I wrote:
We've started running `gmake -j 4' on our 4M/690 to take advantage of
all 4 processors. Unfortunately this seems to crash this system
regularly with this error:
pid 24565, `gmake': Data fault
kernel read fault at addr=0x63707074, pme=0x0
MMU sfsr=126: Invalid Address on supv data fetch at level 1
We've applied these patches to no avail:
100475-01: SunOS 4.1.2: mmap system call on galaxy causes BAD TRAP
100495-01: SunOS 4.1.2: asynch I/O on a sun4m machine causes panics
100542-01: SunOS 4.1.2: MP IPI file system corruption
100575-01: SunOS 4.1.2: MP machines do not perform as well as 4/4XX equivalent
Has anyone else seen this problem?
Thanks to John Lees <lees@pixel.cps.msu.edu> for a very helpful
response. He told me that most of their 690MP problems have been with
their IPI controllers and NC400 network interfaces. Guess what?
That's what we have! Now I've applied these patches:
100475-01: SunOS 4.1.2: mmap system call on galaxy causes BAD TRAP
100495-01: SunOS 4.1.2: asynch I/O on a sun4m machine causes panics
100542-01: SunOS 4.1.2: MP IPI file system corruption
100542-03: SunOS 4.1.2: IPI - Galaxy jumbo patch
100566-01: SunOS 4.1.2: system panics with datafault when using nc400
100575-02: SunOS 4.1.2: MP machines do not perform as well as 4/4XX equivalent
100342-02: SunOS 4.1;4.1.1;4.1.2: NIS client needs long recovery time if server reboots
And we haven't crashed since. I also finally heard from Sun about our
problem, they said that the 100566 patch was the one we really needed.
I still have problems with GNU make dying (sorry for not being clear
earlier, GNU make is what I meant when I said `gmake'), but it doesn't
crash the system anymore.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:43 CDT