Skip site navigation (1)Skip section navigation (2) 
Date:      Sat, 6 Apr 2002 22:45:26 -0400 (AST)
From:      "Marc G. Fournier" <scrappy@hub.org>
To:        freebsd-stable@freebsd.org
Cc:        jkoshy@freebsd.org, <freebsd-ports@freebsd.org>, <mab@red-bean.com>
Subject:   Re: Upgraded libraries breaks aolserver port?
Message-ID:  <20020406223214.L86558-100000@mail1.hub.org>
In-Reply-To: <20020406184751.Q86558-100000@mail1.hub.org>
next in thread | previous in thread | raw e-mail | index | archive | help 
On Sat, 6 Apr 2002, Marc G. Fournier wrote:

>
> Just recently upgraded my machine, and aolserver no longer starts, with an
> error of:
>
> nsthread(20981) error: pthread_cond_timedwait failed in Ns_CondTimedWait: Invalid argument
>
> OS is:
>
> 4.5-STABLE FreeBSD 4.5-STABLE #6: Mon Mar 25 21:01:05 CST 2002
>
> Has anyone else seen this?
>
>
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-stable" in the body of the message
>

As a followup to this, the 'offending' function is in thread/pthread.cpp
... does anything look "wrong" with the function that would cause the
above error in the latest FreeBSD?

int
Ns_CondTimedWait(Ns_Cond *condPtr, Ns_Mutex *mutexPtr, Ns_Time *timePtr)
{
    pthread_cond_t  *cond;
    pthread_mutex_t *lockPtr;
    Thread          *ownerPtr;
    Mutex           *mPtr;
    int              err, status;
    struct timespec  ts;

    if (timePtr == NULL) {
        Ns_CondWait(condPtr, mutexPtr);
        return NS_OK;
    }

    cond = GETCOND(condPtr);
    mPtr = GETMUTEX(mutexPtr);
    lockPtr = mPtr->lock;
    ownerPtr = mPtr->ownerPtr;
    mPtr->ownerPtr = NULL;

    /*
     * Convert the microsecond-based Ns_Time to a nanosecond-based
     * struct timespec.
     */

    ts.tv_sec = timePtr->sec;
    ts.tv_nsec = timePtr->usec * 1000;

    /*
     * As documented on Linux, pthread_cond_timedwait may return
     * EINTR if a signal arrives.  We have noticed that
     * EINTR can be returned on Solaris as well although this
     * is not documented (perhaps, as above, it's possible it
     * bubbles up from _lwp_cond_timedwait???).  Anyway, unlike
     * the ETIME case above, we'll assume the wakeup is truely
     * spurious and simply restart the wait knowing that the
     * ts structure has not been modified.
     */

    do {
        err = pthread_cond_timedwait(cond, lockPtr, &ts);
    } while (err == EINTR);

#ifdef HAVE_ETIME_BUG

    /*
     * See comments above and note that here ETIME is still considered
     * a spurious wakeup, not an indication of timeout because we're
     * not making any assumptions about the nature or this bug.
     * While we're less certain, this should still be ok as properly
     * written condition code should tolerate the wakeup.
     */

    if (err == ETIME) {
        err = 0;
    }
#endif

    if (ERRTIMEDOUT(err)) {
        status = NS_TIMEOUT;
    } else if (err != 0) {
        NsThreadFatal("Ns_CondTimedWait", "pthread_cond_timedwait", err);
    } else {
        status = NS_OK;
    }

    mPtr->ownerPtr = ownerPtr;
    ++mPtr->nlock;

    return status;
}




To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020406223214.L86558-100000>