Slashdot Mirror
SSDs Cause Crisis For Digital Forensics
rifles only writes "Firmware built into many solid state drives (SSDs) to improve their storage efficiency could be making forensic analysis at a later date by police forces and intelligence agencies almost impossible to carry out to legally safe standards, Australian researchers have discovered. They found that SSDs start wiping themselves within minutes after a quick format (or a file delete or full format) and can even do so when disconnected from a PC and rigged up to a hardware blocker." So either SSDs are really hard to erase, or really hard to recover. I'm so confused.
Deleted, should mean deleted.
Problem solved. People need control over their own privacy. Tough luck Digital Forensic folks.
So either SSDs are really hard to erase, or really hard to recover. I'm so confused.
All I know is that if SSDs were really hard to erase, and I was in the business of recovering data that other people didn't want recovered, this is exactly the kind of story that I would tell them so that they would continue using SSDs.
Not that I'm paranoid or anything.
Why the confusion, dear editor? This should be well understood.
If you want to recover, you can't. If you want to erase, you can't. It's Murphy's Law of Data Storage.
When Mindy the undergrad accidentally deletes her term paper and would be really REALLY grateful for a super smart and kinda cute geek to go in and recover the file with Backtrack... then you'll see the downside.
What? I reject your reality and substitute my own!
Deleting a file should tell the OS that I don't need that data. That's what deleting is.
Give me Classic Slashdot or give me death!
The drives have internal overprovisioning and perform internal garbage collection. This means that marked for deletion data has an unknown lifetime and may disappear at any point without interaction from a controller.
The hard to erase bit means that you really can't be sure something is totally erased without a full specific erase command to all flash blocks. Without that a page marked unused but not erased may be nestled in with a bunch of valid pages. As all pages in a block are erased together that marked unused page can hang around for a wile.
On the other side the firmware does garbage collection it actively looks for blocks with many erased pages and then tries to consolidate things so it can create more free blocks. This means if the drive is powered but not connected to a host machine it can still be doing data moves for consilidation and erasing marked for deletion pages.
There are thresholds for the garbage collection so it won't overwork and try for 100% consolidation. Thus you get both the presence of some really sticky stale marked unused pages and some active erasing of others.
Bingo. Forensic tech are used to being able to just plug in a write-blocker and assume the disk will remain intact and unchanged, which is the "legally safe" part they are complaining about. Since SSDs do lots more under the hood than spinning rust drives, they can't guarantee the device is unchanged unless they disassemble it, which might be considered tampering and leaves room for the other side's lawyers to ask "and then you took a soldering iron to a delicate IC?".
It also requires a lot more know-how than "Use this magic cable when copying drives for investigation".
There is an ATA command called "TRIM". If a device supports it, the OS can tell it that a group of sectors is no longer needed, and should be wiped.
It's not hard really. The drives autoclean themselves. So when you delete the inode reference to a given data file, it will be completely much more quickly that on a normal magnetic drive (which won't reclaim the space till it's needed). On the other hand it won't respond to commands that would FORCE a magnetic drive to completely wipe the file.
So if you're in your secret lab and you hear that the evil enemy is an hour away, you just type "rm /*" and wander off to escape. By the time they get there all your data will be completely wiped. On the other hand if they are breaking down the door to your secret lab and you have only seconds left you can't type "shred /home/joeari/secretfile" and expect it to be perma-deleted like you could with a magnetic drive.
They recover deleted sectors more quickly, but can't be forced to do so in a controlled manner.
I don't need a million points of light, just two points of multi-mode fiber and a 10 Gig-E router.
...a foregone conclusion ever since ATA Secure Erase and TRIM were introduced?
Secure Erase basically tells the SSD that all of its cells are now blank (AFAIK implementations actually zero the drive as well but I'm happy to be corrected on that); therefore as soon as anything is written to the disc, it will be written here, there and everywhere. It took about 30s to run on my first vertex and I couldn't find any trace of
TRIM support in the ATA spec, along with kernel/filesystem support, tells the disc that when file A is deleted, cells X, Y, and ABQ are now officially "empty" and that if the controller feels like it, it can zero them out, shunt other data in there, or have a mardi gras for all it cares. The same happens when a drive is formatted; OS tells drive controller "I've just formatted you" and for the sake of preserving performance the controller goes "Brilliant! I can chuck out all this shit I've been saddled with."
