Dmalloc Tutorial
| [Top] | [Contents] | [Index] | [ ? ] |
Debug Malloc Library
Version Version 5.6.5 – December 2020
The debug memory allocation or dmalloc library has been designed as a drop in replacement for the system’s
malloc, realloc, calloc, free and other memory management routines while providing powerful
debugging facilities configurable at runtime. These facilities include such things as memory-leak tracking, fence-post
write detection, file/line number reporting, and general logging of statistics.
The library is reasonably portable having been run successfully on at least the following operating systems: AIX, DGUX, Free/Net/OpenBSD, GNU/Hurd, HPUX, Irix, Linux, OSX, NeXT, OSF/DUX, SCO, Solaris, Ultrix, Unixware, Windows, and Unicos on a Cray T3E. It also provides support for the debugging of threaded programs. See section Using the Library with a Thread Package.
The package includes the library, configuration scripts, debug utility application, test program, and documentation. Online documentation as well as the full source is available at URL http://dmalloc.com/. Details on the library’s mailing list are available there as well.
Please use the github issues URL https://github.com/j256/dmalloc/issues if you have any problems or to request features. Please include the version number of the library that you are using, your machine and operating system types, and the value of the DMALLOC_OPTIONS environment variable.
Gray Watson.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
1. Library License and Copying Information
Copyright 1992 to 2020 by Gray Watson.
Permission to use, copy, modify, and/or distribute this software for any purpose with or without fee is hereby granted, provided that the above copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2. Description of Features and How to Get Started
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.1 How to Install the Library
To configure, compile, and install the library, follow these steps carefully.
- Download the latest version of the library available from http://dmalloc.com/.
- The release files have a ‘.tgz’ file extension which means that they are a tar’d gzip’d directory of files. You will need to ungzip and then untar the release file into your source work directory. You may have to rename the file to ‘.tar.gz’ to get some old zip programs to handle the file correctly.
- You may want to edit or at least review the settings in ‘settings.dist’ to tune specific features of the library. The ‘configure’ script will copy this file to ‘settings.h’ which is where you should be adding per-architecture settings.
- Type sh ./configure to configure the library. You may want to first examine the ‘config.help’ file for some information about configure. You may want to use the --disable-cxx option if you do not want the Makefile to build the C++ version of dmalloc. You may want to use the --enable-threads option to build the threaded version of dmalloc. You may want to use the --enable-shlib option to build the shared versions of the dmalloc libraries. sh ./configure --help lists the available options to configure. Configure should generate the ‘Makefile’ and configuration files automatically.
- You may want to examine the ‘Makefile’ and ‘conf.h’ files created by configure to make sure it did its job correctly.
- You might want to tune the settings in ‘settings.h’ file to tune the library to the local architecture. This file contains relevant settings if you are using pthreads or another thread library. See section Using the Library with a Thread Package. The ‘configure’ script created this file from the ‘settings.dist’ file. Any permanent changes to these settings should made to the ‘settings.dist’ file. You then can run ‘config.status’ to re-create the ‘settings.h’ file.
- The
DMALLOC_SIZEvariable gets auto-configured in ‘dmalloc.h.2’ but it may not generate correct settings for all systems. You may have to alter the definitions in this file to get things to stop complaining when you go to compile about the size arguments to malloc routines. Comments on this please. - Typing make should be enough to build ‘libdmalloc.a’, and ‘dmalloc’ program. If it does not
work, please see if there are any notes in the contrib directory about your system-type. If not and you figure your
problem out, please send me some notes so future users can profit from your experiences.
NOTE: You may experience some errors compiling some of the ‘return.h’ assembly macros which attempt to determine the callers address for logging purposes. See section Issues Important for Porting the Library. You may want to first try disabling any compiler optimization flags. If this doesn’t work then you may need to disable the ‘USE_RETURN_MACROS’ variable in the ‘settings.h’ file.
NOTE: The code is dependent on an ANSI-C compiler. If the configure script gives the ‘WARNING’ that you do not have an ANSI-C compiler, you may still be able to add some sort of option to your compiler to make it ANSI. If there such is an option, please send it to the author so it can be added to the configure script.
- If you use threads and did not add the --enable-threads argument to configure, typing make threads
should be enough to build ‘libdmallocth.a’ which is the threaded version of the library. This may or may not work
depending on the configuration scripts ability to detect your local thread functionality. Feel free to send me mail
with improvements.
See the section of the manual on threads for more information about the operation of the library with your threaded program. See section Using the Library with a Thread Package.
- If you have a C++ compiler installed, the library should have automatically built ‘libdmallocxx.a’ which is
the C++ version of the library. If it was not done automatically, you can build it by typing make cxx. You
should link this library into your C++ programs instead of ‘libdmalloc.a’. See the ‘dmallocc.cc’ C++ file
which contains basic code to overload the
new,new[],delete, anddelete[]C++ operators. My apologies on the minimal C++ support. I am still living in a mostly C world. Any help improving this interface without sacrificing portability would be appreciated. - Typing make light should build and run the ‘dmalloc_t’ test program through a set of light trials. By default this will execute ‘dmalloc_t’ 5 times – each time will execute 10,000 malloc operations in a very random manner. Anal folks can type make heavy to up the ante. Use dmalloc_t --usage for the list of all ‘dmalloc_t’ options.
- Typing make install should install the ‘libdmalloc.a’ library in ‘/usr/local/lib’, the
‘dmalloc.h’ include file in ‘/usr/local/include’, and the ‘dmalloc’ utility in ‘/usr/local/bin’.
You may also want to type make installth to install the thread library into place and/or make installcc to
install the C++ library into place.
