[kernel] Re-add slab_allocated mixin

The return of slab_allocated! Now after the kutil/util/kernel giant
cleanup, this belongs squarely in the kernel, and works much better
there. Slabs are allocated via a bump pointer into a new kernel VMA,
instead of using kalloc() or allocating pages directly.

definitions/memory_layout.yaml

@@ -15,3 +15,6 @@
 
 - name: buffers
   size: 64G
+
+- name: slabs
+  size: 64G

src/kernel/memory_bootstrap.cpp

@@ -23,6 +23,8 @@ using bootproto::allocation_register;
 
 using obj::vm_flags;
 
+uintptr_t g_slabs_bump_pointer;
+
 // These objects are initialized _before_ global constructors are called,
 // so we don't want them to have global constructors at all, lest they
 // overwrite the previous initialization.
@@ -44,6 +46,10 @@ obj::vm_area_guarded g_kernel_buffers {
     mem::buffers_size,
     vm_flags::write};
 
+obj::vm_area_open g_kernel_slabs {
+    mem::slabs_size,
+    vm_flags::write};
+
 void * operator new(size_t size)           { return g_kernel_heap.allocate(size); }
 void * operator new [] (size_t size)       { return g_kernel_heap.allocate(size); }
 void operator delete (void *p) noexcept    { return g_kernel_heap.free(p); }
@@ -106,6 +112,8 @@ memory_initialize_post_ctors(bootproto::args &kargs)
 {
     vm_space &vm = vm_space::kernel_space();
     vm.add(mem::buffers_offset, &g_kernel_buffers);
+    vm.add(mem::slabs_offset, &g_kernel_slabs);
+    g_slabs_bump_pointer = mem::slabs_offset;
 
     g_frame_allocator.free(
         get_physical_page(kargs.page_tables.pointer),

src/kernel/objects/slab_allocated.h

@@ -0,0 +1,57 @@
+#pragma once
+/// \file slab_allocated.h
+/// A parent template class for slab-allocated objects
+
+#include <stdlib.h>
+#include <util/pointers.h>
+#include <util/vector.h>
+
+#include "memory.h"
+
+extern uintptr_t g_slabs_bump_pointer;
+
+template <typename T, unsigned N = 1>
+class slab_allocated
+{
+public:
+    constexpr static unsigned slab_pages = N;
+    constexpr static unsigned slab_size = N * mem::frame_size;
+
+    void * operator new(size_t size)
+    {
+        kassert(size == sizeof(T), "Slab allocator got wrong size allocation");
+        if (s_free.count() == 0)
+            allocate_chunk();
+
+        T *item = s_free.pop();
+        memset(item, 0, sizeof(T));
+        return item;
+    }
+
+    void operator delete(void *p) { s_free.append(reinterpret_cast<T*>(p)); }
+
+private:
+    constexpr static size_t per_block = slab_size / sizeof(T);
+
+    static void allocate_chunk()
+    {
+        s_free.ensure_capacity(per_block);
+
+        uintptr_t start = __atomic_fetch_add(
+                &g_slabs_bump_pointer, slab_size,
+                __ATOMIC_SEQ_CST);
+
+        T *current = reinterpret_cast<T*>(start);
+        T *end = util::offset_pointer(current, slab_size);
+        while (current < end)
+            s_free.append(current++);
+    }
+
+    static util::vector<T*> s_free;
+};
+
+template <typename T, unsigned N>
+util::vector<T*> slab_allocated<T,N>::s_free;
+
+#define DEFINE_SLAB_ALLOCATOR(type) \
+    template<> util::vector<typename type*> slab_allocated<typename type, type::slab_pages>::s_free {};