bossert@dataio.UUCP (John Bossert) (04/08/85)
I've not seen any discussion on this subject for quite some time. Has
anyone fixed this bug on 750's/4.2bsd? Thanks much.
--
John Bossert
Data I/O Corporation
Redmond, WA
uw-beaver!entropy!dataio!{bossert,usenet}eriks@yetti.UUCP (Eriks Rugelis) (04/11/85)
In article <632@dataio.UUCP> bossert@dataio.UUCP (John Bossert) writes: >I've not seen any discussion on this subject for quite some time. Has >anyone fixed this bug on 750's/4.2bsd? Thanks much. >-- > John Bossert > Data I/O Corporation > Redmond, WA > uw-beaver!entropy!dataio!{bossert,usenet} The USENIX 84.2 tape, that contains nothing but bug-fixes from UCB also contains an update that claims to take care of this panic. I'm not convinced that it has the desired effect (since our 'tbuf par fault' panics were reducing in frequency anyway (ANYONE know why?). The installation notes that come with the hardware rev. 7 kit also make reference to this problem and seem to imply that rev. 7 fixes it. We haven't yet had rev. 7 installed so again I have no conclusive evidence. Get DEC to let you read the rev. 7 installation manual and if you have a source license, get USENIX to send you the tape. Hopefully, between the two of them this problem gets resolved. P.S. RE: the earlier posting about 'how to load rev. 7 micro-code in UNIX' Chalk me up as another interested party. I wanna know!! In the rev. 7 kit notes DEC writes that ULTRIX machines should first load the diagnostic supervisor before starting ULTRIX. UUGGHHH!!! Is there SOMEBODY out there that knows which registers to hit, and how to read the micro-code file? Eriks Rugelis USENET: {decvax,linus,ihpn4,allegra}!utzoo!yetti!eriks BITNET: eriks@yuorion
brown@nic_vax.UUCP (04/11/85)
> I've not seen any discussion on this subject for quite some time. Has > anyone fixed this bug on 750's/4.2bsd? Thanks much. > -- > John Bossert > Data I/O Corporation > Redmond, WA > uw-beaver!entropy!dataio!{bossert,usenet} I thought that our 4.2bsd was fixed when we got it, because we looked at the source code and it was already in there. I looked and it is fixed. Our VAX used to tbuf parity fault lots of times with 4.1bsd. We added the DEC unix patch and everything was fine. The file is /sys/vax/machdep.c versions 6.1 (83/08/20) or 6.2 (83/10/02). In version 6.2 the code is at line 811. If you are still having system crashes because of tbuf parity faults, check your source (if you have it) to make sure you have the latest version. If not, get it from someone so that you can recompile the kernal. Mr. Video
lmc@denelcor.UUCP (Lyle McElhaney) (04/13/85)
> In article <632@dataio.UUCP> bossert@dataio.UUCP (John Bossert) writes: > >I've not seen any discussion on this subject for quite some time. Has > >anyone fixed this bug on 750's/4.2bsd? Thanks much. > Chalk me up as another interested party. I wanna know!! In the rev. 7 kit > notes DEC writes that ULTRIX machines should first load the diagnostic > supervisor before starting ULTRIX. UUGGHHH!!! Is there SOMEBODY out > there that knows which registers to hit, and how to read the micro-code file? To reinterate what has gone before, the Rev 7 update for the 750 does some fixing of the tbuf parity fault problem; I have been warned by our CE, however, that some UNIX sites have been seeing cache parity errors after Rev 7 is installed. We had it done a month ago; no crashes yet. The Rev 7 update is hardware and new microcode (Rev 99). In order to make the microcode work, part of the Rev 7 update is to install writable control store, which must of course be loaded. Jim McKie has put together a package containing a new version of boot which knows how to load the microcode, a uuencoded file of the microcode patches, and a fixed standalone uda driver. It worked for us like a charm. If you need it, just ask. Lyle McElhaney -- Lyle McElhaney {hao, stcvax, brl-bmd, nbires, csu-cs} !denelcor!lmc
zemon@fritz.UUCP (Art Zemon) (04/24/85)
This is supposedly solved by upgrading to Rev 7. You also need
to get the "right" L003 board as part of the Rev 7 kit. "Right"
is defined as having the proper gate array chip manufacturer.
There is also supposedly an "official" channel on the East Coast
that your DEC rep can contact to get a "right" L003 board.
Good luck,
--
-- Art Zemon
FileNet Corp.
...! {decvax, ihnp4, ucbvax} !trwrb!felix!zemon