Rename atomic_fetch_add/sub/or/and/xor to atomic_add/sub/or/and/xor
to avoid conflicts with the C/C++ standard library naming. The
atomic_fetch_xxx naming is reserved by the standard; keeping it causes
function name conflicts when source files indirectly include both
<nuttx/atomic.h> and <atomic>/<stdatomic.h>.
Signed-off-by: zhangyu117 <zhangyu117@xiaomi.com>
NuttX implemented fork() and vfork() as the same function. Both were libc
wrappers around a single up_fork() syscall; vfork() differed only by a
trailing waitpid(). Underneath, the child joined the parent's address
environment -- the same addrenv_join() that pthread_create() uses -- and got
a private copy of the stack. So the child shared .data, .bss and the heap
with its parent and ran concurrently with it.
That is not fork(). It is vfork()-with-a-private-stack under fork()'s name,
and the history says so: today's fork() is NuttX's old vfork(), renamed in
c33d1c9c97 (2023) without any change of behaviour. The failure was silent --
a program written against POSIX fork() compiled, ran, and had its child's
writes land in the parent's variables.
Separate them into two primitives, chosen by which function the caller
called rather than by what the hardware happens to be:
fork() child gets its own copy of the parent's memory at the same
virtual addresses; runs concurrently. Only where an address
environment can be duplicated -- elsewhere it is not declared at
all, so calling it is a build error naming the function.
vfork() child shares the parent's memory; parent suspended until the
child _exit()s or exec()s. Implementable everywhere.
Below libc there is still one syscall. up_fork() gains a bool saying which
primitive the caller used, since the per-architecture register snapshot is
the same for both, and passes it to nxtask_setup_fork(), which is the single
place the memory semantics are decided. The argument arrives in the first
argument register and is never touched: each architecture's snapshot takes
some other call-clobbered register for its scratch, so the flag is simply
still there when the C worker is called.
The vfork() parent suspension moves out of libc into nxtask_start_fork(),
released from nxsched_release_tcb() by nxtask_resume_vfork(). Two things
follow: the parent is resumed at exec(), since exec_swap() has already handed
the child's pid to the loaded program by the time the vfork stub exits, and
vfork() no longer depends on CONFIG_SCHED_WAITPID.
Releasing there requires one fix in nxtask_exit(). It raises rtcb->lockcount
directly rather than through sched_lock() while it tears the TCB down, so the
nxsem_post() that wakes the vfork() parent leaves it queued where a blocked
task collects while pre-emption is off -- g_pendingtasks, or g_readytorun on
SMP -- and the matching raw lockcount-- does not publish it the way
sched_unlock() would, leaving the parent stranded with nothing to move it on.
The fix mirrors sched_unlock() for each case: nxsched_merge_pending(), or
nxsched_deliver_task() under CONFIG_SMP. Both are no-ops while pre-emption is
still disabled, and up_exit() re-reads this_task() afterwards, so a change of
the ready-to-run head is honoured. Without it vfork() deadlocks wherever no
other task happens to call sched_unlock() afterwards -- rv-virt:nsh64 and
rv-virt:pnsh64, where NSH is blocked in waitpid() holding the lock, and
qemu-armv8a:citest_smp, which hangs the moment the vfork() test runs.
fork() is built on a new addrenv_fork(), backed by an up_addrenv_fork() hook
that duplicates an address environment into freshly allocated pages mapped at
the same virtual addresses -- unlike up_addrenv_clone(), which copies only
the representation and leaves both pointing at the same page tables. The
child then adopts the parent's stack geometry rather than being given a
relocated copy: a pointer to a stack local taken before fork() must name the
same object in the child that it named in the parent, and the parent's stack
is already in the duplicate, with its contents, at the parent's address.
