antigravity-fhs: crash/coredump on exit

[jys1670]

Nixpkgs version

• Unstable (26.05)

Describe the bug

When closing the fhs-wrapped version of antigravity, the application consistently crashes with a SIGILL/SIGTRAP (ud2) and generates a coredump. While this doesn't impact functionality (as it happens during termination), it heavily spams systemd-coredump logs on every editor shutdown.

I analyzed the coredump with a debugger:

► 0 0x562a03f70e23 (ud2)
 ...
 1a:00d0│     0x7ffd23468420 —▸ 0x7f4db06f3380 (dirTreeCache()::cache) —▸ 0x3de401fcdc00 ◂— 0xefefefefefefefef
 ...
 1f:00f8│     0x7ffd23468448 —▸ 0x7f4db06b81df (std::_Hashtable<...>::_M_rehash(...)+175)
 21:0108│     0x7ffd23468458 —▸ 0x7f4db06f1298 (vtable for std::_Sp_counted_deleter<DirTree*, DirTreeDeleter, ...>+16)
 25:0128│     0x7ffd23468478 —▸ 0x7f4db06b8a1c (std::_Sp_counted_deleter<DirTree*, DirTreeDeleter, ...>::_M_dispose()+812)

1. Bundled @parcel/watcher (native Node addon) maintains a static cache map dirTreeCache()::cache
2. On exit, __cxa_atexit destroys the static map. Electron's memory allocator (PartitionAlloc) frees the memory and poisons the heap with 0xefefefefefefefef to catch dangling pointers.
3. std::shared_ptr<DirTree> drops its reference count to zero. DirTreeDeleter fires and attempts to remove the tree from the destroyed static cache map. It dereferences the poisoned 0xefefefef... pointer, and security assertions force a ud2 trap.

So something is broken in @parcel/watcher.
I am not sure if this issue is related to the Nix wrapping, or if it is a general bug in antigravity.

Steps to reproduce

1. Open some project in antigravity
2. Close antigravity
3. Check coredumpctl list for new entries

Expected behaviour

The application should terminate cleanly without crashing or generating a coredump upon exit

Screenshots

No response

Relevant log output

(typical coredumpctl info)

           PID: 20457 (antigravity)
           UID: 1000 (user)
           GID: 1000 (user)
        Signal: 5 (TRAP)
     Timestamp: Fri 2026-04-10 22:46:33 (9min ago)
  Command Line: /proc/self/exe --type=utility --utility-sub-type=node.mojom.NodeService --lang=en-US --service-sandbox-type=none --ren>
    Executable: /nix/store/i1wk67xzc1yd4q9z13br8jfgn9csxm4d-antigravity-1.22.2/lib/antigravity/antigravity
 Control Group: /user.slice/user-1000.slice/user@1000.service/app.slice/app-org.chromium.Chromium-19268.scope
          Unit: user@1000.service
     User Unit: app-org.chromium.Chromium-19268.scope
         Slice: user-1000.slice
     Owner UID: 1000 (user)
       Boot ID: a24e3f633ffc4481b01991b607f287e2
    Machine ID: 6c486aa9e9d94717af07e362481d60e6
      Hostname: kickshaw
       Storage: /var/lib/systemd/coredump/core.antigravity.1000.a24e3f633ffc4481b01991b607f287e2.20457.1775850393000000.zst (truncated)
  Size on Disk: 6.7M
       Message: Process 20457 (antigravity) of user 1000 dumped core.
                
                Stack trace of thread 20457:
                #0  0x000055bb1d66be23 n/a (n/a + 0x0)
                ELF object binary architecture: AMD x86-64

Additional context

No response

System metadata

> nix-shell -p nix-info --run "nix-info -m"
 - system: `"x86_64-linux"`
 - host os: `Linux 6.18.21-xanmod1, NixOS, 26.05 (Yarara), 26.05.20260409.4c1018d`
 - multi-user?: `yes`
 - sandbox: `yes`
 - version: `nix-env (Nix) 2.31.4`
 - nixpkgs: `/nix/store/hdb0cdayc6nkzm1m796hsb9kpsnkmmm0-nixpkgs-unstable-kickshaw`

Notify maintainers

@xiaoxiangmoe @Zaczero

---

Note for maintainers: Please tag this issue in your pull request description. (i.e. Resolves #ISSUE.)

I assert that this issue is relevant for Nixpkgs

• I assert that this is a bug and not a support request.
• I assert that this is not a duplicate of an existing issue.
• I assert that I have read the NixOS Code of Conduct and agree to abide by it.

Is this issue important to you?

Add a 👍 reaction to issues you find important.

[Zaczero]

It appears to be the parcel-bundler/watcher#216 issue, antigravity bundles 2.5.1 while the fix is in 2.5.2 release. Doesn't seem like a packaging issue.

[jys1670]

It appears to be the parcel-bundler/watcher#216 issue, antigravity bundles 2.5.1 while the fix is in 2.5.2 release. Doesn't seem like a packaging issue.

Do you also have constant crashes like that?
I am just not sure since I haven't seen other crash reports, which should be the case if it's a widespread watcher issue.

[jys1670]

If anyone runs into this issue: indeed, it is caused by the outdated @parcel/watcher that ships with antigravity-ide. I don't understand how such a widespread issue has remained unfixed for so long, given that the coredump spam can be reproduced on other Linux distributions as well.
As a workaround, I replaced @parcel/watcher with a newer version:

parcelWatcherFix = fetchurl {
    url = "https://registry.npmjs.org/@parcel/watcher-linux-x64-glibc/-/watcher-linux-x64-glibc-2.5.6.tgz";
    sha256 = "0bm4av1wd3yp0sim6pg0qkn8hibvh0z52yfgf8hx6cvvdksx5vj3";
  };
# ... overrideAttrs overlay:
postInstall = (oldAttrs.postInstall or "") + ''
  watcherPath="$out/lib/antigravity-ide/resources/app/node_modules/@parcel/watcher/build/Release/watcher.node"
  if [ -f "$watcherPath" ]; then
    tar xzf ${parcelWatcherFix} --strip-components=1 -C "$TMPDIR" package/watcher.node
    cp "$TMPDIR/watcher.node" "$watcherPath"
  fi
'';

[Zaczero]

it's like google doesn't care these days