You may have specified a ‘--prefix=PATH’ option to configure in which case ‘/usr/local’ will have been replaced with ‘PATH’.
See the “Getting Started” section to get up and running with the library. See section Getting Started with the Library.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.2 Getting Started with the Library
This section should give you a quick idea on how to get going. Basically, you need to do the following things to make use of the library:
- Download the latest version of the library from http://dmalloc.com/.
- Run ./configure to configure the library. Follow the installation instructions on how to configure, make, and install the library (i.e. type: make install). See section How to Install the Library.
- Run ./make install to install the library on your system.
- You need to make sure that the library configuration and build process above was able to locate one of the
on_exitfunction,atexitfunction, or had compiler destructor support. If one of these functions or support is available then the dmalloc library should be able to automatically shut itself down when the program exits. This causes the memory statistics and unfreed information to be dumped to the log file. However, if your system has none of the above, then you will need to calldmalloc_shutdownyourself before your program exits. - To get the dmalloc utility to work you need to add an alias for dmalloc to your shell’s runtime configuration
file if supported. The idea is to have the shell capture the dmalloc program’s output and adjust the environment.
After you add the alias to the shell config file you need to log out and log back in to have it take effect, or you can execute the appropriate command below on the command line directly. After you setup the alias, if you enter dmalloc runtime and see any output with DMALLOC_OPTIONS in it then the alias did not take effect.
Bash, ksh, and zsh (http://www.zsh.org/) users should add the following to their ‘.bashrc’, ‘.profile’, or ‘.zshrc’ file respectively (notice the -b option for bourne shell output):
function dmalloc { eval `command dmalloc -b $*`; }If your shell does not support the
commandfunction then try:function dmalloc { eval `\dmalloc -b $*`; }or use a full path to where the dmalloc binary is installed:
function dmalloc { eval `/usr/local/bin/dmalloc -b $*`; }If you are still using csh or tcsh, you should add the following to your ‘.cshrc’ file (notice the -C option for c-shell output):
alias dmalloc 'eval `\dmalloc -C \!*`'
If you are using rc shell, you should add the following to your ‘.rcrc’ file (notice the -R option for rc-shell output):
fn dmalloc {eval `{/usr/local/bin/dmalloc $*}} - Although not necessary, you may want to include ‘dmalloc.h’ in your C files and recompile. This will allow
the library to report the file/line numbers of calls that generate problems. See section Macros Providing File and Line Information. It should be
inserted at the bottom of your include files as to not conflict with wother includes. You may want to ifdef it
as well and compile with cc -DDMALLOC …:
/* other includes above ^^^ */ #ifdef DMALLOC #include "dmalloc.h" #endif
- Another optional task is to compile all of your source with the ‘dmalloc.h’ with the
DMALLOC_FUNC_CHECKcompilation flag. This willallow the library to check all of the arguments of a number of common string and utility routines. See section Checking of Function Arguments.cc -DDMALLOC -DDMALLOC_FUNC_CHECK file.c
- Link the dmalloc library into your program. The dmalloc library should probably be placed at or near the end of the library list.
- Enable the debugging features by typing dmalloc -l logfile -i 100 low (for example). You should not see
any messages printed by the dmalloc utility (see NOTE below). This will:
- Set the malloc logfile name to ‘logfile’ (-l logfile). For programs which change directories, you may want to specify the full path to your logfile.
- Have the library check itself every 100 iterations (-i 100). This controls how fast your program will run. Larger numbers check the heap less and so it will run faster. Lower numbers will be more likely to catch memory problems.
- Enable a number of debug features (low). You can also try runtime for minimal checking or medium or high for more extensive heap verification.
- By default, the low, medium, and high values enable the
error-aborttoken which will cause the library to abort and usually dump core immediately upon seeing an error. See section Generating a Core File on Errors. You can disable this feature by entering dmalloc -m error-abort (-m for minus) to remove theerror-aborttoken and your program will just log errors and continue.
dmalloc --usage will provide verbose usage info for the dmalloc program. See section Dmalloc Utility Program.
You may also want to install the ‘dmallocrc’ file in your home directory as ‘.dmallocrc’. This allows you to add your own combination of debug tokens. See section Format of the Runtime Configuration File.
NOTE: The output from the dmalloc utility should be captured by your shell. If you see a bunch of stuff which includes the string
DMALLOC_OPTIONSthen the alias you should have created above is not working and he environmental variables are not being set. Make sure you’ve logged out and back in to have the alias take effect. - Run your program, examine the logfile that should have been created by
dmalloc_shutdown, and use its information to help debug your program.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.3 Basic Description of Terms and Functions
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.3.1 General Memory Terms and Concepts
Any program can be divided into 2 logical parts: text and data. Text is the actual program code in machine-readable format and data is the information that the text operates on when it is executing. The data, in turn, can be divided into 3 logical parts according to where it is stored: static, stack, and heap.
Static data is the information whose storage space is compiled into the program.
/* global variables are allocated as static data */
int numbers[10];
main()
{
…
}
|
Stack data is data allocated at runtime to hold information used inside of functions. This data is managed by the system in the space called stack space.
void foo()
{
/* this local variable is stored on the stack */
float total;
…
}
main()
{
foo();
}
|
Heap data is also allocated at runtime and provides a programmer with dynamic memory capabilities.
main()
{
/* the address is stored on the stack */
char * string;
…
/*
* Allocate a string of 10 bytes on the heap. Store the
* address in string which is on the stack.
*/
string = (char *)malloc(10);
…
/* de-allocate the heap memory now that we're done with it */
(void)free(string);
…
}
|
It is the heap data that is managed by this library.
