5. Debugging with the Memory Pool System

Memory management errors are some of the most stubborn and difficult to track down, because the effect so often appears at a distant point in the program that is seemingly unrelated to the cause, and by the time the error is revealed, the information needed to reconstruct the cause has long vanished. Immediately after an overwriting error, the block that overran its bounds is fine, and the block that was overwritten may not be visited for a long time. A failure to fix a reference does not necessarily cause the object pointed to by the missed reference to die immediately: there may be other references to that object, or a garbage collection may be delayed. And even if it does die, the space it occupies may not be re-allocated for some time.

5.1. General debugging advice

  1. Compile with debugging information turned on (-g on the GCC or Clang command line).

  2. Build the cool variety of the MPS (by defining the preprocessor constant CONFIG_VAR_COOL, for example by setting -DCONFIG_VAR_COOL on the GCC or Clang command line). This variety contains many internal consistency checks (including such checks on the critical path, which make it too slow for use in production), and can generate profiling output in the form of the telemetry stream.

  3. If your program triggers an assertion failure in the MPS, consult Common assertions and their causes for suggestions as to the possible cause.

  4. Prepare a reproducible test case if possible. The MPS may be asynchronous, but it is deterministic, so in single-threaded applications you should be able to get consistent results.

    However, you need to beware of address space layout randomization: if you perform computation based on the addresses of objects, for example, hashing objects by their address, then ASLR will cause your hash tables to be laid out differently on each run, which may affect the order of memory management operations. See Address space layout randomization below.

    A fact that assists with reproducibility is that the more frequently the collector runs, the sooner and more reliably errors are discovered. So if you have a bug that’s hard to reproduce, or which manifests itself in different ways on different runs, you may be able to provoke it more reliably, or get a more consistent result, by having a mode for testing in which you run frequent collections (by calling mps_arena_collect() followed by mps_arena_release()), perhaps as frequently as every allocation. (This will of course make the system run very slowly, but it ensures that if there are roots or references that are not being scanned then the failure will occur close in time to the cause, making it easier to diagnose.)

  5. Run your test case inside the debugger. Use assert() and abort() in your error handler (rather than exit()) so that you can enter the debugger with the contents of the control stack available for inspection.

  6. If you are using GDB on FreeBSD or Linux, you may want to avoid stopping on barrier (1) hits, because the MPS uses barriers to protect parts of memory, and barrier hits are common and expected. To avoid stopping on a barrier hit, run:

    handle SIGSEGV pass nostop noprint
    

    You can add this command to your .gdbinit if you always want it to be run.

    On macOS and Windows, barrier hits do not use signals and so do not enter the debugger.

  7. Similarly, if you are using GDB on FreeBSD or Linux, and if the client program is multi-threaded, you may want to avoid stopping when the MPS suspends and resumes threads by delivering signals to them. To avoid stopping on thread suspension and resumption, run:

    handle SIGXCPU pass nostop noprint
    handle SIGXFSZ pass nostop noprint
    

    If you have configured these signals as described under Signal and exception handling issues, you will need to adjust the signal names accordingly.

    On macOS and Windows, thread suspension and resumption does not use signals and so does not enter the debugger.

  8. If the client program is stopped in the debugger with the MPS part of the way through execution of an operation in an arena (for example, a crash inside a scan method), it will not be possible to call introspection functions, such as mps_arena_has_addr() or mps_addr_pool() (because the MPS is not re-entrant), and it may not be possible to examine some regions of memory (because they are protected by the MPS).

    If you are in this situation and would like to be able to call MPS functions or examine regions of memory from the debugger, then you can put the arena into the postmortem state by calling mps_arena_postmortem() from the debugger. This unlocks the arena and turns off protection.

    Warning

    After calling mps_arena_postmortem(), MPS-managed memory is not in a consistent state, and so it is not safe to continue running the client program.

5.2. Address space layout randomization

Address space layout randomization (ASLR) makes it hard to prepare a repeatable test case for a program that performs computation based on the addresses of objects, for example, hashing objects by their address. If this is affecting you, you’ll find it useful to disable ASLR when testing.