As soon as hard drives start intelligently erasing/shuffling bits of themselves about so that cells are utilised to their utmost efficiency this was bound to happen. Unlike spinning platters where bad blocks were reallocated only if a) the hard disc knew about it and b) the data could actually be read/recovered, it becomes terribly obvious that data on SSD's is going to be read and written and deleted completely and utterly all over the place, without sequential series of sector found in slackspace like you would on a magnetic drive.
Magnetic drives have no performance penalties for not actually erasing the data, so if you work your way around that double negative you'll see that one of the staples of digital forensics (e.g. recovering files from slack) is a by-product of people trying to make magnetic platters as fast as possible by not actually erasing stuff, because as long as the controller knows that sector is blank then it'll just be overwritten as needed. Technology has now changed sufficiently that the performance gains from new solid state tech are helped by a drive controller that erases stuff as soon as possible, since writing over an occupied cell is slower than writing over a blank one.
I'm sure there'll be new methods to mitigate the change in tech, we're just somewhat on the cusp of a completely new tech. They'll probably come to an agreement that TRIM doesn't actually delete stuff until the amount of free space in the cells reaches a certain threshold or something like that.
Disclaimer: I'm not a digital forensic scientist, but am friends with one and we discussed this problem over some exquisite cocktails a few months back. And I don't think TRIM instructions follow the exact specifications I laid out above (e.g. using Brilliant! as an ACK).
Moderation Total: -1 Troll, +3 Goat
The do not completely erase when you hit them with a commercial erase tool because the commercial erase tool does not give the drive a chance to "take a breath". If you hit it with a DOD wipe instead of overwriting old blocks the drive will give you new until it runs out of spares and after that will slowly overwrite some old. At the end some data will remain.
You need to hit it with short write bursts which are alternated by rest periods so that the controller can reorganise the flash especially if you use the TRIM ATA command to mark stuff as "really deleted". As a result the data will be wiped to a standard which no hard drive can achieve.
I can write a perl 10-liner that feeds DD to do that in about half an hour. So any geek can erase one. However the commercial and most importantly certified by banking and govt tools cannot just yet. They definitely will as there is no rocket science here. Until this happens you will hear both sides of the story depending who you talk to.
By the way the same will apply to overlapping recording and drives with flash cache. A "dumb" wipe will not wipe them. At the same time a script any sysadmin can write in his spare time can wipe them so effectively that no forensics can get the data back.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
> Deleting a file should tell the OS that I don't need that data. That's what deleting is.
It does, and that is, in fact, what it is. So Windows unlinks the file from the directory and removes the blocks from the in-use map.
The drive, however, doesn't know any of this. It updates the requested sectors representing the directory and volume info, and that's about it. It has no idea that blocks somewhere else on the drive are no longer needed, so it will dutifully copy and maintain that data while rewriting blocks.
The idea of the trim command is that it tells the drive that the data is deleted so the drive doesn't have to worry about maintaining it. This was never needed in old drives because the data was basically ignored after being written. In SSDs, the data needs to be continuously copied around to facilitate the erasing (which wipes several thousand blocks at once).
At a guess this is caused by mounting with the discard option, or trim as its called in Windows. It tells the drive you don't need the data stored where a deleted file used to be.
Maybe it's still there if you look with a microscope but who really does that?
Hello there, I'm one of the authors of the paper (Graeme).
This finding isn't related to TRIM in any way, though TRIM poses another nightmare for forensic investigators and may make the idea of examining deleted data/metadata redundant in a few years.
What's happening here is that the drive itself has some code, a garbage collector, that reads the NTFS filesystem metadata and wipes any cells it thinks aren't needed any more, so that the next set of writes over those cells can take place more quickly.
Normally this GC has seemed to take a while to kick in (various benchmarking sites suggest e.g. 30-60 minutes and unpredictable behaviour, e.g. not always kicking in when expected); here, we found that after a quick format has taken place, it reliably can be seen to kick in within minutes and also purges the drive within minutes. This is just one example of a case when the GC can kick in, of course.
Current court-accepted forensics practice is to stick a write blocker between the drive and computer, under the assumption that the drive only modifies data when a PC tells it to: but that doesn't help you when the drive itself nukes all it's data cells within minutes of being powered on.
