There are no other files named ripoff.doc in the archive.
RIPOFF - Cusp-Level Disk Service Routine
G. Michael Uhler
University Computing Center
University of Arizona
March 25, 1978
RIPOFF - Cusp-Level Disk Service Routine Page 2
RIPOFF is intended to be a centralized repository for disk-related
maintenance utilities, and as an alternative to the DEC supplied CUSPs
such as DSKRAT, DSKLST, REDALL, DELFIL, etc. As an added bonus, it
appears to be 5 to 10 times faster than the equivalent DEC utilities
for most non-trivial functions. This document is written as an
explanation of the various functions that RIPOFF provides and is
intended to be read by experienced system personnel who are familiar
with the concepts of the TOPS10 file system.
RIPOFF was originally written by Steve Bush at the University of Texas
Health Sciences Center at Dallas. Since then, a massive rewrite has
been undertaken at the University of Arizona Computing Center. The
original goal was to simply add SFD support to the existing code.
However, it soon became apparent that numerous bugs must be fixed also
(see the revision history for details). We have been running the
current version of RIPOFF for several months with no reported problems
and feel that it is fairly stable. However, we urge every site to be
initially very cautious in the use of RIPOFF, since we cannot possibly
test it under every possible monitor and disk configuration. We
suggest that each function be verified on a scratch pack, preferably
using a DEC standard CUSP, before using that function in production.
For example, use DSKRAT to verify the /V and /S functions, DSKLST to
verify the /P function, etc.
RIPOFF - Cusp-Level Disk Service Routine Page 3
RIPOFF may be run with the following command:
Having found that the user is sufficiently priviliged ([1,2] or
running with JACCT), RIPOFF initializes a number of parameters and
requests the startup option from the user. The answer to this
question may be one of the following (minimal-segment-decoding is
QUICK - Don't ask about off-line devices
LONG - Full startup dialog
HELP - Type help message
The "LONG" response must be used if dismounted structures are to be
accessed. When using this option, any unit whose status indicates
that it does not have a pack logically mounted will result in a
message of the form:
Unit RPA1 has no pack mounted
Type YES to ignore error, NO to consider pack down
If the user replies with a "Y", RIPOFF will ignore the indicated
status and assume that the unit contains a spinning pack. An "N"
response will result in the unit being assigned a "DOWN" status.
When the internal unit data blocks are built for each "UP" unit in the
system, RIPOFF will type a "*" to indicate it's readiness to accept a
command. The command string is intended to tell RIPOFF the following
where the device may be a generic disk name (ALL, D, DS, DSK), a
structure name (DSKB), a specific logical unit in a structure (DSKB1),
a contoller type (RP), a specific controller (RPA), a specific unit
within a controller (RPA1), or even a pack ID (PRIV01).
The basic command string format is one of the following:
DEV:FILE.EXT[PATH]/WABC/XDEF... ; Comments
DEV:BLOCK1<BLOCK2(BLOCK3)/YGHI... ; Comments
DEV:FILE.EXT[PATH]=DEV:FILE.EXT[PATH]/ZJKL... ; Comments
RIPOFF - Cusp-Level Disk Service Routine Page 4
The default command string is ALL:*.*[*,*,*,*,*,*,*]. Any element of
the command string that is left out will also cause a "*" to be
assumed for that element.
Switches typed in the command string immediately following a "/" and
are always exactly one character (A-Z). This one character specifies
the general processor to call. Any following characters (A-Z, 0-9),
up until the next "/", ";", or line terminator, are switch options
whose interpretation is dependent on the switch in question.
In the first example above, the switches are "W" with the "A", "B",
and "C" options, and "X" with the "D", "E", and "F" options.
Similarly, in the second example, the switch is "Y" with the "G", "H",
and "I" options.
For commands where block numbers are relevant, the second format is
more likely (although parts of both can be readily mixed). The usual
interpretation of the second example is from BLOCK1 through BLOCK2
inclusive, by increments of BLOCK3 (e.g., see /R).
The third command format is used to indicate that a scratch area is
being used (e.g., see /I).
