MAX_ORDER

[ 104.490000] [<c0044a0c>] (warn_slowpath_common+0x4c/0x80) from [<c0075874>] (__alloc_pages_nodemask+0x130/0x51c)

__alloc_pages_slowpath

         * In the slowpath, we sanity check order to avoid ever trying to
         * reclaim >= MAX_ORDER areas which will never succeed. Callers may
         * be using allocators in order of preference for an area that is
         * too large.
         */
        if (order >= MAX_ORDER) {
                WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN));
                return NULL;
        }

[ 104.520000] [<c0075874>] (__alloc_pages_nodemask+0x130/0x51c) from [<c0031530>] (__dma_alloc+0x148/0x3ac)
[ 104.550000] [<c0031530>] (__dma_alloc+0x148/0x3ac) from [<c0031810>] (dma_alloc_coherent+0x50/0x5c)
[ 104.580000] [<c0031810>] (dma_alloc_coherent+0x50/0x5c) from [<bf02d39c>]


Linux的存儲管理(一）
http://www.lslnet.com/linux/f/docs1/i31/big5242556.htm

6.2.6内存管理区
http://www.kerneltravel.net/kernel-book/%E7%AC%AC%E5%85%AD%E7%AB%A0%20Linux%E5%86%85%E5%AD%98%E7%AE%A1%E7%90%86/6.2.6.htm

Linux kernel memory management buddy system (linux内核内存管理的伙伴算法)
http://blog.csdn.net/VIV777/archive/2007/07/05/1680363.aspx

第3节 内存的分配和回收
http://www.eefocus.com/article/09-06/74976s.html