From: ADVAX::"mcnc!VM1.NoDak.EDU!cgw%VAXB.ACS.UNT.EDU" 29-JUN-1991 02:21:17.60 To: Multiple recipients of list ANU-NEWS CC: Subj: ANU crashes, almost randomly Received: by ADVAX.DECnet (utk-mail11 v1.5) ; Sat, 29 Jun 91 02:17:13 EDT Received: from mcnc by ge-dab.GE.COM (5.61/GE-DAB 1.15) with UUCP id AA02605 for ; Sat, 29 Jun 91 01:32:44 -0400 Received: from VM1.NoDak.EDU by mcnc.mcnc.org (5.59/MCNC/3-21-91) id AA07556; Tue, 25 Jun 91 10:14:42 -0400 for ARISIA.DNET.ge.com!EVERHART Message-Id: <9106251414.AA07556@mcnc.mcnc.org> Received: from NDSUVM1.BITNET by VM1.NoDak.EDU (IBM VM SMTP R1.2.1MX) with BSMTP id 5751; Tue, 25 Jun 91 09:13:37 CDT Received: from NDSUVM1.BITNET by NDSUVM1.BITNET (Mailer R2.07) with BSMTP id 7372; Tue, 25 Jun 91 09:13:36 CDT Date: Tue, 25 Jun 1991 14:58:58 GMT Reply-To: mcnc!VM1.NoDak.EDU!cgw%VAXB.ACS.UNT.EDU Sender: ANU-NEWS Discussion Comments: Warning -- original Sender: tag was usenet@SOL.ACS.UNT.EDU From: mcnc!VM1.NoDak.EDU!cgw%VAXB.ACS.UNT.EDU Subject: ANU crashes, almost randomly To: Multiple recipients of list ANU-NEWS well, here's a posting, a reposting (because it didn't get posted the first time), some hints about switching from an ANU server to a 'other' server, and some more questions. the reposting: >*phew*. finally out of the debugger, and hopefully, what i came out with >is enough information for someone to fix some bugs. > >first, a little background: > >we run ANU 6.0-3 (14-Mar) with Bill Glass' patches of Mar 17, as a client to a >local node. the remote node is running Cnews 24-Mar-1991. the ANU client is >running on a 6310/VMS 5.3-1. the remote node is a Solbourne 5E/902 running >SunOS 4.0D_Export. we're using Wollongong 5.1. i do not have the SMG patch >installed (can anyone get me that info, please?). in addition, because of the >client-sitation, "Count" numbers are almost always off. > >here are the bugs: > >#1: >run news, immediately type SELECT JUNK and get: > >access violation, reason mask = 04 >break on unhandled exception at NEWSUTILITY\initni\%Line 10376+19 >(which happens to be: "sgap->grp_ia[i] = newsitm;") that one is still evident. >#2: >run news, read a few newsgroups, news crashes. i'm not sure exactly how >many newsgroups you must read before it happens, but i do believe you >have to read more than 5 or so. the way _i_ bring this about is just >enter news, get into some groups i don't read (but that are registered), >and skip items, one-by-one. then i type "select/new" to go to the next >newsgroup and/or "dir/new" to go to a different one, and skip there. >the crash: > >access violation, reason mask = 04 >break on unhandled exception at ITEM_FIND+983 > >running the debug version of news also gave this: >PATCH-NOFREE, No free storage available >but i doubt that's related. so is this one. >#3: >in my debugging, i tried another Cnews site. i couldn't read any >articles, but that was probably because the message #'s in the >newsrc were wrong (i just changed the server domainname to connect >to the node). however, i did get an SMG error: > >unhandled exception preceeding C$CKS+36 >%SMG-F-INVDIS_ID, invalid display-id and, this one, too. however, here's some hints to remember if you ever switch from an anu server to another type of server: don't listen to your boss who says "no, i don't think we need to recompile ANU (hi billy)", and check the code anyway. DO edit NEWSSITE.H and define NNTP_USE_XUPDGROUP, NNTP_USE_XGTITLE, NNTP_USE_XGNOTICE to be 0. then recompile. that solved about 75% of the crashes, but i can still crash ANU by the above bugs. in addition, i've seen #3 happen in our current situation, as well as the other Cnews site. in my checking, i discovered a typo in NEWSUTILITY.C. the patch had it right, but for some reason, it didn't get applied. at this point, i'm about ready to just re-get the 6.0-3 sources, apply the patches if i need to, and try that. [however, i just looked at NEWS_SRC and there are about 12 *.C_REJ files. i never noticed them before, and they could be from old patches. i _think_ the last time i patched news everything went ok. i shall check] >unfortunately, i'm not that familiar w/the routines in NEWSUTILITY, >and, well, i'd rather not be. if someone could try and reproduce these >and debug them, it would be *very* much appreciated! ah, also, if there >are any patches after Bill Glass' Mar 17 patches, could someone point >me to those? many thanks. also, can someone send me (or point me to) the information about the SMG patch to fix the 44th newsgroup bug? >need more information? please mail, don't post. but DO post if you >have patches! :) thanks, -cgw- ------------------------------------------------------------------------------- christopher williams, `gilligan', `dude', cgw@vaxb.acs.unt.edu, +1 817 565 4161 lead programmer/operator, the university of north texas, home of the _VaxCave_! `help stamp out and abolish redundancy!' my other .sig is boring too.