I can imagine a situation where someone connects up the drive to power, (and possibly a write blocker), but not to the PC and makes a cup of coffee for a few minutes - by the time they connect it up for data transfer, it's too late, and they wouldn't even realise it had happened - the drive doesn't exactly make any whirrs or clicks as it wipes itself.
Alternatively, the circumstances we found: in less than the time that would be taken to read an entire image off the disk for forensic study, the GC is racing ahead of you and purging the disk! Yikes. So you can imagine... a forensic investigator takes an image, examines it, presents it to the court, and when the court verifies the original disk that the copy was taken from, nothing is there any more, and the forensic investigator seems to be making it all up....
So this feature has the potential to make it look like the police or forensic investigator themselves tampered with the original disk, as well as potentially destroying evidence that could help establish guilt or help establish innocence. We've put a number of 'interesting legal grey areas' at the back of the paper that we hammered out with reviewers, which are worth being aware of.
Graeme.
Well to be fair, that's only part of the problem. Say you're a stupid criminal. You store all your records of your criminal acts on your computer which happens to have an SSD drive. You've never deleted your records of your criminal acts, but just before the police busted through the door you were deleting some old pictures of your cat, Fluffy. So in order to maintain evidence chain of custody, the police immediately turn off you computer and turn it over to a tech. The tech's first two actions should always be the same. He should plug your drive into a a special read only device that will first do an MD5sum or other fingerprint of the whole contents of the drive, and then do a bit for bit copy of the drive.
The idea here is that the police can prove that nothing was done to alter the data present at the time of seizure. Power was removed from the system and the drive was immediately fingerprinted and copied. Assuming the fingerprints match, there should be no question that the copy on which the actual analysis is done is identical to the original drive. The problem is that you were deleting pictures of Fluffy when the Police came in. It's highly likely that as the drive does its self cleaning routine the fingerprint of the data will change. They fingerprint, get a value, but while they're copying the drive self cleans the bits associated with your cat pics. Bam, the copy has a different fingerprint. Now there's reasonable doubt about the usefulness of the evidence you stupidly left unencrypted in your desktop folder.
Now I'm not saying this is a problem, and that they need to modify the design of SSDs. I didn't get the impression that article said that either. What they are saying is "Hey, we need to come up with another way to do this, becasue what we've been doing will no longer stand up in court."
I don't need a million points of light, just two points of multi-mode fiber and a 10 Gig-E router.
Only a problem if the retard windows admin disabled the recycle bin because it interfered with his OCD.
Do you even lift?
These aren't the 'roids you're looking for.
You wrote: "So if you're in your secret lab and you hear that the evil enemy is an hour away, you just type "rm /*" and wander off to escape. By the time they get there all your data will be completely wiped. On the other hand if they are breaking down the door to your secret lab and you have only seconds left you can't type "shred /home/joeari/secretfile" and expect it to be perma-deleted like you could with a magnetic drive."
Actually, we found something even more interesting when we prepared this paper.
An evil criminal could run a quick format (about 8 seconds effort) if police were at the door, or they could set up a dead man switch to do the same. When the investigator powers up the drive, the drive's internal GC runs quickly (we couldn't get any formal documentation about why this is - presumably it begins so quickly since the filesystem metadata is nice and simple to process) and so after about 3 minutes it begins wiping itself.
Just a few minutes later it has finished, and data and old filesystem metadata are gone (at least, gone as far as logical access to drive sectors is concerned).
I'd encourage anyone who finds the result interesting to take a glance at the original paper, it's written to be accessible to people outside the traditional drive forensics community and there are some fun, free script tools you can use to monitor drive GC behaviour in near-realtime.
Graeme.
I think it's about impoliteness when asking favors. Friends help each other: I'm quite ok fixing my non-technical friend's WLAN, just as he's ok with giving me a hand when I'm moving houses. That's the social norm, and thus asking someone for a favor also indicates how you feel about them.
So if you don't like someone, if you would under normal circumstances not want to spend time with them - then you don't ask them for favors. That would be plain rude.
If you don't have an actual friend to provide tech services - just purchase said services.
On the other hand it won't respond to commands that would FORCE a magnetic drive to completely wipe the file.On the other hand it won't respond to commands that would FORCE a magnetic drive to completely wipe the file.
