Skip to content

cmd: fallback to parsing getent passwd output when $SHELL is not set#1760

Open
Rolv-Apneseth wants to merge 1 commit intocontainers:mainfrom
Rolv-Apneseth:fallback_shell
Open

cmd: fallback to parsing getent passwd output when $SHELL is not set#1760
Rolv-Apneseth wants to merge 1 commit intocontainers:mainfrom
Rolv-Apneseth:fallback_shell

Conversation

@Rolv-Apneseth
Copy link

As suggested in #1615, this PR just adds a fallback if $SHELL is not set for whatever reason.

Not very comfortable with Go, but hopefully the code is alright. Happy to make any changes anyway. And let me know if you want some tests added, I just wasn't sure entirely where to add that.

@gemini-code-assist
Copy link

Summary of Changes

Hello @Rolv-Apneseth, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the application's ability to reliably determine the user's default shell. It addresses scenarios where the $SHELL environment variable might be unset by introducing a new utility function that attempts to read $SHELL and, if unsuccessful, gracefully falls back to parsing the output of the getent passwd command. This change ensures that core functionalities dependent on the user's shell can proceed without error, making the application more resilient to varied environment configurations.

Highlights

  • Robust Shell Detection: Implemented a fallback mechanism to determine the user's default shell when the $SHELL environment variable is not set, improving application robustness.
  • New Utility Function: Introduced a new getUserShell utility function in src/cmd/utils.go to encapsulate the logic for retrieving the shell, first checking $SHELL and then falling back to parsing getent passwd output.
  • Refactored Shell Acquisition: Updated create.go, enter.go, and rootMigrationPath.go to utilize the new getUserShell function instead of directly relying on os.Getenv("SHELL").

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog
  • src/cmd/create.go
    • Replaced direct os.Getenv("SHELL") call with a call to the new getUserShell() function.
  • src/cmd/enter.go
    • Replaced direct os.Getenv("SHELL") call with a call to the new getUserShell() function.
  • src/cmd/rootMigrationPath.go
    • Replaced direct os.Getenv("SHELL") call with a call to the new getUserShell() function.
  • src/cmd/utils.go
    • Added a new getUserShell() function that first checks the $SHELL environment variable.
    • Implemented fallback logic within getUserShell() to execute and parse /bin/getent passwd output if $SHELL is empty.
    • Added error handling for getent command execution and output parsing.
    • Ensured the $SHELL environment variable is set for subsequent operations if determined via getent.
Activity
  • The author initiated this pull request to address an issue identified in failed to get the current user's default shell #1615, providing an initial implementation for a shell fallback mechanism.
  • The author expressed openness to feedback and making further changes, including adding tests if deemed necessary, indicating a collaborative approach to development.
Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature Command Description
Code Review /gemini review Performs a code review for the current pull request in its current state.
Pull Request Summary /gemini summary Provides a summary of the current pull request in its current state.
Comment @gemini-code-assist Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help /gemini help Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

  1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution.

Copy link

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request adds a fallback mechanism to determine the user's shell by parsing getent passwd output when the $SHELL environment variable is not set. This is a good improvement for robustness. The changes are applied consistently across the create, enter, and rootMigrationPath commands. My review includes a couple of suggestions for the new getUserShell function in src/cmd/utils.go to enhance its portability and robustness. Additionally, since you've mentioned being open to adding tests, it would be great to add unit tests for the new getUserShell function to cover both the successful paths and the error cases you're handling. This would improve the long-term maintainability of this new logic.

@Rolv-Apneseth Rolv-Apneseth force-pushed the fallback_shell branch 2 times, most recently from 9c87365 to 262de18 Compare February 19, 2026 16:52
@softwarefactory-project-zuul
Copy link

@dustymabe
Copy link
Collaborator

man.. it sure would be nice if golang/go#59121 were implemented :)

@Rolv-Apneseth
Copy link
Author

man.. it sure would be nice if golang/go#59121 were implemented :)

Yeah that's a shame

Copy link
Member

@debarshiray debarshiray left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for working on this, @Rolv-Apneseth ! Let me first answer your question about how the tests work.

I think something like this that inserts a unset SHELL should work:

$ git diff
diff --git a/test/system/101-create.bats b/test/system/101-create.bats
index 06b64a859dd6..97525ee78d3f 100644
--- a/test/system/101-create.bats
+++ b/test/system/101-create.bats
@@ -60,6 +60,35 @@ teardown() {
   assert_output "true"
 }
 
+@test "create: Smoke test with SHELL unset" {
+  local default_container
+  default_container="$(get_system_id)-toolbox-$(get_system_version)"
+
+  pull_default_image
+  unset SHELL
+
+  run --keep-empty-lines --separate-stderr "$TOOLBX" create
+
+  assert_success
+  assert_line --index 0 "Created container: $default_container"
+  assert_line --index 1 "Enter with: toolbox enter"
+  assert [ ${#lines[@]} -eq 2 ]
+  assert [ ${#stderr_lines[@]} -eq 0 ]
+
+  run podman ps --all
+
+  assert_success
+  assert_output --regexp "Created[[:blank:]]+$default_container"
+
+  run podman inspect \
+        --format '{{index .Config.Labels "com.github.containers.toolbox"}}' \
+        --type container \
+        "$default_container"
+
+  assert_success
+  assert_output "true"
+}
+
 @test "create: With a custom name (using option --container)" {
   pull_default_image

I ran the toolbox binary with and without your commit against this test, and the test passed and failed respectively. So, at least it's roughly working. :)

Feel free to add it to your commit, if you think everything is alright.

There are two broad categories of tests. First, the unit tests in the *_test.go files for the Go code run by meson test .... Second, the system tests using Bats in test/system that run the toolbox binary.

