Restart after changing a setting only shuts down

[Tyriar]

Repro:

1. Set title bar style from custom -> native
2. Select restart, 🐛 it shuts down but doesn't start again

Version: 1.81.0-insider (user setup)
Commit: fda9726
Date: 2023-07-25T08:12:53.439Z
Electron: 22.3.17
ElectronBuildId: 22432899
Chromium: 108.0.5359.215
Node.js: 16.17.1
V8: 10.8.168.25-electron.0
OS: Windows_NT x64 10.0.22621

[Tyriar]

I had screen reader mode on if that matters

[sandy081]

Following API is used to restart

vscode/src/vs/workbench/contrib/relauncher/browser/relauncher.contribution.ts

Line 141 in cdf84c6

() => this.hostService.restart()

[bpasero]

@Tyriar does this reproduce consistently? I think I have seen this once today when I was also testing to change the title style but now I cannot reproduce anymore. I wonder if a pending update could be part of the reason it fails to restart because I think I had a pending update.

//cc @joaomoreno for the update part, I wonder if this has come up before

[Tyriar]

@bpasero it was yesterday, I just tried again and it also did not start up.

[bpasero]

Anything in the logs when you run with --verbose? Does it reproduce running out of sources?

[Tyriar]

All times I reproduced in Insiders.

Logs here, see it terminated instead of relaunching:

[main 2023-07-26T14:39:27.177Z] Lifecycle#onWillShutdown - end 'instanceLockfile'
[main 2023-07-26T14:39:27.180Z] [UtilityProcess id: 1, type: fileWatcher, pid: 11952]: received exit event with code 0
[main 2023-07-26T14:39:27.180Z] [UtilityProcessWorker]: terminated normally with code 0, signal: unknown
[main 2023-07-26T14:39:27.188Z] [File Watcher (node.js)] [raw] ["change"] globalStorage
[main 2023-07-26T14:39:27.189Z] [File Watcher (node.js)] [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\globalStorage
[main 2023-07-26T14:39:27.190Z] StorageMainService: closed workspace storage (11cd0562fd552aa78b92d6894814f6f7)
[main 2023-07-26T14:39:27.190Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.190Z] StorageMainService: closed workspace storage (1d327b9794e29f811edd5e97f31cde5a)
[main 2023-07-26T14:39:27.190Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.190Z] StorageMainService: closed workspace storage (364fa6cda0d57322c8359656b25e5464)
[main 2023-07-26T14:39:27.190Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.190Z] StorageMainService: closed workspace storage (16393023aa659375f2760e3e1aeb56d5)
[main 2023-07-26T14:39:27.190Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.191Z] StorageMainService: closed workspace storage (eef7b424caecd2bf919fa5630ddfb590)
[main 2023-07-26T14:39:27.191Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.191Z] StorageMainService: closed workspace storage (b07b3d603208f714ba143ac906a0984f)
[main 2023-07-26T14:39:27.191Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.203Z] StorageMainService: closed workspace storage (1b927216d3a88da4037d872d0731b799)
[main 2023-07-26T14:39:27.203Z] Lifecycle#onWillShutdown - end 'workspaceStorage'
[main 2023-07-26T14:39:27.216Z] StorageMainService: closed application storage
[main 2023-07-26T14:39:27.216Z] Lifecycle#onWillShutdown - end 'applicationStorage'
[main 2023-07-26T14:39:27.221Z] [UtilityProcess id: 7, type: fileWatcher, pid: 29616]: received exit event with code 0
[main 2023-07-26T14:39:27.221Z] [UtilityProcessWorker]: terminated normally with code 0, signal: unknown
[main 2023-07-26T14:39:27.223Z] [UtilityProcess id: 3, type: fileWatcher, pid: 16348]: received exit event with code 0
[main 2023-07-26T14:39:27.223Z] [UtilityProcessWorker]: terminated normally with code 0, signal: unknown
[main 2023-07-26T14:39:27.274Z] [File Watcher (node.js)] >> normalized [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\globalStorage
[main 2023-07-26T14:39:27.817Z] resolveShellEnv(): skipped (Windows)
[main 2023-07-26T14:39:27.863Z] [File Watcher (node.js)] [raw] ["change"] sync
[main 2023-07-26T14:39:27.863Z] [File Watcher (node.js)] [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:27.864Z] [File Watcher (node.js)] [raw] ["change"] sync
[main 2023-07-26T14:39:27.864Z] [File Watcher (node.js)] [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:27.874Z] resolveShellEnv(): skipped (Windows)
[main 2023-07-26T14:39:27.944Z] [File Watcher (node.js)] >> normalized [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:27.986Z] resolveShellEnv(): skipped (Windows)
[main 2023-07-26T14:39:28.074Z] resolveShellEnv(): skipped (Windows)
[main 2023-07-26T14:39:28.100Z] [File Watcher (node.js)] [raw] ["change"] sync
[main 2023-07-26T14:39:28.100Z] [File Watcher (node.js)] [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:28.100Z] [File Watcher (node.js)] [raw] ["change"] sync
[main 2023-07-26T14:39:28.100Z] [File Watcher (node.js)] [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:28.109Z] [File Watcher (node.js)] Request to start watching: c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\profiles\-3dc91576\snippets (excludes: <none>, includes: <all>)
[main 2023-07-26T14:39:28.112Z] [File Watcher (node.js)] Started watching: 'c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\profiles\-3dc91576\snippets'
[main 2023-07-26T14:39:28.182Z] [File Watcher (node.js)] >> normalized [CHANGED] c:\Users\Daniel\AppData\Roaming\Code - Insiders\User\sync
[main 2023-07-26T14:39:28.869Z] [UtilityProcess id: 2, type: extensionHost, pid: 15392]: received exit event with code 0
[main 2023-07-26T14:39:28.869Z] Extension host with pid 15392 exited with code: 0, signal: unknown.
[main 2023-07-26T14:39:28.948Z] [UtilityProcess id: 7, type: extensionHost, pid: 31448]: received exit event with code 0
[main 2023-07-26T14:39:28.948Z] Extension host with pid 31448 exited with code: 0, signal: unknown.
[main 2023-07-26T14:39:28.958Z] resolveShellEnv(): skipped (Windows)
[main 2023-07-26T14:39:28.962Z] [UtilityProcess id: 4, type: extensionHost, pid: 19664]: received exit event with code 0
[main 2023-07-26T14:39:28.962Z] Extension host with pid 19664 exited with code: 0, signal: unknown.
[main 2023-07-26T14:39:29.104Z] [UtilityProcess id: 3, type: extensionHost, pid: 22360]: received exit event with code 0
[main 2023-07-26T14:39:29.104Z] Extension host with pid 22360 exited with code: 0, signal: unknown.
[main 2023-07-26T14:39:29.124Z] [UtilityProcess id: 5, type: extensionHost, pid: 14296]: received exit event with code 0
[main 2023-07-26T14:39:29.124Z] Extension host with pid 14296 exited with code: 0, signal: unknown.
[main 2023-07-26T14:39:29.124Z] Lifecycle#onWillShutdown - end 'extHostStarter'
[main 2023-07-26T14:39:29.126Z] Lifecycle#app.on(will-quit) - after fireOnWillShutdown
[main 2023-07-26T14:39:29.126Z] Lifecycle#app.on(will-quit) - calling app.quit()
[main 2023-07-26T14:39:29.126Z] Lifecycle#relaunch() - calling app.relaunch()
[main 2023-07-26T14:39:30.138Z] [UtilityProcess type: ptyHost, pid: 18484]: received exit event with code 0

