Using ddrescue to recover data from a USB flash drive

If you’re in the data recovery, forensics or just storage maintenance business (including as an amateur) you probably already know about ddrescue. Released about twenty years ago by Antonio Diaz Diaz, it was a big improvement over the original dd_rescue from Kurt Garloff in 1999. They copy disk images (which are just files in Unix) trying to get as much data extracted when the drive itself has faults.

If you’re using Windows rather than Unix/Linux then you probably want to get someone else to recover your data. This article assumes FreeBSD

The advantage of using either of these over dd is that they expect to find bad blocks in a device and can retry or skip over them. ddrescue is particularly good at retrying, and reducing the block size to recover every last readable scrap – and it treats mechanical drives that are on their last legs as gently as possible.

If you’re new to it, the manual for ddrescue can be found here. https://www.gnu.org/software/ddrescue/manual/ddrescue_manual.html

However, for most use cases the command is simple. Assuming the device you want to copy is /dev/da1 and you’re calling it thumbdrive the command would be:

ddrescue /dev/da1
thumbdrive.img thumbdrive.map

The device would be stored in thumbdrive.img, with on-going state information stored in thumbdrive.map. This state information is important, as it allows ddrescue to pick up where it left off.

However, ddrescue was written before USB flash drives (pen drives, thumb drives or whatever). That’s not to say it doesn’t work, but they have a few foibles of their own. It’s still good enough that I haven’t modified it to cope, using a bit of a shell script to do the necessary.

USB flash drives seem to fail in a different way to Winchester disks. If a block of Flash EPROM can’t be read it’s going to produce a read error – fair enough. But they have complex management software running on them that attempts to make Flash EPROM look like a disk drive, and this isn’t always that great in failure mode. In fact I’ve found plenty of examples where they come across a fault and crash, meaning you have to turn them off and on again to get anything going again (i.e. unplug them and put them back in).

So it doesn’t matter how clever ddrescue is – if it hits a bad block and the USB drive controller crashes the it’s going to be waiting forever for a response and you’ll just have come reset everything and carry on. One of the great features of ddrescue is that it can be stopped and restarted at any time.

However, it is possible to turn a USB device off and on again without unplugging it. Most USB hardware has software control over its power output, and it’s particularly easy on operating systems like FreeBSD to do this from within a shell script. But first you have to figure out what’s where – specifically which device represents your USB drive in /dev and which USB device it is on the system. Unfortunately I can’t find a way of determining it automatically, even on FreeBSD. Here’s how you do it manually; if you’re using a version of Linux it’ll be similar.

When you plug a USB storage device into the system it will appear as /dev/da0 for the first one; /dev/da1 for the second and so on. You can read/write to this device like a file. Normally you’d mount it so you can read the files stored on it, but for data recovery this isn’t necessary.

So how do you know which /dev/da## is your media? This easy way to tell is that it’ll appear on the console on in /var/log/messages if you don’t have a console. You’ll see something like this.

Jun 10 17:54:24 datarec kernel: umass0 on uhub5
kernel: umass0: <vendor 0x13fe USB DISK 3.0, class 0/0, rev 2.10/1.00, addr 2> on usbus1
kernel: umass0 on uhub5
kernel: umass0: on usbus1
kernel: umass0: SCSI over Bulk-Only; quirks = 0x8100
kernel: umass0:7:0: Attached to scbus7
kernel: da0 at umass-sim0 bus 0 scbus7 target 0 lun 0
< USB DISK 3.0 PMAP> Removable Direct Access SPC-4 SCSI device
kernel: da0: Serial Number 070B7126D1170F34
kernel: da0: 40.000MB/s transfers
kernel: da0: 59088MB (121012224 512 byte sectors)
kernel: da0: quirks=0x3
kernel: da0: Write Protected
Please generate and paste your ad code here. If left empty, the ad location will be highlighted on your blog pages with a reminder to enter your code. Mid-Post

So this is telling us that it’s da0 (i.e /dev/da0)

The hardware identification is “<vendor 0x13fe USB DISK 3.0, class 0/0, rev 2.10/1.00, addr 2> on usbus1” which means it’s on USB bus 1, address 2.