No architecture implements up_addrenv_fork() yet, so this commit leaves
fork() unavailable everywhere. That is the intended state. It withdraws
fork() from ARCH_ARM, flat ARCH_ARM64, ARCH_RISCV, ARCH_SIM and ARCH_X86_64,
where until now it named the sharing primitive; per-architecture patches
restore it, with POSIX semantics, as up_addrenv_fork() lands. In the
meantime the sharing primitive is still there under the name that describes
it: vfork() for a child that runs a program, pthread_create() for a second
flow of control that shares memory, posix_spawn() for both at once.
Kconfig: ARCH_HAVE_VFORK inherits ARCH_HAVE_FORK's select lines, conditions
included, so no configuration gains machinery; ARCH_HAVE_FORK is redefined to
mean "can provide POSIX fork() semantics" and now depends on ARCH_ADDRENV.
There is one deliberate departure from "verbatim". ARCH_ARM selected the
fork family unconditionally, BUILD_KERNEL included, and that has never
worked: on a kernel build the architecture's fork entry point sees the
kernel's return address and stack pointer rather than the caller's, so the
child resumes at a kernel address. On qemu-armv7a:knsh master faults in
ostest's fork case with "Child did not run" and then a data abort; without
the condition this change faults the same way through vfork(). ARCH_ARM64
and ARCH_X86_64 already carried "if !BUILD_KERNEL" for exactly this reason --
ARM was the outlier. Conditioning it turns a runtime fault into an honest
absence, which is the whole point of the change; arch/arm takes the condition
off again in the patch that adds its saved-syscall-frame path. Only the
MMU-capable ARM ports are affected, since Cortex-M cannot build BUILD_KERNEL
at all.
Also fixes two latent syntax errors found on the way: a missing comma in
riscv_fork.c and mips_fork.c, both in *_FRAMEPOINTER && !SAVE_GP branches
that are never compiled today.
BREAKING CHANGE: fork() is withdrawn from every architecture. It is no
longer declared in unistd.h, so code that calls it fails to build with an error
naming the function, and the sharing behaviour it used to have is gone rather
than renamed. CONFIG_ARCH_HAVE_FORK no longer means "fork() exists"; it means
"this configuration can provide POSIX fork() semantics", and no architecture
selects it yet.
Quick fix, chosen by why the call was made:
to run a program vfork() + exec*(), or better posix_spawn()
a second flow of control that pthread_create()
shares the caller's memory
a genuinely independent copy keep fork(), and wait for the per-arch patch
of the process that implements up_addrenv_fork() and selects
CONFIG_ARCH_HAVE_FORK
Out-of-tree code that tests CONFIG_ARCH_HAVE_FORK to decide whether a
fork-then-exec path is available wants CONFIG_ARCH_HAVE_VFORK instead, which is
selected in exactly the places CONFIG_ARCH_HAVE_FORK used to be. The full
migration guide is Documentation/guides/fork_vfork_migration.rst.
Assisted-by: Claude Code:claude-opus-5
Signed-off-by: Marco Casaroli <marco.casaroli@gmail.com>
A task does not necessarily own an address environment. tcb->addrenv_own is
set only by addrenv_attach(), which is reached only from addrenv_allocate();
a kernel thread never allocates one, and in a protected build nothing does --
there is a single address space for the whole system and the architecture's
up_addrenv_*() are stubs. addrenv_own is then NULL for every task, always.
That a task may have no address environment is already an expected state.
addrenv_switch() returns OK when tcb->addrenv_curr is NULL and addrenv_drop()
returns early, and every caller of addrenv_select() checks addrenv_own != NULL
before calling in: nxsched_get_stateinfo(), nxtask_argvstr(), proc_groupenv()
and the arm, arm64, risc-v and tricore up_check_tcbstack().
addrenv_take() and addrenv_give() are the only two that dereference
unconditionally. addrenv_join() calls addrenv_take(ptcb->addrenv_own) without
a check, so pthread_create() faults on &((struct addrenv_s *)NULL)->refs
whenever the calling task has no address environment. With
CONFIG_DEBUG_ASSERTIONS off the same access silently corrupts low memory
instead.
Handle NULL in both, the way the rest of the file already does.
addrenv_give() returns a non-zero count for the NULL case so that callers
never conclude an absent address environment has become unreferenced and
should be destroyed.