~  took 20s
❯

[bpasero]

No, it seems to do the right thing:

vscode/src/vs/platform/lifecycle/electron-main/lifecycleMainService.ts

Lines 625 to 626 in 96ebd36

	this.trace('Lifecycle#relaunch() - calling app.relaunch()');
	app.relaunch({ args });

This would indicate that Electrons API fails here.

@deepak1556 any clue?

@Tyriar curious if there is possibly a native crash here, can you run with the --crash-reporter-directory=<folder> arg and see if there is a crash dump?

[Tyriar]

Weird one for you... It works fine when I set a crash reporter dir, but not without it.

Running this worked when I restarted 3 times:

code-insiders.cmd --verbose --crash-reporter-directory=c:\Users\Daniel\vsc-crash

Running this did not work when I restarted 3 times:

code-insiders.cmd --verbose

[Tyriar]

I was not seeing crash dumps before, I do see one which I'll DM to you

[Tyriar]

Just tried with only one window (vscode repo) open and no pty host and it still happens.

[Tyriar]

More tests:

• All builds down to vscode-bisect 1.77 were bad
• It works fine in stable 1.80.1
• It does not work in insiders 1.80.0
• It does not work in user install or archive for insiders

[bpasero]

I am still not able to reproduce even on my native Windows Surface machine. I am inclined to close this as out of scope given its not something stable users see.

@Tyriar curious, how is exploration builds for you?

[deepak1556]

@bpasero or @Tyriar can you forward me the crash dump, I am also unable to repro with insiders.

Couple more follow-up questions,

1. Is the stable build a user install or system install ?
2. Does it work fine with portable mode for archive insiders ?
3. Does it work with user install insiders when you add "disable-chromium-sandbox": true to > Configure runtime arguments ?

[Tyriar]

@deepak1556 I think it was a red herring as there was only 1 crash dump despite 3 restarts. I'll send regardless

Exploration worked 🤷, hopefully that means insiders will when Electron is updated

[deepak1556]

Reopening since issue is present with insiders for @Tyriar

[deepak1556]

Analysis of the bug is here for reference.

The root issue is that vscode has URI like arguments when setting log level for extensions which has the format <extension-id>:<log-level> and if additional arguments follow this uri like argument then the runtime will abort early on start due to an RCE fix made a long time back GHSA-w222-53c6-c86p

@sandy081 The fix here would be to switch from using uri separator : to something like = or \. Similar kind of arguments can be found in the runtime as well https://source.chromium.org/chromium/chromium/src/+/main:base/logging.h;l=103-123, which use = as separator.