Although the above is an example of how to use the malloc and free commands, it is not a good example of why using the heap for runtime storage is useful.
Consider this: You write a program that reads a file into memory, processes it, and displays results. You would like to handle files with arbitrary size (from 10 bytes to 1.2 megabytes and more). One problem, however, is that the entire file must be in memory at one time to do the calculations. You don’t want to have to allocate 1.2 megabytes when you might only be reading in a 10 byte file because it is wasteful of system resources. Also, you are worried that your program might have to handle files of more than 1.2 megabytes.
A solution: first check out the file’s size and then, using the heap-allocation routines, get enough storage to read the entire file into memory. The program will only be using the system resources necessary for the job and you will be guaranteed that your program can handle any sized file.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.3.2 Functionality Supported by All Malloc Libraries
All malloc libraries support 4 basic memory allocation commands. These include malloc, calloc,
realloc, and free. For more information about their capabilities, check your system’s manual pages – in
unix, do a man 3 malloc.
- Function: void *malloc ( unsigned int size )
-
Usage:
pnt = (type *)malloc(size)The malloc routine is the basic memory allocation routine. It allocates an area of
sizebytes. It will return a pointer to the space requested.
- Function: void *calloc ( unsigned int number, unsigned intsize )
-
Usage:
pnt = (type *)calloc(number, size)The calloc routine allocates a certain
numberof items, each ofsizebytes, and returns a pointer to the space. It is appropriate to pass in asizeof(type)value as the size argument.Also, calloc nulls the space that it returns, assuring that the memory is all zeros.
- Function: void *realloc ( void *old_pnt, unsigned int new_size )
-
Usage:
new_pnt = (type *)realloc(old_pnt, new_size)The realloc function expands or shrinks the memory allocation in
old_pnttonew_sizenumber of bytes. Realloc copies as much of the information fromold_pntas it can into thenew_pntspace it returns, up tonew_sizebytes. If there is a problem allocating this memory, 0L will be returned.If the
old_pntis 0L then realloc will do the equivalent of amalloc(new_size). Ifnew_sizeis 0 andold_pntis not 0L, then it will do the equivalent offree(old_pnt)and will return 0L.
- Function: void free ( void *pnt )
-
Usage:
free(pnt)The free routine releases allocation in
pntwhich was returned by malloc, calloc, or realloc back to the heap. This allows other parts of the program to re-use memory that is not needed anymore. It guarantees that the process does not grow too big and swallow a large portion of the system resources.
WARNING: there is a quite common myth that all of the space that is returned by malloc libraries has already been
cleared. Only the calloc routine will zero the memory space it returns.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.4 General Features of the Library
The debugging features that are available in this debug malloc library can be divided into a couple basic classifications:
- file and line number information
-
One of the nice things about a good debugger is its ability to provide the file and line number of an offending piece of code. This library attempts to give this functionality with the help of cpp, the C preprocessor. See section Macros Providing File and Line Information.
- return-address information
-
To debug calls to the library from external sources (i.e. those files that could not use the allocation macros), some facilities have been provided to supply the caller’s address. This address, with the help of a debugger, can help you locate the source of a problem. See section Getting Caller Address Information.
- fence-post (i.e. bounds) checking
-
Fence-post memory is the area immediately above or below memory allocations. It is all too easy to write code that accesses above or below an allocation – especially when dealing with arrays or strings. The library can write special values in the areas around every allocation so it will notice when these areas have been overwritten. See section Diagnosing Fence-Post Overwritten Memory.
NOTE: The library cannot notice when the program reads from these areas, only when it writes values. Also, fence-post checking will increase the amount of memory the program allocates.
- heap-constancy verification
-
The administration of the library is reasonably complex. If any of the heap-maintenance information is corrupted, the program will either crash or give unpredictable results.
By enabling heap-consistency checking, the library will run through its administrative structures to make sure all is in order. This will mean that problems will be caught faster and diagnosed better.
The drawback of this is, of course, that the library often takes quite a long time to do this. It is suitable to enable this only during development and debugging sessions.
NOTE: the heap checking routines cannot guarantee that the tests will not cause a segmentation-fault if the heap administration structures are properly (or improperly if you will) overwritten. In other words, the tests will verify that everything is okay but may not inform the user of problems in a graceful manner.
- logging statistics
-
One of the reasons why the debug malloc library was initially developed was to track programs’ memory usage – specifically to locate memory leaks which are places where allocated memory is never getting freed. See section Tracking Down Non-Freed Memory.
The library has a number of logging capabilities that can track un-freed memory pointers as well as runtime memory usage, memory transactions, administrative actions, and final statistics.
- examining freed memory
-
Another common problem happens when a program frees a memory pointer but goes on to use it again by mistake. This can lead to mysterious crashes and unexplained problems.
To combat this, the library can write special values into a block of memory after it has been freed. This serves two purposes: it will make sure that the program will get garbage data if it trying to access the area again, and it will allow the library to verify the area later for signs of overwriting.
If any of the above debugging features detect an error, the library will try to recover. If logging is enabled then an error will be logged with as much information as possible.
The error messages that the library displays are designed to give the most information for developers. If the error message is not understood, then it is most likely just trying to indicate that a part of the heap has been corrupted.
The library can be configured to quit immediately when an error is detected and to dump a core file or memory-image. This can be examined with a debugger to determine the source of the problem. The library can either stop after dumping core or continue running. See section Generating a Core File on Errors.
