secrets env cmd command causing segfaults when you try to CTRL + C on the subcommand on slow calls

[tegefaulkes]

Describe the bug

During usage recently we've been experiencing segfaults while using the pk secrets env command. I haven't experienced this myself yet so I don't have the exact steps that cause this.

To Reproduce

People need to add details when they experience it.

Expected behavior

Running polykey secrets env should not cause a segfault.

Screenshots

Platform (please complete the following information)

• Device: [e.g. iPhone6]
• OS: [e.g. iOS]
• Version [e.g. 22]

Additional context

Notify maintainers

@tegefaulkes

[linear]

ENG-429 `secrets env` command causing segfaults

[tegefaulkes]

I've created this to track the problem. @CMCDragonkai @brynblack if you run into it at all then please add some details about what happened and why.

[aryanjassal]

Does this still happen? This hasn't been noticed for the past 2 months in almost regular usage.

[tegefaulkes]

I've never seen it happen. @brynblack @CMCDragonkai Have you seen this happen recently?

[aryanjassal]

Now that we are also using secrets env when launching the development shell, we still haven't noticed this issue. This means the issue could be already fixed. I'll be closing it now, and if the issue is encountered again, this can be re-opened with more information.

[CMCDragonkai]

I saw it efore. This happens for when you run a subcommand under secrets env, so not seeing it while running secrets env is not sufficient for closing this issue. There needs to be more tests on subcommands, and specifically it occurs for when you try to CTRL + C the subcommand due to some long running call to secrets env cmd.

[aryanjassal]

Running a potentially long operation like du -h / in the command and immediately cancelling it does not throw a segfault, but I did run into a segfault if i let the command run.

...
e}\0\0\u{14}\0\0\0\0\0\0\0\0\0\0\0\0\0��\0\0word/webSettings.xmlPK\u{1}\u{2}-\0\u{14}\0\u{6}\0\u{8}\0\0\0!\0\u{8}�u�\u{12}\u{1}\0\0�\u{1}\0\0\u{18}\0\0\0\0\0\0\0\0\0\0\0\0\0��\0\0customXml/itemProps2.xmlPK\u{5}\u{6}\0\0\0\0!\0!\0�\u{8}\0\0d�\0\0\0\0"`: file name contained an unexpected NUL byte', library/std/src/env.rs:353:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
fatal runtime error: failed to initiate panic, error 5
zsh: IOT instruction (core dumped)  pk secrets env vault -- 'du -h /'

So this might be related. If this was the exact error you were getting, then this would be useful to reproduce the issue.

[tegefaulkes]

It ran into a null byte where it doesn't expect it. file name contained an unexpected NUL byte. is that the full output of the failure?

[aryanjassal]

No, but there was a lot of junk. I piped the whole output into a file and got this. I was trying to export a .docx file as an environment variable, which ran into this error. So, the error I was getting is not related to this issue.

errlog.txt

[CMCDragonkai]

Running a potentially long operation like du -h / in the command and immediately cancelling it does not throw a segfault, but I did run into a segfault if i let the command run.

...
e}\0\0\u{14}\0\0\0\0\0\0\0\0\0\0\0\0\0��\0\0word/webSettings.xmlPK\u{1}\u{2}-\0\u{14}\0\u{6}\0\u{8}\0\0\0!\0\u{8}�u�\u{12}\u{1}\0\0�\u{1}\0\0\u{18}\0\0\0\0\0\0\0\0\0\0\0\0\0��\0\0customXml/itemProps2.xmlPK\u{5}\u{6}\0\0\0\0!\0!\0�\u{8}\0\0d�\0\0\0\0"`: file name contained an unexpected NUL byte', library/std/src/env.rs:353:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
fatal runtime error: failed to initiate panic, error 5
zsh: IOT instruction (core dumped)  pk secrets env vault -- 'du -h /'

So this might be related. If this was the exact error you were getting, then this would be useful to reproduce the issue.

Definitely had to do with the rust code.