08:34 <RobinMueller[m]> * Hello. I have started a project of writing a bare metal Rust App for a Xilinx Zynq7000 based board. I saw https://git.m-labs.hk/M-Labs/zynq-rs and I am now wondering whether there are any Cortex-A(9) support crates out there? I only saw support for Aarch64.

10:37 <RobinMueller[m]> If there isn´t one yet, I assume this could be useful for other people as well, packaging this as open-source crates similar to cortex-m / cortex-m-rt? zynq-rs already provides a lot of useful baseline code, and there is also https://github.com/Xilinx/embeddedsw/tree/master/lib/bsp/standalone/src/arm/cortexa9 which provides C code as a starting point. I am probably undererstimating cost & effort 😁

10:38 <RobinMueller[m]> * & effort though 😁

10:59 <MrPashaPG[m]> <JamesMunns[m]> "you need to use async delays..." <- I had that embassy_time problem because it wasn't working, otherwise I created a project with embassy on stm32, I know I have to use await.

10:59 <MrPashaPG[m]> Because embassy_time wasn't working even in the main loop, I had it cut and used delay inside esp_hal.

10:59 <MrPashaPG[m]> Now my main problem is that embassy doesn't apply to my esp32 project at all, embassy_time didn't work, I used delay, now I see that task doesn't work either.

11:00 <thejpster[m]> <RobinMueller[m]> "If there isn´t one yet, I assume..." <- I wrote a Cortex-r-rt crate and that probably works for AArch 32 Cortex-A as well. Maybe it needs a different name.

11:01 <thejpster[m]> Jorge also has https://github.com/japaric/ultrascale-plus which might be interesting

11:03 <thejpster[m]> Is https://github.com/rust-embedded/wg/pull/818 approved?

11:35 <JamesMunns[m]> https://github.com/rust-embedded/wg/blob/master/rfcs/0206-voting-majority.md#detailed-design

11:35 <JamesMunns[m]> Counting +1 emojis and not checkboxes isn't something we've done before.

11:37 <JamesMunns[m]> It's been a week, so we need 33%, or three approvals. Might want to ping berkus and nchong to confirm.

11:59 <adamgreig[m]> yea, if they could tick the boxes or approve the PR rather than just the emoji I think we'd be set

12:37 <RobinMueller[m]> <thejpster[m]> "I wrote a Cortex-r-rt crate..." <- Thanks! Where can I find the crate? I also saw the project by japaric, that is an ultrsscale though which has two different CPUs. It might still help though, I would guess that there are similarities between the cortex-a variants

12:38 <thejpster[m]> github.com/rust-embedded/cortex-r. There’s a PR for it - the main branch is a very old placeholder.

12:54 <RobinMueller[m]> <thejpster[m]> "github.com/rust-embedded/cortex..." <- Thanks. I'll have a look, I already saw that there are common registers like MPIDR, and the rt subcrate also contains code/logic which I could probably adapt for cortex-a9.. (e.g. save stack context, call IRQ, restore context)

14:03 <bitts[m]> Seeing a bit of a weird problem here that I hope someone can help with, I include an assembly file with global_asm!(include_str!("boot_armv8_asm.S")); . This works just fine when compiling on mac, but on linux it chokes because the S file contains c-style comments ( /* Like this */ ). It throws a bit different depending on where the comment is, but e.g. "Unexpected token at start of statement"

14:08 <bitts[m]> Does anyone know who is responsible for parsing the asm f.ex.?

14:11 <berkus[m]> bitts[m]: global asm is parsed by rustc to convert it to llvm-acceptable format and then passed on to llvm

14:12 <berkus[m]> (it mostly does argument-related preproc)

14:12 <diondokter[m]> berkus[m]: Most importantly, it's parsed in an agnostic way so any compiler backend can emit the right instructions in the way they need

14:13 <JamesMunns[m]> I wonder if macos uses LLVM binutils like `as`, while on Linux it uses gnu binutils instead by default?