You can confirm this using the usbconfig utility with no arguments:

ugen5.1:  at usbus5, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=SAVE (0mA)
...snip...
ugen1.1: at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=SAVE (0mA)
ugen1.2: at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (300mA)

There it is again, last line.

usbconfig has lots of useful commands, but the ones we’re interested are power_off and power_on. No prizes for guessing what they do. However, unless you specify a target then it’ll switch off every USB device on the system – including your keyboard, probably.

There are two ways of specifying the target, but I’m using the -d method. We’re after device 1.2 so the target is -d 1.2

Try it and make sure you can turn your USB device off and on again. You’ll have to wait for it to come back online, of course.

There are ways of doing this on Linux by installing extra utilities such as hub-ctrl (which you’ll have to install) such as hub-ctrl. You may also be able to do it by writing stuff to /sys/bus/usb/devices/usb#/power/level” – see the manual that came with your favourite Linux distro.

The next thing we need to do is provide an option for ddrescue so that it actually times out if the memory stick crashes. The default is to wait forever. The –timeout=25 or -T 25 option (depending on your optional taste) sees to that, making it exit if it hasn’t been able to read anything for 25 seconds. This isn’t entirely what we’re after, as a failed read would also indicate that the drive hadn’t crashed. Unfortunately there’s no such tweak for ddrescue, but failed reads tend to be quick so you’d expect a good read within a reasonable time anyway.

So as an example of putting it all into action, here’s a script for recovering a memory stick called duracell (because it’s made by Duracell).

#!/bin/sh
while ! ddrescue -T 25 -u /dev/da0 duracell.img duracell.map
do
echo ddrescue returned $?
usbconfig -d 1.2 power_off
sleep 5
usbconfig -d 1.2 power_on
sleep 15
echo Restarting
done

A few notes on the above. Firstly, ddrescue’s return code isn’t defined. However, it appears to do what one might expect so the above loop will drop out if it ever completes. I’ve set the timeout for time since last good read to 25 seconds, which seems about right. Turning off the power for 5 seconds and then waiting for 15 seconds for the system to recognise it may be a bit long – tune as required. I’m also using the -u option to tell ddrescue to only go forward through the drive as it’s easier to read the status when it’s always incrementing. Going backwards and forwards makes sense with mechanical drives, but not flash memory.

Aficionados of ddrescue might want to consider disabling scraping and/or trimming (probably trimming) but I’ve seen it recover data with both enabled. Data recovery is an art, so tweak away as you see fit – I wanted to keep this example simple.

Now this system isn’t prefect. I’m repurposing ddrescue, which does a fine job on mechanical drives, to recover data from a very different animal. I may well write a special version for USB Flash drives but this method does actually work quite well. Let me know how you get on.

Quicken doesn’t like Amex QIF format

If you’re trying to import data into Quicken 98 (or compatible) having downloaded it from the American Express web site you’ll get some odd and undesirable results. There are two incompatibilities.

Firstly the date field (D) needs to have a two-digit year, whereas Amex adds the century.

Secondly, for reasons I haven’t quite figured out yet, the extra information M field sometimes results is a blank entry (other than the date).

You can fix this by editing the QIF file using an editor of your choice, but this gets tedious so here’s a simple program that will do it for you. It reads from stdin and writes to stdout so you’ll probably use it in a script or redirect it from and to a file. I compile it to “amexqif”. If there’s any interest I’ll tweak it to make it more friendly.

It’s hardly a complex program, the trick was figuring out what’s wrong in the first place.

#include <stdio.h>
#include <string.h>

char buf [200];

int main()
{
int lenread;
char *str;
        while ((str = fgets(buf, 200,stdin))) // (()) to placate idiot mode on CLANG
        {
                if ((lenread = strlen(str)) > 1)        // Something other than just the newline
                {
                        if (str[0] == 'M')
                                continue;       // Random stuff in M fields breaks Q98


                        if (str[0] == 'D' && lenread == 12)     // Years in dates must be two digit
                                strcpy (str+7, str+9);
                }
                printf("%s",str);
        }
}

Proper Case in a shell script

How do you force a string into proper case in a Unix shell script? (That is to say, capitalise the first letter and make the rest lower case). Bash4 has a special feature for doing it, but I’d avoid using it because, well, I want to be Unix/POSIX compatible.

