unable to codegen: InputOutput on Linux with Self-Hosted x86 Backend #22642
Labels
Milestone
Comments
jcalabro
added
the
bug
|
Also oddly relevant question, are you sure that there can't possibly be another process running the same build command concurrently? This could include typing Ctrl+C and immediately rerunning the same command, I forget if there's still an issue with the children continuing to run or not. |
|
I can check on the new compiler version tonight and will get back to you It's totally possible that there's another build command running that's botching things. I generally have seen For instance, I'm not working on any zig stuff right now at all, yet these processes are just running in the background: |
|
Could an editor be trying to make use of the compiler watch functionality (maybe check pstree)? |
|
True! I still have hx and vs code open from this morning when I was writing some zig |
|
I don't know enough about zls to confirm it on that side, but I ran a test of running console output |
jacobly0
added
zig build system
|
Seems like the culprit in some way. Good catch, thanks! Is there anything else I can do to help confirm? I tried coming up with a minimal repro, but it appears to be getting some other, but perhaps related issues (i.e. not // main.zig
const std = @import("std");
pub fn main() !void {
std.debug.print("Hello, world!\n", .{});
}#!/usr/bin/env bash
# run.sh
for i in {1..100}; do
zig build-exe main.zig &
done;
waitOutput |
|
Those errors look unrelated, you just missed |
|
True, I forgot, sorry! Here are the actual errors, I also added And lots more like that |
|
Both the dwarf and lld errors are explained by the cache mode being set to incremental. This should be changed to whole cache mode before release for the case of |
jacobly0
added
frontend
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Zig Version
0.14.0-dev.2577+271452d22
Steps to Reproduce and Observed Behavior
Hello! I've been getting this issue on occasion (i.e. see here) and I'm not certain how to reproduce it. However, it's been happening roughly once or twice a day. This is happening across multiple development machines.
Using Linux,
use_llvm = false, anduse_lld = false, I get an error where the compiler appears to be unable to generate some amount of code. An example of a full run is below the fold, however this previously was also happening at different lines of source code. My code is valid and compiles correctly if I re-run the compiler.I'm not sure how to reliably reproduce this, and the bug does not appear to happen when I swap back to using llvm and lld. I'm happy to work with the team to get towards a more reliable repro case, but for the time being, I'd encourage you to try compiling uscope a handful of times. Not a great repro at all, I know, but that's the only thing I have to go on at the moment. Happy to run any commands or do any investigation you need, but I'm not sure what direction to go in to explore that space on my own.
Here's my linux version:
Reveal full log
Expected Behavior
I expect my code to compile and run correctly. My code actually does compile correctly if I re-run the
zig build runcommand.The text was updated successfully, but these errors were encountered: