Kernel: Add API for getting contiguous physcial pages

This will be used to create DMA regions.

kernel/include/kernel/Memory/Heap.h

@@ -21,6 +21,9 @@ namespace Kernel
 		paddr_t take_free_page();
 		void release_page(paddr_t);
 
+		paddr_t take_free_contiguous_pages(size_t pages);
+		void release_contiguous_pages(paddr_t paddr, size_t pages);
+
 		size_t used_pages() const;
 		size_t free_pages() const;
 

kernel/include/kernel/Memory/PhysicalRange.h

@@ -29,10 +29,13 @@ namespace Kernel
 
 	private:
 		unsigned long long* ull_bitmap_ptr() { return (unsigned long long*)m_vaddr; }
+		const unsigned long long* ull_bitmap_ptr() const { return (const unsigned long long*)m_vaddr; }
 		
 		paddr_t paddr_for_bit(unsigned long long) const;
 		unsigned long long bit_for_paddr(paddr_t paddr) const;
 
+		unsigned long long contiguous_bits_set(unsigned long long start, unsigned long long count) const;
+
 	private:
 		const paddr_t m_paddr { 0 };
 		const size_t m_size	{ 0 };

kernel/kernel/Memory/Heap.cpp

@@ -79,6 +79,25 @@ namespace Kernel
 		ASSERT_NOT_REACHED();
 	}
 
+	paddr_t Heap::take_free_contiguous_pages(size_t pages)
+	{
+		LockGuard _(m_lock);
+		for (auto& range : m_physical_ranges)
+			if (range.free_pages() >= pages)
+				if (paddr_t paddr = range.reserve_contiguous_pages(pages))
+					return paddr;
+		return 0;
+	}
+
+	void Heap::release_contiguous_pages(paddr_t paddr, size_t pages)
+	{
+		LockGuard _(m_lock);
+		for (auto& range : m_physical_ranges)
+			if (range.contains(paddr))
+				return range.release_contiguous_pages(paddr, pages);
+		ASSERT_NOT_REACHED();
+	}
+
 	size_t Heap::used_pages() const
 	{
 		LockGuard _(m_lock);

kernel/kernel/Memory/PhysicalRange.cpp

@@ -27,7 +27,7 @@ namespace Kernel
 
 		memset((void*)m_vaddr, 0x00, m_bitmap_pages * PAGE_SIZE);
 		memset((void*)m_vaddr, 0xFF, m_data_pages / 8);
-		for (int i = 0; i < m_data_pages % 8; i++)
+		for (ull i = 0; i < m_data_pages % 8; i++)
 			((uint8_t*)m_vaddr)[m_data_pages / 8] |= 1 << i;
 
 		dprintln("physical range needs {} pages for bitmap", m_bitmap_pages);
@@ -43,6 +43,18 @@ namespace Kernel
 		return (paddr - m_paddr) / PAGE_SIZE - m_bitmap_pages;
 	}
 
+	ull PhysicalRange::contiguous_bits_set(ull start, ull count) const
+	{
+		for (ull i = 0; i < count; i++)
+		{
+			ull off = (start + i) / ull_bits;
+			ull bit = (start + i) % ull_bits;
+			if (!(ull_bitmap_ptr()[off] & (1ull << bit)))
+				return i;
+		}
+		return count;
+	}
+
 	paddr_t PhysicalRange::reserve_page()
 	{
 		ASSERT(free_pages() > 0);
@@ -80,4 +92,61 @@ namespace Kernel
 		m_free_pages++;
 	}
 
+	paddr_t PhysicalRange::reserve_contiguous_pages(size_t pages)
+	{
+		ASSERT(pages > 0);
+		ASSERT(free_pages() > 0);
+
+		if (pages == 1)
+			return reserve_page();
+
+		ull ull_count = BAN::Math::div_round_up<ull>(m_data_pages, ull_bits);
+
+		// NOTE: This feels kinda slow, but I don't want to be
+		//       doing premature optimization. This will be only
+		//       used when creating DMA regions.
+
+		for (ull i = 0; i < ull_count; i++)
+		{
+			if (ull_bitmap_ptr()[i] == 0)
+				continue;
+
+			for (ull bit = 0; bit < ull_bits;)
+			{
+				ull start = i * ull_bits + bit;
+				ull set_cnt = contiguous_bits_set(start, pages);
+				if (set_cnt == pages)
+				{
+					for (ull j = 0; j < pages; j++)
+						ull_bitmap_ptr()[(start + j) / ull_bits] &= ~(1ull << ((start + j) % ull_bits));
+					m_free_pages -= pages;
+					return paddr_for_bit(start);
+				}
+				bit += set_cnt + 1;
+			}
+		}
+
+		ASSERT_NOT_REACHED();
+	}
+
+	void PhysicalRange::release_contiguous_pages(paddr_t paddr, size_t pages)
+	{
+		ASSERT(paddr % PAGE_SIZE == 0);
+		ASSERT(paddr - m_paddr <= m_size);
+		ASSERT(pages > 0);
+
+		ull start_bit = bit_for_paddr(paddr);
+		for (size_t i = 0; i < pages; i++)
+		{
+			ull off = (start_bit + i) / ull_bits;
+			ull bit = (start_bit + i) % ull_bits;
+			ull mask = 1ull << bit;
+
+			ASSERT(!(ull_bitmap_ptr()[off] & mask));
+			ull_bitmap_ptr()[off] |= mask;
+		}
+
+		m_free_pages += pages;
+	}
+
 }