reactor: Always retry waitpid #2531
Merged
+31
−32
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
A possible issue with how Docker Desktop v4.34+ (particulary on macOS) handles `pidfd` functionality causes an assertion failure in Seastar's `reactor::waitpid` method. The issue: - Seastar creates a `pidfd` using `pidfd_open` - It polls this file descriptor until it becomes readable - When readable, call `waitpid` with `WNOHANG`. According to [pidfd_open](https://man7.org/linux/man-pages/man2/pidfd_open.2.html), the file descriptor becomes readable when the process terminates - Seastar expects `waitpid` to either return a positive integer representing the child process that has ended or a negative value to indicate an error. However if `waitpid` returns `0`, then Seastar asserts, crashing the using application. This change adds `reactor::do_waitpid` which loops on calling `waitpid` until it returns a non-zero value. This method is called both when a `pidfd` is in use and when it isn't. The assertion is also removed. Signed-off-by: Michael Boquard <[email protected]>
avikivity
reviewed
Nov 9, 2024
| @@ -2075,6 +2075,32 @@ static auto next_waitpid_timeout(std::chrono::milliseconds this_timeout) { | |||
|
|
|||
| #endif | |||
|
|
|||
| future<int> reactor::do_waitpid(pid_t pid) { | |||
| return do_with(int{}, std::chrono::milliseconds(0), [pid, this](int& wstatus, | |||
There was a problem hiding this comment.
Please use coroutines in new code.
There was a problem hiding this comment.
Ah, you're just moving old code around. ok.
avikivity
reviewed
Nov 9, 2024
| return do_with(pollable_fd(file_desc::from_fd(pidfd.result)), int{}, [pid, this](auto& pidfd, int& wstatus) { | ||
| return pidfd.readable().then([pid, &wstatus, this] { | ||
| return _thread_pool->submit<syscall_result<pid_t>>([pid, &wstatus] { | ||
| return wrap_syscall<pid_t>(::waitpid(pid, &wstatus, WNOHANG)); |
There was a problem hiding this comment.
Presumably, using _thread_pool here (and above) is wrong, since WNOHANG means the kernel won't block.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A possible issue with how Docker Desktop v4.34+ (particulary on macOS) handles
pidfdfunctionality causes an assertion failure in Seastar'sreactor::waitpidmethod.The issue:
pidfdusingpidfd_openwaitpidwithWNOHANG. According to pidfd_open, the file descriptor becomes readable when the process terminateswaitpidto either return a positive integer representing the child process that has ended or a negative value to indicate an error. However ifwaitpidreturns0, then Seastar asserts, crashing the using application.This change adds
reactor::do_waitpidwhich loops on callingwaitpiduntil it returns a non-zero value. This method is called both when apidfdis in use and when it isn't. The assertion is also removed.