Fil-C Makes setjmp/longjmp and ucontext Memory Safe

Fil-C, the memory-safe C dialect, now supports setjmp/longjmp and the ucontext APIs (setcontext, getcontext, makecontext, swapcontext) without stack corruption or capability violations. The feature landed in release 0.680 and is available when building from source.
The Problem with Context APIs
These APIs are notoriously unsafe because misuse can restore a dangling stack. Common bugs include:
- Calling
setjmporgetcontextin a function, then returning — the saved context points to a stack frame that no longer exists. - Exiting the thread, then trying to restore execution on a freed stack.
- Creating a context with
makecontexton a stack, freeing that stack, thenswapcontextorsetcontextto it. - Passing the currently executing context as the second argument to
swapcontext(swapping to itself).
In standard C ("Yolo-C"), these bugs cause silent stack corruption, hard-to-debug crashes, and potential security exploits. In Fil-C, all such cases produce a panic at the point of misuse.
How Fil-C Implements Memory Safety
Fil-C's approach differs for setjmp/longjmp vs ucontext. For setjmp/longjmp, the key challenge is that setjmp returns twice — once when called, again after longjmp. The act of saving the context means the compiler needs to handle volatile-annotated variables correctly, but Fil-C ensures that restoring a context never accesses freed or invalid memory.
For ucontext, Fil-C manages stacks in a way that either makes the operation legal or panics, because dangling stack frames are simply impossible.
Example: setjmp/longjmp in Fil-C
The following program demonstrates the behavior:
#include <setjmp.h>
#include <stdio.h>
int main(int argc, char** argv) {
volatile int x = 42;
jmp_buf jb;
if (setjmp(jb)) {
printf("x = %d\n", x);
return 0;
}
x = 666;
longjmp(jb, 1);
printf("Should not get here.\n");
return 1;
}
This prints x = 666 and exits. Without volatile, the compiler may optimize and print 42 instead. Fil-C does not alter optimization semantics but prevents memory corruption regardless.
For Whom?
Developers using ucontext-based coroutines (e.g., Boost fibers) or setjmp/longjmp for exception handling in C programs that want memory safety.
📖 Read the full source: HN AI Agents
👀 See Also

AppLovin Mediation Cipher Broken: Device Fingerprinting Bypasses ATT
Reverse-engineering revealed that AppLovin's custom cipher uses a constant salt + SDK key, a SplitMix64 PRNG, and no authentication. Decrypted requests carry ~50 device fields (hardware model, screen size, locale, boot time, etc.) even when ATT is denied, enabling deterministic re-identification across apps.

Security Analysis of AI Agents Reveals Broken Trust Model and High Vulnerability Rates
A security analysis of AI agents shows the fundamental trust model is broken, with 49% of MCP packages having security findings and indirect injection achieving 36-98% attack success rates across state-of-the-art models.

Claude Code VS Code Extension Leaks Selection State Across Closed Files and New Sessions
A bug in Claude Code's VS Code extension caches file selection state even after the file is closed, exposing sensitive data (e.g., Supabase service-role keys) to a brand new CLI session. Full repro steps and GitHub issue #58886.

Anthropic's Computer-Use Feature Triggers Governance Lockdown in Real Test
Anthropic shipped computer-use capabilities, and during implementation of governance controls, a risk threshold triggered a LOCKDOWN posture that blocked all mutating operations including the operator's own governance work.