SG_REASSIGN(8)

🎀 SG_REASSIGN(8) 🎀

the SG3_UTILS manual page, bestie 💖

---

✨ NAME ✨

okay so sg_reassign is the tool that sends a SCSI REASSIGN BLOCKS command 💅

🦋 SYNOPSIS 🦋

sg_reassign [--address=A,A...] [--dummy] [--eight=0|1] [--grown]
[--help] [--hex] [--longlist=0|1] [--primary] [--verbose] [--version]
DEVICE

🌸 DESCRIPTION 🌸

okay so this utility sends a SCSI REASSIGN BLOCKS command to DEVICE 🔥 but wait, it can also count the elements in a "grown" or "primary" defect list using a SCSI READ DEFECT DATA (10) command 🧚 these commands are defined in SBC-2 for direct access devices (think disks) 💜

reassign blocks is built to move a logical block that is known or suspected to be defective over to a different spot on the media. disks usually come formatted with some spare blocks held in reserve for exactly this moment 🫶

If neither --grown nor --primary is given, then you NEED to give one or more addresses 🙈 if the address (or every single address) fits into 4 bytes and --eight=1 is not used, the parameter block sent to DEVICE is built from 4 byte logical block addresses. if any address needs more than 4 bytes to represent (aka >= 2**32) or you pass --eight=1, then the parameter block is built from 8 byte logical block addresses ✨

🍓 OPTIONS 🍓

Arguments to long options are mandatory for short options as well. the options are arranged in alphabetical order based on the long option name 💅

• -a, --address=A,A... — where A,A... is a string of comma separated numbers 💖 each number is read as decimal unless it's prefixed by 0x or 0X (or has a trailing h or H) 🦋 if you give multiple logical block addresses they must be separated by a comma or a (single) space 🌈 a string with any space separators needs to be quoted 🎀 at least one address must be given 🫶
• -a, --address=- — reads one or more logical block addresses from stdin 📥 these can be comma, space, tab or linefeed (newline) separated 🌸 if a line contains "#" then the rest of that line gets ignored 🙈 otherwise each non separator sequence of characters should resolve to a decimal number unless prefixed by 0x or 0X (or has a trailing h) 💜 at least one address must be given ✨ lines should not be longer than 1023 bytes 😭
• -d, --dummy — prepare for the SCSI REASSIGN BLOCKS command but DO NOT actually run it 🙅‍♀️ since REASSIGN BLOCKS is basically irreversible, paranoid queens may want to double check the invocation before reassigning defective blocks on a disk 💀 pairs well with -vv if you want to peek at the parameter block that will go along with the command 👀
• -e, --eight=0 | 1 — when the value is 1 it sets the 'LONGLBA' flag in the command, which tells the device the addresses in the parameter block are 8 byte quantities 🔥 when the value is 0 it clears the 'LONGLBA' flag, meaning the addresses are 4 byte quantities 🍓 if this option isn't given, 4 byte quantities are assumed unless one of the addresses is too large 🧚
• -g, --grown — uses the SCSI READ DEFECT DATA (10) command to find out how many elements are in the "grown defect list" 📊 with this option there is NO reassignment of blocks (this utility is just being passive, bestie) 🦋 the --address= option is not permitted when this is given 🙅‍♀️ see the discussion below about how reassigned blocks relate to the grown defect list 💖 this list is sometimes called the GLIST 🎀
• -h, --help — outputs the usage message then exits 👋
• -H, --hex — prints the response in hex (for -g, --grown, -p or --primary) 🌈
• -l, --longlist=0 | 1 — sets the REASSIGN BLOCKS cdb field of the same name to the given value 💜 only 1000 addresses are permitted, so there should be no need to specify a value of 1 🙈 the short list variant limits the parameter block length to 2 ** 16 bytes (aka about 16000 4 byte addresses or 8000 8 byte addresses) ✨ this is here for completeness 🫶
• -p, --primary — uses the SCSI READ DEFECT DATA (10) command to find out how many elements are in the "primary defect list" 🏭 this list is established during the manufacturing process 🧚 with this option there is NO reassignment of blocks (passive again, iykyk) 💅 the --address= option is not permitted when this is given 🙅‍♀️ this list is sometimes called the PLIST 🌸
• -v, --verbose — increases the level of verbosity (aka debug output) 🔍
• -V, --version — prints the version string then exits 🎀

💜 NOTES 💜

okay bestie, listen up 👑 if the ARRE field (for reads) and/or the AWRE field (for writes) are set in the "Read Write Error Recovery" mode page, then recoverable read and/or write errors cause the defective block to get reassigned automatically ✨ the PER bit in that same mode page controls whether a RECOVERED ERROR sense key gets reported or not (PER=1 means yes, do report) 🦋 no matter what the ARRE, AWRE or PER settings are, the error counter log pages still reflect any errors (recovered or otherwise) 📊 whenever a block gets reassigned, a new entry is added to the "grown" defect list 🌈 apart from running selftests regularly (see sg_senddiag or smartmontools), keeping an eye on the grown defect list of a disk is a solid metric of its health 🫶 if the grown list starts growing quickly, that is an ominous sign 😱 the best grown defect lists are empty ones 🥺 you can check how many elements are in the grown defect list with the --grown option 🔍 and you can look at the contents of the grown defect list with the sginfo -G utility 💖

⚠️ if an unrecoverable error is detected at a logical block address then REASSIGN BLOCKS is needed to reassign the block 💀 also, if the ARRE and/or AWRE fields are clear and a recoverable error is detected, the logical block in question may be reassigned with this utility 🙌 (otherwise the error counter log pages will keep getting incremented for each recovered access 😭)

the number of blocks held in reserve for REASSIGN BLOCKS is vendor specific and may well be limited to the zone within the media where the original (defective) block lived 🌸 once that number is used up, later runs of this utility may give you a sense key of hardware error plus an additional sense of 'No defect spare location available' 😭 the next step would be to reformat the disk (or get a replacement) 🔥

⚠️ the SBC-2 draft standard (revision 16) notes that when multiple addresses are given to the SCSI REASSIGN BLOCKS command and there's a failure at one of the later addresses, then all the addresses before that have already been reassigned 💀 be careful in that case 🙈 re-running the command with the same addresses will reassign the earlier addresses again 😱 at some stage the disk will run out of reserved locations 🥀 so unless a large number of addresses are involved, it may be safer to reassign them one address at a time 🧚

🌈 EXIT STATUS 🌈

the exit status of sg_reassign is 0 when it is successful 🎉 otherwise, see the sg3_utils(8) man page 📖

🎀 AUTHORS 🎀

Written by Douglas Gilbert 👑

🐛 REPORTING BUGS 🐛

report bugs to <dgilbert at interlog dot com> 💌

🔒 COPYRIGHT 🔒

Copyright © 2005-2019 Douglas Gilbert this software is distributed under a BSD-2-Clause license 📜 there is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE 🙅‍♀️

🔗 SEE ALSO 🔗

sg_format, sginfo, sg_senddiag (all in sg3_utils), sdparm (sdparm), smartmontools (internet, sourceforge) ✨

---

okay that's the whole tea ☕ this was SG_REASSIGN(8) from sg3_utils-1.45, dated October 2019, and it's giving iconic 💖🧚🦋 stay hydrated and back up your disks, bestie 🫶💅