Signed-off-by: Marco Casaroli <marco.casaroli@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
debug.h is a NuttX-specific, non-POSIX header. Placing it in the
top-level include/ directory creates naming conflicts with external
projects that define their own debug.h.
This commit moves the canonical header to include/nuttx/debug.h,
following the NuttX convention for non-POSIX/non-standard headers,
and updates all in-tree references.
A backward-compatibility shim is left at include/debug.h that
emits a deprecation #warning and re-includes <nuttx/debug.h>,
allowing out-of-tree code to continue building while migrating.
Signed-off-by: Piyush Patle <piyushpatle228@gmail.com>
Replace global enter_critical_section() calls with lightweight spinlock in
addrenv_switch() to reduce interrupt latency, and convert refs counter to
atomic_t for thread-safe reference counting without blocking operations.
Signed-off-by: hujun5 <hujun5@xiaomi.com>
The temporary mappings via addrenv_select() and addrenv_restore() simply
do not work from interrupt, so remove its usage and replace with kmap
which is safe.
The performance penalty in SMP mode is too big for taking the big kernel
lock simply to bump the address environment reference counter; fix this
by using the compiler provided atomic macros.
Most tools used for compliance and SBOM generation use SPDX identifiers
This change brings us a step closer to an easy SBOM generation.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
Store the old environment in a local context so another temporary address
environment can be selected. This can happen especially when a process
is being loaded (the new process's mappings are temporarily instantiated)
and and interrupt occurs.
Instead of using a volatile storage for the address environment in the
binfmt / loadinfo structures, always allocate the address environment
from kheap.
This serves two purposes:
- If the task creation fails, any kernel thread that depends on the
address environment created during task creation will not lose their
mappings (because they hold a reference to it)
- The current address environment variable (g_addrenv) will NEVER contain
a stale / incorrect value
- Releasing the address environment is simplified as any pointer given
to addrenv_drop() can be assumed to be heap memory
- Makes the kludge function addrenv_clear_current irrelevant, as the
system will NEVER have invalid mappings any more
The function is not relevant any longer, remove it. Also remove
save_addrenv_t, the parameter taken by up_addrenv_restore.
Implement addrenv_select() / addrenv_restore() to handle the temporary
instantiation of address environments, e.g. when a process is being
created.
There is currently a big problem in the address environment handling which
is that the address environment is released too soon when the process is
exiting. The current MMU mappings will always be the exiting process's, which means
the system needs them AT LEAST until the next context switch happens. If
the next thread is a kernel thread, the address environment is needed for
longer.
Kernel threads "lend" the address environment of the previous user process.
This is beneficial in two ways:
- The kernel processes do not need an allocated address environment
- When a context switch happens from user -> kernel or kernel -> kernel,
the TLB does not need to be flushed. This must be done only when
changing to a different user address environment.
Another issue is when a new process is created; the address environment
of the new process must be temporarily instantiated by up_addrenv_select().
However, the system scheduler does not know that the process has a different
address environment to its own and when / if a context restore happens, the
wrong MMU page directory is restored and the process will either crash or
do something horribly wrong.
The following changes are needed to fix the issues:
- Add mm_curr which is the current address environment of the process
- Add a reference counter to safeguard the address environment
- Whenever an address environment is mapped to MMU, its reference counter
is incremented
- Whenever and address environment is unmapped from MMU, its reference
counter is decremented, and tested. If no more references -> drop the
address environment and release the memory as well
- To limit the context switch delay, the address environment is freed in
a separate low priority clean-up thread (LPWORK)
- When a process temporarily instantiates another process's address
environment, the scheduler will now know of this and will restore the
correct mappings to MMU
Why is this not causing more noticeable issues ? The problem only happens
under the aforementioned special conditions, and if a context switch or
IRQ occurs during this time.
Detach the address environment handling from the group structure to the
tcb. This is preparation to fix rare cases where the system (MMU) is left
without a valid page directory, e.g. when a process exits.
