| 1 |
2014-08-13 12:36 GMT-05:00 Andrés Becerra Sandoval <andres.becerra@×××××.com>: |
| 2 |
> 2014-08-13 12:21 GMT-05:00 Grant Edwards <grant.b.edwards@×××××.com>: |
| 3 |
>> This is not Gentoo specific, and while I'm doing my prototyping and |
| 4 |
>> development on a Gentoo system, the eventual target is not going to be |
| 5 |
>> running Gentoo -- so feel free to ignore this thread or throw things |
| 6 |
>> at me. |
| 7 |
>> |
| 8 |
>> I'm trying to figure out how to synchronize threads which may be in |
| 9 |
>> different processes. Basically, I want thread A to be able to wake up |
| 10 |
>> any number of other threads B, C, D, ... who are all blocking until A |
| 11 |
>> says "go" (and who may or may not be in other processes). |
| 12 |
>> |
| 13 |
>> Other (mostly embedded) OSes I've used had some sort of "event flag" |
| 14 |
>> API that did exactly what I'm looking for, but I can't seem to find |
| 15 |
>> such a thing in pthreads. |
| 16 |
>> |
| 17 |
>> A condition variable in shared memory is the closest thing I have |
| 18 |
>> found, and my test applications are working OK (so far). But, I'm |
| 19 |
>> unclear on the purpose of the mutex whose address you pass to |
| 20 |
>> pthread_cond_wait(). |
| 21 |
>> |
| 22 |
>> Is it to prevent race conditions when manipulating the condition |
| 23 |
>> variable's internal state? I don't see how that can be the case, since |
| 24 |
>> the signal/broadcast call would have to be aware of it and it isn't. |
| 25 |
>> |
| 26 |
>> The mutex appears to be there to serialize access to some user-defined |
| 27 |
>> variable(s) (outside the condition variable itself) which I don't |
| 28 |
>> have. So all the mutex locking/unlocking and resultant blocking of B, |
| 29 |
>> C, D is just wasted overhead and pointless latency. |
| 30 |
>> |
| 31 |
>> pthread_cond_wait(3) says |
| 32 |
>> |
| 33 |
>> When using condition variables there is always a Boolean predicate |
| 34 |
>> involving shared variables associated with each condition wait that |
| 35 |
>> is true if the thread should proceed. Spurious wakeups from the |
| 36 |
>> pthread_cond_timedwait() or pthread_cond_wait() functions may |
| 37 |
>> occur. Since the return from pthread_cond_timedwait() or |
| 38 |
>> pthread_cond_wait() does not imply anything about the value of this |
| 39 |
>> predicate, the predicate should be re-evaluated upon such return. |
| 40 |
>> |
| 41 |
>> I have no "Boolean predicate" (which presumably comprises the |
| 42 |
>> "user-defined variables outside the condition variable" I mentioned |
| 43 |
>> above), and I don't want spurious wakeups, so a pthreads condition |
| 44 |
>> variable would appear to be the wrong thing to use. |
| 45 |
>> |
| 46 |
>> Is there something like an "event flag" similar to a condition |
| 47 |
>> variable without spurious wakeup problem and without the extra |
| 48 |
>> overhead of the mutex and "Boolean predicate". |
| 49 |
>> |
| 50 |
>> Or am I expected to build my own "event flag" using the aforesaid |
| 51 |
>> "boolean predicate" just to avoid the spurious wakeup problem? [I'm |
| 52 |
>> guessing this is the case...] |
| 53 |
>> |
| 54 |
>> -- |
| 55 |
>> Grant Edwards grant.b.edwards Yow! I'm DESPONDENT ... I |
| 56 |
>> at hope there's something |
| 57 |
>> gmail.com DEEP-FRIED under this |
| 58 |
>> miniature DOMED STADIUM ... |
| 59 |
>> |
| 60 |
>> |
| 61 |
> |
| 62 |
> Hi Grant, |
| 63 |
> |
| 64 |
> The best explanation I have read is this chapter: |
| 65 |
> http://pages.cs.wisc.edu/~remzi/OSTEP/threads-cv.pdf |
| 66 |
> |
| 67 |
> from the book: |
| 68 |
> |
| 69 |
> http://pages.cs.wisc.edu/~remzi/OSTEP/ |
| 70 |
> |
| 71 |
> I know its 17 pages, but it is worth it! |
| 72 |
> |
| 73 |
> |
| 74 |
> -- |
| 75 |
> Andrés Becerra Sandoval |
| 76 |
|
| 77 |
In short: |
| 78 |
|
| 79 |
Withouth the use of the lock, the condition variable and a shared |
| 80 |
variable in concert you can get in trouble! |
| 81 |
|
| 82 |
|
| 83 |
-- |
| 84 |
Andrés Becerra Sandoval |
Archives