diff --git a/musl_src.gni b/musl_src.gni
index 876f5b648c7d848dd4bc5de4b506e86cf7ed5c66..0391b722675b6cf17a807b78131b15f98e56a349 100644
--- a/musl_src.gni
+++ b/musl_src.gni
@@ -1555,7 +1555,7 @@ if (musl_arch == "arm") {
     "src/math/fmaf.c",
     "src/math/sqrt.c",
     "src/math/sqrtf.c",
-    "src/process/vfork.c",
+    "src/process/arm/fork.s",
     "src/setjmp/longjmp.c",
     "src/setjmp/setjmp.c",
     "src/signal/restore.c",
@@ -1647,7 +1647,7 @@ if (musl_arch == "arm") {
     "src/math/sqrtf.c",
     "src/math/sqrtl.c",
     "src/math/truncl.c",
-    "src/process/vfork.c",
+    "src/process/x86_64/fork.s",
     "src/setjmp/longjmp.c",
     "src/setjmp/setjmp.c",
     "src/signal/restore.c",
@@ -2061,6 +2061,7 @@ musl_src_porting_file = [
   "ldso/ns_config.h",
   "ldso/strops.h",
   "src/legacy/ulimit.c",
+  "src/linux/clone.c",
   "src/linux/gettid.c",
   "src/linux/reboot.c",
   "src/linux/tgkill.c",
@@ -2124,6 +2125,11 @@ musl_src_porting_file = [
   "src/ldso/arm/dlvsym.s",
   "src/ldso/riscv64/dlvsym.s",
   "src/ldso/x86_64/dlvsym.s",
+  "src/unistd/getpid.c",
+  "src/process/fork.c",
+  "src/process/vfork.c",
+  "src/process/arm/__vfork.s",
+  "src/process/x86_64/__vfork.s",
 ]
 
 musl_inc_hook_files = [
diff --git a/porting/linux/user/src/env/__init_tls.c b/porting/linux/user/src/env/__init_tls.c
index 153305e0780d50d52ad64d0a5e74ec3d02adab42..9abef853f73b1ae05a19d33b54eaddf52c0be6a0 100644
--- a/porting/linux/user/src/env/__init_tls.c
+++ b/porting/linux/user/src/env/__init_tls.c
@@ -19,7 +19,7 @@ int __init_tp(void *p)
 	if (r < 0) return -1;
 	if (!r) libc.can_do_threads = 1;
 	td->detach_state = DT_JOINABLE;
-	td->tid = __syscall(SYS_set_tid_address, &__thread_list_lock);
+	td->tid = td->cached_pid = __syscall(SYS_set_tid_address, &__thread_list_lock);
 	td->locale = &libc.global_locale;
 	td->robust_list.head = &td->robust_list.head;
 	td->sysinfo = __sysinfo;
diff --git a/porting/linux/user/src/internal/pthread_impl.h b/porting/linux/user/src/internal/pthread_impl.h
index dfc557a727458b792a9529e3093b9d1877db47c5..ff582d65d618b655a83e2f0e535954c030abed00 100644
--- a/porting/linux/user/src/internal/pthread_impl.h
+++ b/porting/linux/user/src/internal/pthread_impl.h
@@ -39,6 +39,7 @@ struct pthread {
 
