mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilbert Song <songzihao1...@gmail.com>
Subject Re: Review Request 71136: Fixed use-after-free bug in `PendingFutureTrackerProcess`.
Date Mon, 22 Jul 2019 20:31:19 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/71136/#review216792
-----------------------------------------------------------


Ship it!




Ship It!

- Gilbert Song


On July 22, 2019, 8:51 a.m., Andrei Budnik wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/71136/
> -----------------------------------------------------------
> 
> (Updated July 22, 2019, 8:51 a.m.)
> 
> 
> Review request for mesos, Benjamin Mahler, Gilbert Song, Meng Zhu, and Qian Zhang.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Previously, the future tracker subscribed on `onAny`, `onAbandoned`,
> `onDiscard` events for the future using callbacks, which accept
> the same iterator to the `pending` list. Since `onAny` and `onDiscard`
> events can overlap, the same callback can be called more than once,
> which leads to accessing invalidated iterator. This patch fixes the
> problem by removing `onDiscard` handler because it does not transition
> a future to the terminal state.
> 
> 
> Diffs
> -----
> 
>   src/common/future_tracker.hpp a3f191a58b7ada58153ca33a8d0409b846faedae 
> 
> 
> Diff: https://reviews.apache.org/r/71136/diff/2/
> 
> 
> Testing
> -------
> 
> I was able to reproduce a segfault observed in our CI by launching `ROOT_LaunchGroupFailure`
test in the loop with `stress` running in parallel:
> ```
> sudo src/mesos-tests --verbose --gtest_repeat=100 --break_on_failure --gtest_filter=ROOT_DOCKER_DockerAndMesosContainerizers/DefaultExecutorTest.ROOT_LaunchGroupFailure/0
> 
> stress -c 16
> ```
> After applying the patch, the test passed successfully.
> 
> 
> Thanks,
> 
> Andrei Budnik
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message