NOTE: do not be surprised if the library catches problems with your system’s routines. It took me hours to finally come to the conclusion that the localtime call, included in SunOS release 4.1, overwrites one of its fence-post markers.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
2.5 How the Library Checks Your Program
This is one of the newer sections of the library implying that it is incomplete. If you have any questions or issues that you’d like to see handled here, please let me know.
The dmalloc library replaces the heap library calls normally found in your system libraries with its own versions. When you make a call to malloc (for example), you are calling dmalloc’s version of the memory allocation function. When you allocate memory with these functions, the dmalloc library keeps track of a number of pieces of debugging information about your pointer including: where it was allocated, exactly how much memory was requested, when the call was made, etc.. This information can then be verified when the pointer is freed or reallocated and the details can be logged on any errors.
Whenever you reallocate or free a memory address, the dmalloc library always performs a number of checks on the pointer to make sure that it is valid and has not been corrupted. You can configure the library to perform additional checks such as detected fence-post writing. The library can also be configured to overwrite memory with non-zeros (only if calloc is not called) when it is allocated and erase the memory when the pointers are freed.
In addition to per-pointer checks, you can configure the library to perform complete heap checks. These complete checks
verify all internal heap structures and include walking all of the known allocated pointers to verify each one in turn.
You need this level of checking to find random pointers in your program which got corrupted but that won’t be freed for
a while. To turn on these checks, you will need to enable the check-heap debug token. See section Description of the Debugging Tokens. By
default this will cause the heap to be fully checked each and every time dmalloc is called whether it is a malloc, free,
realloc, or another dmalloc overloaded function.
Performing a full heap check can take a good bit of CPU and it may be that you will want to run it sporadically. This can be accomplished in a couple different ways including the ’-i’ interval argument to the dmalloc utility. See section Dmalloc Utility Program. This will cause the check to be run every N-th time. For instance, ’dmalloc -i 3’ will cause the heap to be checked before every 3rd call to a memory function. Values of 100 or even 1000 for high memory usage programs are more useful than smaller ones.
You can also cause the program to start doing detailed heap checking after a certain point. For instance, with ’dmalloc
-s 1000’ option, you can tell the dmalloc library to enable the heap checks after the 1000th memory call. Examine the
dmalloc log file produced and use the iteration count if you have LOG_ITERATION_COUNT enabled in your
‘settings.h’ file.
The start option can also have the format ‘file:line’. For instance, if it is set to ‘dmalloc_t.c:126’, dmalloc will start checking the heap after it sees a dmalloc call from the ‘dmalloc_t.c’ file, line number 126. If you use ‘dmalloc_t.c:0’, with a 0 line number, then dmalloc will start checking the heap after it sees a call from anywhere in the ‘dmalloc_t.c’ file.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3. How to Program with the Library
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.1 Macros Providing File and Line Information
By including ‘dmalloc.h’ in your C files, your calls to malloc, calloc, realloc, recalloc, memalign, valloc, strdup, and free are replaced with calls to _dmalloc_malloc, _dmalloc_realloc, and _dmalloc_free with various flags. Additionally the library replaces calls to xmalloc, xcalloc, xrealloc, xrecalloc, xmemalign, xvalloc, xstrdup, and xfree with associated calls.
These macros use the c-preprocessor __FILE__ and __LINE__ macros which get replaced at compilation time
with the current file and line-number of the source code in question. The routines use this information to produce
verbose reports on memory problems.
not freed: '0x38410' (22 bytes) from 'dmalloc_t.c:92' |
This line from a log file shows that memory was not freed from file ‘dmalloc_t.c’ line 92. See section Tracking Down Non-Freed Memory.
You may notice some non standard memory allocation functions in the above list. Recalloc is a routine like realloc that reallocates previously allocated memory to a new size. If the new memory size is larger than the old, recalloc initializes the new space to all zeros. This may or may not be supported natively by your operating system. Memalign is like malloc but should insure that the returned pointer is aligned to a certain number of specified bytes. Currently, the memalign function is not supported by the library. It defaults to returning possibly non-aligned memory for alignment values less than a block-size. Valloc is like malloc but insures that the returned pointer will be aligned to a page boundary. This may or may not be supported natively by your operating system but is fully supported by the library. Strdup is a string duplicating routine which takes in a null terminated string pointer and returns an allocated copy of the string that will need to be passed to free later to deallocate.
The X versions of the standard memory functions (xmalloc, xfree, etc.) will print out an error message to standard error and will stop if the library is unable to allocate any additional memory. It is useful to use these routines instead of checking everywhere in your program for allocation routines returning NULL pointers.
WARNING: If you are including the ‘dmalloc.h’ file in your sources, it is recommended that it be at the end of your include file list because dmalloc uses macros and may try to change declarations of the malloc functions if they come after it.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.2 Getting Caller Address Information
Even though the allocation macros can provide file/line information for some of your code, there are still modules which either you can’t include ‘dmalloc.h’ (such as library routines) or you just don’t want to. You can still get information about the routines that call dmalloc function from the return-address information. To accomplish this, you must be using this library on one of the supported architecture/compilers. See section Issues Important for Porting the Library.
The library attempts to use some assembly hacks to get the return-address or the address of the line that called the dmalloc function. If you have unfreed memory that does not have associated file and line information, you might see the following non-freed memory messages.
not freed: '0x38410' (22 bytes) from 'ra=0xdd2c' not freed: '0x38600' (10232 bytes) from 'ra=0x10234d' not freed: '0x38220' (137 bytes) from 'ra=0x82cc' |
With the help of a debugger, these return-addresses (or ra) can then be identified. I’ve provided a ‘ra_info.pl’ perl script in the ‘contrib/’ directory with the dmalloc sources which seems to work well with gdb. You can also use manual methods for gdb to find the return-address location. See section Translating Return Addresses into Code Locations.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.3 Checking of Function Arguments
One potential problem with the library and its multitude of checks and diagnoses is that they only get performed when a
dmalloc function is called. One solution this is to include ‘dmalloc.h’ and compile your source code with the
DMALLOC_FUNC_CHECK flag defined and enable the check-funcs token. See section Description of the Debugging Tokens.
cc -DDMALLOC -DDMALLOC_FUNC_CHECK file.c |
NOTE: Once you have compiled your source with DMALLOC_FUNC_CHECK enabled, you will have to recompile with it off to disconnect the library. See section How to Disable the library.