Here’s a small program that you can use to check if ASLR is enabled on your system. It outputs addresses from four key memory areas in a program (data segment, text segment, stack and heap):

#include <stdio.h>
#include <stdlib.h>

int data;

int main() {
    void *heap = malloc(4);
    int stack = 0;
    printf("data: %p text: %p stack: %p heap: %p\n",
           &data, (void *)main, &stack, heap);
    return 0;
}

When ASLR is turned on, running this program outputs different addresses on each run. For example, here are four runs on macOS 10.9.3:

data: 0x10a532020 text: 0x10a531ed0 stack: 0x7fff556ceb1c heap: 0x7f9f80c03980
data: 0x10d781020 text: 0x10d780ed0 stack: 0x7fff5247fb1c heap: 0x7fe498c03980
data: 0x10164b020 text: 0x10164aed0 stack: 0x7fff5e5b5b1c heap: 0x7fb783c03980
data: 0x10c7f8020 text: 0x10c7f7ed0 stack: 0x7fff53408b1c heap: 0x7f9740403980

By contrast, here are four runs on FreeBSD 8.3:

data: 0x8049728 text: 0x8048470 stack: 0xbfbfebfc heap: 0x28201088
data: 0x8049728 text: 0x8048470 stack: 0xbfbfebfc heap: 0x28201088
data: 0x8049728 text: 0x8048470 stack: 0xbfbfebfc heap: 0x28201088
data: 0x8049728 text: 0x8048470 stack: 0xbfbfebfc heap: 0x28201088

Here’s the situation on each of the operating systems supported by the MPS:

  • FreeBSD (as of version 10.0) does not support ASLR, so there’s nothing to do.

  • On Windows (Vista or later), ASLR is a property of the executable, and it can be turned off at link time using the /DYNAMICBASE:NO linker option.

  • On Linux (kernel version 2.6.12 or later), ASLR can be turned off for a single process by running setarch with the -R option:

    -R, --addr-no-randomize
           Disables randomization of the virtual address space
    

    For example:

    $ setarch $(uname -m) -R ./myprogram
    
  • On macOS (10.7 or later), ASLR can be disabled for a single process by starting the process using posix_spawn(), passing the undocumented attribute 0x100, like this:

    #include <spawn.h>
    
    pid_t pid;
    posix_spawnattr_t attr;
    
    posix_spawnattr_init(&attr);
    posix_spawnattr_setflags(&attr, 0x100);
    posix_spawn(&pid, argv[0], NULL, &attr, argv, environ);
    

    The MPS provides the source code for a command-line tool implementing this (tool/noaslr.c). We’ve confirmed that this works on macOS 10.9.3, but since the technique is undocumented, it may well break in future releases. (If you know of a documented way to achieve this, please contact us.)

5.3. Example: underscanning

An easy mistake to make is to omit to fix a reference when scanning a formatted object. For example, in the Scheme interpreter’s scan method, I might have forgotten to fix the first element of a pair:

case TYPE_PAIR:
case TYPE_PROMISE:
  /* oops, forgot: FIX(CAR(obj)); */
  FIX(CDR(obj));
  base = (char *)base + ALIGN_OBJ(sizeof(pair_s));
  break;

This means that as far as the MPS is concerned, the first element of the pair is unreachable and so dead, so after collecting the region of memory containing this object, the space will be reused for other objects. So CAR(obj) might end up pointing to the start of a valid object (but the wrong one), or to the middle of a valid object, or to an unused region of memory, or into an MPS internal control structure.

The reproducible test case is simple. Run a garbage collection by calling (gc) and then evaluate any expression:

$ gdb -q ./scheme
Reading symbols from ./scheme...done.
(gdb) run
Starting program: /home/grees/github.com/mps/example/scheme/scheme
MPS Toy Scheme Example
The prompt shows total allocated bytes and number of collections.
Try (vector-length (make-vector 100000 1)) to see the MPS in action.
You can force a complete garbage collection with (gc).
If you recurse too much the interpreter may crash from using too much C stack.
13248, 0> (gc)
Collection started.
  Why: Client requests: immediate full collection.
  Clock: 1819
Collection finished.
    live 5272
    condemned 16384
    not_condemned 0
    clock: 1987
13272, 1> foo
scheme: scheme.c:1421: lookup_in_frame: Assertion `TYPE(CAAR(frame)) == TYPE_SYMBOL' failed.

Program received signal SIGABRT, Aborted.
__GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
51  ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.

What’s going on?

