07:35 <yruama_lairba[m]> GrantM11235: using atomic generate a lot of DSB and DMB memory barrier instruction, i read in embedednomicon it tells that is not required under some circumstance

07:39 <yruama_lairba[m]> * i read the embedednomicon again, it's no so clear https://docs.rust-embedded.org/embedonomicon/dma.html#dont-we-need-a-memory-barrier

07:47 <yruama_lairba[m]> anyway, i think you can greatly reduce the number of memory barrier when you read a chunk of memory

07:56 <yruama_lairba[m]> if i understand correctly, i only need one "Acquire" or DSB before reading the chunk, and one "Release" or DMB after the write

10:04 <yruama_lairba[m]> i don't know what i did before, but Atomic tyep don't generate DSB/DMB instruction anymore with relaxed oredring

16:27 <askrodney[m]> Hi, I'm running into an unexpected SIGINT following the rust-embedded/discovery micro:bit book. OK to describe the issue, and ask for help here?

17:25 <firefrommoonligh> I am guessing that is a different definition of the word from the typical one

17:25 <firefrommoonligh> s/word/acronym/

18:01 <therealprof[m]> <askrodney[m]> "Hi, I'm running into an unexpect..." <- 👋 Sure, ask away.

18:02 <askrodney[m]> therealprof[m]: thanks...

18:02 <askrodney[m]> askrodney[m]: Hi, I'm running through the discovery micro:bit book. I am getting an unexpected SIGINT on [Chapter 5.3](https://docs.rust-embedded.org/discovery/microbit/05-led-roulette/debug-it.html), when I try to debug. I am running an Intel Mac with micro:bit v2, and earlier test (from the book) runs fine... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/GwXJDUpfpCMpGssLNuilqEnW>)

19:56 <JamesMunns[m]> askrodney what happens if you put one of these in the body of the loop?... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/JhfqWOUnASnIqVjRgDUicUnH>)

19:57 <JamesMunns[m]> there used to be an issue that empty loops would get miscompiled away (it's not allowed in C++ I think, but is in Rust, sometimes LLVM gets confused)

19:59 <JamesMunns[m]> The other thing I would suggest is setting `lto = "off"` here: https://github.com/rust-embedded/discovery/blob/master/microbit/Cargo.toml#L11-L14

20:00 <JamesMunns[m]> The other thing that goes wrong some times is that LTO breaks the debug information that GDB uses, and it's possible that GDB is just getting confused and crashing (not the program).

20:00 <dirbaio[m]> they're building with debug, so no lto already

20:00 <JamesMunns[m]> I would actually try the LTO thing first, I think that's more likely than loop {} being an issues with a modern compiler.

20:02 <JamesMunns[m]> ah, good catch

20:02 <JamesMunns[m]> then yeah, maaaaybe the loop thing? But that also usually doesn't get miscompiled without optimizations...