So why isn't a "wipe everything, including the spares" command part of the SATA standard command set? I can think of many times I'd like to completely erase a drive. Basically, any time I run the equivalent of "fdisk, o" in whatever OS I'm using, I'm explicitly telling the drive, "I value nothing here". That would be a prime time to optionally tell a drive to totally erase itself.
There are even non-security-related reasons for doing this. IIRC, the reiserfs recovery tools would scan a partition to look for data laid out in reiserfs-formatted chunks and undelete them. That's normally exactly what you want. However, suppose you're using reiserfs and running a VM that also uses reiserfs on its virtual disk, which lives inside a file on your main filesystem. The recovery tools would see those structures, too, and merge them back into your directory structure. This is typically disastrous.
When programming in low-level languages, it's nice to know that the only thing in your process's heap is data you've explicitly put there. The same holds true for hard drives. While we're obviously pretty good at keeping the wheat separated from the chaff, it'd be nice if you could guarantee that there was no chaff to start with.
Dewey, what part of this looks like authorities should be involved?
You often get laid for being a dick. Hell, if you're in the right place at the right time, with low enough standards, you can get laid just for *having* a dick.
Learn about Photography Basics.
Plaintiff's Attorney: "Sir, what are the chances of the drive automatically generating the exact sequence of bits required to form this email?"
Expert Witness: "Billions to one, certainly."
Defendant's Attorney: "And how many times will that this 2KB email fit on the drive?"
Expert Witness: "Well, it's a 2TB drive, so... about a billion, give or take."
Defendant's Attorney: "So, assuming the data on the drive is random, then it's safe to say there are at least two billion opportunities on this drive to produce this email?"
Expert Witness: "That's not what I meant..."
Defendant's Attorney: "Yes or no?"
Expert Witness: "Well, yes, but..."
Defendant's Attorney: "No further questions"
Learn about Photography Basics.
It used to be the drive remapping meant that if you deleted something (or even overwrote it), there was no guarantee it would be gone - SSD controllers do wear leveling to avoid having part of the drive get used excessively. Forensic analysts could go pull the raw data from the NAND flash.
The problem was that the wear leveling algorithms need a "free block" pool to work well. Drives that have been used heavily deplete the free block pool, and the drive slows down. For a long time, SSDs would have no knowledge of whether a file had been deleted or not.
The ATA TRIM command was added for just this purpose - with TRIM, when the OS deletes a file that references a block, it can tell the SSD controller that those blocks are free. The SSD controller will then begin erasing those blocks in its free time. (SSDs can be written one block at a time, but must be erased one page at a time. A page consists of multiple blocks. Oh, and I may have page/block swapped here.) So you get a lot of performance improvement by having a bunch of pre-erased pages - these can just have individual blocks written without a read/erase/modify/write on a whole page.
Pre-TRIM, SSDs were probably great for forensic analysts. Post-TRIM, SSDs are not. Oh, and I think the latest ATA standard added a "sanitize" command to make life easier for information assurance types, for whom SSDs have always been a pain.
retrorocket.o not found, launch anyway?
"The drive, however, doesn't know any of this. It updates the requested sectors representing the directory and volume info, and that's about it. It has no idea that blocks somewhere else on the drive are no longer needed, so it will dutifully copy and maintain that data while rewriting blocks."
Actually, this is no longer correct. SSDs (such as the one in this study) are quite capable of examining the filesystem stored on the drive, independently, and the concept of 'dutifully' and ignorantly maintaining deleted data goes out of the window as a result.
The main point of the paper is not that this 'analyse-and-purge' drive GC behaviour happens (which is well known among SSD enthusiasts), but that it fouls up long-accepted court and forensics assumptions about what computer drives are. It can result in loss of evidence (of guilt or innocence) very easily in court cases in a manner which might be incorrectly attributed to malicious human intention.
Graeme.
Yep, it's the answer that immediately springs to mind when you approach the problem as a techie, and we touched on in the paper (see point 16, page 13).
Keep in mind though that this is extremely non-trivial - e.g. not many court officers will know how to do it - and that a court may have real trouble with a forensics guy telling them that for 'this particular case', they need to not use the standard procedure to preserve the data unmodified, but instead to rip open the disk and muck about with how it processes data.