Several characters have special meaning in the command string. They
are as follows:
" Followed by a delimiter, ASCII characters, and another
delimiter, e.g., "/TEXT/
' Followed by a delimiter, SIXBIT characters, and another
delimiter, e.g., '/TEXT/. The single quote is also used to
indicate division within expressions. See below.
# A cluster number follows, i.e. #1234 represents cluster
$ A filename or extension follows. RIPOFF assumes that any
command string atom that begins with a digit is a number.
Therefore the $ must be used to "protect" filenames or
extensions that start with a number, e.g., LSX012.$01E
( Relative or incremental number follows.
) Matched with an opening "(", ignored.
* Used to indicate wildcards in the filename, extension, or
path. Also used to indicate multiplication within
expressions. See below.
, Separates octal half-words.
,, Separates octal half-words.
. Filename has preceded, extension (maybe null) follows.
/ Processor switch follows
: Device name preceded
; Terminates command scan, comments follow
< Initial block or cluster number preceded, final block or
cluster number follow.
> Same as <
== Word number preceded, new contents follow (see /EC)
= Scratch area file spec preceded
! Run preceding file
[ Path follows.
RIPOFF - Cusp-Level Disk Service Routine Page 5
] When matched with opening "[", ignored.
^ New radix (^D, ^O, or ^B) for next expression, then
In addition to the above characters, four (actually two new ones) more
should be considered. RIPOFF will accept an expression anywhere a
number is valid. Such an expression is evaluated strictly
left-to-right with no parenthesis allowed. Valid binary operators in
an expression are "+" (addition), "-" (subtraction), "*"
(multiplication), and "'" (division).
By default, RIPOFF writes it's output to device TTY:. However, if a
logical device LST:, exists, RIPOFF will use that instead. If it is a
directory device, the listing will go to the file RIP0.LST. If this
file already exists, RIPOFF will create RIP1.LST,...,RIP9.LST, until a
non-existent file is found. This insures that previous outputs (as
long as there aren't too many) are never lost.
Note that if the output device is not TTY:, the top of every page of
the listing file contains the RIPOFF version number, the current time
and date, the page number, and the command string that it is currently
executing. Thus, all listings show what command created them and
when. If a ";" appears in the command string, the rest of the command
is considered comments. These comments will also appear on the top of
each page, thereby further documenting the listing.
Each function is covered separately below.
RIPOFF - Cusp-Level Disk Service Routine Page 6
/A - Alphabetize directories
This switch causes the entries in the specified directories to be
sorted by filename, extension, or creation date/time. The /A options
are as follows:
/A - Same as /AF
/AF - Sort by filename, then extension
/AE - Sort by extension, then filename
/AT - Sort by creation date/time
/AM - Sort the MFD by PPN
/AX - (Xlist) When OR'ed with one of the above options, e.g.,
/AFX, the sorted directories are not listed on the TTY.
This is useful if a lot of directories are to be done,
since printing the listing slows down the whole process.
Note that on all sorts except /AM, the MFD is avoided. Conversely,
/AM only sorts the MFD.
This switch should be used with care. Do not try to sort a directory
when the structure is mounted and the user is logged in, since the
monitor may have part of the directory in core. Preferably, the
structure should be dismounted, (see /SL) but directories can probably
be sorted with the structure mounted if the user is not logged in.
RIPOFF will not allow the /AM switch to be used on a mounted
*DSKB:/A ; Sort all directories on DSKB except the
*DSKB:/AM ; Sort the MFD of DSKB. Note that this
; can never be done with the structure
*DSKB:[10,*]/AE ; Sort all the 10 project UFD's by
*DSKB:[10,12,RIPOFF,NEW]/AT ; Sort the specified directory by creation
RIPOFF - Cusp-Level Disk Service Routine Page 7
/C - Convert disk parameters
This switch accepts a block number, cluster number, CFP,
cylinder-surface-sector, or a universal date/time and converts them to
equivalent values. This is quite useful when one has one value and
needs to know the equivalent values. The /C options are as follows:
/CB - Convert from a structure block number
/CC - Convert from a structure cluster number. Note that this
switch is the only place in RIPOFF where the cluster number
is specified without the "#" preface, i.e., 1234/CC instead
/CD - Convert from a structure CFP
/CP - Convert from unit cylinder, surface, and sector
/CT - Convert from a universal date/time. This isn't exactly a
disk parameter but it's a useful conversion anyway.
