Unclear how to use `#[interrupt]` macro without deprecated `--reexport-interrupt` flag

[qwandor]

According to the documentation of svd2rust, the --reexport-interrupt flag is deprecated. However according to the documentation of cortex-m-rt, the #[interrupt] macro should only be used via the svd2rust reexport, and indeed it fails to compile when used directly, with a confusing error.

How is the interrupt macro supposed to be used without the --reexport-interrupt flag passed to svd2rust? This should at least be documented clearly in the svd2rust documentation.

[burrbull]

There have been planned some changes in cortex-m-rt 0.8 like moving interrupt in independent crate. But it is not done yet.
Ask @adamgreig.

[robamu]

I also stumbled into the issue, wondering why I could not use the interrupt macro, and I had to rebuild with the --reexport-interrupt flag.

[royb3]

You can use it without the flag. But you have to import your PAC's Interrupt enum as interrupt.

use hal::pac::Interrupt as interrupt;
use cortex_m_rt::interrupt;

That's basically what the --reexport-interrupt flag does as well. Not sure if there's any value in doing this over using the deprecated flag for now.

[Bnyro]

To save any future reader time with debugging:

If you are trying to use this with rtic and come across some linker errors regarding the interrupt vector table, e.g. link.x:49: symbol not found: DefaultHandler_:

Make sure to include the import to use cortex_m_rt::interrupt;. I thought rtic would do that by itself, but apparently it doesn't.

E.g., this would look like

Details

use your_pac as _;
use cortex_m_rt::interrupt;
#[rtic::app(

device = your_pac,

dispatchers = [UARTE0],

)]

mod app {}

For reference, that's what --reexport-interrupt does:

#[cfg(feature = "rt")]
pub use self::Interrupt as interrupt;
#[cfg(feature = "rt")]
pub use cortex_m_rt::interrupt;

Sorry if this looks obvious to most of you, but understanding this earlier would have probably saved me hours :)