Project 3a: Malloc and Free

Make sure to read the general instructions for the project here.

Changelog

  1. Clarified Mem_Available()
  2. Changed Mem_Free to return -1 if ptr is NULL or invalid. Added example.
  3. Changed Header max size from 16 to 32 bytes
  4. Added Extra Credit
  5. Added Test Information
  6. Clarified that Mem_Alloc() can use all the memory mmaped by Mem_Init()

Objectives

There are three objectives to this part of the assignment:

  • To understand the nuances of building a memory allocator.
  • To do so in a performance-efficient manner.
  • To understand how to optimize performance for a given workload.
  • To create shared libraries.
  • To have fun with a contest! (scroll down for details)

Overview

In this project, you will be implementing a memory allocator for the heap of a user-level process. Your functions will be to build your own malloc() and free().

Memory allocators have two distinct tasks. First, the memory allocator asks the operating system to expand the heap portion of the process's address space by calling either sbrk or mmap. Second, the memory allocator doles out this memory to the calling process. This involves managing a free list of memory and finding a contiguous chunk of memory that is large enough for the user's request; when the user later frees memory, it is added back to this list.

This memory allocator is usually provided as part of a standard library and is not part of the OS. To be clear, the memory allocator operates entirely within the virtual address space of a single process and knows nothing about which physical pages have been allocated to this process or the mapping from logical addresses to physical addresses; that part is handled by the operating system.

When implementing this basic functionality in your project, we have a few guidelines. First, when requesting memory from the OS, you must use mmap() (which is easier to use than sbrk()). Second, although a real memory allocator requests more memory from the OS whenever it can't satisfy a request from the user, your memory allocator must call mmap() only one time (when it is first initialized).

Classic malloc() and free() are defined as follows:

  • void *malloc(size_t size): malloc() allocates size bytes and returns a pointer to the allocated memory. The memory is not cleared (i.e., set to all zeroes).
  • void free(void *ptr): free() frees the memory space pointed to by ptr, which must have been returned by a previous call to malloc(). Otherwise, or if free(ptr) has already been called before, undefined behaviour occurs. If ptr is NULL, no operation is performed.

For simplicity, your implementations of Mem_Alloc(int size) and Mem_Free(void *ptr) should basically follow what malloc() and free() do; see below for details.

You will also provide two supporting functions: Mem_Available() and Mem_Dump(), described below.

Program Specifications

For this project, you will be implementing several different routines as part of shared libraries. Note that you will not be writing a main() routine for the code that you handin (but you should implement one for your own testing). We have provided the prototypes for these functions in the file mem.h (which is available at ~cs537-1/public/mem.h); you should include this header file in your code to ensure that you are adhering to the specification exactly. You should not change mem.h in any way! We now define each of these routines more precisely.

  • int Mem_Init(int sizeOfRegion): Mem_Init() is called one time by a process using your routines. sizeOfRegion is the number of bytes that you should request from the OS using mmap().

    Note that you may need to round up this amount so that you request memory in units of the page size (see the man pages for getpagesize()). Because of rounding up, you may request more memory than sizeOfRegion: you may use all of this memory for allocating memory to the user. Note also that you need to use this allocated memory for your own data structures as well; that is, your infrastructure for tracking the mapping from addresses to memory objects has to be placed in this region as well. You are not allowed to malloc(), or any other related function, in any of your routines! Similarly, you should not allocate global arrays. However, you may allocate a few global variables (e.g., a pointer to the head of your free list.)

    Return 0 on a success (when call to mmap is successful). Otherwise, return -1. Cases where Mem_Init() should return a failure: Mem_Init() is called more than once; sizeOfRegion is less than or equal to 0.

  • void *Mem_Alloc(int size): Mem_Alloc() is similar to the library function malloc(). Mem_Alloc() takes as input the size in bytes of the object to be allocated and returns a pointer to the start of that object. The function returns NULL if there is not enough contiguous free space within the memory (this may be more than sizeOfRegion if it is not a multiple of the page size) allocated by Mem_Init() to satisfy this request.

    For performance reasons, Mem_Alloc() should return 8-byte aligned chunks of memory. For example if a user allocates 1 byte of memory, your Mem_Alloc() implementation should return 8 bytes of memory so that the next free block will be 8-byte alligned too. To figure out whether you return 8-byte aligned pointers, you could print the pointer this way printf("%p", ptr) . The last digit should be a multiple of 8 (i.e. 0 or 8).

  • int Mem_Free(void *ptr): Mem_Free() frees the memory object that ptr points to. Just like with the standard free(), if ptr is NULL, then no operation is performed . The function returns 0 on success, and -1 otherwise. Return -1 if the ptr is invalid. The ptr is invalid if it is NULL or not a value returned by Mem_Alloc(). For example, if Mem_Alloc() returns pointers to address 0 and 16, then calling Mem_Free() with 8 should return -1.

    Coalescing: Mem_Free() should make sure to coalesce free space. Coalescing rejoins neighboring freed blocks into one bigger free chunk, thus ensuring that big chunks remain free for subsequent calls to Mem_Alloc().

  • int Mem_Available(): This should print out the number of bytes that can be allocated in the future by calls to Mem_Alloc(). Note that this should not include space used to store headers, bitmaps, etc. Note that Mem_Available() returns the total amount of usable space, not the largest contiguous space. If Mem_Alloc() was called immediately after Mem_Init() with the result of Mem_Available(), the call should succeed.

  • void Mem_Dump(): This is just a debugging routine for your own use. Have it print the regions of free memory to the screen.