WARNING: You should be sure to have ‘dmalloc.h’ included at the end of your include file list because dmalloc uses macros and may try to change declarations of the checked functions if they come after it.
When this is defined dmalloc will override a number of functions and will insert a routine which knows how to check its
own arguments and then call the real function. Dmalloc can check such functions as bcopy, index,
strcat, and strcasecmp. For the full list see the end of ‘dmalloc.h’.
When you call strlen, for instance, dmalloc will make sure the string argument’s fence-post areas have not been
overwritten, its file and line number locations are good, etc. With bcopy, dmalloc will make sure that the
destination string has enough space to store the number of bytes specified.
For all of the arguments checked, if the pointer is not in the heap then it is ignored since dmalloc does not know anything about it.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.4 Generating a Core File on Errors
If the error-abort debug token has been enabled, when the library detects any problems with the heap memory, it
will immediately attempt to dump a core file. See section Description of the Debugging Tokens. Core files are a complete copy of the program and
it’s state and can be used by a debugger to see specifically what is going on when the error occurred. See section Using Dmalloc With a Debugger. By default, the low, medium, and high arguments to the library utility enable the error-abort
token. You can disable this feature by entering dmalloc -m error-abort (-m for minus) to remove the
error-abort token and your program will just log errors and continue. You can also use the error-dump
token which tries to dump core when it sees an error but still continue running. See section Description of the Debugging Tokens.
When a program dumps core, the system writes the program and all of its memory to a file on disk usually named ‘core’. If your program is called ‘foo’ then your system may dump core as ‘foo.core’. If you are not getting a ‘core’ file, make sure that your program has not changed to a new directory meaning that it may have written the core file in a different location. Also insure that your program has write privileges over the directory that it is in otherwise it will not be able to dump a core file. Core dumps are often security problems since they contain all program memory so systems often block their being produced. You will want to check your user and system’s core dump size ulimit settings.
The library by default uses the abort function to dump core which may or may not work depending on your operating
system. If the following program does not dump core then this may be the problem. See KILL_PROCESS definition
in ‘settings.dist’.
If abort does work then you may want to try the following setting in ‘settings.dist’. This code tries to
generate a segmentation fault by dereferencing a NULL pointer.
#define KILL_PROCESS { int *_int_p = 0L; *_int_p = 1; }
|
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.5 Additional Non-standard Routines
The library has a number of variables that are not a standard part of most malloc libraries:
- Variable: int dmalloc_errno
-
This variable stores the internal dmalloc library error number like errno does for the system calls. It can be passed to
dmalloc_strerror()(see below) to get a string version of the error. It will have a value of zero if the library has not detected any problems.
- Variable: char* dmalloc_logpath
-
This variable can be used to set the dmalloc log filename. The env variable
DMALLOC_LOGFILEoverrides this variable.
Additionally the library provides a number of non-standard malloc routines:
- Function: void dmalloc_shutdown ( void )
-
This function shuts the library down and logs the final statistics and information especially the non-freed memory pointers. The library has code to support auto-shutdown if your system has the
on_exit()call,atexit()call, or compiler destructor support (see ‘conf.h’). If you do not have these, thendmalloc_shutdownshould be called right beforeexit()or as the last function inmain().main() { … dmalloc_shutdown(); exit(0); }
- Function: int dmalloc_verify ( char * pnt )
-
This function verifies individual memory pointers that are suspect of memory problems. To check the entire heap pass in a NULL or 0 pointer. The routine returns DMALLOC_VERIFY_ERROR or DMALLOC_VERIFY_NOERROR.
NOTE: ‘dmalloc_verify()’ can only check the heap with the functions that have been enabled. For example, if fence-post checking is not enabled, ‘dmalloc_verify()’ cannot check the fence-post areas in the heap.
- Function: unsigned-int dmalloc_debug ( const unsigned int flags )
-
This routine sets the debug functionality flags and returns the previous flag value. It is helpful in server or cgi-bin programs where environmental variables cannot be used. See section Debugging Memory in a Server or Cgi-Bin Process. For instance, if debugging should never be enabled for a program, a call to
dmalloc_debug(0)as the first call inmain()will disable all the memory debugging from that point on.NOTE: you cannot add or remove certain flags such as signal handlers since they are setup at initialization time only.
NOTE: you can also use
dmalloc_debug_setupbelow.
- Function: unsigned-int dmalloc_debug_current ( void )
-
This routine returns the current debug functionality value value. This allows you to save a copy of the debug dmalloc settings to be changed and then restored later.
- Function: void dmalloc_debug_setup ( const char * options_str )
-
This routine sets the global debugging functionality as an option string. Normally this would be passed in in the DMALLOC_OPTIONS environmental variable. This is here to override the env or for circumstances where modifying the environment is not possible or does not apply such as servers or cgi-bin programs. See section Debugging Memory in a Server or Cgi-Bin Process.