(gdb) backtrace
#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1  0x00007ffff78058b1 in __GI_abort () at abort.c:79
#2  0x00007ffff77f542a in __assert_fail_base (fmt=0x7ffff797ca38 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", assertion=assertion@entry=0x555555604050 "TYPE(CAAR(frame)) == TYPE_SYMBOL", file=file@entry=0x555555603bb2 "scheme.c", line=line@entry=1421, function=function@entry=0x555555605390 <__PRETTY_FUNCTION__.3661> "lookup_in_frame") at assert.c:92
#3  0x00007ffff77f54a2 in __GI___assert_fail (assertion=0x555555604050 "TYPE(CAAR(frame)) == TYPE_SYMBOL", file=0x555555603bb2 "scheme.c", line=1421, function=0x555555605390 <__PRETTY_FUNCTION__.3661> "lookup_in_frame") at assert.c:101
#4  0x000055555555d0d8 in lookup_in_frame (frame=0x7ffff5900bb8, symbol=0x7ffff5883418) at scheme.c:1421
#5  0x000055555555d16c in lookup (env=0x7ffff5880180, symbol=0x7ffff5883418) at scheme.c:1440
#6  0x000055555555d2d1 in eval (env=0x7ffff5880180, op_env=0x7ffff5880198, exp=0x7ffff5883418) at scheme.c:1487
#7  0x0000555555564f29 in start (argc=0, argv=0x7fffffffe0d0) at scheme.c:4341
#8  0x0000555555565717 in main (argc=0, argv=0x7fffffffe0d0) at scheme.c:4489
(gdb) frame 4
#4  0x000055555555d0d8 in lookup_in_frame (frame=0x7ffff5900bb8, symbol=0x7ffff5883418) at scheme.c:1421
1421            assert(TYPE(CAAR(frame)) == TYPE_SYMBOL);
(gdb) print (char *)symbol->symbol.string
$1 = 0x7ffff5883428 "foo"

The backtrace shows that the interpreter is in the middle of looking up the symbol foo in the environment. The Scheme interpreter implements the environment as a list of frames, each of which is a list of bindings, each binding being a pair of a symbol and its value, as shown here:

Diagram: The environment data structure in the Scheme interpreter.

The environment data structure in the Scheme interpreter.

In this case, because the evaluation is taking place at top level, there is only one frame in the environment (the global frame). And it’s this frame that’s corrupt:

(gdb) list
1416        static obj_t lookup_in_frame(obj_t frame, obj_t symbol)
1417        {
1418          while(frame != obj_empty) {
1419            assert(TYPE(frame) == TYPE_PAIR);
1420            assert(TYPE(CAR(frame)) == TYPE_PAIR);
1421            assert(TYPE(CAAR(frame)) == TYPE_SYMBOL);
1422            if(CAAR(frame) == symbol)
1423              return CAR(frame);
1424            frame = CDR(frame);
1425          }
(gdb) print CAAR(frame)->type.type
$3 = 13

The number 13 is the value TYPE_PAD. So instead of the expected symbol, CAAR(frame) points to a padding object.

You might guess at this point that the symbol had not been fixed, and since you know that the frame is referenced by the car of the first pair in the environment, that’s the suspect reference. But in a more complex situation this might not yet be clear. In such a situation it can be useful to look at the sequence of events leading up to the detection of the error. See Telemetry.

5.4. Example: allocating with wrong size

Here’s another kind of mistake: an off-by-one error in make_string() leading to the allocation of string objects with the wrong size:

static obj_t make_string(size_t length, const char *string)
{
  obj_t obj;
  mps_addr_t addr;
  size_t size = ALIGN_OBJ(offsetof(string_s, string) + length /* oops, forgot +1 */);
  do {
    mps_res_t res = mps_reserve(&addr, obj_ap, size);
    if (res != MPS_RES_OK) error("out of memory in make_string");
    obj = addr;
    obj->string.type = TYPE_STRING;
    obj->string.length = length;
    if (string) memcpy(obj->string.string, string, length+1);
    else memset(obj->string.string, 0, length+1);
  } while(!mps_commit(obj_ap, addr, size));
  total += size;
  return obj;
}

Here’s a test case that exercises this bug:

(define (church n f a) (if (eqv? n 0) a (church (- n 1) f (f a))))
(church 1000 (lambda (s) (string-append s "x")) "")

And here’s how it shows up in the debugger:

$ gdb -q ./scheme
Reading symbols from ./scheme...done.
(gdb) run < test.scm
Starting program: /home/grees/github.com/mps/example/scheme/scheme < test.scm
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
MPS Toy Scheme Example
The prompt shows total allocated bytes and number of collections.
Try (vector-length (make-vector 100000 1)) to see the MPS in action.
You can force a complete garbage collection with (gc).
If you recurse too much the interpreter may crash from using too much C stack.
13248, 0> church
14104, 0> scheme: scheme.c:4067: obj_skip: Assertion `0' failed.