14:14 <berkus[m]> <bitts[m]> "Seeing a bit of a weird problem..." <- that's weird, but it seems like .S files are perhaps treated differently from .s files

14:14 <berkus[m]> try renaming it to lowercase

14:14 <JamesMunns[m]> Oh no lol

14:14 <JamesMunns[m]> Case insensitive file systems strike again?

14:14 <berkus[m]> the .S file will go through preproc i think

14:15 <berkus[m]> JamesMunns[m]: well, it's case agnostic on mac, but definitely sensitive on linux :)

14:16 <berkus[m]> however, i have .S files and they seem to be cross-building just fine on all 3 systems...

14:17 <JamesMunns[m]> berkus[m]: Do they have block comments?

14:17 <berkus[m]> JamesMunns[m]: Yup, mix of both block and line comments.

14:17 <thejpster[m]> global asm needs {{ and }} in things like stmfd sp! doesn’t it? Because {foo} is an argument to be replaced?

14:17 <berkus[m]> https://github.com/metta-systems/vesper/blob/develop/machine/src/arch/aarch64/exception/vectors.S

14:18 <thejpster[m]> So I wouldn’t pass .S files to global asm. I’d use cc-rs.

14:18 <bitts[m]> Ah, we just realized that it is only a problem when using a macro... I guess that is slightly more corner case and less weird

14:19 <dirbaio[m]> try global_asm!(include_str!("boot_armv8_asm.S"), options(raw))

14:19 <bitts[m]> Renaming from uppercase to lower case S did not help..

14:20 <berkus[m]> bitts[m]: Yeah, red herring

14:21 <bitts[m]> options(raw) did also not help, but removing the inline comments on macro invocations is acceptable workaround

14:22 <bitts[m]> It also choked on lines starting with /* which confused me a bit, but that was just a side effect of the first error

14:23 <berkus[m]> bitts[m]: can you show the syntax that choked? you can prolly rename the source parts if it's private

14:26 <bitts[m]> Adding a ; or changing to // save all CPU regs both work

14:27 <berkus[m]> hmm what's the last line of the macro?

14:27 <bitts[m]> .end

14:27 <bitts[m]> * .endm

14:27 <berkus[m]> okay, one non-empty line before that

14:28 <bitts[m]> stp x30, xzr, [\stackPtr, #-16]!

14:28 <berkus[m]> innnttttrsting

14:35 <RobinMueller[m]> thejpster: Thanks for the hints to the cortex-r crate. I just created a PR with some minimal changes :) . I think renaming might make a lot of sense for the cortex-r crate at least.. The CP15 registers seem to be identical. I will try to get familiar with QEMU.. I think it has both a cortex-A8 versatile base board and a xilinx-zynq-a9 config (https://www.qemu.org/docs/master/system/arm/xlnx-zynq.html)

14:37 <RobinMueller[m]> * created a draft PR with

15:01 <MrPashaPG[m]> Why does the rust analyzer give a syntax error even though the program is working properly and has no problems? How can I fix this problem?

15:06 <vollbrecht[m]> MrPashaPG[m]: please dont crosspost questions. Most of the people are subscribed over the same channels.

15:29 <burrbull[m]> <MrPashaPG[m]> "Why does the rust analyzer..." <- this is not a syntax error.

15:29 <burrbull[m]> is this a new project or example on cloned hal repository?

15:29 <burrbull[m]> possibly you should write vscode (your editor) config with rust-analyzer settings where list features that should be enabled

15:30 <MrPashaPG[m]> <vollbrecht[m]> "please dont crosspost questions...." <- Oh yes, I'm sorry.

15:37 <MrPashaPG[m]> <burrbull[m]> "this is not a syntax error...." <- What settings can I make to fix this problem? If you have encountered this problem and fixed it, I would appreciate it if you could guide me on what to do because this [link](https://github.com/esp-rs/esp-hal/issues/3148) states that this problem is for rust-analyzer.