Some examples:
/* * debug tokens high, threaded lock-on at 20, * log to dmalloc.%p (pid) */ dmalloc_debug_setup("debug=0x4f46d03,lockon=20,log=dmalloc.%p"); /* * turn on some debug tokens directly and log to the * file 'logfile' */ dmalloc_debug_setup( "log-stats,log-non-free,check-fence,log=logfile");
- Function: int dmalloc_examine ( const DMALLOC_PNT pnt, DMALLOC_SIZE * user_size_p, DMALLOC_SIZE * total_size_p, char ** file_p, int * line_p, DMALLOC_PNT * ret_addr_p, unsigned long * user_mark_p, unsigned long * seen_p )
-
This function returns the size of a pointer’s allocation as well as the total size given including administrative overhead, file and line or the return-address from where it was allocated, the last pointer when the pointer was "used", and the number of times the pointer has been "seen". It will return DMALLOC_NOERROR or DMALLOC_ERROR depending on whether pnt is good or not.
NOTE: This function is certainly not provided by most if not all other malloc libraries.
- Function: void dmalloc_track ( const dmalloc_track_t track_func )
-
Register an allocation tracking function which will be called each time an allocation occurs. Pass in NULL to disable. To take a look at what information is provided, see the dmalloc_track_t function typedef in dmalloc.h.
- Function: unsigned-long dmalloc_mark ( void )
-
Return to the caller the current "mark" which can be used later to log the pointers which have changed since this mark with the
dmalloc_log_changedfunction. Multiple marks can be saved and used.This is very useful when using the library with a server which does not exit. You can then save a mark before a transaction or event happens and then check to see what has changed using the
dmalloc_log_changedfunction below. See section Debugging Memory in a Server or Cgi-Bin Process.If you
LOG_ITERATIONenabled in your ‘settings.h’ file then the entries in the log file will be prepended with the number of memory transactions that the library has handled so far. You can also enableLOG_PNT_ITERATIONin ‘settings.h’ to store the memory transaction number with each pointer.
- Function: unsigned-long dmalloc_memory_allocated ( void )
-
Return to the caller the total number of bytes that have been allocated by the library. This is not the current in use but the total number of bytes returned by allocation functions.
- Function: unsigned-int dmalloc_page_size ( void )
-
Return to the caller the memory page-size being used by the library. This should be the same value as the one returned by the
getpagesize()function, if available.
- Function: unsigned-long dmalloc_count_changed ( const unsigned long mark, const int not_freed_b, const int free_b )
-
Count the pointers that have changed since the mark which was returned by
dmalloc_mark. Ifnot_freed_bis set to non-0 then count the pointers that have not been freed. Iffree_bis set to non-0 then count the pointers that have been freed.This can be used in conjunction with the
dmalloc_mark()function to help servers which never exit ensure that transactions or events are not leaking memory. See section Debugging Memory in a Server or Cgi-Bin Process.unsigned long mark = dmalloc_mark() ; … assert(dmalloc_count_changed(mark, 1, 0) == 0) ;
- Function: void dmalloc_log_stats ( void )
-
This routine outputs the current dmalloc statistics to the log file.
- Function: void dmalloc_log_unfreed ( void )
-
This function logs the unfreed-memory information to the log file. This is also useful to log the currently allocated points to the log file to be compared against another dump later on.
- Function: void dmalloc_log_changed ( const unsigned long mark, const int not_freed_b, const int freed_b, const int details_b )
-
Log the pointers that have changed since the mark which was returned by
dmalloc_mark. Ifnot_freed_bis set to non-0 then log the pointers that have not been freed. Iffree_bis set to non-0 then log the pointers that have been freed. Ifdetails_bset to non-0 then log the individual pointers that have changed otherwise just log the summaries.This can be used in conjunction with the
dmalloc_mark()function to help servers which never exit find transactions or events which are leaking memory. See section Debugging Memory in a Server or Cgi-Bin Process.
- Function: void dmalloc_vmessage ( const char * format, va_list args )
-
Write a message into the dmalloc logfile using vprintf-like arguments.
- Function: void dmalloc_message ( const char * format, ... )
-
Write a message into the dmalloc logfile using printf-like arguments.
- Function: void dmalloc_get_stats ( DMALLOC_PNT * heap_low_p, DMALLOC_PNT * heap_high_p, unsigned long * total_space_p, unsigned long * user_space_p, unsigned long * current_allocated_p, unsigned long * current_pnt_np, unsigned long * max_allocated_p, unsigned long * max_pnt_np, unsigned long * max_one_p)
-
This function return a number of statistics about the current heap. The pointers
heap_low_pandheap_high_pwill be set to the low and high spots in the heap.total_space_pwill be set to the total space in the heap including user space, administrative space, and overhead.user_space_pwill be set to the space given to the user process (allocated and free space).current_allocated_pwill be set to the current allocated space given to the user process.current_pnt_npwill be set to the current number of pointers allocated by the user process.max_allocated_pwill be set to the maximum allocated space given to the user process.max_pnt_npwill be set to the maximum number of pointers allocated by the user process.max_on_pwill be set to the maximum space allocated with one call by the user process.
- Function: const-char* dmalloc_strerror ( const int error_number )
-
This function returns the string representation of the error value in
error_number(which probably should be dmalloc_errno). This allows the logging of more verbose memory error messages.You can also display the string representation of an error value by a call to the ‘dmalloc’ program with a ‘-e #’ option. See section Dmalloc Utility Program.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.6 Description of the Internal Error Codes
The following error codes are defined in ‘error_val.h’. They are used by the library to indicate a detected
problem. They can be caused by the user (‘ERROR_TOO_BIG’) or can indicate an internal library problem
(‘ERROR_SLOT_CORRUPT’). The ‘dmalloc’ utility can give you the string version of the error with the -e
argument:
$ dmalloc -e 60 dmalloc: dmalloc_errno value '60' = 'pointer is not on block boundary' |
Here are the error codes set by the library. They are non contiguous on purpose because I add and delete codes all of the time and there are sections for various error-code types.
