[
https://issues.apache.org/jira/browse/TS-2137?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Peach resolved TS-2137.
-----------------------------
Resolution: Fixed
The reported assertion is no longer in the code. Let's mark this as resolved
and open a new bug if we find any other problems.
> Use eventfd instead of pthread signal/wait in ATS
> -------------------------------------------------
>
> Key: TS-2137
> URL: https://issues.apache.org/jira/browse/TS-2137
> Project: Traffic Server
> Issue Type: Improvement
> Components: Core, Logging, Stats
> Reporter: Yunkai Zhang
> Assignee: Yunkai Zhang
> Fix For: 4.2.0
>
> Attachments:
> 0001-TS-2137-Forgot-to-add-header-guide-in-EventNotify.h.patch,
> 0001-TS-2137-Use-eventfd-instread-of-pthread-signal-wait-.V2.patch,
> 0001-TS-2137-Use-eventfd-instread-of-pthread-signal-wait-.V3.patch,
> 0001-TS-2137-Use-eventfd-instread-of-pthread-signal-wait-.V4.patch,
> 0001-TS-2137-Use-eventfd-instread-of-pthread-signal-wait-.patch,
> eventfd_vs_pthread_cond_benchmark.tar.gz
>
>
> pthread_cond_signal/wait is used in several places in ATS, including but not
> limited:
> 1) Logging system.
> 2) ProtectedQueue in event system.
> 3) RecProcess in stats system.
> As we known, pthread_cond_signal() needs to take a lock. As such it'll cause
> more context switches than eventfd.
> I wrote a simple benchmark program to compare the speed of eventfd and
> pthread_cond notify mechanism.
> You can get the benchmark program from the attachment:
> https://issues.apache.org/jira/secure/attachment/12598570/eventfd_vs_pthread_cond_benchmark.tar.gz,
> and test it by yourself if interested.
> What the program does is:
> 1) Create 10 producer threads and 1 consumer thread.
> 2) Producer threads will notify consumer concurrently in a loop.
> 3) The consumer thread will receive notification up to a given loop time.
> Here is my testing result, the command line argument is loop time(1000000) of
> consumer:
> {code}
> [eventfd_vs_pthread_cond_benchmark]$ ls
> main.cc Makefile
> [eventfd_vs_pthread_cond_benchmark]$ make
> g++ -g -o eventfd_test main.cc -DHAVE_EVENTFD -lpthread
> g++ -g -o pthread_cond_test main.cc -lpthread
> [eventfd_vs_pthread_cond_benchmark]$ ls
> eventfd_test main.cc Makefile pthread_cond_test
> [eventfd_vs_pthread_cond_benchmark]$ time ./eventfd_test 1000000
> real 0m11.644s
> user 0m1.517s
> sys 1m31.179s
> [eventfd_vs_pthread_cond_benchmark]$ time ./pthread_cond_test 1000000
> real 0m57.438s
> user 0m30.152s
> sys 6m7.289s
> {code}
> We can see that eventfd is about 5 times faster than pthread_cond notify
> mechanism.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
