mckenzie@june.cs.washington.edu (Neil McKenzie) (05/09/89)
I just installed the upgrade from A/UX 1.0.1 to A/UX 1.1 on my 4 Mbyte Mac IIx ('030 version of the MacII). There seems to be a serious reliability problem with the new kernel running on my computer. I can make it crash by simply doing the following from the shell: % rlogin localhost ... % exit The machine then dies and prints the error message: panic: sbflush 2 It is necessary to perform a cold restart at this point. I've also seen similar problems when I run the new (color supported) X11, running innocuous programs like "plaid" from a remote client. The machine bombs and various messages come up: Double panic: mget Kernel bus error (then a register dump) Double panic: kernel memory management error I tried playing with kconfig to increase different kernel parameters (as is described in Chapter 2 of the "X Window System User's Guide for A/UX", but this has no effect; the machine crashes the same way. Fortunately I saved a copy of the 1.0.1 kernel and I am just using that instead; it is highly robust. Running the new X11 with the old kernel is fine but just monochrome. Both the new kernel and the new X11 server are necessary to make X11 work in color. Questions: Are there others (including people at Apple) who have seen these bugs? Is there a fix in the works? This problem is pretty serious -- serious enough to make this whole upgrade unusable, since color X11 is the whole point of the upgrade, for my purpose. --Neil McKenzie (mckenzie@june.cs.washington.edu)
chuq@Apple.COM (Chuq Von Rospach) (05/09/89)
>I just installed the upgrade from A/UX 1.0.1 to A/UX 1.1 on my >4 Mbyte Mac IIx ('030 version of the MacII). There seems to be >a serious reliability problem with the new kernel running on my >computer. I can make it crash by simply doing the following from >the shell: > % rlogin localhost > ... > % exit >The machine then dies and prints the error message: > > panic: sbflush 2 I can't reproduce this. >I've also seen similar problems when I run the new (color supported) >X11, running innocuous programs like "plaid" from a remote client. >The machine bombs and various messages come up: > > Double panic: mget > > Kernel bus error (then a register dump) > Double panic: kernel memory management error I can't reproduce this, either. Both of these error messages are somewhat indicative of a sytem problem -- I would strongly suggest either a latent hardware problem (very possibly on the ethernet card) that got tickled by the upgrade or a corrupted kernel. Other possibilities might be a memory problem that passes self-test but occasionally flips a bit or a flakey 68030 chip. Since a 1.0.1 kernel works, I'd try building a new 1.1 kernel from scratch and see whether it works. It sounds like something was corrupted to me. Chuq Von Rospach =|= Editor,OtherRealms =|= Member SFWA/ASFA chuq@apple.com =|= CI$: 73317,635 =|= AppleLink: CHUQ [This is myself speaking. No company can control my thoughts.] Bookends. What a wonderful thought.
phil@Apple.COM (Phil Ronzone) (05/10/89)
In article <8150@june.cs.washington.edu> mckenzie@uw-june.UUCP (Neil McKenzie) writes: >I just installed the upgrade from A/UX 1.0.1 to A/UX 1.1 on my >4 Mbyte Mac IIx ('030 version of the MacII). There seems to be >a serious reliability problem with the new kernel running on my >computer. I can make it crash by simply doing the following from >the shell: > > % rlogin localhost > ... > % exit > >The machine then dies and prints the error message: > > panic: sbflush 2 > I can't reproduce any of these errors. I would suspect flakey hardware. I just had an old Mac II that has been running without poweroff or failure for almost two years -- and it started getty flakey. I finally took the cover off and "stroked" the SIMMs and wiggled all the connectors and reseated the PMMU. Now it behaves again. If the bug is repeatable, check your Ethernet card to see if it has all the revs on it. +-----------------------------------------------------------------------------+ |Philip K. Ronzone, Apple Computer, 10440 Bubb Rd, MS 58A, Cupertino, CA 95014| |{amdahl,decwrl,sun,voder,nsc,mtxinu,dual,unisoft,...}!apple!phil | +-----------------------------------------+-----------------------------------+ | All "IMHOs" disclaimed and copyrighted. | Self defense is a human right ... | +-----------------------------------------+-----------------------------------+
randy@violet.berkeley.edu (Randy Ballew) (05/10/89)
>>I've also seen similar problems when I run the new (color supported) >>X11, running innocuous programs like "plaid" from a remote client. >>The machine bombs and various messages come up: >> >> Double panic: mget >> >> Kernel bus error (then a register dump) >> Double panic: kernel memory management error > >I can't reproduce this, either. Both of these error messages are somewhat >indicative of a sytem problem -- I would strongly suggest either a latent >hardware problem (very possibly on the ethernet card) that got tickled by >the upgrade or a corrupted kernel. Other possibilities might be a memory >problem that passes self-test but occasionally flips a bit or a flakey 68030 >chip. > >Since a 1.0.1 kernel works, I'd try building a new 1.1 kernel from scratch >and see whether it works. It sounds like something was corrupted to me. I experienced a similar problem: MC68030, AUX 1.1, color X11, 8Mb, 80Mb external disk drive. Kernel was configured according to the instructions in "X Window System User's Guide for A/UX:" NBUF=1000 NINODE=600 NFILE=400 NPROC=100 MAXUP=100 The system was left running X and an xterm or two (one with an rlogin) over the weekend, on Monday it was crashed with a screenful of memory management error messages and references to XMacII. The machine was completely wedged, I was unable to bring up MacOS. Not being a MacWizard, I was stumped. Fortunately, I was able to locate our local Mac expert, who was able to boot the machine from a floppy. One unsolved mystery: we had to disconnect the SCSI cable before MacOS would come up. We were then able to do a sanity restore by "zapping the PRAM," and resetting all its parameters. The machine is running X again, I'm waiting for a repeat performance... any tips? Another problem: the A/UX version[s] of cpp seem unable to cope with certain X software. I'm trying to get our hacked up X11R3 HP widgets to compile (they're compiled and working on Sun 3.5 and 4.0, Ultrix 2.2 and 3.0). I switched to /usr/lib/big after /bin/cpp choked in <stdio.h> (!!). Here's where I'm hung now: randy@scrapple [73] % cd MACII randy@scrapple [74] % make TextEdit.o /bin/cc -O -c -I.. -DmacII -B /usr/lib/big/ ../TextEdit.c ../TextEdit.c: 433: token too long *** Error code 1 Stop. Line 433 is about half way through the static declaration of the widget's translations, which I'd rather not modify. Unless someone has a less painful solution, my next shot is to make a GNU cpp. Unfortunately, gcc version 1.35 has no configuration files for AUX. Before I reinvent a wheel, has anyone successfully made gcc for AUX? If so, any tips? Thanks in advance... Randy Randy Ballew randy@violet.berkeley.edu ucbvax!violet!randy