Should we consider goto-based error handling?

[haochengxia]

Context: Our codebase uses a lot of malloc without error handling in pure C style.

One proposal: like the Linux kernel (e.g., mux.c)

    parent_names = kcalloc(num_parents, sizeof(char *), GFP_KERNEL);
    if (!parent_names)
        goto cleanup;

cleanup:
	kfree(parent_names);

Consider this when reviewing PR #184: Is it necessary at this stage?

[1a1a11a]

We could, I tried to write our own malloc a long time ago, but the maintenance becomes increasingly hard over time. However, I agree that memory leakage is a significant issue at present.

I am not against using goto to clean up, as an alternative, we could also crash if memory allocation fails.

Perhaps we could add a test to detect memory leaks if this is the concern.

[haochengxia]

Following your suggestion, we went ahead and added a lightweight leak detection task to CI in #195.

More thorough enhancements can be progressively applied to the codebase later.