/CU - Convert from a block number on a unit.
*DSKB:1234/CB ; Convert block 1234 on DSKB
*DSKB:102/CC ; Convert from cluster 102 on DSKB. Note
; that the cluster number has no "#"
; before it
*DSKB:102/CD ; Convert from CFP 102 on DSKB
*DSKB:1<16(10)/CP ; Convert from cylinder 1, surface 16,
; sector 10 on DSKB
*RPA1:1<16(10)/CP ; Do the same for physical unit RPA1
*124720,,345632/CT ; Convert from the specified universal
*RPA1:1234/CU ; Convert from block 1234 on unit RPA1
RIPOFF - Cusp-Level Disk Service Routine Page 8
/D - Delete files
This switch is used to delete files. RIPOFF is capable of deleting a
file through almost any error that may exist. The user has the option
of specifying monitor RENAME only, DELFIL-type delete only, or a
combination of the two in which RIPOFF will try the monitor RENAME
first and then the RIPOFF internal subroutine to delete the directory
entry if the RENAME fails. Note that the internal subroutine will not
release the SAT bits and therefore create lost blocks. The /D options
are as follows:
/D - Delete the file(s). (Try monitor RENAME first then
/DM - Delete with monitor RENAME only
/DR - Delete with internal subroutine only. (Don't even try
/DU - Delete all files in the specified UFD and then delete the
UFD. Note that this option is available with entire UFD's
only, not with SFD's.
/DN - Delete null directories
/DB - Delete only those files that are bad (RIB errors) in the
/DT - Asks for a creation date/time and an access date/time and
only deletes those files created before the first date/time
and not accessed since the second.
/DA - Prints each file to be deleted and asks for confirmation
before attempting to delete. May be OR'ed with all except
the /DU or /DN options.
*FILE.*[7,7]/D ; Delete ALL:FILE.EXT[7,7]
*DSKB:*.*[7,7]/D ; Deletes DSKB:*.*[7,7]. Note that if [7,7]
; contains any non-empty SFD's, the monitor
; RENAME will fail on the SFD and RIPOFF
; will zap the SFD with the internal
; subroutine. To avoid this, use either
; /DA or /DM
*DSKB:[7,7]/DU ; Delete all files in [7,7] and lower
; directories and then delete the UFD. This
; option will not create unnecessary lost
; blocks as in the above example. Note that
; in this case only, the path spec [7,7]
; really implies [7,7,*,*,*,*,*]
*DSKB:[7,7,FOO]/DB ; Delete all bad files in DSKB:[7,7,FOO]
*/DB ; Delete all bad files in the entire system
*DSKB:'/():-/.WRD[7,7,FOO]/D ; Delete weird file in [7,7,FOO]
RIPOFF - Cusp-Level Disk Service Routine Page 9
*DSKB:/DN ; Delete all null directories on DSKB
*DSKB:[10,*,*,*,*,*,*]/DN ; Delete all null directories in project 10
*DSKB:*.TMP/DT ; Delete all TMP files on DSKB that meet the
; date/time criteria
RIPOFF - Cusp-Level Disk Service Routine Page 10
/E - Edit disk blocks
This switch allows any disk block to be modified. This provides an
easy way to change parameters in a HOME block or fix a RIB that is
unacceptable to FILSER. One block at a time may be read into core,
examined, changed, and then written back out. The new contents of a
word may be typed in in any format that is acceptable to the command
scanner, i.e., ASCII, SIXBIT, octal, decimal, or binary. The type out
format is normally 12 octal digits, but may also be ASCII, SIXBIT, or
universal date/time. Note that it is possible to write the block onto
a completely different place from which it was read. However, it is
assumed that it will always be written back to the same place. If
this is not the case, a warning message is issued and the user is
forced to confirm the choice. A write is illegal without a previous
read. The /E options are:
/ER - Read the block into core
/EW - Write the block back to disk
/ERS - Read the same block as the last read/write
/EWS - Write the same block as the last read/write
/EC - Change the contents of a word in the block in core
/ET - Type the contents of a word in octal
/ETA - Type the contents of a word in ASCII
/ET7 - Same as /ETA
/ET6 - Type the contents of a word in SIXBIT
/ETU - Type the contents of a word as though it were in universal
/ETS - Type the same word as that specified in the last /ET or
/ETL - Type the word previous to the one specified in the last
/ET or /EC command.
