`build_assert` relies on the compiler to optimize out its error path.
Functions using it with its arguments must thus always be inlined,
otherwise the error path of `build_assert` might not be optimized out,
triggering a build error.
Cc: stable@vger.kernel.org
Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
---
rust/kernel/cpufreq.rs | 2 ++
1 file changed, 2 insertions(+)
diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
index f968fbd22890..0879a79485f8 100644
--- a/rust/kernel/cpufreq.rs
+++ b/rust/kernel/cpufreq.rs
@@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
..pin_init::zeroed()
};
+ // Always inline to optimize out error path of `build_assert`.
+ #[inline(always)]
const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
let src = name.to_bytes_with_nul();
let mut dst = [0; CPUFREQ_NAME_LEN];
--
2.52.0On Mon, 08 Dec 2025 11:47:01 +0900
Alexandre Courbot <acourbot@nvidia.com> wrote:
> `build_assert` relies on the compiler to optimize out its error path.
> Functions using it with its arguments must thus always be inlined,
> otherwise the error path of `build_assert` might not be optimized out,
> triggering a build error.
>
> Cc: stable@vger.kernel.org
> Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
> Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
> Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
> Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
> ---
> rust/kernel/cpufreq.rs | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
> index f968fbd22890..0879a79485f8 100644
> --- a/rust/kernel/cpufreq.rs
> +++ b/rust/kernel/cpufreq.rs
> @@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
> ..pin_init::zeroed()
> };
>
> + // Always inline to optimize out error path of `build_assert`.
> + #[inline(always)]
> const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
> let src = name.to_bytes_with_nul();
> let mut dst = [0; CPUFREQ_NAME_LEN];
>
This change is not needed as this is a private function only used in
const-eval only.
I wonder if I should add another macro to assert that the function is
only used in const eval instead? Do you think it might be useful to have
something like:
#[const_only]
const fn foo() {}
or
const fn foo() {
const_only!();
}
? If so, I can send a patch that adds this feature.
Implementation-wise, this will behave similar to build_error, where a
function is going to be added that is never-linked but has a body for
const eval.
Best,
Gary
On Mon, 08 Dec 2025 11:47:01 +0900
Alexandre Courbot <acourbot@nvidia.com> wrote:
> `build_assert` relies on the compiler to optimize out its error path.
> Functions using it with its arguments must thus always be inlined,
> otherwise the error path of `build_assert` might not be optimized out,
> triggering a build error.
>
> Cc: stable@vger.kernel.org
> Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
> Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
> Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
> Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
> ---
> rust/kernel/cpufreq.rs | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
> index f968fbd22890..0879a79485f8 100644
> --- a/rust/kernel/cpufreq.rs
> +++ b/rust/kernel/cpufreq.rs
> @@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
> ..pin_init::zeroed()
> };
>
> + // Always inline to optimize out error path of `build_assert`.
> + #[inline(always)]
> const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
> let src = name.to_bytes_with_nul();
> let mut dst = [0; CPUFREQ_NAME_LEN];
>
> This change is not needed as this is a private function only used in
> const-eval only.
>
> I wonder if I should add another macro to assert that the function is
> only used in const eval instead? Do you think it might be useful to have
> something like:
>
> #[const_only]
> const fn foo() {}
>
> or
>
> const fn foo() {
> const_only!();
> }
>
> ? If so, I can send a patch that adds this feature.
>
> Implementation-wise, this will behave similar to build_error, where a
> function is going to be added that is never-linked but has a body for
> const eval.
I already applied this from V2, should I drop this change ?
--
viresh
On Mon, 15 Dec 2025 10:36:55 +0530
Viresh Kumar <viresh.kumar@linaro.org> wrote:
> On Mon, 08 Dec 2025 11:47:01 +0900
> Alexandre Courbot <acourbot@nvidia.com> wrote:
>
> > `build_assert` relies on the compiler to optimize out its error path.
> > Functions using it with its arguments must thus always be inlined,
> > otherwise the error path of `build_assert` might not be optimized out,
> > triggering a build error.
> >
> > Cc: stable@vger.kernel.org
> > Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
> > Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
> > Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
> > Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
> > ---
> > rust/kernel/cpufreq.rs | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
> > index f968fbd22890..0879a79485f8 100644
> > --- a/rust/kernel/cpufreq.rs
> > +++ b/rust/kernel/cpufreq.rs
> > @@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
> > ..pin_init::zeroed()
> > };
> >
> > + // Always inline to optimize out error path of `build_assert`.
> > + #[inline(always)]
> > const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
> > let src = name.to_bytes_with_nul();
> > let mut dst = [0; CPUFREQ_NAME_LEN];
> >
>
> > This change is not needed as this is a private function only used in
> > const-eval only.
> >
> > I wonder if I should add another macro to assert that the function is
> > only used in const eval instead? Do you think it might be useful to have
> > something like:
> >
> > #[const_only]
> > const fn foo() {}
> >
> > or
> >
> > const fn foo() {
> > const_only!();
> > }
> >
> > ? If so, I can send a patch that adds this feature.
> >
> > Implementation-wise, this will behave similar to build_error, where a
> > function is going to be added that is never-linked but has a body for
> > const eval.
>
> I already applied this from V2, should I drop this change ?
>
Thinking again about this I think `#[inline(always)]` is fine to keep as
it can also be used to indicate "this function shall never be codegenned".
However I do still think the comment is confusing per-se as there is no
"optimization" for this function at all.
RE: the patch I am fine either without this patch picked or having this
patch in and fix the comment later.
