Revert hrev43455 and instead just align the size.

While this isn't really correct, it works for the use case it is
intended and doesn't open the can of worms we get when trying to do
memalign() across platforms (due to build tools use of KMessage).

src/system/kernel/messaging/KMessage.cpp

@@ -6,8 +6,6 @@
 
 #include <util/KMessage.h>
 
-#include <malloc.h>
-	// for memalign()
 #include <stdlib.h>
 #include <string.h>
 
@@ -610,8 +608,7 @@ KMessage::ReceiveFrom(port_id fromPort, bigtime_t timeout,
 		return error;
 
 	// allocate a buffer
-	uint8* buffer = (uint8*)memalign(kMessageBufferAlignment,
+	uint8* buffer = (uint8*)malloc(_Align(messageInfo->size));
-		messageInfo->size);
 	if (!buffer)
 		return B_NO_MEMORY;
 
@@ -830,7 +827,7 @@ KMessage::_InitFromBuffer(bool sizeFromBuffer)
 			fBufferCapacity = size;
 		}
 
-		void* buffer = memalign(kMessageBufferAlignment, fBufferCapacity);
+		void* buffer = malloc(_Align(fBufferCapacity));
 		if (buffer == NULL)
 			return B_NO_MEMORY;
 
@@ -964,7 +961,7 @@ KMessage::_AllocateSpace(int32 size, bool alignAddress, bool alignSize,
 	// reallocate if necessary
 	if (fBuffer == &fHeader) {
 		int32 newCapacity = _CapacityFor(newSize);
-		void* newBuffer = memalign(kMessageBufferAlignment, newCapacity);
+		void* newBuffer = malloc(_Align(newCapacity));
 		if (!newBuffer)
 			return B_NO_MEMORY;
 		fBuffer = newBuffer;