[13434] in Athena Bugs

home help back first fref pref prev next nref lref last post

NetProb ticket #3323, sun4 [7.7R]: afs

daemon@ATHENA.MIT.EDU (mbarker@MIT.EDU)
Thu Apr 20 09:48:02 1995

From: mbarker@MIT.EDU
Date: Thu, 20 Apr 1995 09:47:56 -0400
To: afsdev@MIT.EDU, yandros@MIT.EDU
Cc: bugs@MIT.EDU


thanks for the report.  I'm passing it on to afsdev for information.

mike

>From yandros@MIT.EDU Thu Apr 20 06:18:31 1995
>Received: by gregor.MIT.EDU (5.57/4.7) id AA12583; Thu, 20 Apr 95 06:18:25 -04>00
>Received: from DEATHTONGUE.MIT.EDU by MIT.EDU with SMTP
>	id AA05535; Thu, 20 Apr 95 06:18:24 EDT
>From: yandros@MIT.EDU
>Received: by deathtongue.MIT.EDU (5.0/4.7) id AA20216; Thu, 20 Apr 1995 06:18:>22 -0400
>Date: Thu, 20 Apr 1995 06:18:22 -0400
>Message-Id: <9504201018.AA20216@deathtongue.MIT.EDU>
>To: bugs@MIT.EDU
>Subject: sun4 [7.7R]: afs
>X-Orgs: MIT SIPB VWA DIMINS 
>X-Shabby: not
>Content-Length: 552
>
>
>System name:		deathtongue
>Type and version:	SPARC/IPX 7.7R
>Display type:		cgthree
>
>What were you trying to do?
>
>  stress-test AFS.  OK, not really, but it might as well have been
>  what I was trying to do.
>
>What went wrong?
>
>  I got this message:
>
>    06:11 afs: FlushVcache page screwup
>
>  in my console.
>
>What should have happened?
>
>  AFS shouldn't be screwing up when FlushVCacheing? :)
>  
>  Since this is clearly a ``hey, this happened!'' debugging message, I
>  thought that I'd tell you ``hey, this happened!''
>
>Anything else you'd like to add?
>
>
>NETPROB-NOTE:
>Note:  mbarker    Thu Apr 20 09:44:25 1995
>error message is in afs/afs_cache.c
afs_FlushVCache(avc)
    register struct vc>ache *avc;
...
    /* eval, while holding all locks, whether there are any page>s
     * left, or any spurious references.  If both are false, and
     * we ha>ve vcache lock, then there can be no references added
     * while we hold the >vcache lock.
     * If "i" is set, we can discard this entry.
     */
    i = (>avc->v.v_pages == (struct page *) 0) && (avc->v.v_count == 1);
    if ((avc->v.>v_pages && !(avc->states & CHasPages)) ||
        (!avc->v.v_pages && (avc->sta>tes & CHasPages))) {
        printf("afs: FlushVcache page screwup\n");
       >i = 0;
    }
...>
------

home help back first fref pref prev next nref lref last post