     /* Part 2 -- implementation details, non-ABI. */
     int tid;
+    int cached_pid;
     int errno_val;
     volatile int detach_state;
     volatile int cancel;
diff --git a/porting/linux/user/src/linux/clone.c b/porting/linux/user/src/linux/clone.c
index 23202cb758ff12495d4a46a0dc46284470b1a07a..bab0752a70a2b27a5f003d12a1481a666b7929f4 100644
--- a/porting/linux/user/src/linux/clone.c
+++ b/porting/linux/user/src/linux/clone.c
@@ -48,19 +48,36 @@ int clone(int (*func)(void *), void *stack, int flags, void *arg, ...)
 	tls  = va_arg(ap, void *);
 	ctid = va_arg(ap, pid_t *);
 	va_end(ap);
-	if (!(flags & (CLONE_VM | CLONE_VFORK)) && func) {
-		clone_args = (struct __clone_args *)malloc(sizeof(struct __clone_args));
-		if (clone_args == NULL) {
-			errno = ENOMEM;
-			return -1;
+
+	pthread_t self = __pthread_self();
+	pid_t parent_pid = self->cached_pid;
+	self->cached_pid = 0;
+	pid_t caller_tid = self->tid;
+
+	if (!(flags & (CLONE_VM | CLONE_VFORK))) {
+		self->tid = -1;
+		if (func) {
+			clone_args = (struct __clone_args *)malloc(sizeof(struct __clone_args));
+			if (clone_args == NULL) {
+				errno = ENOMEM;
+				return -1;
+			}
+			clone_args->func = clone_func;
+			clone_args->arg = arg;
+			clone_func = __start_child;
 		}
-		clone_args->func = clone_func;
-		clone_args->arg = arg;
-		clone_func = __start_child;
 	}
 	ret = __syscall_ret(__clone(clone_func, stack, flags, (void *)clone_args, ptid, tls, ctid));
 	if (!(flags & (CLONE_VM | CLONE_VFORK)) && func) {
 		free(clone_args);
 	}
+
+	if (ret != 0) {
+		self->cached_pid = parent_pid;
+		self->tid = caller_tid;
+	} else if (self->tid == -1) {
+		self->tid = __syscall(SYS_gettid);
+		self->cached_pid = self->tid;
+	}
 	return ret;
 }
diff --git a/porting/linux/user/src/process/arm/__vfork.s b/porting/linux/user/src/process/arm/__vfork.s
new file mode 100644
index 0000000000000000000000000000000000000000..5774d7dd49ba59fb4de3298a58c95b2c1fe5db86
--- /dev/null
+++ b/porting/linux/user/src/process/arm/__vfork.s
@@ -0,0 +1,10 @@
+.syntax unified
+.hidden __vfork
+.type __vfork,%function
+__vfork:
+	mov ip, r7
+	mov r7, 190
+	svc 0
+	mov r7, ip
+	.hidden __syscall_ret
+	b __syscall_ret
diff --git a/porting/linux/user/src/process/fork.c b/porting/linux/user/src/process/fork.c
new file mode 100644
index 0000000000000000000000000000000000000000..45c0bc882e3433d8d550477663dfae13baad3956
--- /dev/null
+++ b/porting/linux/user/src/process/fork.c
@@ -0,0 +1,38 @@
+#include <unistd.h>
+#include <string.h>
+#include <signal.h>
+#include "syscall.h"
+#include "libc.h"
+#include "pthread_impl.h"
+
+static void dummy(int x)
+{
+}
+
+weak_alias(dummy, __fork_handler);
+
+pid_t fork(void)
+{
+	pid_t ret;
+	sigset_t set;
+	__fork_handler(-1);
+	__block_all_sigs(&set);
+#ifdef SYS_fork
+	ret = __syscall(SYS_fork);
+#else
+	ret = __syscall(SYS_clone, SIGCHLD, 0);
+#endif
+	if (!ret) {
+		pthread_t self = __pthread_self();
+		self->tid = __syscall(SYS_gettid);
+		self->cached_pid = self->tid;
+		self->robust_list.off = 0;
+		self->robust_list.pending = 0;
+		self->next = self->prev = self;
+		__thread_list_lock = 0;
+		libc.threads_minus_1 = 0;
+	}
+	__restore_sigs(&set);
+	__fork_handler(!ret);
+	return __syscall_ret(ret);
+}
diff --git a/porting/linux/user/src/process/vfork.c b/porting/linux/user/src/process/vfork.c
new file mode 100644
index 0000000000000000000000000000000000000000..4901f538d43d9433e0620e71693a00d74f8b555a
--- /dev/null
+++ b/porting/linux/user/src/process/vfork.c
@@ -0,0 +1,27 @@
+#define _GNU_SOURCE
+#include <unistd.h>
+#include <signal.h>
+#include "syscall.h"
+#include "pthread_impl.h"
+
+hidden pid_t __vfork(void)
+{
+	/* vfork syscall cannot be made from C code */
+#ifdef SYS_fork
+	return syscall(SYS_fork);
+#else
+	return syscall(SYS_clone, SIGCHLD, 0);
+#endif
+}
+
+pid_t vfork(void)
+{
+	pthread_t self = __pthread_self();
+	pid_t parent_pid = self->cached_pid;
+	self->cached_pid = 0;
+	pid_t ret = __vfork();
+	if (ret != 0) {
+		self->cached_pid = parent_pid;
+	}
+	return ret;
+}
diff --git a/porting/linux/user/src/process/x86_64/__vfork.s b/porting/linux/user/src/process/x86_64/__vfork.s
new file mode 100644
index 0000000000000000000000000000000000000000..684dd315eb552c486e2294c5d8a3be065ed587ad
--- /dev/null
+++ b/porting/linux/user/src/process/x86_64/__vfork.s
@@ -0,0 +1,10 @@
+.hidden __vfork
+.type __vfork,@function
+__vfork:
+	pop %rdx
+	mov $58,%eax
+	syscall
+	push %rdx
+	mov %rax,%rdi
+	.hidden __syscall_ret
+	jmp __syscall_ret
diff --git a/porting/linux/user/src/thread/pthread_create.c b/porting/linux/user/src/thread/pthread_create.c
index 7e20ecb07fd46576ae158a643016afdb15c14274..8555750adbf1951d27575ddf4ada6c516d24bf47 100644
--- a/porting/linux/user/src/thread/pthread_create.c
+++ b/porting/linux/user/src/thread/pthread_create.c
@@ -10,6 +10,8 @@
 #include <stddef.h>
 #include <stdarg.h>
 
