home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.unix.programmer:4662 comp.unix.internals:1810
- Path: sparky!uunet!sun-barr!olivea!spool.mu.edu!darwin.sura.net!gatech!mailer.cc.fsu.edu!sun13!delta!mueller
- From: mueller@delta.cs.fsu.edu (Frank Mueller)
- Newsgroups: comp.unix.programmer,comp.unix.internals
- Subject: Re: spurious early wakeups by SIGALRM under SunOS 4.1
- Message-ID: <10736@sun13.scri.fsu.edu>
- Date: 15 Sep 92 15:28:20 GMT
- References: <10728@sun13.scri.fsu.edu> <1992Sep14.223515.6843@noao.edu>
- Sender: news@sun13.scri.fsu.edu
- Followup-To: comp.unix.programmer
- Organization: Florida State University Computer Science
- Lines: 29
-
- In article <1992Sep14.223515.6843@noao.edu>, jdavis@bordeaux.kpno.noao.edu (Jim Davis) writes:
- > In article <10728@sun13.scri.fsu.edu> mueller@pi.fsu.edu (Frank Mueller) writes:
- > >Has anybody else experienced any spurious early wakeups by the timer signal
- > >which was requested by a setitimer (real-time timer) under SunOS 4.1.1?
- > >
- > >Any ideas why this happens? Is it a known bug?
- >
- > Yup. Here's the patch README. BTW, mail to pi.fsu.edu bounced with a 'Host
- > unknown' error...
- >
- > Patch-ID# 100458-02
- [stuff deleted]
-
- Thanks, Jim, but that didn't fix it. Our problem is that the SIGALRM is coming
- in *early*. When we set up setitimer to send a signal in 60 sec. (from now, i.e.
- using gettimeofday), the signal comes in and we check on the current time (again
- gettimeofday) just to find out that the signal came up to ~20ms early. We were
- under the impression that the semantics of setitimer guarantees that you are
- woken up by SIGALRM *no earlier* than the specified time. These results vary
- considerably from machine to machine, from always waking up early, over sporadi-
- cally doing so, to never doing so.
-
- Has anybody else an idea? (I wonder if Sun knows... Or if they even care... Maybe
- it's a feature:-)
-
- Frank
- mueller@cs.fsu.edu
-
- P.S. My return address should be fixed now.
-