Courts have accepted practices that are used worldwide; what you propose is not accepted practice; therefore courts will have a problem, at least in the short term.
Also, given the number of variations of controllers and firmware out there too, I'm sure you'll agree there's tremendous potential for a court forensics officer to get it wrong and accidentally nuke the drive data. e.g. how do you find out which firmware was in use, without powering on the drive (which activates the GC)...
Graeme.
"On the SSD, the TRIM command caused the device's firmware to wipe the entire device." Close, but not exactly right. Here, the drive's firmware undertook the operation by itself using a NTFS-aware garbage collector; no ATA TRIM command from the PC was needed. We used a non-TRIM OS for just this reason. Graeme.
Not likely in this situation - here, the GC isn't killing the data for the hell of it, it's killing it because it needs to reset the cell preemptively in order to maintain drive performance at a high level. If you preserve the data in the manner of an old HDD, you nuke performance by forcing the controller to do a reset before future writes.
Graeme.
-----
Hey there
Yep, it's the answer that immediately springs to mind when you approach the problem as a techie, and we touched on in the paper (see point 16, page 13).
Keep in mind though that this is extremely non-trivial - e.g. not many court officers will know how to do it - and that a court may have real trouble with a forensics guy telling them that for 'this particular case', they need to not use the standard procedure to preserve the data unmodified, but instead to rip open the disk and muck about with how it processes data.
Courts have accepted practices that are used worldwide; what you propose is not accepted practice; therefore courts will have a problem, at least in the short term.
Also, given the number of variations of controllers and firmware out there too, I'm sure you'll agree there's tremendous potential for a court forensics officer to get it wrong and accidentally nuke the drive data. e.g. how do you find out which firmware was in use, without powering on the drive (which activates the GC)...
Graeme.
p.s. Also keep in mind that data is stored on the chips in a highly non-linear manner; reassembling the logical image may not be easy, since you need to work out how the flash translation layer was being represented. This is getting wildly beyond what a typical court forensics officer is usually trained to do, namely, rip an image of the drive with a write-blocker attached and scan it for jpegs/whatever.
Wow. Thanks for dropping by. That's one thing I really enjoy about slashdot - seeing an author come by to post a clarification. Indeed, that really helps me to understand the process better.
The diversity and expression of human opinion is essential to human survival.
Hello, I'm one of the authors of the paper. To explain the apparent paradox in rough terms:
Drive data was traditionally purged manually, by having the computer tell the drive to write something else over the top of the old data. In the absence of such an overwrite, magnetically stored data persists. However, if you try that trick on an SSD, it may not work. The logical address you try to overwrite may be remapped on the fly, so that your 'overwrite' goes to some other physical cell rather than the one which stored the data. From a logical viewpoint, it looks like the overwrite worked - you can't access the data any more through your computer's OS. But from the drives point of view, the data is still there, lurking in some physical cell that is presently out of use as far as the logical sector list is concerned. A cunning firmware or a hacker with a soldering iron might still get at it.
However, separately to this, modern SSD drives use tricks to try and automatically improve their performance, and one of these tricks is to pre-empetively wipe data cells that contain data no longer referenced by the filesystem. Here, the drive is actively attempting to permanently purge everything it can from the drive, all of it's own accord, in the interests of accelerating future writes by having a pool of completely unused cells available.
Summary:
- If you're a computer telling a drive to zero over some data, the drive may lie to you a bit, and not bother to zero it.
- If you're a drive, you do whatever the heck you like, and you see the physical layer directly (unlike the computer). That means the drive can open up the NTFS metadata, looking for data cells which could be preemptively reset, and nuking that data out of existence (when it might traditionally have been recoverable to an expert).
In summary. If your drive wants to nuke something, (and we've shown, they really DO want to nuke everything they can at a few minutes notice), it gets nuked. If your PC wants to nuke something, it may or may not get nuked by attempting an overwrite.
Finally, separate to this is TRIM, which is a hybrid of the two situations - an ATA command by which the OS can signal to the drive that it would like the corresponding physical cell for a particular logical sector address to be nuked, thank you very much.
Hope that clears things up.
Graeme.
Trust me, dd can really make a mess of things, if by intention or mistake.
