roc: `examples/cli/file.roc` segmentation fault

Recent PRs modifying examples have seemed to break some stuff:

[jan@framey roc]$ neofetch
             .',;::::;,'.                jan@framey 
         .';:cccccccccccc:;,.            ---------- 
      .;cccccccccccccccccccccc;.         OS: Fedora release 36 (Thirty Six) x86_64 
    .:cccccccccccccccccccccccccc:.       Host: Laptop AA 
  .;ccccccccccccc;.:dddl:.;ccccccc;.     Kernel: 5.19.8-200.fc36.x86_64 
 .:ccccccccccccc;OWMKOOXMWd;ccccccc:.    Uptime: 12 hours, 3 mins 
.:ccccccccccccc;KMMc;cc;xMMc:ccccccc:.   Packages: 2216 (rpm), 42 (nix-default), 22 (flatpak) 
,cccccccccccccc;MMM.;cc;;WW::cccccccc,   Shell: bash 5.1.16 
:cccccccccccccc;MMM.;cccccccccccccccc:   Resolution: 2256x1504 
:ccccccc;oxOOOo;MMM0OOk.;cccccccccccc:   DE: GNOME 42.4 
cccccc:0MMKxdd:;MMMkddc.;cccccccccccc;   WM: Mutter 
ccccc:XM0';cccc;MMM.;cccccccccccccccc'   WM Theme: Adwaita 
ccccc;MMo;ccccc;MMW.;ccccccccccccccc;    Theme: Adwaita [GTK2/3] 
ccccc;0MNc.ccc.xMMd:ccccccccccccccc;     Icons: Adwaita [GTK2/3] 
cccccc;dNMWXXXWM0::cccccccccccccc:,      Terminal: kitty 
cccccccc;.:odl:.;cccccccccccccc:,.       CPU: 11th Gen Intel i5-1135G7 (8) @ 4.200GHz 
:cccccccccccccccccccccccccccc:'.         GPU: Intel TigerLake-LP GT2 [Iris Xe Graphics] 
.:cccccccccccccccccccccc:;,..            Memory: 5277MiB / 15781MiB 
  '::cccccccccccccc::;,.
                                                                 
                                                                 


[jan@framey roc]$ git switch main
Already on 'main'
Your branch is up to date with 'origin/main'.
[jan@framey roc]$ git pull --prune
Already up to date.
[jan@framey roc]$ roc dev examples/helloWorld.roc 
roc: /lib64/libtinfo.so.6: no version information available (required by roc)
🔨 Rebuilding platform...
Hello, World!
[jan@framey roc]$ roc dev examples/cli/form.roc 
roc: /lib64/libtinfo.so.6: no version information available (required by roc)
🔨 Rebuilding platform...
What's your first name?
J
What's your last name?
V
Hi, J V! 👋
[jan@framey roc]$ 
[jan@framey roc]$ roc dev examples/cli/args.roc
roc: /lib64/libtinfo.so.6: no version information available (required by roc)
🔨 Rebuilding platform...
An internal compiler expectation was broken.
This is definitely a compiler bug.
Please file an issue here: https://github.com/roc-lang/roc/issues/new/choose
thread 'main' panicked at 'Undefined Symbol in relocation, (+6c36, Relocation { kind: Relative, encoding: Generic, size: +20, target: Symbol(SymbolIndex(+305)), addend: +fffffffffffffffc, implicit_addend: false }): Ok(Symbol { name: "", address: +0, size: +0, kind: Section, section: Section(SectionIndex(+9)), scope: Compilation, weak: false, flags: Elf { st_info: +3, st_other: +0 } })', crates/linker/src/elf.rs:1288:33
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
[jan@framey roc]$ 
[jan@framey roc]$ roc dev examples/cli/file.roc 
roc: /lib64/libtinfo.so.6: no version information available (required by roc)
🔨 Rebuilding platform...
cwd: /home/jan/_code/_roc/roc
Dir.list...
Segmentation fault (core dumped)
[jan@framey roc]$

About this issue

Original URL
State: open
Created 2 years ago
Comments: 19 (13 by maintainers)

Most upvoted comments

We have test coverage for “file” and “args” so I’m unsure why commits that far back would be seen as broken now.

There is/was no test coverage for file.roc. I think that’s because it used to be under the interactive folder which was exempt from needing to have tests.

Anton-4 on Oct 10, 2022

We have test coverage for “file” and “args” so I’m unsure why commits that far back would be seen as broken now.

For args.roc, it is the know surgical linker bug #3609. Just made a PR to at least print a better error message when this comes up.

Just a note; in the cli_run tests the args example is tested with the legacy linker, which is why it does not fail on CI.

Anton-4 on Oct 10, 2022

For args.roc, it is the know surgical linker bug #3609. Just made a PR to at least print a better error message when this comes up.

bhansconnect on Oct 9, 2022

Ok, yeah, very not sure why it is broken, but I figured out a minimal diff to fix. Simply collect in a vector and then convert to a RocList. Still really confused why this cause the dir_entries to break rather than the RocList, but I guess we are probably overwriting memory we don’t own and breaking the dir_entries. So probably a RocList in rust bug.

--- a/examples/cli/cli-platform/src/lib.rs
+++ b/examples/cli/cli-platform/src/lib.rs
@@ -407,16 +407,17 @@ pub extern "C" fn roc_fx_dirList(
 ) -> RocResult<RocList<RocList<u8>>, WriteErr> {
     println!("Dir.list...");
     match std::fs::read_dir(path_from_roc_path(roc_path)) {
-        Ok(dir_entries) => RocResult::ok(
-            dir_entries
+        Ok(dir_entries) => {
+            let vec = dir_entries
                 .map(|opt_dir_entry| match opt_dir_entry {
                     Ok(entry) => os_str_to_roc_path(entry.path().into_os_string().as_os_str()),
                     Err(_) => {
                         todo!("handle dir_entry path didn't resolve")
                     }
                 })
-                .collect::<RocList<RocList<u8>>>(),
-        ),
+                .collect::<Vec<RocList<u8>>>();
+            RocResult::ok(RocList::from(&vec[..]))
+        }
         Err(_) => {
             todo!("handle Dir.list error");
         }

bhansconnect on Oct 9, 2022

Anyway, too tired too investigate this more now. At least we now have some specific CLs to investigate.

bhansconnect on Oct 9, 2022

I used the newest version of the nix environment the entire time, so it could be related to rust versioning changes that only now made an issue present. Or something else of that nature.

bhansconnect on Oct 9, 2022

I am gonna try and bisect

bhansconnect on Oct 9, 2022