Conversation
0fe3a7c to
2abb7dd
Compare
ec6c7ef to
667de8f
Compare
ac4ddc1 to
42c98c1
Compare
fd3ccb5 to
edb22b0
Compare
|
@thejpster @jonathanpallant @adamgreig Reviewers, please double check the new crate (in cortex-m/macros). I'm assuming that the cortex-m crate using the new crate will need to be updated to use an actual version once it is published. Can we just publish the new crate once this is landed so that I can make a PR to depend on the published version? |
edb22b0 to
e53eba0
Compare
|
You can depend on a version and a path at the same time. |
Can I depend on a path in a published crate, like "cortex-m"? |
|
As long as you have both. It uses the path if valid, and falls back to the version if not. |
e53eba0 to
63bee36
Compare
|
I have updated the dep spec for the new crate to include the version number. I use an exact version like the |
The `cm7-r0p1` feature only works with the plateform feature `armv7em`. I added a compile error to help someone figure that out more redily if they accidentally combine those features.
This will allow getting all the asm in shape for edition 2024.
63bee36 to
fe62ca1
Compare
|
I removed the |
fe62ca1 to
dd25b47
Compare
dd25b47 to
94fb96a
Compare
Inline asm has been supported in stable rust for some time, so I removed the separate asm build infra and added that code to the normal crate code. I also crated a mock of the asm functions that were needed to run clippy with a non-arm target (like x86_64).
The purpose of this attribute is to allow platforms that do not support the asm instructions to compile code code that calls those functions. This is needed so that clippy can be run once in CI and when testing locally.
This is the first step of removing the asm::inner module entirely.
This removes the inner module entirely.
The more I think about this, the more I think this is more correct. Considering that the argument is owned, by the function, it seems more correct to just recognize that in the signature. It is not a breaking change since the arg is passed by value since it's a primitive type, so I think this is better. In principle, I would prefer to merge this with the commit that moves the asm code into the crate directly instead of this separate commit.
94fb96a to
1ef6c56
Compare
I used `any(armv6m, armv7m, ...)` in many places where I could have said `cortex_m` instead for the cfg macro calls. Let's change that.
e0d3ab8 to
3188915
Compare
This is a draft for updating the asm toolchain. It's based on my #628, so it shouldn't be landed before that is.