Naked function must contain a single asm block #73

New Issue

pca006132 · 2021-01-14T16:40:17+08:00

pca006132 commented

2021-01-14 16:40:17 +08:00

https://github.com/rust-lang/rust/issues/32408

Warning on latest nightly, would become hard error later.
But I'm not sure how should we run other functions from interrupt with asm block only, maybe we should mark the functions as extern "C" and call them from asm?

https://github.com/rust-lang/rust/issues/32408 Warning on latest nightly, would become hard error later. But I'm not sure how should we run other functions from interrupt with asm block only, maybe we should mark the functions as `extern "C"` and call them from asm?

astro commented

2021-01-14 23:24:45 +08:00

maybe we should mark the functions as extern "C" and call them from asm?

That seems to be the way forward.

Perhaps we can create a shared abstraction (macro_rules) that works with our memory model for all our naked fns.

> maybe we should mark the functions as extern "C" and call them from asm? That seems to be the way forward. Perhaps we can create a shared abstraction (macro_rules) that works with our memory model for all our naked fns.

pca006132 commented

2021-01-15 14:10:43 +08:00

It seems that the new compiler also break our code regarding core1.

The experiments break even if changes to the memcpy is removed, and only works when I comment out boot::Core1::start(false); and related core1 code.

Not sure what is happening here.

My fork with updated compiler: https://git.m-labs.hk/pca006132/zynq-rs

It seems that the new compiler also break our code regarding core1. The experiments break even if changes to the `memcpy` is removed, and only works when I comment out `boot::Core1::start(false);` and related core1 code. Not sure what is happening here. My fork with updated compiler: https://git.m-labs.hk/pca006132/zynq-rs

pca006132 referenced a pull request that will close this issue

2021-01-15 16:52:32 +08:00

Updated rustc for fast memcpy #74

sb10q closed this issue

2021-01-15 17:40:02 +08:00

pca006132 reopened this issue

2021-01-18 09:52:47 +08:00

pca006132 commented

2021-01-18 09:54:48 +08:00

Using extern "C" function instead of naked function would break kernel code. Getting weird interrupts after several restarts.

Reopening to see if there is any better solution.

~~I'm not sure using a naked function to call an extern "C" function by bl ... could work.~~ (It does not work)

Using `extern "C"` function instead of naked function would break kernel code. Getting weird interrupts after several restarts. Reopening to see if there is any better solution. ~~I'm not sure using a naked function to call an `extern "C"` function by `bl ...` could work.~~ (It does not work)

pca006132 referenced this issue from M-Labs/artiq-zynq

2021-01-18 16:48:21 +08:00

Naked IRQ #121

pca006132 commented

2021-01-18 17:00:20 +08:00

Maybe we should initialize SP for other interrupt handlers as well. idk why it works now, considering the SP should point to 0 and idk how the registers are pushed to the stack...

Maybe we should initialize `SP` for other interrupt handlers as well. idk why it works now, considering the `SP` should point to 0 and idk how the registers are pushed to the stack...

pca006132 referenced this issue

2021-01-28 11:06:14 +08:00

libcortex_a9: added interrupt_handler macro #77

sb10q referenced this issue from a commit

2021-01-28 12:34:14 +08:00

libcortex_a9: added interrupt_handler macro

sb10q closed this issue

2021-01-28 12:34:14 +08:00

Sign in to join this conversation.