-
1 (ERROR_NONE) no error No error. It is good coding practice to set the no-error code to be non-0 value because it forces you to set it explicitly.
-
2 (INVALID_ERROR) Invalid error number. If the library outputs this error then your dmalloc utility may be out of date with the library you linked against. This will be returned with all error codes not listed here.
-
10 (ERROR_BAD_SETUP) initialization and setup failed Bad setup value. This is currently unused but it is intended to report on invalid setup configuration information.
-
11 (ERROR_IN_TWICE) malloc library has gone recursive Library went recursive. This usually indicates that you are not using the threaded version of the library. Or if you are then you are not using the ‘-o’ "lock-on" option. See section Using the Library with a Thread Package.
-
13 (ERROR_LOCK_NOT_CONFIG) thread locking has not been configured Thread locking has not been configured. This indicates that you attempted to use the ‘-o’ "lock-on" option without linking with the thread version of the library. You should probably be using
-ldmallocthnot-ldmallocwhen you are linking. Or you should include…/lib/libdmallocth.aon your compilation line.-
20 (ERROR_IS_NULL) pointer is null Pointer is null. The program passed a NULL (0L) pointer to
freeand you have theerror-free-nulltoken enabled.-
21 (ERROR_NOT_IN_HEAP) pointer is not pointing to heap data space Pointer is not pointing to heap data space. This means that the program passed an out-of-bounds pointer to
freeorrealloc. This could be someone trying to work with a wild pointer or trying to free a pointer from a different source thanmalloc.-
22 (ERROR_NOT_FOUND) cannot locate pointer in heap Cannot locate pointer in heap. The user passed in a pointer which the heap did not know about. Either this pointer was allocated by some other mechanism (like
mmaporsbrkdirectly) or it is a random invalid pointer.In some rare circumstances, sometimes seen with shared libraries, there can be two separate copies of the dmalloc library in a program. Each one does not know about the pointers allocated by the other.
-
23 (ERROR_IS_FOUND) found pointer the user was looking for This indicates that the pointer specified in the address part of the environmental variable was discovered by the library. See section Environment Variable Name and Features. This error is useful so you can put a breakpoint in a debugger to find where a particular address was allocated. See section Using Dmalloc With a Debugger.
-
24 (ERROR_BAD_FILE) possibly bad .c filename pointer A possibly invalid filename was discovered in the dmalloc administrative sections. This could indicate some corruption of the internal tables. It also could mean that you have a source file whose name is longer than 100 characters. See
MAX_FILE_LENGTHin the ‘settings.dist’ file.-
25 (ERROR_BAD_LINE) possibly bad .c file line-number A line-number was out-of-bounds in the dmalloc administrative sections. This could indicate some corruption of the internal tables. It also could mean that you have a source file containing more than
30000lines of code. SeeMAX_LINE_NUMBERin the ‘settings.dist’ file.-
26 (ERROR_UNDER_FENCE) failed UNDER picket-fence magic-number check This indicates that a pointer had its lower bound picket-fence magic number overwritten. If the
check-fencetoken is enabled, the library writes magic values above and below allocations to protect against overflow. Most likely this is because a pointer below it went past its allocate and wrote into the next pointer’s space.-
27 (ERROR_OVER_FENCE) failed OVER picket-fence magic-number check This indicates that a pointer had its upper bound picket-fence magic space overwritten. If the
check-fencetoken is enabled, the library writes magic values above and below allocations to protect against overflow. Most likely this is because an array or string allocation wrote past the end of the allocation.Check for improper usage of
strcat,sprintf,strcpy, and any other functions which work with strings and do not protect themselves by tracking the size of the string. These functions should always be replaced with:strncat,snprintf,strncpy, and others.-
28 (ERROR_WOULD_OVERWRITE) use of pointer would exceed allocation This error is generated by the function pointer checking code usually enabled with the
check-funcstoken. Dmalloc overloads a number of string and memory copying functions and verifies that the buffers (if allocated in the heap) would not be overwritten by the function.-
30 (ERROR_NOT_START_BLOCK) pointer is not to start of memory block This indicates that the user passed in a pointer to be freed or reallocated that was not at the start of the allocation. You would get this error, for example, if you allocate and get pointer
Xbut then try to freeX+1.-
40 (ERROR_BAD_SIZE) invalid allocation size This error indicates that a size value in the internal structures of the library were corrupted. This could be a random pointer problem, pointer overflow, or some other corruption.
-
41 (ERROR_TOO_BIG) largest maximum allocation size exceeded An allocation asked for memory larger than the configured maximum. This is a user configured setting. See
LARGEST_ALLOCATIONin the ‘settings.dist’ file. It is used to protect against wild allocation sizes. If you have super large allocation sizes then you should tune theLARGEST_ALLOCATIONvalue appropriately.-
43 (ERROR_ALLOC_FAILED) could not grow heap by allocating memory The library could not allocate more heap space and the program has run out of memory. This could indicate that you’ve overflowed some system imposed limit. On many operation systems, the
ulimitcall can tune system defaults. The library uses a lot more memory compared to the system malloc library because it stores a lot more information about the allocated pointers.NOTE: This also may be due to an inability of your operating system to use the
mmapsystem call to allocate memory. You may need to force theUSE_MMAPsetting to be 0. Please use the forums at URL http://dmalloc.com/ to report issues with this.-
45 (ERROR_OVER_LIMIT) over user specified allocation limit The library has allocated more memory than was specified in the memory-limit environmental variable. See section Environment Variable Name and Features.
