.TH CRASH 8 VAX-11
.UC
.tr |
.SH NAME
crash \- what to do when the system crashes
.SH DESCRIPTION
This section gives at least a few clues about how to proceed if the
system crashes.
It can't pretend to be complete.
.PP
.I "What to do first.||"
(Someday the LSI-11 will do this automatically.)
If the reason for the crash is not evident
(see below for guidance on `evident')
you may want to try to dump the system if you feel up to
debugging.
At the moment a dump can be taken only on magnetic tape.
Before you do anything, be sure that a clean tape is mounted with a ring-in
on the tape drive if you plan to make a dump.
.PP
Write the date and time on the console log.
Use the console commands to examine the registers, program status long word,
and the top several locations on the stack.
A suggested command sequence, which is executed by the ``@DUMP''
console command script, is:
.DS
.nf
	E PSL<return>
	E R0/NE:F<return>
	E SP<return>
	E/V @ /NE:40<return>
.fi
.DE
If hardware problems dictate a special set of commands be executed when
the system crashes, a sequence of commands can be saved using the console
command ``LINK'' to be reexecuted with ``PERFORM'' (which can be
abbreviated ``P'').
If a dump is to be taken on magnetic tape (this is a good idea
in most any case where the cause of the crash is not immediately obvious)
then the following commands will (should) be executed:
.DS
.nf
	D PSL 0<return>
	D PC 80000200<return>
	C<return>
.fi
.DE
These commands are actually part of the standard ``@DUMP'' script.
This should write a copy of all of memory
on the tape, followed by two EOF marks.
Caution:
Any error is taken to mean the end of memory has been reached.
This means that you must be sure the ring is in,
the tape is ready, and the tape is clean and new.
.PP
If there are not 40 locations active on the kernel stack when the
procedure is begun, then the console may begin to print error diagnostics.
You can stop this by hitting ``^C'' (control-C), and then give the
last three commands above.
.PP
If the dump fails, you can try again,
but some of the registers will be lost.
See below for what to do with the tape.
.PP
.I "How to bring it back up.||"
To restart after a crash, follow the directions in
.IR bproc (8);
if the virtual memory subsystem is suspected as the cause of the crash,
then a version of the system other than ``vmunix'' should be booted;
this version of the system will leave the paging areas temporarily intact
for use by the post-mortem analysis program
.I analyze.
On Ernie Co-vax at UCB, the backup system is ``unix''.
When this system is running, check its root file system (currently
.I /dev/rrm0a
but this is likely to change), and then read the core tape into the
file
.I /vmcore,
with the command:
.IP
cp /dev/mt0 /vmcore
.LP
With the system still in single-user mode, run the analysis program
.I analyze,
i.e.:
.IP
analyze \-s /dev/drum /vmcore /vmunix
.LP
and save the output.
Be sure to boot up
``vmunix''
before coming up multi-user.
.PP
Do a
.I sync,
and proceed to check and fix all file systems,
performing a
.I dcheck
and
.IR  icheck (1)
on all file systems which could have been in use at the time
of the crash.
If any serious file system problems are found, they should be repaired.
When you are satisfied with the health of your disks,
log out by typing an EOT (<control-D>).
The command sequence in /etc/rc will be executed and the system will
be in multi-user mode.
.PP
To even boot \s8UNIX\s10 at all,
three files (and the directories leading to them)
must be intact.
First,
the initialization program
.I /etc/init.vm
must be present and executable.
If it is not,
then the cpu will loop at location 0x13.
For
.I init.vm
to work correctly,
.I /dev/console
and
.I /bin/sh
must be present.
If either does not exist,
the symptom is best described
as thrashing.
.I Init
will go into a
.I fork/exec
loop trying to create a
Shell with proper standard input and output.
.PP
If you cannot get the system to boot,
a runnable system must be obtained from
a backup medium.
The root file system may then be doctored as
a mounted file system as described below.
If there are any problems with the root
file system,
it is probably prudent to go to a
backup system to avoid working on a
mounted file system.
.PP
.I "Repairing disks.||"
The first rule to keep in mind is that an addled disk
should be treated gently;
it shouldn't be mounted unless necessary,
and if it is very valuable yet
in quite bad shape, perhaps it should be dumped before
trying surgery on it.
This is an area where experience and informed courage count for much.
.PP
The problems reported by
.I icheck
typically fall into two kinds.
There can be
problems with the free list:
duplicates in the free list, or free blocks also in files.
These can be cured easily with an
.I "icheck \-s."
If the same block appears in more than one file
or if a file contains bad blocks,
the files should be deleted, and the free list reconstructed.
The best way to delete such a file is to use
.IR  clri (1),
then remove its directory entries with
.IR rm (1).
(Do not use
.IR mv (1).)
If any of the affected files is really precious,
you can try to copy it to another device
first.
.PP
.I Dcheck
may report files which
have more directory entries than links.
Such situations are potentially dangerous;
.I clri
discusses a special case of the problem.
All the directory entries for the file should be removed.
If on the other hand there are more links than directory entries,
there is no danger of spreading infection, but merely some disk space
that is lost for use.