It’s actually very easy once you’ve realised tr won’t do it all for you. The tr utility has no concept on where in the input stream it is, but combining tr with cut works a treat.

I came across this problem when I was writing a few lines to automatically create directory layouts for interpreted languages (in this case the Laminas framework). Languages of this type like capitalisation of class names, but other names have be lower case.

Before I get started, I note about expressing character ranges in tr. Unfortunately different systems have done it in different ways. The following examples assume BSD Unix (and POSIX). Unix System V required ranges to be in square brackets – e.g. A-Z becomes “[A-Z]”. And the quotes are absolutely necessary to stop the shell globing once you’ve introduced the square brackets!

Also, if you’re using a strange character set, consider using \[:lower:\] and \[:upper:\] instead of A-Z if your version of tr supports it (most do). It’s more compatible with foreign character sets although I’d argue it’s not so easy on the eye!

Anyway, these examples use A-Z to specify ASCII characters 0x41 to 0x5A – adjust to suit your tr if your Unix is really old.

To convert a string ($1) into lower case, use this:

lower=$(echo $1 | tr A-Z a-z)

To convert it into upper case, use the reverse:

upper=$(echo $1 | tr a-z A-Z)

To capitalise the first letter and force the rest to lower case, split using cut and force the first character to be upper and the rest lower:

proper=$(echo $1 | cut -c 1 | tr a-z A-Z)$(echo $1 | cut -c 2- | tr A-Z a-z)

A safer version would be:

proper=$(echo $1 | cut -c 1 | tr "[:lower:]" "[:upper:]")$(echo $1 | cut -c 2- | tr "[:upper:]" [":lower:"])

This is tested on FreeBSD in /bin/sh, but should work on all BSD and bash-based Linux systems using international character sets.

You could, if you wanted to, use sed to split up a multi-word string and change each word to proper case, but I’ll leave that as an exercise to the reader.

Digital Postage Stamps

The Royal Mail hasn’t just lost your item, it’s lost the plot completely.

While the news media has been obsessed with what civil servants might have been doing after work in Downing Street they have overlooked the latest bonkers development from the Post Office – “digital stamps”.

The gimmick is that every new stamp will have a 2D barcode on one side. According to the Royal Mail’s Nick Landon, “Introducing unique barcodes on our postage stamps allows us to connect the physical letter with the digital world and opens up the possibilities for a range of new innovative services in future.” This was followed by promises that it would be possible to link the codes to videos, and by scanning them with an App you could send “birthday messages” and other videos.

Just because something’s possible, it doesn’t mean its a good idea Nick! But what’s the harm, eh?

Well look a bit further – from the start of 2023 you won’t be able to use any of your existing stamps. That’s right – they’re being withdrawn. In a statement Royal Mail has said:

“Mail posted with non-barcoded Definitive stamps after 31 January 2023, will be treated in the same way as if there is insufficient postage on an item….Any item that has insufficient postage is subject to a surcharge. Surcharge fees can be found on our website.”

What you’re supposed to do now is find all your “old fashioned” stamps and post them off to the Royal Mail, who will send you the new digital ones in return. What a waste of time and money – theirs and ours. Why not just accept the old stamps people have paid for until they run out? I’ve asked but received no further comment.

So let’s just assume the Royal Mail hasn’t completely lost its senses and there’s a better reason for this than using an App to “send” Shaun the Sheep videos, or to make money by cancelling stamps already paid for that people won’t get around to replacing in time.

One answer would be to make the stamps machine readable. Possibly, but that’d also make them much easier to forge. You could machine-read an existing stamp anyway; barcode technology is quicker and more forgiving, which is also its weakness.

