porphano@lehi3b15.csee.lehigh.edu (Paul Orphanos) (07/17/90)
Now that I got slip running (thanks to those who e-mailed with help), I'm having another problem. Slip is crashing my system (data panic) with ring and silo buffer overflows. As far as I can tell, the kernel is not servicing the serial interrupts fast enough to keep the hardware buffer from overflowing, and thus the data panics. This is the only reasonable explanation I can think of. Has anyone seen this before? I sent tracebacks to my sun rep two weeks ago, and I've gotten nothing back yet. I'm working on how to get slip to auto connect when the system crashes, but even on a light load (no user processes running, and a simple terminal session on the slip line), the system crashes. This makes it sort of dumb to keep reconnecting if it's going to crash many times a day. configuration: sparc 1, gx graphics, 19" color, 16mb ram, 64mb swap, 800mb disk, sunos 4.0.3c this is getting rediculous...any help is worshipped, paul orphanos porphano@lehi3b15.csee.lehigh.edu (a stable system) porphano@calvin.sfc.lehigh.edu (<--if my slip line would stay up)
"Mark_Weiser.PARC"@xerox.com (07/20/90)
I have seen this happen when when SLIP was running and I was using the raw console. Don't do that! Use a window system, or use one of the new console emulators recently mentioned in sun-spots. The serial lines and the raw console both use the forth interpreter in the sparcstation rom, and it can't keep up with both.
guy@uunet.uu.net (Guy Harris) (07/24/90)
>The serial lines and the raw console both use the forth interpreter >in the sparcstation rom, and it can't keep up with both. If that's true (specifically the part about the serial lines), somebody *quite majorly* rewhacked the CPU serial port driver, and not for any good reason I can think of....
jms@tardis.tymnet.com (Joe Smith) (07/28/90)
In article <9974@brazos.Rice.edu> porphano@lehi3b15.csee.lehigh.edu (Paul Orphanos) writes: >having another problem. Slip is crashing my system (data panic) with ring >and silo buffer overflows. As far as I can tell, the kernel is not >servicing the serial interrupts fast enough to keep the hardware buffer >from overflowing, and thus the data panics. This is the only reasonable >explanation I can think of. Our Sun-3 was having problems very similar to what your Sun-4 is having. In particular, I think you need the new heap_kmem and zs_async modules. Current Original Name 59245 8 31849 6 sun3/OBJ/heap_kmem.o 6449 24 42794 24 sun3/OBJ/machdep.o 54647 5 59441 5 sun3/OBJ/mcp.o 4825 16 50132 16 sun3/OBJ/mcp_async.o 4445 12 55088 12 sun3/OBJ/mti.o 37552 30 11819 30 sun3/OBJ/str_io.o 26187 3 25352 3 sun3/OBJ/tty.o 43549 7 64885 7 sun3/OBJ/uipc_mbuf.o 51155 12 36901 12 sun3/OBJ/zs_async.o The patches to the above files fixed our problems with "bus error 80" due to improper allocation of buffers at interrupt level when heavy traffic was present on the serial ports. This patch also fixed the problem with X.25 circuits getting hung. ----- Contents of README from the tape ---- serial I/O patch tape This tape contains all the changes to the serial drivers and streams code since SunOS 4.0.3 FCS. The bugs fixed with this patch are: 1020129 ALM-2 hangs system when encountering Framing Errors 1022156 ALM-2 parallel printer output purged by close 1022741 ALM-1 error codes incompletely reported 1024746 system with serial console watchdog resets 1025254 UUCICO can wedge port and eventually system 1025622 Panic bus error in streams close code Kernel object files are included for Sun-2, Sun-3, Sun-3X, and Sun-4 architectures. Note that the Sun-2 does not support the ALM-2, therefore no mcp driver files are provided. Files: ------ README install sun2/mti.o sun2/str_io.o sun2/zs_async.o sun3/mcp.o sun3/mcp_async.o sun3/mti.o sun3/str_io.o sun3/zs_async.o sun3x/mcp.o sun3x/mcp_async.o sun3x/mti.o sun3x/str_io.o sun3x/zs_async.o sun4/mcp.o sun4/mcp_async.o sun4/mti.o sun4/str_io.o sun4/zs_async.o -- Joe Smith (408)922-6220 | SMTP: jms@tardis.tymnet.com or jms@gemini.tymnet.com BT Tymnet Tech Services | UUCP: ...!{ames,pyramid}!oliveb!tymix!tardis!jms PO Box 49019, MS-C41 | BIX: smithjoe | 12 PDP-10s still running! "POPJ P," San Jose, CA 95161-9019 | humorous dislaimer: "My Amiga speaks for me."