fix(memtrack): track successful posix_memalign allocations

[not-matthias]

Problem

posix_memalign was attached to the shared memalign uprobe handler (UPROBE_ARG_RET), which assumes the standard allocator ABI. But posix_memalign(void** memptr, size_t alignment, size_t size) is inverted:

• Returns int 0 on success (nonzero errno on failure), not a pointer.
• Delivers the allocated pointer through the memptr out-parameter, not the return register.
• Size lives in PARM3 (PARM2 is the alignment).

The shared handler gated success on a non-NULL return and used the return register as the address. Since successful posix_memalign returns 0, every successful allocation was treated as a failure and silently dropped.

Fix

Dedicated uprobe/uretprobe pair for posix_memalign:

• uprobe stashes memptr (PARM1) and size (PARM3) keyed by tid.
• uretprobe gates on RC == 0, reads the address back from *memptr via bpf_probe_read_user, then submits an aligned-alloc event.

Test

Added posix_memalign_test to the existing rstest table in tests/c_tests.rs with a fixture + snapshot confirming the allocation is now tracked.

[greptile-apps]

Greptile Summary

Introduces ABI-correct tracking for successful posix_memalign allocations.

• Adds dedicated entry and return probes that preserve the output pointer location and allocation size.
• Attaches the dedicated probes to standard and allocator-prefixed posix_memalign symbols.
• Adds successful and invalid-alignment integration fixtures and updates allocator snapshots.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failures remain.

Important Files Changed

Filename	Overview
crates/memtrack/src/ebpf/c/allocator.h	Adds dedicated ABI-aware entry and return probes for posix_memalign.
crates/memtrack/src/ebpf/memtrack/allocator.rs	Routes discovered posix_memalign symbols to the new dedicated probe pair.
crates/memtrack/tests/c_tests.rs	Registers success and invalid-alignment integration cases for allocation tracking.
crates/memtrack/testdata/alloc_cpp/src/main.cpp	Extends allocator compatibility coverage with a posix_memalign allocation.
crates/memtrack/testdata/posix_memalign_test.c	Exercises successful posix_memalign tracking and cleanup.
crates/memtrack/testdata/posix_memalign_einval.c	Verifies failed posix_memalign calls do not emit allocation events.

Sequence Diagram

sequenceDiagram
    participant App as Tracked process
    participant Entry as posix_memalign uprobe
    participant Allocator as posix_memalign
    participant Return as posix_memalign uretprobe
    participant Events as Memtrack event stream
    App->>Entry: "posix_memalign(&ptr, alignment, size)"
    Entry->>Entry: Store memptr and size by TID
    Entry->>Allocator: Continue call
    Allocator-->>Return: Return status
    alt "status == 0"
        Return->>App: "Read allocated address from *memptr"
        Return->>Events: Submit AlignedAlloc(size, address)
    else "status != 0"
        Return-->>Return: Drop failed allocation
    end

Loading

_{Reviews (5): Last reviewed commit: "test(memtrack): exercise posix_memalign ..." | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will not alter performance

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

✅ 17 untouched benchmarks

---

_{Comparing cod-3238-memtrack-drops-all-successful-posix_memalign-allocations (e56e9ae) with main (8018219)}

[GuillaumeLagrange]

LGTM, I'm guessing you've double checked that we do not double dip and cannot count allocations twice because of this change