Perhaps they’re worried about counterfeit stamps? Printing a barcode isn’t difficult. Unless…

I’ve looked at the stamps and they’ve got what’s probably a 47×16 matrix. Allowing for ECC and alignment marks that’s still going to be something like a 480-bit number – enough to give every stamp printed its own serial number from now until the end of time. This would also explain how scanning one could be used to deliver a unique video message to the recipient. If this is the plan – every stamp is unique – they could spot when the same stamp passed through their scanners twice, thus spotting when a forgery has been used.

The flaw in this brilliant plan is that the Royal Mail will have no way of telling if the stamp its currently scanning is the original or the forgery. If a forger has used your stamp number before you did, I predict an almighty row.

Reply-To: gmail spam and Spamassassin

Over the last few months I’ve noticed huge increase is spam with a “Reply To:” field set to a gmail address. What the miscreants are doing is hijacking a legitimate mail server (usually a Microsoft one) and pumping out spam advertising a service of some kind. These missives only work if the mark is able to reply, and as even a Microsoft server will be locked down sooner or later, so they’ll never get the reply.

The reason for sending this way is, of course, spam from a legitimate mail server isn’t going to be blacklisted or blocked. SPF and other flags will be good. So these spams are likely to land in inboxes, and a few marks will reply based on the law of numbers.

To get the reply they’re using the email “Reply-To:” field, which will direct the reply to an alternative address – one which Google is happy to supply them for nothing.

The obvious way of detecting this would be to examine the Reply-To: field, and if it’s gmail whereas the original sender isn’t, flag it as highly suspect.

I was about to write a Spamassassin rule to do just this, when I discovered there is one already – and it’s always been there. The original idea came from Henrik Krohns in 2009, but it’s time has now definitely arrived. However, in a default install, it’s not enabled – and for a good reason (see later). The rule you want is FREEMAIL_FORGED_REPLYTO, and it’s found in 20_freemail.cf

Enabling FREEMAIL_FORGED_REPLYTO in Spamassassin

If you check 20_freemail.cf you’ll see the rules require Mail::SpamAssassin::Plugin::FreeMail, The FreeMail.pm plugin is part of the standard install, but it’s very likely disabled. To enable this (or any other plugin) edit the init.pre file in /usr/local/etc/mail/spamassassin/ Just add the following to the end of the file:

# Freemail checks
#
loadplugin Mail::SpamAssassin::Plugin::FreeMail FreeMail.pm

You’ll then need to add a list of what you consider to be freemail accounts in your local.cf (/usr/local/etc/mail/spamassassin/local.cf). As an example:

freemail_domains aol.* gmail.* gmail.*.* outlook.com hotmail.* hotmail.*.*

Note the use of ‘*’ as a wildcard. ‘?’ matches a single character, but neither match a ‘.’. It’s not a regex! There’s also a local.cf setting “freemail_whitelist”, and other things documented in FreeMail.pm.

Then restart spamd (FreeBSD: service spamd restart) and you’re away. Except…

The problem with this Rule

If you look at 20_freemail.cf you’ll see the weighting is very low (currently 0.1). If this is such a good rule, why so little? The fact is that there’s a lot of spam appearing in this form, and it’s the best heuristic for detecting it, but it’s also going to lead to false positives in some cases.

Consider those silly “contact forms” beloved by PHP Web Developers. They send an email from a web server but with a “faked” reply address to the person filling in the form. This becomes indistinguishable from the heuristic used to spot the spammers.

If you know this is going to happen you can, of course add an exception. You can even have the web site use a local submission port and send it to a local mailbox without filtering. But in a commercial hosting environment this gets a bit complicated – you don’t know what Web Developers are doing. (How could you? They often don’t).

If you have control over your users, it’s probably safe to up the weighting. I’d say 3.0 is a good starting point. But it may be safer to leave it at 0.1 and examine the results for what would have been false positives.

It’s the LAW (GDPR as an excuse)

GDPR

In the 2000s it was “It’s necessary for our QA procedure”. Now it’s GDPR. Basically, the technical sounding response to shut people up when they complain. As a qualified ISO-9000 auditor I used to had a lot of fun calling their bluff in the first case.

