add a custom allocator for aligned memory

With this allocator we can use the STL container, such as std::vector,
and be sure that we can use aligned load instruction when necessary.

Refs: scality#222

src/simd/allocator.h

@@ -0,0 +1,165 @@
+/*
+ * Copyright 2017-2018 Scality
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ *    this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/** @file allocator.h
+ *
+ * Provide a custom memory allocator for SIMD.
+ *
+ * This allocator always returns memory that is suitably aligned to be loaded
+ * efficiently by a Register object.
+ */
+
+#ifndef __QUAD_SIMD_SIMD_ALLOCATOR_H__
+#define __QUAD_SIMD_SIMD_ALLOCATOR_H__
+
+#include <cassert>
+#include <cstdint>
+#include <limits>
+
+#include "simd/definitions.h"
+
+namespace quadiron {
+namespace simd {
+
+/// Check that the given address is properly aligned.
+template <typename T>
+inline bool addr_is_aligned(const T* addr)
+{
+    // Without SIMD, there is no specific alignment constraint.
+    if (INSTRUCTION_SET == InstructionSet::NONE) {
+        return true;
+    }
+    const std::uintptr_t address = reinterpret_cast<std::uintptr_t>(addr);
+    return (address & (ALIGNMENT - 1)) == 0;
+}
+
+/** Custom allocator to take advantage of SIMD processing.
+ *
+ * This allocator always return memory that is suitably aligned for the current
+ * SIMD instruction set. Thanks to this property, you can safely use the aligned
+ * load from the Register class in order to increase performance.
+ */
+template <typename T>
+class AlignedAllocator {
+  public:
+    using value_type = T;
+
+    AlignedAllocator() noexcept {}
+    // No state, => nothing to copy.
+    template <class U>
+    AlignedAllocator(AlignedAllocator<U> const& other) noexcept
+    {
+    }
+
+    value_type* allocate(std::size_t count)
+    {
+        // Guard against overflow!
+        if (count > max_size()) {
+            throw std::bad_alloc();
+        }
+
+        // No SIMD: default allocator is good enough!
+        if (INSTRUCTION_SET == InstructionSet::NONE) {
+            return static_cast<value_type*>(
+                ::operator new(count * sizeof(value_type)));
+        }
+
+        // Overallocate just enough to have room for alignment adjustment.
+        const std::size_t size = count * sizeof(value_type) + ALIGNMENT;
+        unsigned char* ptr = static_cast<unsigned char*>(::operator new(size));
+
+        // Align the allocated memory.
+        const std::uintptr_t address = reinterpret_cast<std::uintptr_t>(ptr);
+        const unsigned offset = ALIGNMENT - (address % ALIGNMENT);
+        assert(offset >= 1); // We need a byte to store the offset itself.
+        unsigned char* aligned_ptr = ptr + offset;
+
+        // Store the offset just before the aligned memory.
+        assert(offset <= std::numeric_limits<unsigned char>::max());
+        *(aligned_ptr - 1) = static_cast<unsigned char>(offset);
+
+        // Return the aligned pointer.
+        //
+        // Clang analyser think that we leak `ptr`, whereas we can re-compute
+        // it from `aligned_ptr` and free it in `deallocate`.
+        // NOLINTNEXTLINE(clang-analyzer-cplusplus.NewDeleteLeaks)
+        return reinterpret_cast<value_type*>(aligned_ptr);
+    }
+
+    void deallocate(value_type* ptr, std::size_t count) noexcept
+    {
+        // No SIMD: default allocator is good enough!
+        if (INSTRUCTION_SET == InstructionSet::NONE) {
+            ::operator delete(ptr);
+            return;
+        }
+
+        if (ptr == nullptr) {
+            return;
+        }
+        // Respect strict aliasing rules: read through a character type.
+        unsigned char* raw = reinterpret_cast<unsigned char*>(ptr);
+        // Get the alignment offset stored just before the aligned pointer.
+        const unsigned offset = *(raw - 1);
+        ::operator delete(raw - offset);
+    }
+
+    std::size_t max_size() const noexcept
+    {
+        const std::size_t max_size = std::numeric_limits<std::size_t>::max();
+        return (max_size - ALIGNMENT) / sizeof(value_type);
+    }
+
+    // Our allocator is stateless.
+    using propagate_on_container_copy_assignment = std::true_type;
+    using propagate_on_container_move_assignment = std::true_type;
+    using propagate_on_container_swap = std::true_type;
+};
+
+template <class T, class U>
+bool operator==(AlignedAllocator<T> const&, AlignedAllocator<U> const&) noexcept
+{
+    // Our allocator is stateless: Any instance of our allocator can deallocate
+    // the memory from another instance.
+    return true;
+}
+
+template <class T, class U>
+bool operator!=(
+    AlignedAllocator<T> const& x,
+    AlignedAllocator<U> const& y) noexcept
+{
+    return !(x == y);
+}
+
+} // namespace simd
+} // namespace quadiron
+
+#endif

src/simd/simd.h

@@ -31,6 +31,7 @@
 #ifndef __QUAD_SIMD_SIMD_H__
 #define __QUAD_SIMD_SIMD_H__
 
+#include "simd/allocator.h"
 #include "simd/definitions.h"
 
 namespace quadiron {