aboutsummaryrefslogtreecommitdiff
path: root/doc/sg_write_long.8
diff options
context:
space:
mode:
Diffstat (limited to 'doc/sg_write_long.8')
-rw-r--r--doc/sg_write_long.8176
1 files changed, 176 insertions, 0 deletions
diff --git a/doc/sg_write_long.8 b/doc/sg_write_long.8
new file mode 100644
index 00000000..2d5560d7
--- /dev/null
+++ b/doc/sg_write_long.8
@@ -0,0 +1,176 @@
+.TH SG_WRITE_LONG "8" "January 2016" "sg3_utils\-1.42" SG3_UTILS
+.SH NAME
+sg_write_long \- send SCSI WRITE LONG command
+.SH SYNOPSIS
+.B sg_write_long
+[\fI\-\-16\fR] [\fI\-\-cor_dis\fR] [\fI\-\-help\fR] [\fI\-\-in=IF\fR]
+[\fI\-\-lba=LBA\fR] [\fI\-\-pblock\fR] [\fI\-\-verbose\fR]
+[\fI\-\-version\fR] [\fI\-\-wr_uncor\fR] [\fI\-\-xfer_len=BTL\fR]
+\fIDEVICE\fR
+.SH DESCRIPTION
+.\" Add any additional description here
+.PP
+Send the SCSI WRITE LONG (10 or 16 byte) command to \fIDEVICE\fR. The buffer
+to be written to the \fIDEVICE\fR is filled with
+.B 0xff
+bytes or read from the \fIIF\fR file. This buffer includes the logical
+data (e.g. 512 bytes) and the ECC bytes.
+.PP
+This utility can be used to generate a MEDIUM ERROR at a specific logical
+block address. This can be useful for testing error handling. Prior to
+such a test, the
+.B sg_dd
+utility could be used to copy the original contents of the logical
+block address to some safe location. After the test the
+.B sg_dd
+utility could be used to write back the original contents of the
+logical block address. An alternate strategy would be to read the "long"
+contents of the logical block address with
+.B sg_read_long
+utility prior to testing and restore it with this utility after testing.
+.PP
+.B Take care:
+If recoverable errors are being injected (e.g. only one or a few bits
+changed so that the ECC is able to correct the data) then care should
+be taken with the settings in the "read write error recovery" mode page.
+Specifically if the ARRE (for reads) and/or AWRE (for writes) are set
+then recovered errors will cause the lba to be reassigned (and the old
+location to be added to the grown defect list (PLIST)). This is not easily
+reversed and uses (one of the finite number of) the spare sectors set
+aside for this purpose. If in doubt it is probably safest to clear the
+ARRE and AWRE bits. These bits can be checked and modified with the
+sdparm utility. For example: "sdparm \-c AWRE,ARRE /dev/sda" will clear
+the bits until the disk is power cycled.
+.PP
+In SBC\-4 revision 7 all uses of SCSI WRITE LONG (10 and 16 byte) commands
+were made obsolete apart from the case in which the WR_UNCOR bit is set.
+The SCSI READ LONG (10 and 16 byte) commands were made obsolete in the
+same revision.
+.SH OPTIONS
+Arguments to long options are mandatory for short options as well.
+.TP
+\fB\-S\fR, \fB\-\-16\fR
+send a SCSI WRITE LONG (16) command to \fIDEVICE\fR. The default action (in
+the absence of this option) is to send a SCSI WRITE LONG (10) command.
+.TP
+\fB\-c\fR, \fB\-\-cor_dis\fR
+sets the correction disabled (i.e 'COR_DIS') bit. This inhibits various
+other mechanisms such as automatic block reallocation, error recovery
+and various informational exception conditions being triggered.
+This bit is relatively new in SBC\-3 .
+.TP
+\fB\-h\fR, \fB\-\-help\fR
+output the usage message then exit.
+.TP
+\fB\-i\fR, \fB\-\-in\fR=\fIIF\fR
+read data (binary) from file named \fIIF\fR and use it for the SCSI WRITE
+LONG command. If \fIIF\fR is "\-" then stdin is read. If this option is
+not given then 0xff bytes are used as fill.
+.TP
+\fB\-l\fR, \fB\-\-lba\fR=\fILBA\fR
+where \fILBA\fR is the logical block address of the sector to overwrite.
+Defaults to lba 0 which is a dangerous block to overwrite on a disk that is
+in use. Assumed to be in decimal unless prefixed with '0x' or has a
+trailing 'h'. If \fILBA\fR is larger than can fit in 32 bits then the
+\fI\-\-16\fR option should be used.
+.TP
+\fB\-p\fR, \fB\-\-pblock\fR
+sets the physical block (i.e 'PBLOCK') bit. This instructs \fIDEVICE\fR
+to use the given data (unless \fI\-\-wr_uncor\fR is also given) to write
+to the physical block specified by \fILBA\fR. The default action
+is to write to the logical block corresponding to the given lba.