With data protection it might seem more clear cut than having an encyclopaedic knowledge of ISO9000:2000. After all DPA 2018 (that which implemented GDPR) isn’t that dissimilar to its predecessors, and has a much tighter scope. However, it’s more open for interpretation and we’re waiting for some test cases.

However, what it doesn’t cover are situations like this:

Dear Mr Leonhardt, 
Hope you're well; It is law to speak to the account holder.
Kind regards, Salvin Tingh
Morrisons Online Customer Service Team

I won’t bore you with the full details of what led to this attempted put-down, but briefly I emailed Morrisons about a mistake they’d made on an order. On receiving no response I called (and they sorted it out efficiently, over the phone). A week later I got an email response, and I said it was too late but it was sorted out, thanks very much. A week later, another reply that suggested they hadn’t read the first one. I said “Sorted, thanks, and I’ll just use the ‘phone in future”.

Next week’s reply was along the lines that they couldn’t verify I was the customer. I replied that perhaps they should have tried (they know my email address and telephone number), but don’t worry it’s sorted. A week later the above arrived (name changed to protect the guilty).

Leaving aside the principles of good customer service – if you need to check someone’s identity before solving a problem then do so – one might wonder what law he might be talking about. You see, data protection laws are not as wide-ranging as people think.

Basically, the law relates to sensitive information about an identifiable individual. Stronger protections exist depending on the sensitivity of the information (e.g. race, religion, biometrics and the usual stuff). But if it’s not sensitive information about an identifiable information it’s definitely out-of-scope.

In this case, Mr Tingh was dealing with a customer’s problem. He wasn’t being asked to divulge sensitive information to a possible third party. It’s possible (and desirable) that company procedures required that he make sure it really was the customer complaining, but that’s hardly “the law”. And had I been an imposter claiming I hadn’t received my sausage, the worst that would happen was someone else got a couple of quid refunded unexpectedly. Does Morrisons get that kind of thing often, one wonders?

And it also begs the question, if they were so concerned about whether a customer complaint about an order, emailed in with the full paperwork, really was from the household in question they need only pick up the phone; or check the email address? Neither of these is fool proof, but in the circumstances one might have thought this good enough. Did he want me to visit the shop show the manager my passport?

But to reiterate, The Data Protection Act (colloquially referred to as GDPR) is there to protect information pertaining to an individual. A company would have a duty to ensure it was talking to the right person if giving out sensitive information, but when someone is reporting the non-delivery of a vegan sausage to the suppler there is no sensitive information involved. They only need to check your identity if its really necessary.

Other protections in the DPA include transparent use of an individual’s data, not storing more than is necessary or for longer than necessary, and ensuring it’s accessible to the individual concerned, not leaked and is accurate (corrected if needs be). The European GDRP added provisions for portability, forcing companies to make your data available to competing services at your request.

So when someone tries to fob you off with “data protection”, stop and think if the above actually apply. And if you’re trying to fob someone off, don’t try to bluff a data security expert.

EU to Ban Russia Today (RT)

I’ve just heard Ursula von der Leyen (President of the European Commission) announce to the European Parliament that Kremlin propaganda network Russia Today (now restyled RT) is to have its license to operates revoked (i.e. it’s being banned). This is a terrible idea.

There are many media organisations committed to undermining the western “establishment” using a heavily slanted narrative. However, allowing them to exist is what makes us different from Putin’s Russia. By banning one of them we’d be playing the same game as Putin, the CPC and every other repressive regime around the world.

Putin won’t invade Ukraine

Putin’s Russia isn’t about to invade Ukraine, and neither will China be marching in to Taiwan. They’ll be doing nothing until the Winter Olympics are over.

It’s not complicated. Putin wants to assert Russian influence over the former Soviet Union and has no intention of allowing one of the Ukrainian factions to take the country in the direction of Poland and other Warsaw Pact allies. This has nothing to do with Russia’s security – to believe otherwise you’d have to think the West had designs for an invasion of Russia. This is nonsense. Putin probably isn’t paranoid.

Western media paints the Ukrainian situation as a noble government in Kiev and a bunch of evil separatists. It’s not. It’s an ongoing nasty civil war, with no end in sight. Because of the geopolitical situation, no one is doing anything about it.

