kernel: Update standard_error to use prim_tty nif #9116
Merged
Conversation
This file contains hidden or 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
CT Test Results 4 files 200 suites 1h 51m 51s ⏱️ For more details on these failures, see this check. Results for commit ba6446b. ♻️ This comment has been updated with latest results. To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass. See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally. Artifacts// Erlang/OTP Github Action Bot |
dd4251d to
ba6446b
Compare
frazze-jobb
previously approved these changes
Dec 11, 2024
ba6446b to
d79b98a
Compare
standard_error is started before the on_load handlers in init are run, so prim_tty will not be completely loaded when running in embedded mode. So we call it during init of prim_tty instead.
The primary reason that we do this is to make writing to standard_error as synchronous as possible. When using the fd driver the write was done on an async thread, which meant that there was no simple way to get an acknowledgement when the write had completed. Now that we use a NIF we can much more easily get a check when the write is done so that we know when io:format(standard_error, ...) has returned we have also written to that fd.
d79b98a to
ead7424
Compare
frazze-jobb
approved these changes
Dec 17, 2024
NelsonVides
added a commit
to NelsonVides/otp
that referenced
this pull request
Oct 16, 2025
Fix a race condition where the standard_error process could call prim_tty:init/1 before being registered, causing a crash when prim_tty's writer and reader processes attempt to derive their own registered names from the parent process. The issue was introduced in commit ead7424 (PR erlang#9116), which changed prim_tty's reader and writer processes to dynamically derive their registered names from their parent process name (e.g., 'standard_error' -> 'standard_error_reader', 'standard_error_writer'). This was done to support multiple TTY instances (stdout and stderr) with distinct process names. However, the standard_error:start/0 function had a race condition: 1. spawn(fun server/0) - creates process, starts executing immediately 2. register(?NAME, Id) - parent registers the spawned process 3. server() calls prim_tty:init() - may execute before step 2 When prim_tty:init/1 spawns writer/reader processes, they call set_name(Parent, Postfix) which does: erlang:process_info(Parent, registered_name) If this executes before the parent completes registration, it returns [] instead of {registered_name, standard_error}, causing a pattern match failure: "no match of right hand side value: []" at prim_tty.erl:674. The fix moves the register/2 call into the spawned process itself, ensuring it completes before prim_tty:init/1 is called. This eliminates the race condition entirely. Closes erlang#10174
frazze-jobb
pushed a commit
to NelsonVides/otp
that referenced
this pull request
Oct 23, 2025
Fix a race condition where the standard_error process could call prim_tty:init/1 before being registered, causing a crash when prim_tty's writer and reader processes attempt to derive their own registered names from the parent process. The issue was introduced in commit ead7424 (PR erlang#9116), which changed prim_tty's reader and writer processes to dynamically derive their registered names from their parent process name (e.g., 'standard_error' -> 'standard_error_reader', 'standard_error_writer'). This was done to support multiple TTY instances (stdout and stderr) with distinct process names. However, the standard_error:start/0 function had a race condition: 1. spawn(fun server/0) - creates process, starts executing immediately 2. register(?NAME, Id) - parent registers the spawned process 3. server() calls prim_tty:init() - may execute before step 2 When prim_tty:init/1 spawns writer/reader processes, they call set_name(Parent, Postfix) which does: erlang:process_info(Parent, registered_name) If this executes before the parent completes registration, it returns [] instead of {registered_name, standard_error}, causing a pattern match failure: "no match of right hand side value: []" at prim_tty.erl:674. The fix moves the register/2 call into the spawned process itself, ensuring it completes before prim_tty:init/1 is called. This eliminates the race condition entirely. Closes erlang#10174
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
The primary reason that we do this is to make writing to standard_error as synchronous as possible. When using the fd driver the write was done on an async thread, which meant that there was no simple way to get an acknowledgement when the write had completed. Now that we use a NIF we can much more easily get a check when the write is done so that we know when io:format(standard_error, ...) has returned we have also written to that fd.