It is sufficient to copy the file (if it has any entries and is useful)
then use
.I clri
on its inode and remove any directory
entries that do exist.
.PP
Finally,
there may be inodes reported by
.I dcheck
that have 0 links and 0 entries.
These occur on the root device when the system is stopped
with pipes open, and on other file systems when the system
stops with files that have been deleted while still open.
A
.I clri
will free the inode, and an
.I "icheck -s"
will
recover any missing blocks.
.PP
.I "Why did it crash?||"
UNIX types a message
on the console typewriter when it voluntarily crashes.
Here are some of the possible panic messages,
with enough information to provide
a hope at least of the remedy.
The message has the form `panic: ...',
`Trap from kernel mode', or `ILL I/E VEC'
(possibly accompanied by other information).
In rare cases the system will ``panic'' but the console message
will not appear; if this happens, you can trace the message easily
through the variable
.I panicstr
in the system.
Left unstated in all cases
is the possibility that hardware or software
error produced the message in some unexpected way.
.HP 5
blkdev
.br
The
.I getblk
routine was called with a nonexistent major device as argument.
Definitely hardware or software error.
.HP 5
devtab
.br
Null device table entry for the major device used as argument to
.I getblk.
Definitely hardware or software error.
.HP 5
iinit
.br
An I/O error reading the super-block for the root file system
during initialization.
.HP 5
out of inodes
.br
A mounted file system has no more i-nodes when creating a file.
Sorry, the device isn't available;
the
.I icheck
should tell you.
.HP 5
no fs
.br
A device has disappeared from the mounted-device table.
Definitely hardware or software error.
.HP 5
no imt
.br
Like `no fs', but produced elsewhere.
.HP 5
no inodes
.br
The in-memory inode table is full.
Try increasing NINODE in param.h.
Shouldn't be a panic, just a user error.
.HP 5
IO error in swap
.br
An unrecoverable I/O error during a swap.
Really shouldn't be a panic.
.HP 5
out of swap
.br
A program needs to be swapped out, and there is no more swap space.
It has to be increased.
This really shouldn't be a panic.
.HP 5
out of text
.br
A pure procedure program is being executed,
and the table for such things is full.
This shouldn't be a panic.
.HP 5
trap from kernel mode
.br
An unexpected trap has occurred within the system.
The trap type can be determined by examining the top word of the
stack (the trap type) with the console commands.
The trap types are:
.TP 10
0
reserved addressing mode
.br
.ns
.TP 10
1
privileged instruction
.br
.ns
.TP 10
2
BPT
.br
.ns
.TP 10
3
XFC
.br
.ns
.TP 10
4
reserved operand
.br
.ns
.TP 10
5
CHMK (system call)
.br
.ns
.TP 10
6
arithemtic trap
.br
.ns
.TP 10
7
reschedule trap (software level 3)
.br
.ns
.TP 10
8
segmentation fault
.br
.ns
.TP 10
9
protection fault
.br
.ns
.TP 10
10
trace pending (TP bit)
.HP 5
ILL I/E VEC, HALTED AT xx
.br
an illegal interrupt or exception has occurred.  The possible addresses are
.ns
.TP 10
4
machine check (hardware error).
.br
.ns
.TP 10
8
kernel stack not valid
.br
.ns
.TP 10
C
power failure
.PP
In some of these cases it is
possible for octal 20 to be added into the trap type;
this indicates that the processor was in user mode when the trap occurred.
If you wish to examine the stack after such a trap,
either dump the system, or use the console to examine memory;
the required address mapping is described below.
.PP
There are also a large number of panics if internal consistency
checks in the paging subsystem fail.  These can be caused by hardware
(e.g. if disk or tape problems cause a data structure to be mutilated)
but are most often caused by software problems.
Refer to a system listing to locate these and other panics not discussed above.
.PP
.I "Interpreting dumps.||"
All file system problems
should be taken care of before attempting to analyze dumps.
As mentioned above, the dump tape should be read into the file
.IR /vmcore ;
.IR  cp (1)
will do.
At this point, you should execute
.I "ps \-alxk"
and
.I who
to print the process table and the users who were on
at the time of the crash.
Use
.IR adb (1)
to examine
.IR /vmcore .
The location
.I dumpstack\-80000000
is the bottom of a stack onto which were pushed the stack pointer
.BR sp ,
.B PCBB
(containing the physical address of a
.IR u_area ),
.BR MAPEN ,
.BR IPL ,
and registers
.BR r13 \- r0
(in that order).
.BR r13 (fp)
is the system frame pointer and the stack is used in standard
.B calls
format.  Use
.IR  adb (1)
to get a reverse calling order.
In most cases this procedure will give
an idea of what is wrong.
A more complete discussion
of system debugging is impossible here.
See, however,
.IR analyze (1)
for some more hints.
.SH "SEE ALSO"
analyze(1m), clri(1), icheck(1), dcheck(1), bproc(8)
.br
.I "VAX 11/780 System Maintenance Guide"
for more information about machine checks.
.SH BUGS