fix: stm32h7rs usb_refck panic message now compiles

[friedman-ionq]

With the following in Cargo.toml:

embassy_stm32 = {"0.2", features = [   "memory-x",
    "stm32h7s3l8",
    "defmt",
    "time-driver-any",
    "exti",
    "unstable-pac",
] }

the crate would not compile. This fixes the issue.

[Dirbaio]

Could you add this feature combination to ci.sh so it'd have been caught? thanks!

[friedman-ionq]

Could you add this feature combination to ci.sh so it'd have been caught? thanks!

Sure. Turns out the minimum repeatable feature is just "stm32h7s3l8". Which is surprising, because there already is a test case. Digging deeper.

[Dirbaio]

maybe it's due to defmt. When it's enabled the panic! macro is overridden, so maybe it works with and fails without.

[friedman-ionq]

maybe it's due to defmt. When it's enabled the panic! macro is overridden, so maybe it works with and fails without.

$ ./ci.sh
thread 'main' panicked at src/bin/cargo-batch.rs:136:85:
called `Result::unwrap()` on an `Err` value: the package 'embassy-stm32-tests' does not contain this feature: defmt
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

[Dirbaio]

the embassy-stm32-tests crate doesn't have it, embassy-stm32 does.

[friedman-ionq]

the embassy-stm32-tests crate doesn't have it, embassy-stm32 does.

``~/work/embassy (main)$ git diff
diff --git a/ci.sh b/ci.sh
index e32105b16..46354f096 100755
--- a/ci.sh
+++ b/ci.sh
@@ -169,6 +169,7 @@ cargo batch
--- build --release --manifest-path embassy-stm32/Cargo.toml --target thumbv6m-none-eabi --features stm32u031r8,defmt,exti,time-driver-any,time
--- build --release --manifest-path embassy-stm32/Cargo.toml --target thumbv6m-none-eabi --features stm32u073mb,defmt,exti,time-driver-any,time
--- build --release --manifest-path embassy-stm32/Cargo.toml --target thumbv6m-none-eabi --features stm32u083rc,defmt,exti,time-driver-any,time \

• --- build --release --manifest-path embassy-stm32/Cargo.toml --target thumbv7em-none-eabihf --features stm32h7s3l8,defmt,time-driver-any,exti,unstable-pac
  --- build --release --manifest-path embassy-nxp/Cargo.toml --target thumbv8m.main-none-eabihf
  --- build --release --manifest-path cyw43/Cargo.toml --target thumbv6m-none-eabi --features ''
  --- build --release --manifest-path cyw43/Cargo.toml --target thumbv6m-none-eabi --features 'log' \

does not repro the bug when I execute `./ci.sh` on main.

[Dirbaio]

I cannot reproduce the build failure this is supposed to fix. Could you provide the exact build command to reproduce? What cargo build --features xxxxx do I have to run from the embassy-stm32 dir in a fresh repo clone?

Hertz values should always be printable with {}, if that's not working then we should fix that, not add .0 as a workaround.

[friedman-ionq]

I cannot reproduce the build failure this is supposed to fix. Could you provide the exact build command to reproduce? What cargo build --features xxxxx do I have to run from the embassy-stm32 dir in a fresh repo clone?

Hertz values should always be printable with {}, if that's not working then we should fix that, not add .0 as a workaround.

I agree with that latter sentiment -- and then point out https://github.com/embassy-rs/embassy/blob/main/embassy-stm32/src/rcc/h.rs#L795 as an example where this workaround had previously been added. In any event, the command that I added to build.rs failed when I executed it in the embassy-stm32 directory.

[Dirbaio]

can you try again to see if it fails now on latest main?

[friedman-ionq]

can you try again to see if it fails now on latest main?

Now, it is not only passing on main, but also on 0.2. Perhaps it is breaking because of a dependency?

[Dirbaio]

no idea! Since it works i'm going to close this, feel free to reopen if you do find the compile issue.