Processing a complex syntax with Rust’s declarative macros

I am by no means a Rust expert, and I’m no expert on declarative macros. I was merely solving an issue I had: there was a lot of redundancy in the way my error types were coded. I didn’t want to repeat the same coding patterns all the time, yet the types also seemed too heterogenous for declarative macros. And going with procedural macros wasn’t worth the effort in this case.

After some experimentation I figured out how declarative macros could work in this scenario after all. I doubt that this approach is very original, and maybe I solved something in a sub-optimal way. I couldn’t find any helpful online sources however, which is why I’m documenting this in my own article. Edit: After publishing this article, I found that the pattern used here is called TT munching.

Documenting KeePass KDBX4 file format

I’ve been playing around with KeePass databases. One aspect was rather surprising: given how many open source products use this format, it is remarkably underdocumented. At best, you can find outdated and incomplete descriptions by random people. The KeePass developers themselves never bothered providing complete documentation. All you get is a semi-intelligible list of changes from KDBX 3.1 to KDBX 4 and from KDBX 4 to KDBX 4.1. With the starting point not being documented, these are only moderately useful.

And so it’s not surprising that the implementations I looked at aren’t actually implementing the same file format. They all probably manage to handle common files in the same way, but each of them has subtle differences when handling underdocumented format features.

I’ll try to explain the format and the subtle details here. For that, I looked at the source code of KeePass, KeePassXC, keepass-rs Rust library and the kxdbweb JavaScript library. Let’s hope this documentation helps whoever else needs to work with that file format, and studying source code will no longer be required.

I can only document the latest version of the format (KDBX 4.1), though I’ll try to highlight changes wherever I’m aware of them.

The end of PfP: Pain-free Passwords

Seven years ago I created a password manager. And a few days ago I pushed out the last release for it, notifying users that nothing else will come now. Yes, with the previous release being from 2019, this might have been obvious. Now it’s official however, PfP: Pain-free Passwords is no longer being developed.

Screenshot of a message titled “PfP: Pain-free Passwords is being discontinued.” The text says: “All good things must come to an end. It has been seven years since PfP was first introduced, back then under the name EasyPasswords. It has features that no other password manager can match, and I still like it. Unfortunately, developing a good password manager is lots of effort. I notice that I’ve been neglecting PfP, and this won’t change any more. So I’m now making official what has been somewhat obvious already: PfP is no longer being developed. Does this mean that you can no longer use it? You can. But whatever breaks now stays broken. Sync to Google Drive in particular is already broken and non-trivial to fix unfortunately. What can you use instead? Not LastPass please. Maybe Bitwarden or 1Password. Definitely KeePass or a clone if you don’t mind it being less intuitive. Personally, I’ll be using KeePassXC. How to get your passwords over? On the “All Passwords” page, there is a new button for CSV export. It should be possible to import the resulting file into any password manager. In KeePassXC you’ll need to check “First line has field names,” otherwise the default import settings will do.”

I certainly learned a lot from this adventure, and I really like the product which came out of this. Unfortunately, a password manager is a rather time-consuming hobby, and my focus has been elsewhere for a while already.

This doesn’t mean that PfP is going away. In fact, it will probably work well for quite a while until a browser change or something like that makes it unusable. Sync functionality in particular depends on third parties however, and this one already started falling apart.

Converting incoming emails on the fly with OpenSMTPD filters

This little adventure began with me being annoyed at DMARC aggregate reports. My domain doesn’t have enough email traffic to justify routing DMARC emails to some third-party analytics service, yet I want to take a brief glance at them. And the format of these emails makes that maximally inconvenient: download the attachment, unpack it, look through some (always messy but occasionally not even human-readable) XML code. There had to be a better way.

This could have been a Thunderbird extension, processing the email attachment in order to produce some nicer output. Unfortunately, Thunderbird extensions no longer have this kind of power. So I went for another option: having the email server (OpenSMTPD) convert the email as it comes in.

Since I already had the implementation details of OpenSMTPD filters figured out, this wasn’t as complicated as it sounds. The resulting code is on GitHub but I still want to document the process for future me and anyone else who might have a similar issue.

Veraport: Inside Korea’s dysfunctional application management

Note: This article is also available in Korean.

As discussed before, South Korea’s banking websites demand installation of various so-called security applications. At the same time, we’ve seen that these applications like TouchEn nxKey and IPinside lack auto-update functionality. So even in case of security issues, it is almost impossible to deliver updates to users timely.

And that’s only two applications. Korea’s banking websites typically expect around five applications, and it will be different applications for different websites. That’s a lot of applications to install and to keep up-to-date.

Luckily, the Veraport application by Wizvera will take care of that. This application will automatically install everything necessary to use a particular website. And it will also install updates if deemed necessary.

Laptop with Veraport logo on the left, three web servers on the right. First server is labeled “Initiating server,” the arrow going from it to the laptop says “Get policy from banking.example.” Next web server is labeled “Policy server,” the arrow pointing from the laptop to it says “Installation policy?” and the arrow back “Install app.exe from download.example.” The final web server is labeled “Download server” and an arrow points to it from the laptop saying “Give me app.exe.”