You must provide these routines in a shared library. Placing the routines in a shared library instead of a simple object file makes it easier for other programmers to link with your code. There are further advantages to shared (dynamic) libraries over static libraries. When you link with a static library, the code for the entire library is merged with your object code to create your executable; if you link to many static libraries, your executable will be enormous. However, when you link to a shared library, the library's code is not merged with your program's object code; instead, a small amount of stub code is inserted into your object code and the stub code finds and invokes the library code when you execute the program. Therefore, shared libraries have two advantages: they lead to smaller executables and they enable users to use the most recent version of the library at run-time. To create a shared library named libmem1.so, use the following commands (assuming your library code is in a single file "mem.c"):

gcc -c -fpic mem.c -Wall -Werror
gcc -shared -o libmem1.so mem.o

To link with this library, you simply specify the base name of the library with "-lmem1" and the path so that the linker can find the library "-L.".

gcc -lmem1 -L. -o myprogram mymain.c -Wall -Werror

Of course, these commands should be placed in a Makefile. Before you run "myprogram", you will need to set the environment variable, LD_LIBRARY_PATH, so that the system can find your library at run-time. Assuming you always run myprogram from this same directory, you can use the command:

setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:.

If the setenv command returns an error "LD_LIBRARY_PATH: Undefined variable", do not panic. The error implies that your shell has not defined the environment variable. In this case, you simply need to run:

setenv LD_LIBRARY_PATH .

Note that setenv is what you use in tcsh; if you are using bash, you'll have to commands such as export.

Handing In Code

You should handin whatever files you use to implement your library, along with a Makefile that will generate the required libraries. The Makefile should have the all target, so that if we simply do make, we should get all the required libraries. If your Makefile requires us to do make my_project_name, you will fail the tests.

Workloads

Performance can often be improved drastically if the implementation is tailored to the workload. In this project, you will have a chance to understand this first-hand!

Your library will be tested with three workloads. You should submit the code for three libraries - libmem1.so, libmem2.so, and libmem3.so.

Workload 1: All calls to Mem_Alloc() will have size as 16 bytes. Mem_Alloc() can return NULL otherwise.

Workload 2: All calls to Mem_Alloc() will have size as one among 16, 80, and 256 bytes. Mem_Alloc() can return NULL otherwise.

Workload 3: Calls to Mem_Alloc() may have any size.

Unix Hints

In this project, you will use mmap to map zero'd pages (i.e., allocate new pages) into the address space of the calling process. Note there are a number of different ways that you can call mmap to achieve this same goal; we give one example here:

// open the /dev/zero device
int fd = open("/dev/zero", O_RDWR);

// sizeOfRegion (in bytes) needs to be evenly divisible by the page size
void *ptr = mmap(NULL, sizeOfRegion, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
if (ptr == MAP_FAILED) { perror("mmap"); exit(1); }

// close the device (don't worry, mapping should be unaffected)
close(fd);
return 0;

Reading

Chapter 16 from the free operating systems book.

Allocation Bitmap

Bitwise Operations

Note

If you use a header for each allocated block, the maximum size of such a header is 32 bytes.

Extra Credit

For exta credit, you should make your libraries work with multiple threads. In the tests, if we instantiate multiple threads that call Mem_Alloc() and Mem_Free() concurrently, the library should still work correctly. Note that Mem_Init() will still be called only once.

Grading

Your implementation will be graded on functionality. Each library will have different tests. In particular, we will be testing things like:

  • Allocating memory correctly - suppose you Mem_Init() 1024 bytes, you should be able to use much more of this for allocating memory in Workload 1, rather than in Workload 3 (can you figure out why?)
  • Freeing allocated memory successfully
  • Coalescing memory successfully - calls to Mem_Alloc() should succeed if enough memory has been freed. For example, if we start with 1024 bytes, allocate 512 bytes in units of 16 bytes, and then free all of them, we should be able to reallocate the 512 bytes.

Test Information

For each Workload library, we will be testing that a user is able to allocate a certain number of bytes. This should constrain the size of the headers in each library. In these tests, we don't fragment memory, so the user should be able to ideally use all of the space.

If we Mem_Init(4096), then the user should be able to allocate:

  • 4032 bytes for Workload 1.

  • 3840 bytes for Workload 2.

  • 1344 bytes for Workload 3.

If we Mem_Init(1024*1024), then the user should be able to allocate:

  • 983040 bytes for Workload 1.

  • 786432 bytes for Workload 2.

  • 349504 bytes for Workload 3.

To run the tests, execute: 

 python ~cs537-2/testing/p3a/MemTest.pyc dir

where dir is the directory containing your code. To help you debug, we are releasing the C code containing the actual tests. To access the tests, check in ~cs537-2/testing/p3a/3aTests/. There is a folder containing tests for each workload. For example, bucket_wl1 contains the tests for Workload 1.

Contest!

We will be holding a contest among students of each section. The contest will proceed as follows. To participate, students need to handin the code for libcontest1.so and libcontest2.so.

The tests for the contest are available at /u/c/s/cs537-2/testing/p3a/3aTests/contest.

Time. We will link your libcontest1.so with time.c. The program has to run without triggering any assertions. We will measure the time taken for the program to run. This will be measured as T.

Space. We will link your libcontest2.so with space.c. The program has to run without triggering any assertions. The code will output the total number of bytes successfully allocated. This will be measured as N.

There are three winners for time in each section: the people who get the least three values of T.

There are three winners for space in each section: the people who get the highest three values of N.

Two extra rules concerning winners:

  • A student may win in only one of the contests
  • Only one grad student can win in each contest - the other two winners must be undergrad students

Prizes will be CS 537 T-Shirts to show off your hard-won glory! Stay tuned for more details about the T-Shirts!