<JamesMunns[m]>
Let me know if you want me to type up an advisory notice, like we did for the -rt bug. the cortex-m repo maybe isn't the perfect spot for that announcement, but it seems fine to add it there as well.
<ragarnoy[m]>
does anyone have experience creating bindings for c libraries that target embedded stuff ? I'm having issues using bindgen but I'm encountering the following error
<ragarnoy[m]>
/usr/bin/ld: cannot find -lacconeer_a121: No such file or directory
<ragarnoy[m]>
/usr/bin/ld: skipping incompatible lib/libacconeer_a121.a when searching for -lacconeer_a121
spinfast[m] has joined #rust-embedded
<spinfast[m]>
whats the arch of the static lib? was it linked with the same linker you are trying to build with?
<spinfast[m]>
like I'd check those things first, i've had problems in the past and ended up telling cargo to use gnu's ld rather than llvm's
<spinfast[m]>
* than llvm's lld
<ragarnoy[m]>
no, it's using gcc-arm-none-eabi but I can't recompile it
<ragarnoy[m]>
I only have the static library and the headers
<spinfast[m]>
and you told cargo to build for armv7m-none-eabihf as well presumably? if its not an arch mismatch, its lld not understanding the .a
<PeterHansen[m]>
dngrsspookyvisio: Not likely related, as it shows it occurring even with opt-level=3, which is (apparently) one way to avoid trigger the outliner bug.
<dngrsspookyvisio>
seems llvm 17 needs to spend a bit more time in the oven
<PeterHansen[m]>
Also, so far anyway, it seems folks think this is on cortex-m only, though I don't actually know of any evidence about that one way or the other.
<PeterHansen[m]>
we're the oven ;-)
<dngrsspookyvisio>
as far as I'm concerned, the beta channel is
<JamesMunns[m]>
dngrsspookyvisio: note that this already affects stable releases 1.73 and 1.74 :p
<dngrsspookyvisio>
yeah, that's what I meant
therealprof[m] has joined #rust-embedded
<therealprof[m]>
<JamesMunns[m]> "Started writing an advisory here..." <- Where would I run with your PSA? 😅
<dngrsspookyvisio>
(in the sense that currently stable seems a little less stable than it should be, there was also an ICE in 1.74.0)
<JamesMunns[m]>
I could probably do the posting, but I figured I'd check in with the team for review/feedback before going all out on it therealprof
<therealprof[m]>
@jamesmunns:beeper.com Not sure about the visibility of those discussions. I'm perfectly fine with you going all out on it.
<JamesMunns[m]>
I guess anyone lemme know today if they have feedback, and I'll plan on posting the advisory publicly tomororw?
<JamesMunns[m]>
s/tomororw/tomororow/
<therealprof[m]>
I've read it and think it's a great summary. Just not sure whether posting that on c-m "discussions" would do your efforts justice. I think this might actually warrant a post on https://blog.rust-lang.org/inside-rust/
thejpster[m] has joined #rust-embedded
<thejpster[m]>
James Munns: you could also try rust@arm.com
<JamesMunns[m]>
thejpster[m]: Not sure what for? Issue seems to be with LLVM and has been tentatively fixed upstream
<JamesMunns[m]>
unless you think they would be able to share the advisory through their channels
<thejpster[m]>
because I think Arm should know if the target they don't currently bother to maintain generates Bad Code.
<thejpster[m]>
and I guess it applies to clang using LLVM 17 too? Or is it a Rust specific thing?
<JamesMunns[m]>
Appears to apply to all LLVM compilers that use that opt pass, so I believe it applies to clang17 as well, same opt levels
<JamesMunns[m]>
Not sure if it's exactly llvm17, but introducing llvm17 from whatever tip we were on before to now is the bisection range
<JamesMunns[m]>
so could have been late 16 too, I guess? Not sure what our llvm version rev was before.
<thejpster[m]>
I sure am glad we don't support Ferrocene on that target yet
<ragarnoy[m]>
<spinfast[m]> "and you told cargo to build..." <- just to be sure, you meant thumbv7m-none-eabihf here right ?
<thejpster[m]>
(also I wonder if the compiler test suite would have found it ... I don't know what opt-level they run the tests at)
<thejpster[m]>
James Munns: I don't know how to build regular rustc by changing LLVM, but on Ferrocene there's a config.toml option to build LLVM from source, and LLVM is just a git submodule so you can go into its folder and do a git checkout of a different branch / commit.
<spinfast[m]>
ragarnoy[m]: For rust yes
<thejpster[m]>
on the assumption that regular rustc always builds LLVM from source, I would change the checkout for src/llvm-project and then ./x.py build as usual.
notgull has joined #rust-embedded
DirkVanHaerenbor has joined #rust-embedded
<DirkVanHaerenbor>
Hi! stupid question that I'd rather not ask. I've previously joined an embedded project that use c, whereas all my previous embedded stuff has been in Rust. It's a weird, I know.
<DirkVanHaerenbor>
I was impressed with just how good probe-rs works. Is there an equivalent for plain old c projects? I am pretty new tot that area
<spinfast[m]>
not really a true equivalent no, you can flash and get gdb with pyocd but nothing like probe-rs run
jr-oss has quit [Ping timeout: 264 seconds]
<JamesMunns[m]>
fwiw: you CAN use probe-rs(-cli) with C projects, to handle flashing and running, and I think it SHOULD support "plain" RTT comms as well for those projects.
<JamesMunns[m]>
you wont have defmt, that's pretty Rust-specific, but you can get a good chunk of the value back :)
Noah[m]1 has joined #rust-embedded
<Noah[m]1>
yup, what James says :)
<Noah[m]1>
probe-rs is nothing rust specific :)
<Noah[m]1>
We offer a GDB server too & vscode if you need it then :)
<DirkVanHaerenbor>
oh that's very cool. I could even call it from cmake then
<Noah[m]1>
DirkVanHaerenbor: yep that's the correct choice :)
<Noah[m]1>
sorry, I borked some links when I revamped the webpage
<Noah[m]1>
should make aliases
notgull has quit [Ping timeout: 268 seconds]
jr-oss has joined #rust-embedded
<thejpster[m]>
dirbaio: is embedded-io going to hit 1.0 at the same time as embedded-hal 1.0? Because embedded-hal removes the serial traits in favour of embedded-io.
<thejpster[m]>
if so, will there be a 1.0 rc release?
<DirkVanHaerenbor>
Noah: thanks!
<Noah[m]1>
Dirk Van Haerenborgh: np, feel free to also drop by in #probe-rs:matrix.org :)
<dirbaio[m]>
<thejpster[m]> "dirbaio: is embedded-io going to..." <- not the 28th
<dirbaio[m]>
but yeah we should do it soon
<dirbaio[m]>
for now the recommendation is to implement the 0.6 traits
Guest7282 has left #rust-embedded [Error from remote client]
Ekho has quit [Quit: CORE ERROR, SYSTEM HALTED.]
Ekho has joined #rust-embedded
Rahix has quit [Read error: Connection reset by peer]