If this sounds like a lot of power: that’s because it is. And so Veraport already made the news as the vehicle of an attack by North Korean hackers.

Back then everybody was quick to shift the blame to the compromised web servers. I now took a deeper dive into how Veraport works and came to the conclusion: its approach is inherently dangerous.

As of Veraport (released on February 28), all the reported security issues seem to be fixed. Getting users to update will take a long time however. Also, the dangerous approach of allowing Veraport customers to distribute arbitrary software remains of course.

LastPass breach update: The few additional bits of information

Half a year after the LastPass breach started in August 2022, information on it remains sparse. It took until December 2022 for LastPass to admit losing their users’ partially encrypted vault data. This statement was highly misleading, e.g. making wrong claims about the protection level provided by the encryption. Some of the failures to protect users only became apparent after some time, such as many accounts configured with a dangerously low password iterations setting, the company hasn’t admitted them to this day.

Screenshot of an email with the LastPass logo. The text: Dear LastPass Customer, We recently notified you that an unauthorized party was able to gain access to a third-party cloud-based storage service which is used by LastPass to store backups. Earlier today, we posted an update to our blog with important information about our ongoing investigation. This update includes details regarding our findings to date, recommended actions for our customers, as well as the actions we are currently taking.

Despite many questions being raised, LastPass maintained strict radio silence since December. Until yesterday they published an article with details of the breach. If you were hoping to get answers: nope. If you look closely, the article again carefully avoids making definitive statements. There is very little to learn here.

TL;DR: The breach was helped by a lax security policy, an employee was accessing critical company data from their home computer. Also, contrary to what LastPass claimed originally, business customers using Federated Login Services are very much affected by this breach. In fact, the attackers might be able to decrypt company data without using any computing resources on bruteforcing master passwords.

Update (2023-02-28): I found additional information finally explaining the timeline here. So the breach affects LastPass users who had an active LastPass account between August 20 and September 16, 2022. The “Timeline of the breach” section has been rewritten accordingly.

South Korea’s banking security: Intermediate conclusions

Note: This article is also available in Korean.

A while back I wrote my first overview of South Korea’s unusual approach to online security. After that I published two articles on specific applications. While I’m not done yet, this is enough information to draw some intermediate conclusions.

The most important question is: all the security issues and bad practices aside, does this approach to banking security make sense? Do these applications have the potential to make people more secure when rolled out mandatorily nation-wide?

Message on stating: [IP Logger] program needs to be installed to ensure safe use of the service. Do you want to move to the installation page?

TL;DR: I think that the question above can be answered with a clear “no.” The approaches make little sense given actual attack scenarios, they tend to produce security theater rather than actual security. And while security theater can sometimes be useful, the issues in question have proper solutions.

Automating Windows installation in a VM

I recently switched from VirtualBox to KVM for my virtualization needs. While this approach has clear advantages such as not requiring custom kernel drivers, the downside is that snapshots aren’t currently supported for Windows 11. And since I don’t want applications I analyze to corrupt my main Windows VM, I decided that I should run these in freshly created Windows VMs.

The issue with this approach is: setting up a new Windows VM is fairly time-consuming. Not only is it necessary to answer a number of questions during installation, installing the proper guest tools for KVM is non-trivial as well. And I also rely on some further applications for my work.

Luckily, Windows installation supports answer files which make this task far easier. With the right answer file and some scripts, setting up a Windows VM is a fully automated task that takes half an hour and none of my attention. The necessary information was rather scattered and often outdated, so I decided to put it all into one blog post.

Screenshot of Windows Setup installing Windows

Weakening TLS protection, South Korean style

Note: This article is also available in Korean.

Normally, when you navigate to your bank’s website you have little reason to worry about impersonations. The browser takes care of verifying that you are really connected to the right server, and that your connection is safely encrypted. It will indicate this by showing a lock icon in the address bar.

So even if you are connected to a network you don’t trust (such as open WiFi), nothing can go wrong. If somebody tries to impersonate your bank, your browser will notice. And it will refuse connecting.

Screenshot of an error message: Did Not Connect: Potential Security Issue. Firefox detected a potential security threat and did not continue to because this website requires a secure connection.

This is achieved by means of a protocol called Transport Layer Security (TLS). It relies on a number of trusted Certification Authorities (CAs) to issue certificates to websites. These certificates allow websites to prove their identity.

When investigating South Korea’s so-called security applications I noticed that all of them add their own certification authorities that browsers have to trust. This weakens the protection provided by TLS considerably, as misusing these CAs allows impersonating any website towards a large chunk of South Korean population. This puts among other things the same banking transactions at risk that these applications are supposed to protect.

Password strength explained

The conclusion of my blog posts on the LastPass breach and on Bitwarden’s design flaws is invariably: a strong master password is important. This is especially the case if you are a target somebody would throw considerable resources at. But everyone else might still get targeted due to flaws like password managers failing to keep everyone on current security settings.

There is lots of confusion about what constitutes a strong password however. How strong is my current password? Also, how strong is strong enough? These questions don’t have easy answers. I’ll try my best to explain however.

If you are only here for recommendations on finding a good password, feel free to skip ahead to the Choosing a truly strong password section.