/ETN - Type the word following the one specified in the last /ET
or /EC command.
Note that the S, L, and N switches may be OR'ed together with any
combination of the A, 7, 6, or U format.
*DSKB:4027/ER ; Read block 4027 of DSKB
*2/ET ; Type word 2 of the block in octal
*/ET6 ; Type same word in SIXBIT. Note that /ETx
; with no word number specified is the same
; as /ETSx
*/ETL6 ; Type word 1 in SIXBIT
*0<12/ET ; Type words 0 through 12 in octal
RIPOFF - Cusp-Level Disk Service Routine Page 11
*35/ETU ; Type word 35 as if it were in universal
; date/time format
*1=='/FOO//EC ; Change word 1 to SIXBIT/FOO/. Note that
; the // is intentional
*5==10/EC ; Change word 5 to an octal 10
*/EWS ; Write the block back to the place from
; which it came. (In this example, 4027
; on DSKB)
RIPOFF - Cusp-Level Disk Service Routine Page 12
/F - Find RIBs of files
This switch causes RIPOFF to find the RIB of a file and print the
relative block within the unit and the logical block within the file
structure of that RIB. RIPOFF first tries to LOOKUP the file and
print the information from the CFP. If this fails, or if the /FD
option is selected, the structure is searched cluster-by-cluster until
all RIBs matching the specifications have been found. The /F options
are as follows:
/F - Do directory search, then structure search
/FE - Do only directory search. The file exists.
/FD - Do only structure search; the file has been deleted.
/FD2 - Enable RIPOFF to find 2nd RIBs by forcing it to read every
block instead of every cluster.
*DSKB:FOO.BAZ[10,10]/FE ; Find the RIB of the specified file by doing
; only a directory search
*DSKB:FOO.BAZ[10,12,RIPOFF]/FE ; Same as above only in an SFD
*DSKB:FOO.BAZ[10,10]/FD ; Find the RIB by doing a structure search.
; Note that specifying more of a path than
; just the PPN with the /FD switch is futile
; since the monitor does not keep enough
; information in each RIB to tell which level
; the file belongs in.
RIPOFF - Cusp-Level Disk Service Routine Page 13
/H - Type RIPOFF help file
This switch causes the file RIPOFF.HLP to be typed on the terminal.
The file is looked for on HLP:, SYS:, and finally DSK:. RIPOFF.HLP
contains a summary of all switches and options that are acceptable to
RIPOFF. The /H options are as follows:
/H - Type the RIPOFF help file
*/H ; Find the help file and type it
RIPOFF - Cusp-Level Disk Service Routine Page 14
/I - Initialize files from RIBs alone
** Special Note **
This is the only RIPOFF function with which I am not satisfied. There
are inherent problems with recovering an entire structure containing
SFDs from the RIBs alone given the design of the TOPS10 file system.
There is simply not enough information in the RIB of a file to
determine at what level of nesting it existed. It is therefore quite
difficult to totally rebuild a structure with the SFD structure
intact. I will be looking at this problem in the next few months and
would appreciate any feedback from other users.
It is sometimes useful to rescue files which have been deleted or
lost. When a file is deleted, although it's block space on the disk
is marked free for use and directory pointers to that file are
deleted, the data itself remains physically on the disk until another
file is allocated over it, which may not be immediately. Therefore,
if action is taken quickly enough, it is possible to recover deleted
files. RIPOFF will search the entire file structure for RIBs matching
the filenames, extensions, and PPNs in the command string. Note that
I say "PPNs" instead of "PATHs" since the RIB only contains the PPN in
which the file existed, not the full path. When the entire structure
is passed over once, RIPOFF will transfer all files found to a scratch
area (disk or magtape). When all files have been transferred to the
scratch area, it may be rewound and the files restored to the original
structure (or any other structure for that matter).
The /I command string is of the third form in the examples given at
the beginning of this document. The scratch area device and file spec
are specified to the left side of the equal sign. The files to be
saved or restored are specified to the right side of the equal sign.
Note that if a number is typed in parenthesis within the scratch area
device and file spec, this will be interpreted as the number of
buffers to use for the device. The default is 15 decimal.
The scratch area is used for RIPOFF's internal needs. If it is a
magtape, RIPOFF will correctly handle the change of reels, if
necessary, on both the save and restore. If it is a disk file, there
must be enough space available to total the combined allocated sizes
of all files to be saved. RIPOFF will not delete the scratch area
disk file when it is through.
The /I options are as follows:
/I - Same as /ISR
/IS - Save specified files on scratch area
/IR - Restore specified files from scratch area
/IP - Print directory listing of scratch area
/ID - When OR'ed with /IS, causes RIPOFF to save only those files
that are free in the SATs.
/IE - When OR'ed with /IS, causes RIPOFF to save only those files
that are marked in the SATs.
/IA - Same as /IDE, i.e., save or restore files regardless of the
RIPOFF - Cusp-Level Disk Service Routine Page 15
/IT - Asks for date/time limits and only saves/restores files
created between these limits. May be OR'ed with any other
/I2 - Enable 2nd RIB recovery by forcing RIPOFF to search every
block for RIBs instead of just the first block of every
cluster. May be OR'ed with with all but /IF saves.
/IO - Allow overwritting. The possibility exists that another
file may have overwritten the file that you wanted before
RIPOFF got to it. However, if the amount of overwritten
data is small compared to the size of the file, it may be
worth getting a few clusters worth of garbage to save a
large file. Normally, RIPOFF will stop restoring the file
if it finds overwritten data since this is obviously a
security violation. However, if this option is selected,
RIPOFF will restore the file, overwritten data and all.
This option should never be used until a normal restore is
tried and fails.
/IX - Don't print every file saved/restored, only the
/IF - Failsafe mode. Don't use any RIB search logic, simply copy
all files from or to the scratch area. This function may
go away in the next release in favor of a strictly
pack-to-pack copy function.
*MTA1:=DSKB:/IS ; Save all (see special note above)
; files from DSKB on MTA1
; Restore FOO.MAC to [10,7] using
; DSKA:FOO.BAZ[1,2] as a scratch area
*MTA1:=DSKB:FOO.BAZ[10,7]/IR ; Restore FOO.BAZ to [10,7]. Note that
; MTA1 may contain a save of the entire
*DSKA:FOO.BAZ[1,2]=/IP ; Print directory listing of scratch area
RIPOFF - Cusp-Level Disk Service Routine Page 16
/L - Lock RIPOFF in core
This switch locks the job in core with the LOCK UUO. If the job
cannot be locked because of insufficient privileges, (shouldn't ever
happen in [1,2]) the user will be notified immediately. Otherwise,
RIPOFF will attempt a LOCK every 2 seconds for 16 seconds. At the end
of this period the error code will be given if the lock was
unsuccessful. Note that a number of routines in RIPOFF expand core.
Since a CORE UUO is illegal when locked, the job is first unlocked,
the core area expanded, then the job is relocked. If the relock
fails, a message will be typed and execution will continue unlocked.
The /L options are as follows:
/L - Lock job in core
/LU - Unlock job
*/L ; Lock job in core
*/LU ; Unlock job
RIPOFF - Cusp-Level Disk Service Routine Page 17
/P - Print information, blocks, and files
This switch allows the user to observe the disk system. Specifically,
the /P switch may be used to print disk statistics, SAT's, BAT's,
units, etc. There are two distinct subsets of the /P switch. The
first subset is similar to the DSKLST program and allows the user to
print various information about the disk system. The second subset
allows the user to print the contents of blocks or files in any of
several formats. Multiple switch options may be mixed within either
subset, but options from both subsets may not be mixed. The /P disk
list options are as follows:
/PB - Print the unit's BAT blocks in octal, then interpret the
block in readable format
/PE - Print error summary. This is similar to the /PF option,
except the file information is not listed. However, any
file that is found to have read errors or a bad RIB is
listed, and a summary is given of all disk errors,
including those listed in the RIBSTS word of the RIB.
/PF - Print file information. Produces a complete directory
listing of the specified disk by user directory. Specific
filenames, extension, or path specifications may be given
in the command string in which case only the files that
match the command string will be listed.
/PL - Print listing. This is similar to the /PF option in that
it prints the file information. However, it is used to
look for files and is therefore similar to the monitor
DIRECT command. Unlike the /PF option, null UFD's are not
listed. Thus the /PL option may be used to get a listing
of all occurances of a particular filename, extenstion,
etc. The /PL option may have another function. A command
of the form DEV:FILE.EXT[path](len) will only print those
files whose allocated length is greater than "len".
/PP - Print performance summary. This option histograms the file
length in blocks and the number of RIB pointers in each
file. The second histogram will give a breakdown on how
fragmented the structure is, and therefore how badly
refreshing is needed.
/PQ - Print quick. This is similar to the /PL option except no
information beyond the filename, extension and path is
printed. Therefore, it is similar to the DIRECT/F monitor
/PS - Print SAT blocks. This option print the SAT blocks of all
units specified in octal and gives the total remaining
space in each block.
/PU - Print physical units for each pack in a structure.
/PV - Print vital statistics. This option prints the HOME block
of every specified unit in octal and then interprets each
interesting entry into readable format.
The block list subset consists of the following options:
/PA - Print in ASCII format. This option prints the specified
blocks/files in ASCII and is very similar to the monitor
RIPOFF - Cusp-Level Disk Service Routine Page 18
TYPE or PRINT commands.
/PD - Print in directory format. This option prints the
specified block as if it were a directory. The first word
of each word pair is listed as a SIXBIT filename, the left
half of the second word as a SIXBIT extension, and the
right half is interpreted as a CFP.
/PO - Print in octal. This option prints the specified
blocks/files in octal with a header for each block giving
the block number.
/PR - Print as a RIB. This option prints the specified block as
if it were a RIB. In addition, if a filename is given in
the command string, the RIB of the specified file is
printed. This option prints all non-zero information in
the RIB preamble in a readable format (much like
DIRECT/DETAIL) and then interprets and prints the retrieval
pointers from the RIB.
/P6 - Print in SIXBIT. This option prints the specified
blocks/files in SIXBIT with a header for each block.
/P7 - Print in ASCII. This option prints the specified
blocks/files in ASCII with a header for each block. It is
very similar to the /PA option except for the inclusion of
the header information.
*DSKB:FOO.BAZ[10,7,FOO]/PR ; Print the RIB of the specified file
*DSKB:1234/PR ; Print the specified block as a RIB
*DSKB:1234<1234+5/PO ; Print blocks 1234 through 1241 in octal
*DSKB:FOO.*/PL ; Find and print all files named FOO.*
; on DSKB
*10,7.UFD/PD ; Prints ALL:[10,7] in a readable format
*FOO.SFD[10,7]/PD ; Prints ALL:[10,7,FOO] directories in a
; readable format
*/PV ; Print vital statistics for all units
; on the system
*DSKB:/PB ; Print the BAT blocks on DSKB
RIPOFF - Cusp-Level Disk Service Routine Page 19
/R - Read disk blocks
This switch simply reads all blocks specified in the command string
and reports any errors. This function is similar to the REDALL CUSP.
The /R options are as follows:
/R - Read specified blocks
*DSKB:1234<4567(10) ; Read blocks 1234 through 4567 skipping 10
; blocks between each read
*DSKB:/R ; Read all blocks on DSKB (0 through highest
; block on unit using 1 as an increment)
*DSKB:40000/R ; Read blocks 40000 through the highest
; block on the unit using 1 as an increment
RIPOFF - Cusp-Level Disk Service Routine Page 20
/S - Manipulate SAT's
This switch allows the user to observe and manipulate the SAT blocks
on a structure. Since the monitor keeps SATs in core for every
mounted structure and rewrites these SATs often, RIPOFF will not allow
SATs to be written back to a mounted structure. The /S options are as
/SL - Lock up a structure in preparation for removing it. The
specified structure is first locked using the .FSLOK STRUUO
function and then removed with the .FSREM STRUUO function.
The user may specify the number of seconds between the
.FSLOK and the .FSREM by placing the desired number in the
command string. The default value is 60 seconds.
/SR - Read the SATs of the specified structure into core. If you
plan to change the SATs, do a /SL first.
/SW - Write the in-core copy of the SATs back out. Note that
this may not be done unless the structure is dismounted.
/SP - Print the SAT blocks of the specified structure. Produces
the same type of listing as /PS
/ST - Type the status of the SAT bits specified by the block or
cluster numbers given in the command string.
/SF - Free the SAT bits specified by the block or cluster numbers
given in the command string. Note that this only changes
the in-core copy of the SATs.
/SM - Mark the SAT bits specified by the block or cluster numbers
given in the command string. Note that this only changes
the in-core copy of the SATs.
*DSKB:/SL ; Lock up DSKB and remove it 60 seconds
*DSKB:5/SL ; Lock up DSKB and remove it 5 seconds
*DSKB:/SR ; Read in the SATs
*DSKB:1234/ST ; Type the status of the cluster containing
; block 1234
*DSKB:#100<#200/ST ; Type the status of clusters 100 through 200
*DSKB:#1234/SM ; Mark cluster 1234 in the SAT
*DSKB:#100/SF ; Free cluster 100 in the SAT
*DSKB:/SW ; Write the SATs back out to DSKB
RIPOFF - Cusp-Level Disk Service Routine Page 21
/U - Create specified UFDs/SFDs
This switch performs a function similar to the CREDIR CUSP in that it
creates the directories specified in the command string. Quotas will
all be set to infinity and if a directory already exists, no attempt
will be made to create over the top of it. Note that this switch uses
monitor ENTERs to create the specified directories and as a result,
requires that the specified structure be mounted. The /U options are
/U - Create the specified directories
*DSKB:[10,12,RIPOFF,NEW]/U ; Create all non-existent directories
; in the command string
*DSKB:[10,7]/U ; Create the [10,7] UFD if it doesn't exist
RIPOFF - Cusp-Level Disk Service Routine Page 22
/V - Verify files
This switch verifies the integrety of the files specified in the
command string. The RIBs of every file are checked, each file is
checksummed and compared with the checksum in the RIB, the
corresponding SAT bits are checked to insure that they are marked, and
at least one block of every group is read and checked for hardware
If *.*[*,*,*,*,*,*,*] is specified in the command string, or left out
and assumed by default, the disk SATs are read into core, and RIPOFF
builds it's own SATs as each file is read. These two SATs are then
compared (similar to DSKRAT) and any clusters which are multiply used,
free, or lost are reported. In addition, any such errors may be fixed
by rewriting RIPOFF's copy of the SAT's back to the disk. This
function may only be done if the structure is dismounted. The /V
options are as follows:
/V - Verify the specified files
/VA - Read all blocks. This option forces RIPOFF to read all
blocks of the files specified in the command string and
report any which are hardware unreadable. Also, if all
files were specified, RIPOFF will read all other blocks on
the disk that were not contained in a file. This is
similar to a DSKRAT followed by a REDALL.
/VQ - Quick option. This option reads no more blocks than are
absolutely necessary to do the job. /V normally reads at
least the first block of every group; /VA reads all blocks
of the file; /VQ does neither.
/VF - Fix the SATs. If all files are specified, the RIPOFF copy
of the SATs will be written back to the disk, thus fixing
all errors. Note that the F option may also be combined
with /VA and /VQ. RIPOFF will not allows this option to be
selected if the structure is mounted.