+This bit is relatively new in SBC\-3 .
+.TP
+\fB\-v\fR, \fB\-\-verbose\fR
+increase the degree of verbosity (debug messages).
+.TP
+\fB\-V\fR, \fB\-\-version\fR
+output version string then exit.
+.TP
+\fB\-w\fR, \fB\-\-wr_uncor\fR
+sets the "write uncorrected" (i.e 'WR_UNCOR') bit. This instructs the
+\fIDEVICE\fR to flag the given lba (or the physical block that contains it
+if \fI\-\-pblock\fR is also given) as having an unrecoverable error
+associated with it. Note: no data is transferred to \fIDEVICE\fR,
+other than the command (i.e. the cdb). In the absence of this option, the
+default action is to use the provided data or 0xff
+bytes (\fI\-\-xfer_len=BTL\fR in length) and write it to \fIDEVICE\fR.
+This bit is relatively new in SBC\-3 .
+.TP
+\fB\-x\fR, \fB\-\-xfer_len\fR=\fIBTL\fR
+where \fIBTL\fR is the byte transfer length (default to 520). If the
+given value (or the default) does not match the "long" block size of the
+device, nothing is written to \fIDEVICE\fR and the appropriate xfer_len value
+may be deduced from the error response which is printed (to stderr).
+.SH NOTES
+Various numeric arguments (e.g. \fILBA\fR) may include multiplicative
+suffixes or be given in hexadecimal. See the "NUMERIC ARGUMENTS" section
+in the sg3_utils(8) man page.
+.PP
+The 10 byte SCSI WRITE LONG command limits the logical block address
+to a 32 bit quantity. For larger LBAs use the \fI\-\-16\fR option for the
+SCSI WRITE LONG (16) command.
+.SH EXAMPLES
+This section outlines setting up a block with corrupted data, checking the
+error condition, then restoring useful contents to that sector.
+.PP
+First, if the data in a sector is important, save it with the sg_read_long
+utility:
+.PP
+ sg_read_long \-\-lba=0x1234 \-\-out=0x1234_1.img \-x \fIBTL\fR /dev/sda
+.PP
+This utility may need to be executed several time in order to determine
+what the correct value for \fIBTL\fR is.
+Next use this utility to "corrupt" that sector. That might be done with:
+.PP
+ sg_write_long \-\-lba=0x1234 \-x \fIBTL\fR /dev/sda
+.PP
+This will write a sector (and ECC data) of 0xff bytes. Some disks may
+reject this (at least one of the author's does). Another approach is
+to copy the 0x1234_1.img file (to 0x1234_2.img in this example) and
+change some values with a hex editor. Then write the changed image with:
+.PP
+ sg_write_long \-\-lba=0x1234 \-\-in=0x1234_2.img \-x \fIBTL\fR /dev/sda
+.PP
+Yet another approach is to use the \fI\-\-wr_uncor\fR option, if supported:
+.PP
+ sg_write_long \-\-lba=0x1234 \-\-wr_uncor /dev/sda
+.PP
+Next we use the sg_dd utility to check that the sector is corrupted. Here is an
+example:
+.PP
+ sg_dd if=/dev/sda blk_sgio=1 skip=0x1234 of=. bs=512 count=1 verbose=4
+.PP
+Notice that the "blk_sgio=1" option is given. This is to make sure that
+the sector is read (and no others) and the error is fully reported.
+The "blk_sgio=1" option causes the SG_IO ioctl to be used by sg_dd rather
+than the block subsystem.
+.PP
+Finally we should restore sector 0x1234 to a non\-corrupted state. A sector
+full of zeros could be written with:
+.PP
+ sg_dd if=/dev/zero of=/dev/sda blk_sgio=1 seek=0x1234 bs=512 count=1
+.PP
+This will result in a sector (block) with 512 bytes of 0x0 without a
+MEDIUM ERROR since the ECC and associated data will be regenerated and
+thus well formed. The 'blk_sgio=1' option is even more important in this
+case as it may stop the block subsystem doing a read before write (since
+the read will most likely fail).
+Another approach is to write back the original contents:
+.PP
+ sg_write_long \-\-lba=0x1234 \-\-in=0x1234_1.img \-x \fIBTL\fR /dev/sda
+.PP
+.SH EXIT STATUS
+The exit status of sg_write_long is 0 when it is successful. Otherwise see
+the sg3_utils(8) man page.
+.SH AUTHORS
+Written by Saeed Bishara. Further work by Douglas Gilbert.
+.SH "REPORTING BUGS"
+Report bugs to <dgilbert at interlog dot com>.
+.SH COPYRIGHT
+Copyright \(co 2004\-2016 Douglas Gilbert
+.br
+This software is distributed under the GPL version 2. There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+.SH "SEE ALSO"
+.B sg_read_long, sg_dd (both in sg3_utils), sdparm(sdparm)