Kernel: Guard the all processes list with a Spinlock rather than a Mutex

There are callers of processes().with or processes().for_each that
require interrupts to be disabled. Taking a Mutexe with interrupts
disabled is a recipe for deadlock, so convert this to a Spinlock.

Kernel/GlobalProcessExposed.cpp

@@ -916,7 +916,7 @@ KResult ProcFSRootDirectory::traverse_as_directory(unsigned fsid, Function<bool(
         InodeIdentifier identifier = { fsid, component.component_index() };
         callback({ component.name(), identifier, 0 });
     }
-    processes().for_each_shared([&](Process& process) {
+    processes().for_each([&](Process& process) {
         VERIFY(!(process.pid() < 0));
         u64 process_id = (u64)process.pid().value();
         InodeIdentifier identifier = { fsid, static_cast<InodeIndex>(process_id << 36) };

Kernel/Process.cpp

@@ -44,7 +44,7 @@ static void create_signal_trampoline();
 
 RecursiveSpinlock g_profiling_lock;
 static Atomic<pid_t> next_pid;
-static Singleton<MutexProtected<Process::List>> s_processes;
+static Singleton<SpinlockProtected<Process::List>> s_processes;
 READONLY_AFTER_INIT HashMap<String, OwnPtr<Module>>* g_modules;
 READONLY_AFTER_INIT Memory::Region* g_signal_trampoline_region;
 
@@ -55,7 +55,7 @@ MutexProtected<String>& hostname()
     return *s_hostname;
 }
 
-MutexProtected<Process::List>& processes()
+SpinlockProtected<Process::List>& processes()
 {
     return *s_processes;
 }
@@ -86,7 +86,7 @@ UNMAP_AFTER_INIT void Process::initialize()
 NonnullRefPtrVector<Process> Process::all_processes()
 {
     NonnullRefPtrVector<Process> output;
-    processes().with_shared([&](const auto& list) {
+    processes().with([&](const auto& list) {
         output.ensure_capacity(list.size_slow());
         for (const auto& process : list)
             output.append(NonnullRefPtr<Process>(process));
@@ -138,7 +138,7 @@ void Process::register_new(Process& process)
 {
     // Note: this is essentially the same like process->ref()
     RefPtr<Process> new_process = process;
-    processes().with_exclusive([&](auto& list) {
+    processes().with([&](auto& list) {
         list.prepend(process);
     });
 }
@@ -301,7 +301,7 @@ bool Process::unref() const
     // NOTE: We need to obtain the process list lock before doing anything,
     //       because otherwise someone might get in between us lowering the
     //       refcount and acquiring the lock.
-    auto did_hit_zero = processes().with_exclusive([&](auto& list) {
+    auto did_hit_zero = processes().with([&](auto& list) {
         auto new_ref_count = deref_base();
         if (new_ref_count > 0)
             return false;
@@ -418,7 +418,7 @@ void Process::crash(int signal, FlatPtr ip, bool out_of_memory)
 
 RefPtr<Process> Process::from_pid(ProcessID pid)
 {
-    return processes().with_shared([&](const auto& list) -> RefPtr<Process> {
+    return processes().with([&](const auto& list) -> RefPtr<Process> {
         for (auto& process : list) {
             if (process.pid() == pid)
                 return &process;
@@ -696,7 +696,7 @@ void Process::die()
         m_threads_for_coredump.append(thread);
     });
 
-    processes().with_shared([&](const auto& list) {
+    processes().with([&](const auto& list) {
         for (auto it = list.begin(); it != list.end();) {
             auto& process = *it;
             ++it;

Kernel/Process.h

@@ -815,13 +815,13 @@ static_assert(sizeof(Process) == (PAGE_SIZE * 2));
 
 extern RecursiveSpinlock g_profiling_lock;
 
-MutexProtected<Process::List>& processes();
+SpinlockProtected<Process::List>& processes();
 
 template<IteratorFunction<Process&> Callback>
 inline void Process::for_each(Callback callback)
 {
     VERIFY_INTERRUPTS_DISABLED();
-    processes().with_shared([&](const auto& list) {
+    processes().with([&](const auto& list) {
         for (auto it = list.begin(); it != list.end();) {
             auto& process = *it;
             ++it;
@@ -835,7 +835,7 @@ template<IteratorFunction<Process&> Callback>
 inline void Process::for_each_child(Callback callback)
 {
     ProcessID my_pid = pid();
-    processes().with_shared([&](const auto& list) {
+    processes().with([&](const auto& list) {
         for (auto it = list.begin(); it != list.end();) {
             auto& process = *it;
             ++it;
@@ -876,7 +876,7 @@ inline IterationDecision Process::for_each_thread(Callback callback)
 template<IteratorFunction<Process&> Callback>
 inline void Process::for_each_in_pgrp(ProcessGroupID pgid, Callback callback)
 {
-    processes().with_shared([&](const auto& list) {
+    processes().with([&](const auto& list) {
         for (auto it = list.begin(); it != list.end();) {
             auto& process = *it;
             ++it;

Kernel/Syscalls/kill.cpp

@@ -65,7 +65,7 @@ KResult Process::do_killall(int signal)
     KResult error = KSuccess;
 
     // Send the signal to all processes we have access to for.
-    processes().for_each_shared([&](auto& process) {
+    processes().for_each([&](auto& process) {
         KResult res = KSuccess;
         if (process.pid() == pid())
             res = do_killself(signal);