To run the unit tests locally, ensure that all the optional dependencies listed during meson setup are present, and do a git submodule init and git submodule update dance. You can either run all the tests with:

$ TMPDIR=/var/tmp TOOLBX=/path/to/toolbox bats /path/to/test/system

... or a particular file with:

$ TMPDIR=/var/tmp TOOLBX=/path/to/toolbox bats /path/to/test/system/101-create.bats

The latter might be easier to start with because there are a lot of tests and many of them do a lot of I/O. So, it may take a while to run all of them. :)

There's also a README.md in the test/system directory that might be of help.

Copy link
Member

@debarshiray debarshiray left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The changes look mostly good to me, other than some of these details.

I have one question arising out of curiosity. Did you check that the SHELL environment variable is really missing from the CoreOS command line environment? I wanted to try it for myself but I never got around to it.

The only other time I saw it go missing was in the GitHub Actions workflow environment when trying to run the Toolbx tests on a Ubuntu 22.04 host. See:
https://github.com/orgs/community/discussions/59413
8c28dc2660d61025
c22b09d095c7a8ac

The log-in sequence jumps through various hoops, and our future selves will be grateful, if we documented these deviations from the usual. :)

}

// However, fallback to parsing `getent passwd` output if it is not set for whatever reason
logrus.Debug("$SHELL environment variable was empty - falling back to getent passwd")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The comment looks superfluous, because the debug log says the same. :)

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

True, I'll remove it

src/cmd/utils.go Outdated
logrus.Debug("$SHELL environment variable was empty - falling back to getent passwd")

if currentUser == nil {
return "", errors.New("failed to get the current user's default shell")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Did you actually hit any scenario where currentUser was nil inside this function?

I am asking because I expect currentUser to be set to a valid *user.User very early in the lifecycle of the process. Look for the setUpGlobals() function. It's called in the init() of the cmd package, which should be very early. So, it would have to be a programmer error for currentUser to be nil.

If all that is true, then it will be better to use a panic() to express a mandatory pre-condition, because returning an error makes it look like a possible runtime scenario. See the getCurrentUserHomeDir() function for an example.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah, makes sense. I'll use a panic.

src/cmd/utils.go Outdated
if currentUser == nil {
return "", errors.New("failed to get the current user's default shell")
}
cmd := exec.Command("getent", "passwd", currentUser.Uid)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Instead of directly using os/exec, it will be better to go through one of the github.com/containers/toolbox/pkg/shell wrappers because they hook up the standard error and output streams of the child process to toolbox --verbose.

Something like this should work:

import (
    ...
    "github.com/containers/toolbox/pkg/shell"
    ...
)
...
...
    var stderr strings.Builder
    var stdout strings.Builder
    if err := shell.Run("getent", nil, &stdout, &stderr, "passwd", currentUser.Uid); err != nil {
        errString := stderr.String()
        logrus.Debugf("Getting entry for user %s from the passwd database failed: %s", currentUser.Uid, errString)
        return "", fmt.Errorf("failed to get the current user's default shell: %w", err)
    }
...

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah, didn't know about that, thank you

src/cmd/utils.go Outdated
func getUserShell() (string, error) {
// $SHELL is supposed to get set during login
userShell := os.Getenv("SHELL")
if userShell != "" {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wonder if these two lines can be combined for brevity and to restrict the scope of userShell. Like:

    if userShell := os.Getenv("SHELL"); userShell != "" {

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure, yes

src/cmd/utils.go Outdated
userShell = split[len(split)-1]

// Ensure $SHELL is set moving forward
os.Setenv("SHELL", userShell)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Was it really necessary to set the SHELL environment variable? It's a bit unexpected for a getSomething() function to have the side effect of setting something. :)

If it's not really necessary, then let's not set it. The SHELL environment variable is not used in any hot code path, and only once at the beginning for the create and enter commands.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No I suppose not, I just assumed it would make sense if $SHELL is also forwarded to the container itself as it might cause issues with other programs? Without that line, env | grep SHELL within the container has no output.

@debarshiray
Copy link
Member

The changes look mostly good to me, other than some of these details.

I have one question arising out of curiosity. Did you check that the SHELL environment variable is really missing from the CoreOS command line environment? I wanted to try it for myself but I never got around to it.

The only other time I saw it go missing was in the GitHub Actions workflow environment when trying to run the Toolbx tests on a Ubuntu 22.04 host. See: https://github.com/orgs/community/discussions/59413 8c28dc2660d61025 c22b09d095c7a8ac

The log-in sequence jumps through various hoops, and our future selves will be grateful, if we documented these deviations from the usual. :)

I was beginning to wonder if SHELL disappears only on OpenShift pods created by oc debug node, but not when someone is directly logging into CoreOS without OpenShift. However, the original issue at #1615 mentions Fedora CoreOS 38. :)

Signed-off-by: Rolv Apneseth <rolv.apneseth@gmail.com>
Signed-off-by: Rolv-Apneseth <rolv.apneseth@gmail.com>
@Rolv-Apneseth
Copy link
Author

Let me first answer your question about how the tests work.

Thank you for the breakdown of the testing setup and writing a test! It looks good to me.

I was beginning to wonder if SHELL disappears only on OpenShift pods created by oc debug node, but not when someone is directly logging into CoreOS without OpenShift. However, the original issue at #1615 mentions Fedora CoreOS 38. :)

I linked the other issue as it was related, and that's where your suggestion for a solution was, but I'm actually here from this issue, which is indeed an OpenShift debug pod. I haven't personally been able to reproduce the issue on CoreOS locally. Where would you like these occurrences documented?

And thank you for the thorough review. Hopefully I've addressed all your points but just let me know if more changes are required.

@softwarefactory-project-zuul
Copy link

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants