[Bug?]: misconfigured yarn global script issue when executed with parallel flag

Self-service

• I'd be willing to implement a fix

Describe the bug

Yarn version I am on is 4.3.1 and it seems to be utilizing 100% of the cpu when a misconfigured global script is used.

To reproduce

I have created a repro here https://github.com/twlite/yarn-global-script-repro with a simple setup. This seems to occur only with the workspace.

Steps

• create a simple workspace with at least 2 packages
• add a root global script that also executes global script (:), preferably same name (in my original case, it was build:check)
• make sure that script does not exist in any of the packages inside the workspace
• execute the command yarn workspaces foreach --all -pt run <globalScriptName>
• The script never stops and uses 100% of the cpu (my fan went jet mode)

Environment

System:
    OS: Windows 11 10.0.22631
    CPU: (28) x64 Intel(R) Core(TM) i7-14700K
  Binaries:
    Node: 22.2.0 - ~\AppData\Local\Temp\xfs-4bfc2d56\node.CMD
    Yarn: 4.3.1 - ~\AppData\Local\Temp\xfs-4bfc2d56\yarn.CMD
    npm: 10.5.2 - ~\AppData\Local\Volta\tools\image\npm\10.5.2\bin\npm.CMD
    pnpm: 9.4.0 - C:\Program Files\Volta\pnpm.EXE
    bun: 1.1.18 - ~\.bun\bin\bun.EXE

Additional context

I wanted to take a screenshot but it failed due to the stress on my cpu.

Start by cloning the linked reproduction and running the documented yarn workspaces foreach --all -pt run <globalScriptName> command with the root global script setup. Trace the workspace execution path that handles the missing package script, and consider the issue resolved when the command terminates without looping or consuming 100% CPU.