Ben Fiedler

'changeme' is valid base64

Base64-encoding is ubiquitous in our modern world. Many programs communicate in base64-encoded messages, since these have nice properties: They consist of a very limited subset of ASCII characters, and can thus be displayed as text. E-Mails for example are commonly encoded in base64 in transfer, which can be seen in the Content-Transfer-Encoding header.

Where I volunteer we have a tradition of putting changeme as a placeholder for many template variables indicating that they should be overwritten later. Thus, if we later still see changeme anywhere we know that somewhere somebody must have forgotten to specify a variable1.

Sometimes these template variables should correspond to base64-encoded content, i.e. the base64-decoding of said variable should produce meaningful content. But what happens if somebody forgets to define said variable, and it is substituted with changeme instead? I always thought that it would fail to decode the string since the probability that changeme is actually valid base64 encoding must be very low. But lo and behold, this is not the case:

% echo changeme | base64 -d
r�[1m

Running hexdump on the output we can identify the characters:

% echo changeme | base64 -d | hexdump -C
00000000  72 16 a7 81 e9 9e      |r.....|
00000006

Mostly nonsense, but the second byte is interesting: It is an ASCII control character 2, meaning trips up all sorts of programs which expect to read printable ASCII characters, such as JSON deserializers.

This bug actually occurred when I was migrating some applications between different Kubernetes clusters: In the old cluster the application ran without problems, but I forgot to specify some variable in the new cluster, and it lead to Go complaining about non-ASCII characters in a JSON string.

I guess the moral of the story is: If you use dummy values be sure that they cannot be misinterpreted.

Articles from blogs I follow

What's cooking on Sourcehut? January 2021

Another year begins, and hopefully with better prospects for us all. SourceHut has emerged from 2020 relatively unscathed, thankfully, and I hope the same is true of most of our users. A body which, by the way, today numbers 19,647 strong, up 623 from Decemb…

via Blogs on Sourcehut on

New PGP Key Fingerprint

New PGP Key Fingerprint This morning I got an encrypted email, and in the process of trying to decrypt it I discovered that I had lost my PGP key. I have no idea how I lost it. As such, I have created a new PGP key and replaced the one on my website with it. …

via Christine Dodrill's Blog on

Status update, January 2021

Hello from the future! My previous status update was last year, but it feels like it was only a month ago. I hope you didn't miss my crappy jokes too much during the long wait. One of the advancements that I would like to mention this month is the genera…

via Drew DeVault's blog on