288089/topic/x/near/539103602
> > > Signed-off-by: Onur Özkan <work@onurozkan.dev> ---
> > >  Makefile      |  7 +++++
> > >  rust/Makefile | 73
> > > +++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files
> > > changed, 80 insertions(+)
> > > 
> > > diff --git a/Makefile b/Makefile
> > > index cf37b9407821..7812cdc72938 100644
> > > --- a/Makefile
> > > +++ b/Makefile
> > > @@ -1716,6 +1716,8 @@ help:
> > >  	@echo  '		    is formatted, printing a diff
> > > otherwise.' @echo  '  rustdoc	  - Generate Rust
> > > documentation' @echo  '		    (requires kernel
> > > .config)'
> > > +	@echo  '  rustcheck       - Check that the Rust code
> > > builds'
> > > +	@echo  '                    (requires kernel .config)'
> > >  	@echo  '  rusttest        - Runs the Rust tests'
> > >  	@echo  '                    (requires kernel .config;
> > > downloads external repos)' @echo  '  rust-analyzer	  -
> > > Generate rust-project.json rust-analyzer support file' @@ -1821,6
> > > +1823,11 @@ PHONY += rustdoc rustdoc: prepare
> > >  	$(Q)$(MAKE) $(build)=rust $@
> > >  
> > > +# Checking Rust sources.
> > > +PHONY += rustcheck
> > > +rustcheck: prepare0
> > 
> > Why do you let rustcheck depend on prepare0 instead of prepare?
> > 
> 
> Because "prepare" does more job which isn't necessary (therefore waste
> of time) for "rustcheck".
> 
> > > +	$(Q)$(MAKE) $(build)=rust $@
> > > +
> > >  # Testing target
> > >  PHONY += rusttest
> > >  rusttest: prepare
> > > diff --git a/rust/Makefile b/rust/Makefile
> > > index bfa915b0e588..b45878870207 100644
> > > --- a/rust/Makefile
> > > +++ b/rust/Makefile
> > > @@ -265,6 +265,79 @@ rusttest-kernel: $(src)/kernel/lib.rs
> > > rusttestlib-ffi rusttestlib-kernel \ rusttestlib-uapi
> > > rusttestlib-pin_init FORCE +$(call if_changed,rustc_test)
> > >  
> > > +## Check-only compilation (similar to `cargo check`)
> > > +quiet_cmd_rustc_check_library = $(RUSTC_OR_CLIPPY_QUIET) CHECK $<
> > > +      cmd_rustc_check_library = \
> > > +        OBJTREE=$(abspath $(objtree)) \
> > 
> > Have you tried this?
> > 
> >         OBJTREE=$(CURDIR) \
> > 
> > or
> > 
> >         OBJTREE=$(abs_output) \
> > 
> > I'd favor using one of these instead.
> > 
> 
> I don't have a strong opinion on any of them. I followed the existing
> approach from the other use cases used in the same file.

well, you're right, it's not really consistent right now.

Acked-by: Nicolas Schier <nsc@kernel.org>