Some factions in Ukraine, which is a big place, have neo Nazi roots. They’re not the separatists. The Russians have a history with the Ukrainian Nazis, and seeing them on the streets of Kiev is going to have an effect of their thinking. The President, government and most Ukrainians are certainly not Nazi sympathisers, of course – far from it.

My guess would be that Putin has had enough of the instability on his boarders and plans to do something about it. This would be logical, but it will be the people of the region that suffer if this takes the form of military action. I fear that it will, but after the sport is over.

Note to history

When this was written, the BBC was reporting that Russia was pulling its troops and tanks back from the border and Putin’s insistence that he had no plans to invade. The headline is mocking the BBC and other media outlets. As predicted, Russia declared the pro Russian areas of Ukraine independent of Kiev the day after the Olympics ended, and at time of writing, is believed to have sent (more) troops in to “defend” them – i.e invaded Ukraine.

Nothing new with Intel SDSi

Intel’s latest wheeze for its CPUs is Software Defined Silicone (SDSi). The deal is that you buy the CPU at one price and then pay extra for a license to enable more stuff.

If you want the geeky stuff about how it’s supposed to work in Linux, see here. https://github.com/intel/intel-sdsi

Basically, the CPU has an interface that you can access if you have an Authentication Key Certificate (AKC) and have purchased a Capability Activation Payload (CAP) code. This will then enable extra stuff that was previously disabled. Quite what the extra stuff is remains to be seen – it could be extra instructions or enabling extra cores on a multi-core chip, or enabling more of the cache. In other words, you buy extra hardware that’s disabled, and pay extra to use it. What’s even more chilling is that you could be continuously paying licenses for the hardware you’ve bought or it’ll stop working.

It’s not actually defining the silicone in software like a FPGA, as you’d expect from euphemistic name. Software Defined Uncrippling would be more honest, but a harder sell.

But this is nothing new. I remember IBM doing this with disk drives in the 1970’s. If you upgraded your drive to double the capacity an IBM tech turned up and removed a jumper, enabling the remaining cylinders. Their justification was that double the capacity meant double the support risk – and this stuff was leased.

Fast forward 20 years to Intel CPUS. Before the Intel 80486 chips you could provide whatever input clock you wanted to your 80386, just choosing how fast it went. Intel would guarantee the chip to run at a certain speed, but that was the only limiting factor. Exceed this speed at your own risk.

The thing was that the fast and slow CPUs were theoretically identical. It’s often the case with electronic components. However, manufacturing tolerances mean that not all components end up being the same, so they’re batch tested when the come off the line. Those that pass the toughest test get stamped with a higher speed and go in the fast bucket, where they’re sold for more. Those that work just fine at a lower speed go into the slower bucket and sell for less. Fair enough. Except…

It’s also the nature of chip manufacture that the process improves over time, so more of the output meets the higher test – eventually every chip is a winner. You don’t get any of the early-run slow chips, but you’re contracted to sell them anyway. The answer is to throw some of the fast chips into the slow bucket and sell them cheap, whilst selling others at premium price to maintain your margins.

In the early 1990’s I wrote several articles about how to take advantage of this in PCW, after real-world testing of many CPUs. It later became known as overclocking. I also took the matter up with Intel at the time, and they explained that their pricing had nothing to do with manufacturing costs, and everything to do with supply and demand. Fair enough – they were honest about it. This is why AMD gives you more bang-per-buck – they choose to make things slightly better and cheaper because that maximises their profits too.

With the introduction of the 80486, the CPU clock speed was set in the package so the chip would only run at the speed you paid for. SDSi is similar, except you can adjust the setting by paying more at a later date. It also makes technical sense – producing large quantities of just one chip has huge economies of scale. The yield improves, and you just keep the fab working. In order to have a product range you simply knobble some chips to make them less desirable. And using software to knobble them is the ultimate, as you can decide at the very last minute how much you want to sell the chip for, long after it’s packaged and has left the factory.

