What should I use to erase my hard disk before selling my computer?
dban/any linux livecd
single zero pass
>>51751755
piss on it. the sodium will destroy it.
>>51751806
A single pass is good enough?
And then I can reinstall the OS as if the hard disk was new?
>>51751827
PISS ON IT.
>>51751815
I heard you had to drink some ferromagnetic liquid so your piss is gonna destroy the hard disk through its own magnetic field.
Can anyone confirm this?
>>51751827
yes, pretty much
>>51751827
I'd secure ata erase the drive myself. It's actually faster than overwriting with zeros.
>>51751847
Ok thanks for the advice
Definitely going to try it out
>>51751843
just a tea cup worth.
>>51751850
What's this? Never heard of it
>>51751856
It was for >>51751862
But oh well, if you take out the sarcastic vibe in it it makes sense for your answer too
>>51751875
It's a feature of SSDs with Hardware encryption.
The SSD simply throws the current key away and generates a new one, making all the data useless(because the drive can't decrypt it anymore without the old key)
aka irrelevant
>>51752218
Can I do this with HDD, too? Even though I've never encrypted them?
>>51753119
rotating keys to make old data unreadable is something you can only do with encrypted data, and has nothing to do with media (ssd/hdd)
it applies just the same to hdds, IF it was encrypted
sell it without the HDD
scratch it with a nail
fire a coule .44 rounds into it
set fire to it
bury it in your neighbor's backyard
>>51753142
Are we talking about something like bitlocker or something at BIOS/hardware level?
Anyway I've never encrypted them but I don't have sensitive data so I'll wipe them with a couple of 0s wipe
>>51753173
it doesn't matter how the data is encrypted, the effect is the same (only the method may differ)
since your data isn't currently encrypted, you need to do a zero pass, it would take longer to encypt everything then discard the key
>>51753226
Ok thanks
https://www.youtube.com/watch?v=-bpX8YvNg6Y
>>51751806
>>51751850
Both good solutions.
Either - for /dev/sda being the device you want to recycle:
# time dd if=/dev/zero of=/dev/SDA bs=1M conv=notrunc
or, ask the firmware to do it:
# hdparm -I /dev/sda
## if it says frozen, instead of not frozen, try suspending and resuming - if that doesn't unfreeze, try dd above
# hdparm --security-set-pass password /dev/sda
# time hdparm --security-erase password /dev/sda
## if supported, in hdparm -I, try --security-erase-enhanced instead, which will also try to wipe bad sectors.
Advantages of hdparm: if firmware supports it, it'll wipe any HPA (host protected area) and will, with enhanced, try to wipe any bad/remapped sector areas too.
SSDs should be wiped using the hdparm method, but beware: SSDs cannot be securely wiped as the failure mode for flash memory is that cells can no longer be erased or written. If the drive is nearly new, it should be okay - but no guarantees. Always encrypt them with software if you can to avoid this.
Hard disk self-encryption is shit, don't use it.
If you'd encrypted the data, zeroise the key first.
One zero pass is enough against any adversary on hard disks which use perpendicular recording, SMR or HAMR (that is, almost any drive still spinning). Probabilistic data recovery has never been demonstrated in the wild, even in anti-terrorism cases by well-funded nation-states, unfortunately. Gutmann's patterns were for MFM/RLL (pre-IDE) drives. Magnets don't erase hard disks, no commercial degausser works, the coercivity is too high.
If you really positively need it dead, do both and then remove the platters, controller board and heads and destroy them as absolutely as you can. If a drive may have been implanted/infected, do this. If you really are worried, start destroying anything with memory, which includes microcontrollers, Guardian/GCHQ style.
>>51753519
/dev/sda, not SDA, obviously. Sorry.
Keys should be held on EEPROM where possible, not Flash. EEPROM, properly constructed, has a zeroisation mode that works, the failure mode is failure to program, not erase.
Battery-backed fast-fade (low retention) SRAM or DRAM works too, but you need to zeroise on environmental traps like low temperature.
CPU L1/L2/L3 cache works a treat as long as it's on-die and not just on-package: this approach is taken by TRESOR.