Program received signal SIGABRT, Aborted.
__GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
51  ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) backtrace
#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1  0x00007ffff78058b1 in __GI_abort () at abort.c:79
#2  0x00007ffff77f542a in __assert_fail_base (fmt=0x7ffff797ca38 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", assertion=assertion@entry=0x555555603e0f "0", file=file@entry=0x555555603c32 "scheme.c", line=line@entry=4067, function=function@entry=0x555555605568 <__PRETTY_FUNCTION__.4815> "obj_skip") at assert.c:92
#3  0x00007ffff77f54a2 in __GI___assert_fail (assertion=0x555555603e0f "0", file=0x555555603c32 "scheme.c", line=4067, function=0x555555605568 <__PRETTY_FUNCTION__.4815> "obj_skip") at assert.c:101
#4  0x000055555556434b in obj_skip (base=0x7ffff58af898) at scheme.c:4067
#5  0x00005555555e15bf in amcSegScanNailedRange (totalReturn=0x7fffffffd4a0, moreReturn=0x7fffffffd4a4, ss=0x7fffffffd5f0, amc=0x7ffff57f73f8, board=0x7ffff57f8a40, base=0x7ffff58ae000, limit=0x7ffff58afff0) at ../../code/poolamc.c:1278
#6  0x00005555555e177b in amcSegScanNailedOnce (totalReturn=0x7fffffffd4a0, moreReturn=0x7fffffffd4a4, ss=0x7fffffffd5f0, seg=0x7ffff57f89a8, amc=0x7ffff57f73f8) at ../../code/poolamc.c:1325
#7  0x00005555555e187c in amcSegScanNailed (totalReturn=0x7fffffffd5d0, ss=0x7fffffffd5f0, pool=0x7ffff57f73f8, seg=0x7ffff57f89a8, amc=0x7ffff57f73f8) at ../../code/poolamc.c:1355
#8  0x00005555555e1bd3 in amcSegScan (totalReturn=0x7fffffffd5d0, seg=0x7ffff57f89a8, ss=0x7fffffffd5f0) at ../../code/poolamc.c:1413
#9  0x0000555555599188 in SegScan (totalReturn=0x7fffffffd5d0, seg=0x7ffff57f89a8, ss=0x7fffffffd5f0) at ../../code/seg.c:778
#10 0x000055555558c926 in traceScanSegRes (ts=1, rank=1, arena=0x7ffff7ff7000, seg=0x7ffff57f89a8) at ../../code/trace.c:1148
#11 0x000055555558cb2d in traceScanSeg (ts=1, rank=1, arena=0x7ffff7ff7000, seg=0x7ffff57f89a8) at ../../code/trace.c:1223
#12 0x000055555558e24d in TraceAdvance (trace=0x7ffff7ff7ac0) at ../../code/trace.c:1664
#13 0x000055555558e792 in TracePoll (workReturn=0x7fffffffd7b0, collectWorldReturn=0x7fffffffd78c, globals=0x7ffff7ff7008, collectWorldAllowed=1) at ../../code/trace.c:1784
#14 0x000055555557c990 in ArenaPoll (globals=0x7ffff7ff7008) at ../../code/global.c:760
#15 0x00005555555690ed in mps_ap_fill (p_o=0x7fffffffd940, mps_ap=0x7ffff57f7928, size=24) at ../../code/mpsi.c:1111
#16 0x0000555555559c93 in make_pair (car=0x7ffff58afbf0, cdr=0x7ffff5880000) at scheme.c:457
#17 0x000055555555d605 in eval_list (env=0x7ffff58afed0, op_env=0x7ffff58afee8, list=0x7ffff5883828, message=0x5555556041f8 "eval: badly formed argument list") at scheme.c:1564
#18 0x000055555555d94b in eval_args_rest (name=0x555555604e02 "string-append", env=0x7ffff58afed0, op_env=0x7ffff58afee8, operands=0x7ffff5883828, restp=0x7fffffffdb00, n=0) at scheme.c:1637
#19 0x0000555555562870 in entry_string_append (env=0x7ffff58afed0, op_env=0x7ffff58afee8, operator=0x7ffff5882340, operands=0x7ffff5883828) at scheme.c:3396
#20 0x000055555555d420 in eval (env=0x7ffff58afed0, op_env=0x7ffff58afee8, exp=0x7ffff5883810) at scheme.c:1511
#21 0x000055555555dbc8 in entry_interpret (env=0x7ffff58af9e0, op_env=0x7ffff58af9f8, operator=0x7ffff5883700, operands=0x7ffff5883670) at scheme.c:1713
#22 0x000055555555d420 in eval (env=0x7ffff58af9e0, op_env=0x7ffff58af9f8, exp=0x7ffff5883598) at scheme.c:1511
#23 0x0000555555564fab in start (argc=0, argv=0x7fffffffe0d0) at scheme.c:4341
#24 0x0000555555565799 in main (argc=0, argv=0x7fffffffe0d0) at scheme.c:4489
(gdb) frame 4
#4  0x000055555556434b in obj_skip (base=0x7ffff58af898) at scheme.c:4067
4067            assert(0);
(gdb) list
4062            break;
4063          case TYPE_PAD1:
4064            base = (char *)base + ALIGN_WORD(sizeof(pad1_s));
4065            break;
4066          default:
4067            assert(0);
4068            fflush(stdout);
4069            fprintf(stderr, "Unexpected object on the heap\n");
4070            abort();
4071          }

