process: implement thread-backed fork()/vfork()/execve()/waitpid() (opt-in, off by default)

[gburd]

What

Adds a thread-backed fork()/vfork()/execve()/waitpid() implementation to OSv, gated entirely behind a new CONFIG_fork kconfig option that defaults to OFF. With the flag off (the default), none of this code is compiled and OSv is byte-for-byte the current single-address-space kernel — fork()/vfork() return ENOSYS exactly as today. The feature is opt-in for workloads that need fork().

Why

OSv is a single-address-space unikernel, so a literal copy-on-write fork() is not its model, and fork() has historically been a stub. But a large class of real Linux programs use fork() — most commonly fork()+exec() to spawn a child program, plus system()/popen(). This provides the useful, compatible subset without disturbing the default OSv model.

How

With CONFIG_fork enabled:

• fork()/vfork() create a child OSv thread that resumes in fork()'s caller and returns 0 in the child / the child pid in the parent (the classic twin return). The child runs on a private copy of the parent's user stack (so parent and child have independent locals after the return) and gets its own fresh OSv per-thread TLS block (own errno, etc.). Arch halves in arch/x64/fork.cc and arch/aarch64/fork.cc.
• execve() launches the target as a fresh OSv application (its own ELF namespace) and does not return — making fork()+exec() work.
• waitpid()/wait4()/wait() reap a child's exit status via a pid→child registry; SIGCHLD is raised to the parent on child exit (and SIGCHLD/SIGURG/SIGWINCH now correctly default to ignore rather than powering off the VM).
• exit()/_exit() in a fork child ends only that child, not the whole unikernel.
• pthread_atfork prepare/parent/child handlers are now actually run around fork() (they were a no-op stub; glibc/musl register these internally, e.g. to reset the malloc arena lock in the child).
• sys_clone() routes the non-CLONE_THREAD (fork) case here when enabled.

Validation

tst-fork (new) passes 10/10 on both x86-64 and aarch64: twin return, private-stack isolation (parent's local intact after the child mutates its own copy), fork()+exec(), vfork(), and waitpid() reaping exit codes. Built and confirmed in both configurations: CONFIG_fork off builds with the fork code fully excluded and the kernel healthy; CONFIG_fork on builds and passes tst-fork.

Scope / honest limitations (documented in documentation/fork.md)

This base does not give the child private memory: it shares the parent's heap and globals (only the stack is copied). That is fine for fork()+exec(), system(), and children that only read shared state before exec/exit. A follow-up adds per-child copy-on-write address spaces (behind the same flag) for programs that need memory-isolated multi-process fork(). Also documented as not carried by this base: deep-call-chain child unwind, fork()-as-memory-snapshot (e.g. Redis BGSAVE), and a separate pre-existing fault in execve()'s new-ELF-namespace path.

Off by default, so there is no risk to existing OSv builds; this simply makes fork() available to those who opt in.

[gburd]

Added a fix for a fork-child lifecycle leak found while validating execve(): a forked child was created as an attached thread, so it held a shared_ptr to the app's application_runtime that nothing ever released (the parent reaps the child via waitpid/the pid registry, never join()). The runtime refcount therefore never hit zero, ~application_runtime never fired, and the loader's application::join() blocked forever -> OSv hung at shutdown after a successful fork+exec. Fix: create the fork child detached and dispose() it after its exit bookkeeping so the reaper releases application_runtime and OSv shuts down cleanly. Also confirmed execve() itself works (launches the target program via application::run in a new ELF namespace) and re-enabled tst-fork's fork+exec test to exec a real payload; added tst-execve. Validated on x86-64/KVM: tst-execve 3/3, tst-fork 10/10, 5x repeat clean, clean shutdown, no non-fork regressions.