ssh_tunnel_vision
Joined 31 March 2024
Posts in the forum
-
Replied in Lab writeup: the box where the 'vulnerability' was a forgotten .bak file
A tidy habit for the threat-model crowd: give each credential the **smallest scope** that works, so when one leaks the damage is small. A database user that can only read two tables is a very differe…
-
Replied in How a pin tumbler lock binds: it is all about tolerances (locksport, own locks only)
Good framing at the end. Threat model: a pin tumbler on a door stops casual opportunists and slows determined ones. If your actual risk is a determined person, the question is the door, the frame, th…
-
Replied in Android's app sandbox is just Linux users (and I think that is neat)
Keep in mind what the sandbox does and does not do. It protects apps from *each other*. It does not protect you from the permissions you grant an app. "Allow access to contacts" is a decision, and th…
-
Replied in Volatile first: the order of evidence on a suspect Linux box (in your own lab)
Contention point: isolate or keep it connected? Pulling the network preserves the box but loses live connection state, and some malware (in a lab, say) behaves differently when it cannot phone home. …
-
Replied in Container vs VM: what is actually shared, and what is isolated?
Yes. Shared kernel means shared attack surface: every syscall a container can make is code in the host kernel. A VM boundary has a much smaller interface (the hypervisor), which is why people who nee…
-
Replied in Container vs VM: what is actually shared, and what is isolated?
Your feeling is right. A container is a normal process (or a group of them) on the host kernel, wrapped in: - **namespaces**: separate views of PIDs, mounts, network, hostname, IPC, users, cgroups, …
-
Replied in X-Forwarded-For: el encabezado que cualquiera puede escribir
Also worth saying that "the first value is the real client" is the classic bug, and the mirror one is just as bad: trusting the *last* value when your own proxy chain has two layers. Draw the chain, …
-
Replied in DNS over TLS to your own resolver: what it hides and what it does not
Yes, that is the one that bites everybody: fail-open fallback. Decide up front: should name resolution fail, or should it degrade to plaintext? That is a policy question, not a technical one. I choos…
-
Started DNS over TLS to your own resolver: what it hides and what it does not
Threat model first, config second. Classic DNS is plaintext on UDP/TCP port 53. Anyone on the path between you and the resolver (the café network, a shared router) can read which names you look up a…
-
Replied in Permission denied? Just chmod -R 777 and move on with your life
`777` means every user on the system can read, write and execute everything in the tree. On a web server, that includes the account the web server runs as. Any bug in the application that can write a…
-
Replied in What is actually shared between a container and the host?
And to answer the second question honestly: it is a smaller barrier than a VM boundary. A VM exposes a small hardware-like interface to the host; a container exposes the whole system call surface of …
-
Replied in HTTP/2 over TCP vs HTTP/3 over QUIC: the head-of-line blocking story
From a threat-model view: 0-RTT early data is replayable, so anything non-idempotent (a POST that changes state) must not be accepted in it. Your servers have to enforce that, not hope the client beh…
-
Replied in Order of volatility: what to capture first and why
Add: write the output of your tools to a separate medium. If you dump RAM onto the disk you are trying to preserve, you just overwrote part of the evidence. Threat model your own investigation.
-
Replied in Compilaciones reproducibles: mismo código, mismos bytes (y por qué te importa)
(In English, sorry.) Reproducibility is great, but remember what it proves: that the binary matches the source. It does not prove the source is benign. Pair it with review of what goes into the sourc…
-
Replied in I wanted to rename a host. Day four.
Do not forget `~/.ssh/known_hosts` on every client, and any firewall rule or allow-list that used the name. Names in access rules are a quiet way to lose access, or a quiet way to keep it for the wro…
-
Replied in Threat-modelling your dependencies: what pinning does and does not buy you
Right. "Two people to publish" is the cheapest control that exists, and few small projects have it. Trust is a process, not a person.
-
Started Threat-modelling your dependencies: what pinning does and does not buy you
Every project I audit has a `requirements.txt` or a lockfile and a feeling of safety. So let me do the threat model out loud. **What you trust when you add a dependency** 1. The author of that packa…