+pid_t getpid(void);
+
 void log_print(const char* info,...)
 {
     va_list ap;
@@ -389,6 +391,7 @@ int __pthread_create(pthread_t *restrict res, const pthread_attr_t *restrict att
 	new->stack_size = stack - stack_limit;
 	new->guard_size = guard;
 	new->self = new;
+	new->cached_pid = getpid();
 	new->tsd = (void *)tsd;
 	new->locale = &libc.global_locale;
 	if (attr._a_detach) {
@@ -424,7 +427,7 @@ int __pthread_create(pthread_t *restrict res, const pthread_attr_t *restrict att
 
 	__tl_lock();
 	libc.threads_minus_1++;
-	ret = __clone((c11 ? start_c11 : start), stack, flags, args, &new->tid, TP_ADJ(new), &__thread_list_lock);
+	ret = clone((c11 ? start_c11 : start), stack, flags, args, &new->tid, TP_ADJ(new), &__thread_list_lock);
 
 	/* All clone failures translate to EAGAIN. If explicit scheduling
 	 * was requested, attempt it before unlocking the thread list so
diff --git a/porting/linux/user/src/unistd/getpid.c b/porting/linux/user/src/unistd/getpid.c
new file mode 100644
index 0000000000000000000000000000000000000000..4dbfcf8d1877546b53e7e85155d6c7f4453a628a
--- /dev/null
+++ b/porting/linux/user/src/unistd/getpid.c
@@ -0,0 +1,17 @@
+#include <unistd.h>
+#include "pthread_impl.h"
+#include "syscall.h"
+
+static pid_t __get_cached_pid()
+{
+	return __pthread_self()->cached_pid;
+}
+
+pid_t getpid(void)
+{
+	pid_t cached_pid = __get_cached_pid();
+	if (cached_pid != 0) {
+		return cached_pid;
+	}
+	return __syscall(SYS_getpid);
+}