*DSKB:FOO.BAZ[10,7,FOO]/V ; Verify the integrity of the specified
*DSKB:FOO.BAZ[10,7,FOO]/VA ; Verify and read all blocks of the file
*DSKB:/VF ; Verify all files and fix the SATs. DSKB
; may not be mounted if the F option is used
*DSKB:/VQF ; Faster than the above example but will
; not catch checksum errors
RIPOFF - Cusp-Level Disk Service Routine Page 23
/W - Do word searches
This switch allows the user to search for a particular pattern in a
range of disk blocks or files. A mask word may be specified that is
used to mask each word that is a candidate before the comparison with
the word being searched for. The /W options are as follows:
/WM - Set search mask for the search
/WS - Start the search
/WT - Type current values of the search mask and search word
/WW - Set word to be searched for
*777777/WM ; Set mask for the right-half of the
*606060/WW ; Set word to be searched for to 0,,606060
*/WT ; Type current values of mask and word
*DSKB:100<500/WS ; Search blocks 100 through 500 on DSKB for
; the specified pattern masked with the mask
*DSKB:FOO.BAZ[10,7]/WS ; Search the specified file
RIPOFF - Cusp-Level Disk Service Routine Page 24
/X - Exit and close listing files
This switch closes all listing files and releases all channels before
exiting to the monitor. Note that if the user types a ^C to RIPOFF
when the listing is being written to a file, RIPOFF will ask if the
user wishes to close the listing file before exiting. This is
intended to avoid losing the listing if the user forgets to use /X to
exit. The /X options are as follows:
/X - Exit to the monitor
/XQ - Exit and run QUEUE
*/X ; Exit to the monitor
*/XQ ; Exit and run QUEUE
RIPOFF - Cusp-Level Disk Service Routine Page 25
RIPOFF error messages
A large number of warning or error messages can result from an invalid
or non-applicable command string (/LU without a previous /L, /EW when
no data has been /ER'ed, etc.). These messages are rather explicit in
meaning and need not be discussed at length.
As a protection against inadvertently wiping out file structures, two
extra precautions are taken. If an OUTPUT UUO is attempted at any
time, a warning of the following form is typed:
If answered by a "N", RIPOFF will stop immediately and exit. A "Y"
will allow the output to continue. This question will only be asked
once for every command string, i.e., the write enable is good for one
and only one command string.
In addition, if any attempt is made to alter the status of files from
[1,1], [1,4], [1,2], [2,5], or [10,1], the following message occurs:
Access files from [path]?
This insures that the fumble-fingered user doesn't accidentally
destroy a structure. Additional PPNs may be added by adding them to
the table VIPS.
Hardware and software errors are reported to the listing file. Note
that the entire listing file (if not TTY) is tabbed over one tab from
the left margin. All error messages are not, and are thus readily
identified. All RIPOFF disk I/O falls through one major low-level
subroutine, BLKRED/BLKWRT. At this level, a complete summary of all
hardware error conditions is given as in the following example:
FOO.BAZ[10,7,FOO] Read error on RPA0, block 1234
= Cylinder 1 surface 16 sector 10
Status = 400000,,140000 = IO.BKT+IO.DTE
Coni = 200014 (Exception)+(*Done*)+(PI channel=4)
The right half status bits are from the monitor GETSTS UUO. The left
half bits are internally defined and are as follows:
IO.FAC Bit 0 File is active on RIPOFF software channel
IO.CKS Bit 1 File has a software checksum error
IO.WRT Bit 2 File is being written on this channel
The CONI word is obtained from the DEVSTS UUO. All bits of the status
and CONI words are interpreted into English. Stars around phrases in
the CONI interpretation indicate that the bit causes an interrupt.
Software errors result in messages similar to the hardware errors
except the CONI word is suppressed since the hardware status is not at
fault. The only software-type errors that currently exist are RIB
RIPOFF - Cusp-Level Disk Service Routine Page 26
error and software checksum error.