Jump to content

Spurious wakeup

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by Msnicki (talk | contribs) at 13:28, 12 May 2020 (wakeups allowed without being signaled in some definitions of cv's). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

A spurious wakeup happens when a thread wakes up from waiting on a condition variable that's been signaled, only to discover that the condition it was waiting for is no longer satisfied. It's called spurious because the thread has seemingly been awakened for no reason. But spurious wakeups don't happen for no reason, they happen because in between the time when the condition variable was signaled and when the waiting thread finally ran, another thread ran and changed the condition. There was a race condition between the threads, with the typical result that sometimes, the thread waking up on the condition variable runs first, winning the race, and sometimes it runs second, losing the race.

On many systems, especially multiprocessor systems, the problem of spurious wakeups is exacerbated because if there are several threads waiting on the condition variable when it's signaled, the system may decide to wake them all up, treating every signal( ) to wake one thread as a broadcast( ) to wake all of them, thus breaking any possibly expected 1:1 relationship between signals and wakeups.[1] If there are ten threads waiting, only one will win and the other nine will experience spurious wakeups.

Some definitions of condition variables also allow for a wait to return without being signaled, thus offering implementation flexibility in dealing with error conditions and races inside the operating system. In the Solaris implementation of condition variables, a spurious wakeup may occur without the condition being signaled if the process is signaled; the wait system call aborts and returns EINTR .[2] The Linux pthread implementation of condition variables guarantees it will not do that.[3][4]

Because spurious wakeups can happen whenever there's a race and in some implementations, even in the absence of a race, when a thread wakes on a condition variable, it should check that the condition it sought is still satisfied. If it's not, meaning it lost the race, it should go back to sleeping on the condition variable, waiting for another opportunity.

References

  1. ^ Raymond Chen (February 1, 2018). "Spurious wake-ups in Win32 condition variables". Retrieved May 9, 2020.
  2. ^ "Interrupted Waits on Condition Variables (Solaris Threads Only)". Oracle Corporation. Retrieved May 9, 2020.
  3. ^ "pthread_cond_wait(3) - Linux man page". die.net. Retrieved May 9, 2020. These functions shall not return an error code of [EINTR].
  4. ^ "pthread_cond_timedwait, pthread_cond_wait - wait on a condition". The Open Group. 2018. Retrieved May 9, 2020.