drivers: fix utube kick in the ready buffer mode - #258
Open
bigbes wants to merge 1 commit into
Open
Conversation
Two defects in method.kick() of the utube driver, both in the
'ready_buffer' storage mode. utubettl does the same thing correctly via
update_ready().
1. When the utube had no entry in the ready buffer, kick() called
put_ready(self, task[3])
while the helper is put_ready(self, id, utube). The utube name was
passed as the task id and the utube became nil, so the insert failed
on the space format and the error was swallowed by the pcall inside
put_ready(). The kicked task stayed READY but was never returned by
take(), because take_ready() only looks at the ready buffer.
2. When the utube already had a younger task in the buffer, kick()
replaced it with
self.space_ready_buffer:insert({task[1], task[2]})
that is {id, status} instead of {id, utube}. The status string landed
in the utube field, so the buffer no longer had an entry under the
real utube name and a later put() added a second one. With two entries
for one utube the driver could take two tasks of the same utube at
once, which breaks the main utube guarantee, and take_ready() could
spin forever on an entry whose utube already had a TAKEN task.
Introduce update_ready() mirroring the utubettl helper and use it in
kick().
Also commit the transaction before the early returns of kick(): with no
buried tasks kick() returned while the transaction opened by
begin_if_not_in_txn() was still running, and it leaked into whatever the
caller fiber did next. This part is within the scope of #231.
Closes #256
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two defects in method.kick() of the utube driver, both in the 'ready_buffer' storage mode. utubettl does the same thing correctly via update_ready().
When the utube had no entry in the ready buffer, kick() called
put_ready(self, task[3])while the helper isput_ready(self, id, utube). The utube name was passed as the task id and the utube became nil, so the insert failed on the space format and the error was swallowed by the pcall inside put_ready(). The kicked task stayed READY but was never returned by take(), because take_ready() only looks at the ready buffer.When the utube already had a younger task in the buffer, kick() replaced it with
insert({task[1], task[2]}), that is {id, status} instead of {id, utube}. The status string landed in the utube field, so the buffer no longer had an entry under the real utube name and a later put() added a second one. With two entries for one utube the driver could take two tasks of the same utube at once, which breaks the main utube guarantee, and take_ready() could spin forever on an entry whose utube already had a TAKEN task.Introduce update_ready() mirroring the utubettl helper and use it in kick().
Also commit the transaction before the early returns of kick(): with no buried tasks kick() returned while the transaction opened by begin_if_not_in_txn() was still running, and it leaked into whatever the caller fiber did next. This part is within the scope of #231.
Closes #256