1 Freeing With Explicit Free Lists Freeing With a LIFO Policy (Case - PDF document

Today  Explicit free lists Dynamic Memory Allocation:  Segregated free lists  Garbage collection Advanced Concepts  Memory-related perils and pitfalls CSci 2021: Machine Architecture and Organization April 27th-29th, 2020 Your instructor: Stephen McCamant Based on slides originally by: Randy Bryant, Dave O’Hallaron 1 2 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Explicit Free Lists Keeping Track of Free Blocks  Method 1: Implicit free list using length — links all blocks Free Allocated (as before) Size a Size a 5 Next 4 6 2 Prev Payload and padding  Method 2: Explicit free list among the free blocks using pointers Size a Size a 5 4 6 2  Method 3: Segregated free list  Maintain list(s) of free blocks, not all blocks  Different free lists for different size classes  The “next” free block could be anywhere  So we need to store forward/back pointers, not just sizes  Method 4: Blocks sorted by size  Still need boundary tags for coalescing  Can use a balanced tree (e.g. Red-Black tree) with pointers within each  Luckily we track only free blocks, so we can use payload area free block, and the length used as a key 3 4 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Explicit Free Lists Allocating From Explicit Free Lists conceptual graphic  Logically: Before A B C  Physically: blocks can be in any order After (with splitting) Forward (next) links A B 4 4 4 4 6 6 4 4 4 4 C Back (prev) links = malloc(…) 5 6 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition 1

Freeing With Explicit Free Lists Freeing With a LIFO Policy (Case 1) conceptual graphic  Insertion policy : Where in the free list do you put a newly Before freed block? free( )  LIFO (last-in-first-out) policy  Insert freed block at the beginning of the free list Root  Pro: simple and constant time  Con: studies suggest fragmentation is worse than address ordered  Insert the freed block at the root of the list  Address-ordered policy  Insert freed blocks so that free list blocks are always in address order: After addr(prev) < addr(curr) < addr(next)  Con: requires search Root  Pro: studies suggest fragmentation is lower than LIFO 7 8 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Freeing With a LIFO Policy (Case 2) Freeing With a LIFO Policy (Case 3) conceptual graphic conceptual graphic Before Before free( ) free( ) Root Root  Splice out successor block, coalesce both memory blocks and  Splice out predecessor block, coalesce both memory blocks, and insert the new block at the root of the list insert the new block at the root of the list After After Root Root 9 10 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Freeing With a LIFO Policy (Case 4) Explicit List Summary conceptual graphic  Comparison to implicit list: Before free( )  Allocate is linear time in number of free blocks instead of all blocks  Much faster when most of the memory is full Root  Slightly more complicated allocate and free since needs to splice blocks in and out of the list  Some extra space for the links (2 extra words needed for each block)  Does this increase internal fragmentation?  Splice out predecessor and successor blocks, coalesce all 3 memory blocks and insert the new block at the root of the list  Most common use of linked lists is in conjunction with After segregated free lists  Keep multiple linked lists of different size classes, or possibly for different types of objects Root 11 12 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition 2

Keeping Track of Free Blocks Today  Method 1: Implicit list using length — links all blocks  Explicit free lists  Segregated free lists 5 4 6 2  Garbage collection  Memory-related perils and pitfalls  Method 2: Explicit list among the free blocks using pointers 5 4 6 2  Method 3: Segregated free list  Different free lists for different size classes  Method 4: Blocks sorted by size  Can use a balanced tree (e.g. Red-Black tree) with pointers within each free block, and the length used as a key 13 15 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Segregated List (Seglist) Allocators Seglist Allocator  Given an array of free lists, each one for some size class  Each size class of blocks has its own free list 1-2  To allocate a block of size n :  Search appropriate free list for block of size m > n  If an appropriate block is found: 3  Split block and place fragment on appropriate list (optional) 4  If no block is found, try next larger class  Repeat until block is found 5-8  If no block is found:  Request additional heap memory from OS (using sbrk() ) 9-inf  Allocate block of n bytes from this new memory  Place remainder as a single free block in largest size class.  Often have separate classes for each small size  For larger sizes: One class for each two-power size 16 17 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Seglist Allocator (cont.) More Info on Allocators  To free a block:  D. Knuth, “ The Art of Computer Programming ”, 2 nd edition,  Coalesce and place on appropriate list Addison Wesley, 1973  The classic reference on dynamic storage allocation  Advantages of seglist allocators  Higher throughput  Wilson et al, “ Dynamic Storage Allocation: A Survey and  log time for power-of-two size classes Critical Review ”, Proc. 1995 Int’l Workshop on Memory  Better memory utilization Management, Kinross, Scotland, Sept, 1995.  First-fit search of segregated free list approximates a best-fit  Comprehensive survey search of entire heap.  Available from CS:APP student site (csapp.cs.cmu.edu)  Extreme case: Giving each block its own size class is equivalent to best-fit. 18 19 Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition Bryant and O’Hallaron, Computer Systems: A Programmer’s Perspective, Third Edition 3