Fixed VectActive::from() returning incorrect value

[skibon02]

Fixes #499

Fixed VectActive::from(u8) now returns correct value for Interrupt variant
Refactored SCB::vect_active() to use VectActive::from(u8) and get rid of code duplication

[newAM]

Can you rebase this? The CI failure is from an old MSRV

[skibon02]

@newAM, Could you please explain a little more? I forked the repo and made this commit from the v0.7.x branch.

[newAM]

Oh, sorry, I didn't see that this was targeting v0.7.x! That makes more sense then. The problem is that serde_json isn't pinned, and the latest version requires edition 2021. I think specifying >=1.00,<=1.0.100 for serde_json should fix the build failure.

[skibon02]

Oh, unwrap_unchecked seems to not be available on rustc version 1.38.0. It forces me to write a reaction to the case, when icsr register returns unexpected value. Is panic appropriate in this case? Or maybe we should execute udf instruction?

Assuming correct implementation for VectActive::from for all cortex-m processors, this problem can only appear when incorrect target was selected by user, and it is impossible to check this in compile time.

[skibon02]

Adding these ugly lines in Cargo.toml to fix deps versions did help compiling on 1.38

Or should we better just bump MSRV to 1.60?

[jonathanpallant]

cortex-m-rt says:

This crate is guaranteed to compile on stable Rust 1.59.0 and up

[skibon02]

But this is cortex-m, not cortex-m-rt
latest update 0.7.7 was published 1 year ago

From v0.7.x branch Readme:

cortex-m/README.md

Line 14 in bb4a782

This crate is guaranteed to compile on stable Rust 1.38 and up. It might compile with older versions but that may change in any new patch release.

[jonathanpallant]

But how many people use cortex-m without using cortex-m-rt?

[skibon02]

cortex-m has twice as many downloads compared to cortex-m-rt
It seems that there isn't much sense in supporting these old versions of the Rust compiler. Moreover, a significant portion of people even uses Nightly.

Should I bump the MSRV to 1.60 and remove fixed dependency versions?

[jonathanpallant]

1.59 makes sense, so they are the same.

unwrap_unchecked first appeared in 1.58 so that's fine.