⛤ Sol Defectus Æternus ⛤ ❍ 🝶

Joined July 2026
Replying to @WildMediaOnly
RPGs are Bearable Arms. *points to cannon*
2
1,033
I have achieved a magic trick! My system has a series of function pointers in an arbitrary stack; but it needs to init the thread stack. Instead of using a global variable, I write the address as a function pointer, Preemptively extract it, and replace with an empty function.
8
Now the hard part. In order to use the threads, I need to modify the %rsp and %rbp to point to the generated stack for the program image. but this function is global, and I cannot simply pass a parameter into it. Need a secure way to get my true address into this function.
7
This comment is the mark of truly professional software.
6
Replying to @specialopsmag
This guy may be flipping the bird
5
11,458
Session handling for nerds and weirdos
3
Core session is always 0, and always comes first. Containerized sessions are given an ID that's generated via session count and PID. To actually interface, via signals, with the container, I also have a PidFD, which is more reliable than the PID itself. Pretty happy with this!
18
Confirmed! We can see that process 1 acquires a lock via the kernel sem_lock() function. Process 1 then goes and waits for zero. We then see Process hit DEBUG5, the sem_lock(), and never gets passed it. Now to see why wf0 deadlocks when preceded by increment ops
5