I was cloning one drive (A) to a larger one (B). It needed to preserve the partition table, MBR, etc.
dd of=/dev/sda if=/dev/sdb bs=1024
About 10 seconds after I hit enter, I saw the mistake, and aborted it. Ya, I dd'd an empty drive over the populated one. How much harm could that possibly do, right? :)
I've done plenty of drive recoveries, with about an 80% success rate. The only ones I can't do are drives that don't spin. I don't have the facilities to transfer the platters.
Two weeks and several recovery attempts, I had recovered about 25% of what I wanted. The partition table, and the MFT were destroyed, and files were on four partitions with two different filesystems (NTFS and ext3). That was in 10 seconds. If you were to say" dd if=/dev/random of=/dev/sda bs=1024", it's rather unlikely even the best would find anything.
From what I understand, the best chance at recovering data from an overwritten media is to look at sectors marked as bad. They are automatically skipped, so they don't get overwritten by anything. Who knows what will be in them though. Most likely on most users machines, it will be pieces of the browser cache, or Windows Update files.
That's not to say if some agency named "No Such Agency", were really bent on finding out what I had on there, they couldn't recover something. Many agencies can't go beyond files that currently exist, or are in the "Recycle Bin". Some go as far as undeleting from an otherwise functional filesystem. Is it worth weeks and thousands of dollars for whatever organization to recover what you have hiding on your desktop? Probably not. If you have secrets, don't hide them on your PC at home, even if you are protected by a nice wood front door with fancy security locks. One brick through the window, and a fast getaway, will defeat any door lock and alarm system. If you kept all of your important information on some random colocated server in another country, AND you have a second person with access to clear your data, it would be untouchable by most agencies. Say I had a server colocated in China, with the normal protections (encrypted virtual filesystem, etc, etc.), and only used my desktop to manipulate them through a terminal session (VNC, SSH, etc) the most that could be found is that I connected to a server in an untouchable place, and by the time they managed to get to it, my 2nd party could have removed it.
Serious? Seriousness is well above my pay grade.
"How would it do this without maintaining firmware for all possible types of file systems?"
It wouldn't. But, how many different filesystems are in realworld use on 95%+ of systems? HFS+, ext2/3, NTFS, FAT32?
I can imagine that it does this for simple FAT16 or FAT32, but I doubt they deal with NTFS or EXT3 or HFS.
It does work on NTFS, and if you set up an experimental rig similar to the one we describe at the back of the paper (page 16), hopefully you can satisfy your doubts :-)
2 - Defragging: similarly, if you're moving data around in dead space without safely duplicating it or having a filename pointing to the blocks in use at any given time, you're not being careful. Also, which defraggers have random 3-minute gaps in operation that would even allow GC to kick in?
I think it is time to start bringing the discussion to a close, as it appears that we do share at least some common ground.
I will comment only on this one question. Your implication that the 3-minute rule somehow makes the GC "safe" is missing my point entirely. Yes, in practice there are checks and balances such as you describe, that make the GC unlikely to screw up. But, in my view, "unlikely" is not good enough. I want, and need, perfect (logical) block storage and retrieval. This should and must be the design goal. Of course, this goal is impossible to achieve in practice. For example, if the firmware (such as older, pre-SSD firmware) is designed with the goal of providing logical block storage, but fails in this task because of some honest bug, then I can understand that. At least, in this case, the programming code was written unambiguously with the correct goal (and no other directly conflicting goals) in mind, even if this goal was not achieved in practice due to an unintentional bug.
However, when a manufacturer deliberately designs firmware with the goal of deleting logical sectors, no matter how well-intentioned or well-implemented, this design goal (by definition) must come into conflict with the original, core goal of reliable (logical) data retrieval. I do not care what happens in the underlying physical layer, but I do care very greatly about data accuracy at the logical layer. The existence of certain checks and balances to prevent data loss is better than no checks and balances, but it is not better than the REAL alternative, namely, firmware that stores and retrieves logical blocks correctly, and is designed for this and only this purpose, without any other directly contradictory design goals.
No one, not even rocket scientists, has ever succeeded in writing bug-free software. But one should make an effort to minimize the number of opportunities for data loss bugs to arise. Firmware-based logical-sector garbage collection fundamentally and irreconcilably contradicts every reliability design principle known to man. That is why I consider the idea to be so abhorrent.