Want To Ensure Your Personal Android Data Is Truly Wiped? Turn On Encryption

MojoKid writes We've been around the block enough times to know that outside of shredding a storage medium, all data is recoverable. It's just matter of time, money, and effort. However, it was still sobering to find out exactly how much data security firm Avast was able to recover from Android devices it purchased from eBay, which included everything from naked selfies to even a completed loan application. Does this mean we shouldn't ever sell the old handset? Luckily, the answer is no. Avast's self-serving study was to promote its Anti-Theft app available on Google Play. The free app comes with a wipe feature that overwrites all files, thereby making them invisible to casual recovery methods. That's one approach. There's another solution that's incredibly easy and doesn't require downloading and installing anything. Before you sell your Android phone on eBay, Craigslist, or wherever, enable encryption and wait for it to encrypt the on board storage. After that, perform a wipe and reset as normal, which will obliterate the encryption key and ensure the data on your device can't be read. This may not work on certain devices, which will ask you to decrypt data before wiping but most should follow this convention just fine.

Comments

Free space

[narced]

What is not addressed is whether or not this wipes the free space as well. Recovering deleted files is easy, and if the encryption doesn't fill the device then encrypt then this trick can leave some stuff behind.

Full-disk wipe or only current data?

[Gaygirlie]

I'm under the impression that turning on encryption works by file-by-file basis, not full-disk encryption, and as such it might still be possible to find at least some old files there if the locations haven't been overwritten by new data. If it indeed works as I have the impression of then turning encryption on is still possibly inadequate a safety method.

Re:Full-disk wipe or only current data?

[Anonymous Coward]

According to the android documentation it is full-disk encryption based on dm-crypt.

Re:Full-disk wipe or only current data?

[swillden]

Who gives a shit what the documentation says. Actual implementation is what matters.

Absolutely. So, look at the source: https://android.googlesource.c...

That file contains the code that generates the master key, derives the key encryption key used to protect it (using scrypt), stores the protected master key, and configures dm_crypt with the master key.

Some functions to look at:

- create_encrypted_random_key(), which creates the master key (reading from /dev/urandom).
- encrypt_master_key(), which derives a KEK from your password and uses it to encrypt the master key.
- decrypt_master_key(), which does the reverse.
- create_crypto_blk_dev(), which creates dm_crypt block device.
- cryptfs_setup_volume(), which mounts an encrypted block device.
- cryptfs_enable_inplace(), which encrypts an existing file system.

Do you really trust a mobile platform to be faithful to the documentation when you're trying to wipe a partition (which could easily be implemented directly but isn't) by first encrypting all data and then throwing away the key?

The device doesn't know you're trying to wipe. It knows that you (a) requested full disk encryption and then later (b) requested a wipe. So it can't optimize (a) away. I suppose it's possible it could just lie and tell you "Yep, I'm encrypting" even though it isn't, but that's the sort of thing that would definitely get noticed by security analysts and gleefully published.

"It's just matter of time, money, and effort."

[retroworks]

It's well established that plenty of consumers discard or donate hard disks without taking any precautions, and are playing roulette with their identity. It's also well established that hundreds of millions of tons of this equipment is replaced, resold, stolen or discarded, and most people who wind up with the secondary device lack either the time, money, or effort to scavenge data off the phone. If in fact someone is in the identity theft business by buying phones on ebay, they'd profile themselves pretty well after a dozen phone purchases (what do these data-theft-victims have in common?). And who knows how many phones they'd have to buy which had been wiped in some way (and required more time, money and effort)?

This isn't a bad article in that it birddogs simple things you can do before selling your used phone, and if it elevates the perception of risk in order to get people to do something easy, that's appropriate. But in response to people who are shooting and burning their devices to be "100% sure" that no one spends the time, money and effort to follow them... that's appropriate if you are a high risk target. If you have stuff on your phone of interest to the FBI or KGB, the amount of time+money+effort may be less than or = the amount of risk. Your call.

But there is a lot of hyperbole out there about the percentage of identity theft which is traced to secondary market devices, and the billions of dollars in secondary market sales on sites like ebay represent time+money+effort interest in new product makers to spend fanning flames. Again it's appropriate that the article raises concerns and then points to simple efforts a consumer can take to increase the barrier-to-entry to their personal data. But the army of ebay buyers getting their porn fixes by buying and then de-encrypting cell phones to retrieve ugly selfies seems exaggerated. Warn people about sharks if they are swimming in shark infested waters, don't tell people that most swimmers will be attacked by sharks.