All good? Well not by me. This only works if you’re in a near monopoly position in the first place. Microsoft scalps its customers with licenses and residual income, and Intel wants in on that game. It’s nothing about being best, it’s about holding your customers to ransom for buying into your tech in the first place. This hasn’t hurt Microsoft’s bottom line, and I doubt it’ll hurt Intel’s either.

STARTTLS is not a protocol

As regular readers will know, I’m not a fan of STARTTLS but today I realised that some people are confused as to what it even means. And there’s a perfectly good reason for this – some graphical email software is actually listing STARTTLS as a protocol for talking to mail servers and people are jumping to conclusions.

So what is STARTTLS all about if you go back to basics?

Originally, when only nice people had access to computers, network traffic was unencrypted. If you had physical access to the network you could pretty much read anything you wanted to, as everything connected to the same network saw the same data. This isn’t true now, but encryption you data is a good idea just in case it can be intercepted – and if it’s going over the Internet that’s definitely the case.

In the mid 1990s, the original mass-market web browser, Netscape, decided to do something about it and they (or more specifically their chief scientist Taher Elgamal, invented a protocol called Secure Sockets Layer (SSL) to protect HTTP (web) traffic. Actually, several times as the first couple of attempts weren’t very secure at all.

SSL didn’t really fit in with the OSI model; it runs on top of the transport protocol (usually TCP) but under the presentation layer, which would logically handle encryption but doesn’t usually. To use it you need an SSL layer added to the stack to transparently do the deed on a particular port.

But, as a solution to the encryption problem, SSL took off and pretty much every major protocol has an SSL port along with its original cleartext one. So clear HTTP is on port 80, HTTPS is on port 443. Clear POP3 is on port 110, encrypted on 995. Clear IMAP is on port 143, encrypted on 993.

As is the way of genius ideas in cybersecurity, even the third version of SSL was found to be full of holes. SSL version 3.1, which was renamed TLS, continued plugging the leaks and by TLS 1.2 it’s considered pretty much secure now. TLS 1.3, which interoperates with TLS 1.2, simply deprecates certain cyphers and hashes on the suspicion they might be insecure; although anyone into cybersecurity should tell you that everything is secure only until it’s broken.

Unfortunately, because different levels of TLS use different cyphers and reject others, TLS levels are by no means interoperable. And neither is it the case that a newer version is more secure; bugs have been introduced and later fixed. This’d be fine if everything and everyone used the same version of TLS, but in the real world this isn’t practical – old hardware, in particular, bakes in old versions of SSL or TLS and if you decided to deprecate older cyphers and not work with them, you loose the ability to talk to your hardware.

But apart from this, things were going along pretty well; and then someone had the bright idea of operating encrypted and unencrypted connections on the same port by hacking it at the application layer instead. This was achieved by modifying the application protocol to include a STARTTLS command. If this is received, the application then negotiates a TLS connection. If the receiving host didn’t understand what STARTTLS meant it’d send back and error, and things could continue unencrypted.

In other words, if you’re implementing an SMTP server with STARTTLS, this keyword is added to the protocol and the SMTP server does something about it when it sees it.

What could go wrong?

Well quite a lot of things, actually. Because TLS doesn’t fit in to the OSI model, it’s actually very difficult to deal with the situation where a TLS connection is requested and agreed to but the TLS layer fails to agree on a cypher with an older or newer version on the other end. There’s no mechanism for passing this to the application to say “okay, let’s revert to Plan A”, and the connection tends to hang.

There’s also a problem with name-based virtual servers must all use the same host certificate because the TLS connection must be established before the application layer headers are transferred.

But perhaps my biggest gripe is that enabling STARTTLS makes encryption optional. You’re not enforcing encryption when you need to, and even if you think you are, STARTTLS connection are obviously vulnerable to a man-in-the-middle attack. You have no idea how many times TLS has been turned on and off between the two endpoints.

You might be tempted to think that optional encryption is better than none at all, but in reality it means you don’t care – and if you don’t care, don’t bother. It just leads to a false sense of security. And it can lead to interoperability problems. My advice is to use “always TLS” ports for sensitive data and turn off the old port.