The object being skipped is corrupt:

(gdb) print obj->type.type
$1 = -175623000

What happened to it? It’s often helpful in these situations to have a look at nearby memory.

(gdb) x/20gx obj
0x7ffff58af898:     0x00007ffff58834a8      0x00007ffff58af7c8
0x7ffff58af8a8:     0x0000000000000000      0x00007ffff58af890
0x7ffff58af8b8:     0x00007ffff58af660      0x0000000000000000
0x7ffff58af8c8:     0x00007ffff58836e8      0x00007ffff5880000
0x7ffff58af8d8:     0x0000000000000000      0x00007ffff58af5d0
0x7ffff58af8e8:     0x00007ffff58af8c0      0x0000000000000000
0x7ffff58af8f8:     0x00007ffff58af5b8      0x00007ffff58af8d8
0x7ffff58af908:     0x0000000000000000      0x00007ffff5880090
0x7ffff58af918:     0x00007ffff58af8f0      0x0000000000000000
0x7ffff58af928:     0x00007ffff58834f0      0x00007ffff5880000

You can see that this is a region of memory containing pairs, which have type TYPE_PAIR = 0 and consist of three words. But what’s that at the start of the region, where obj's tag should be? It looks like a pointer. So what’s in the memory just below obj? Let’s look at the previous region of memory:

(gdb) x/30gx (mps_word_t*)obj-28
0x7ffff58af7b8:     0x00007ffff5883840      0x00007ffff5880000
0x7ffff58af7c8:     0x0000000000000005      0x00000000000000b8
0x7ffff58af7d8:     0x7878787878787878      0x7878787878787878
0x7ffff58af7e8:     0x7878787878787878      0x7878787878787878
0x7ffff58af7f8:     0x7878787878787878      0x7878787878787878
0x7ffff58af808:     0x7878787878787878      0x7878787878787878
0x7ffff58af818:     0x7878787878787878      0x7878787878787878
0x7ffff58af828:     0x7878787878787878      0x7878787878787878
0x7ffff58af838:     0x7878787878787878      0x7878787878787878
0x7ffff58af848:     0x7878787878787878      0x7878787878787878
0x7ffff58af858:     0x7878787878787878      0x7878787878787878
0x7ffff58af868:     0x7878787878787878      0x7878787878787878
0x7ffff58af878:     0x7878787878787878      0x7878787878787878
0x7ffff58af888:     0x7878787878787878      0x0000000000000000
0x7ffff58af898:     0x00007ffff58834a8      0x00007ffff58af7c8

In this region we can see a string starting at address 0x7ffff58af7c8. Its type is TYPE_STRING = 5, its length is 0xb8 = 184, and its contents is 184 repetitions of the byte 0x78 (ASCII for “x”) resulting from the repeated (string-append s "x") in the test code. The string is followed by a pair (with type 0) at the address 0x7ffff58af890, which is one word before obj. So it looks as though obj should be pointing at the pair, but its value is one word too low.

The value in obj comes from skipping the string object:

(gdb) print obj_skip(0x7ffff58af7c8)
$3 = (void *) 0x7ffff58af898

So either obj_skip() has skipped one word too far, or the string object is one word too short. This should be enough evidence to track down the cause.

5.5. What next?

If you tracked down all your bugs, then the next step is the chapter Tuning the Memory Pool System for performance.

But if you’re still struggling, please contact us and see if we can help.