Tear your mail in 8 pieces and someone could dig it out of the trash and tape it together, but the time+money+effort that represents is significant. I remember people selling paper shredding equipment in the 1990s who described armies of Iranian students or Chinese peasants who could be buying torn paper and taping it back together. If they know it's the President of the USA's mail, they no doubt will expend that time+money+effort... Presidents should assume they are swimming in a shark tank. For most of us, ebay resales are a swimming pool, and warnings of shark attacks get tiresome.

Re:And then throw it in a fire

[plover]

This.

What is the value of a used device? Compare that to the risk of the data on that device going to a malevolent third party.

I've had people saying "oh, look at all these hard drives, you should totally sell them on ebay and I bet you could get $10 apiece for them!" Adding up the time I would waste running DBAN or sdelete or whatever, and keeping track of which ones have been wiped, and double checking to make sure everything is really gone, it's not worth the time.

A big hammer and a punch, driven deeply through the thin aluminum cover and down the platter area, takes about a second and leaves nothing anybody would bother trying to recover. You can quickly look at a drive and say "yes, this drive has been taken care of", or "hey, there's no jagged hole here, this drive isn't destroyed." The aluminum cover contains the shards if the platters are glass. I don't care who handles them after destruction. There's no worries about toxic smoke. And if you have to inventory them before shipping them to a recycler, the serial numbers are still readable.

Smashing a phone wouldn't destroy the data on the chips, so a fire is a somewhat safer option.

Re:And then throw it in a fire

[wisnoskij]

Data cannot be destroyed, that is a fundamental axiom of physics. Someone might read your data based on the smoke your chimney emits.

Unsafe Advice

[bill_mcgonigle]

Any marginal blocks mapped out before you encrypt will remain unencrypted and may be available to a determined attacker. Same goes for hard drives, and SATA secure erase is not provably trustworthy. Always encrypt your storage before you put any data on it. If you do not trust your hardware AES to not be backdoored then use software crypto.

Re:srm -v -z

[wiredlogic]

The "special values" were from Guttmann's paper on wiping MFM/RLL drives. It is pointless on any modern magnetic drive or solid state memory. He points out in his newer paper on solid-state memories that multi-level flash (now used everywhere other than the most performance critical applications) is particularly hard to recover data from. Furthermore, the wear-leveling strategies used in flash mass storage devices negates any attempt to securely wipe them short of physical destruction. You're just practicing cult cargo voodoo.

Re:srm -v -z

[Immerman]

>Furthermore, the wear-leveling strategies used in flash mass storage devices negates any attempt to securely wipe them short of physical destruction.

Well, it confounds it at any rate. But completely filling the device's memory 33 times in a row is pretty likely to overwrite everything at least once or twice - even the hidden "failure reserve" space if it's included in the wear leveling (and if it's not, then it doesn't yet hold any sensitive data, so there's no problem). Guttmann's values may be irrelevant to today's storage media, but that many repeated rewrites of anything still mostly does the job.

I don't know that I'd trust it to wipe vital military secrets, but it should do a good enough job for most anything in the civilian realm.

Re:srm -v -z

[Immerman]

Not quite - modern magnetic drives still have tracks wider than the read-write head so that atomic-level alignment isn't necessary. There may be far less "overwrite" than there once was, but if a newly recorded track is not *perfectly* aligned with the last recording then there may well be several percent of the previously recorded track that remains unaltered (consider the worst case scenario case that the previous recording in this track was written at the smallest radius allowed by actuator tolerances, while this pass is at the maximum radius allowed). Now, recovering that data will probably require removing the platters and analyzing them with much higher resolution read heads, but it can be done.

I was more addressing the problems with flash though - in order to disguise degradation modern flash drives typically include more capacity than is addressable by the host system. Fill it to the brim so there are zero bytes free, and there's still several percent of the total drive capacity that is sitting unused in the reserve pool. The only way to overwrite that (barring a OS-accessible "secure wipe" command implemented on the drive) is to generate sufficient churn that the internal wear leveling algorithms cycle through every byte of the reserve capacity at least once. And since you probably don't know the exact algorithm used or wear levels of the drive to begin with, more is better - after all you have to tease out the most heavily used page currently sitting in the reserve.