Best,
Gary
On 15-12-25, 11:14, Gary Guo wrote: > Thinking again about this I think `#[inline(always)]` is fine to keep as > it can also be used to indicate "this function shall never be codegenned". > > However I do still think the comment is confusing per-se as there is no > "optimization" for this function at all. > > RE: the patch I am fine either without this patch picked or having this > patch in and fix the comment later. Thanks Gary. I will keep the patch then and apply add-ons later. -- viresh
On Mon, Dec 8, 2025 at 2:55 PM Gary Guo <gary@garyguo.net> wrote: > > ? If so, I can send a patch that adds this feature. Sounds like `consteval` in C++20, which is useful from time to time. If we add it, then the attribute form may make a bit more "sense" conceptually (and we already also added the `export` one). Cheers, Miguel
On Mon Dec 8, 2025 at 10:55 PM JST, Gary Guo wrote:
> On Mon, 08 Dec 2025 11:47:01 +0900
> Alexandre Courbot <acourbot@nvidia.com> wrote:
>
>> `build_assert` relies on the compiler to optimize out its error path.
>> Functions using it with its arguments must thus always be inlined,
>> otherwise the error path of `build_assert` might not be optimized out,
>> triggering a build error.
>>
>> Cc: stable@vger.kernel.org
>> Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
>> Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
>> Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
>> Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
>> ---
>> rust/kernel/cpufreq.rs | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
>> index f968fbd22890..0879a79485f8 100644
>> --- a/rust/kernel/cpufreq.rs
>> +++ b/rust/kernel/cpufreq.rs
>> @@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
>> ..pin_init::zeroed()
>> };
>>
>> + // Always inline to optimize out error path of `build_assert`.
>> + #[inline(always)]
>> const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
>> let src = name.to_bytes_with_nul();
>> let mut dst = [0; CPUFREQ_NAME_LEN];
>>
>
> This change is not needed as this is a private function only used in
> const-eval only.
... for now. :)
>
> I wonder if I should add another macro to assert that the function is
> only used in const eval instead? Do you think it might be useful to have
> something like:
>
> #[const_only]
> const fn foo() {}
>
> or
>
> const fn foo() {
> const_only!();
> }
>
> ? If so, I can send a patch that adds this feature.
>
> Implementation-wise, this will behave similar to build_error, where a
> function is going to be added that is never-linked but has a body for
> const eval.
It could be useful in the general sense, but for this particular case
the rule "if you do build_assert on a function argument, then always
inline it" also covers us in case `copy_name` gets used outside of const
context, so isn't it the preferable workaround?
On Tue, 09 Dec 2025 09:52:13 +0900
"Alexandre Courbot" <acourbot@nvidia.com> wrote:
> On Mon Dec 8, 2025 at 10:55 PM JST, Gary Guo wrote:
> > On Mon, 08 Dec 2025 11:47:01 +0900
> > Alexandre Courbot <acourbot@nvidia.com> wrote:
> >
> >> `build_assert` relies on the compiler to optimize out its error path.
> >> Functions using it with its arguments must thus always be inlined,
> >> otherwise the error path of `build_assert` might not be optimized out,
> >> triggering a build error.
> >>
> >> Cc: stable@vger.kernel.org
> >> Fixes: c6af9a1191d0 ("rust: cpufreq: Extend abstractions for driver registration")
> >> Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
> >> Reviewed-by: Daniel Almeida <daniel.almeida@collabora.com>
> >> Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
> >> ---
> >> rust/kernel/cpufreq.rs | 2 ++
> >> 1 file changed, 2 insertions(+)
> >>
> >> diff --git a/rust/kernel/cpufreq.rs b/rust/kernel/cpufreq.rs
> >> index f968fbd22890..0879a79485f8 100644
> >> --- a/rust/kernel/cpufreq.rs
> >> +++ b/rust/kernel/cpufreq.rs
> >> @@ -1015,6 +1015,8 @@ impl<T: Driver> Registration<T> {
> >> ..pin_init::zeroed()
> >> };
> >>
> >> + // Always inline to optimize out error path of `build_assert`.
> >> + #[inline(always)]
> >> const fn copy_name(name: &'static CStr) -> [c_char; CPUFREQ_NAME_LEN] {
> >> let src = name.to_bytes_with_nul();
> >> let mut dst = [0; CPUFREQ_NAME_LEN];
> >>
> >
> > This change is not needed as this is a private function only used in
> > const-eval only.
>
> ... for now. :)
>
> >
> > I wonder if I should add another macro to assert that the function is
> > only used in const eval instead? Do you think it might be useful to have
> > something like:
> >
> > #[const_only]
> > const fn foo() {}
> >
> > or
> >
> > const fn foo() {
> > const_only!();
> > }
> >
> > ? If so, I can send a patch that adds this feature.
> >
> > Implementation-wise, this will behave similar to build_error, where a
> > function is going to be added that is never-linked but has a body for
> > const eval.
>
> It could be useful in the general sense, but for this particular case
> the rule "if you do build_assert on a function argument, then always
> inline it" also covers us in case `copy_name` gets used outside of const
> context, so isn't it the preferable workaround?
In this particular case the `copy_name` shouldn't be used at all
outside const eval. It's specificially for building a table during
const eval. It's a bug if it's outside, hence I think
`#[inline(always)]` adds confusion to the reader of this code.
I get that you want to have a general rule of "if you're using
something with `build_assert!`, then use `#[inline(always)]`", but I
think applying that rule here is detrimental.
Hence I suggested adding a marker to indicate const-eval only function,
so we can either say const-eval-only functions are fine without inline
markers, or perhaps just use normal panicking-assertion inside these
functions (as `build_assert!` behave identical to just `assert!` in
const-eval).
Best,
Gary
© 2016 - 2025 Red Hat, Inc.