-
60 (ERROR_NOT_ON_BLOCK) pointer is not on block boundary The user tried to free or realloc a pointer that was not pointing to a block boundary. You would get this error, for example, if you allocate and get pointer
Xbut then try to freeX+1.-
61 (ERROR_ALREADY_FREE) tried to free previously freed pointer The user tried to free a pointer than has already been freed. This is a very common mistake and can lead to serious problems. It can be because a destructor is being called twice for some reason. Although tracking down the specific source is highly recommended, it is good to set pointers to NULL (0L) after you free them as a rule.
-
67 (ERROR_FREE_OVERWRITTEN) free space has been overwritten If either the
free-blankorcheck-blanktokens are enabled then the library will overwrite memory when it is freed with the “dmalloc-free” byte (hex 0xdf, octal 0337, decimal 223). If the program writes into this space, then the library will detect the write and trigger this error. This could indicate that the program is using a pointer after it has been freed.-
70 (ERROR_ADMIN_LIST) bad admin structure list An internal corruption in the library’s administrative structures has been detected. This could be a random pointer problem, pointer overflow, or some other corruption.
-
72 (ERROR_ADDRESS_LIST) internal address list corruption An internal corruption in the library’s administrative structures has been detected. This could be a random pointer problem, pointer overflow, or some other corruption.
-
73 (ERROR_SLOT_CORRUPT) internal memory slot corruption An internal corruption in the library’s administrative structures has been detected. This could be a random pointer problem, pointer overflow, or some other corruption.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.7 How to Disable the library
If you would like to disable the library’s detailed checking features during a particularly allocation intensive section of code, you can do something like the following:
unsigned int dmalloc_flags; … /* turn off all debug flags and save a copy of old value */ dmalloc_flags = dmalloc_debug(0); /* section of a lot of allocations */ … /* end of section */ /* restore the dmalloc flag setting */ dmalloc_debug(dmalloc_flags); |
When you are finished with the development and debugging sessions, you may want to disable the dmalloc library and put in its place either the system’s memory-allocation routines, gnu-malloc, or maybe your own. Attempts have been made to make this a reasonably painless process. The ease of the extraction depends heavily on how many of the library’s features your made use of during your coding.
Reasonable suggestions are welcome as to how to improve this process while maintaining the effectiveness of the debugging.
-
If you want to totally disable the dmalloc library then you will need to recompile all the C files that include
‘dmalloc.h’ while defining
DMALLOC_DISABLE. This will cause the dmalloc macros to not be applied. See section Macros Providing File and Line Information.cc -g -DDMALLOC_DISABLE file.c
An alternative is to surround the
dmalloc.hinclusion or any direct dmalloc references with an#ifdef DMALLOCand then just remove the -DDMALLOC.#ifdef DMALLOC #include "dmalloc.h" #endif main() { … #ifdef DMALLOC dmalloc_verify(0L); #endif return 0; }// to get dmalloc information $ cc -DDMALLOC main.c // without dmalloc information $ cc main.c
-
If you compiled any of your source modules with
DMALLOC_FUNC_CHECKdefined then you must first recompile all those modules without the flag enabled.If you have disabled dmalloc with the
DMALLOC_DISABLEDflag or never included ‘dmalloc.h’ in any of your C files, then you will not need to recompile your sources when you need to disable the library.If you get unresolved references like
_dmalloc_mallocor_dmalloc_bcopythen something was not disabled as it should have been.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.8 Using the Library with C++
For those people using the C++ language, the library tries to configure and build ‘libdmallocxx.a’ library. This library should be linked into your C++ programs instead of ‘libdmalloc.a’.
Dmalloc is not as good with C++ as C because the dynamic memory routines in C++ are new() and delete() as
opposed to malloc() and free(). Since new and delete are usually not used as functions but rather as
x = new type, there is no easy way for dmalloc to pass in file and line information unfortunately. The
‘libdmallocxx.a’ library provides the file ‘dmallocc.cc’ which effectively redirects new to the more
familiar malloc and delete to the more familiar free.
NOTE: The author is not a C++ hacker so feedback in the form of other hints and ideas for C++ users would be much appreciated.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.9 Using Dmalloc With a Debugger
Here are a number of possible scenarios for using the dmalloc library to track down problems with your program.
You should first enable a logfile filename and turn on a set of debug features. You can use dmalloc -l logfile
low to accomplish this. If you are interested in having the error messages printed to your terminal as well, enable
the print-messages token by typing dmalloc -p print-messages afterwards. See section Dmalloc Utility Program.
Now you can enter your debugger (I use the excellent GNU debugger gdb), and put a break-point in
dmalloc_error() which is the internal error routine for the library. When your program is run, it will stop
there if a memory problem is detected.
If you are using GDB, I would recommend adding the contents of ‘dmalloc.gdb’ in the ‘contrib’ subdirectory to
your ‘.gdbinit’ file in your home directory. This enables the dmalloc command which will prompt you for the
arguments to the dmalloc command and will set a break point in dmalloc_error() automatically.
If you are using shared libraries, you may want to execute the following commands initially to load in dmalloc and other library symbols:
(gdb) sharedlibrary (gdb) add-shared-symbol-files |
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.9.1 Diagnosing General Problems with a Debugger
If your program stops at the dmalloc_error() routine then one of a number of problems could be happening.
Incorrect arguments could have been passed to a malloc call: asking for negative number of bytes, trying to realloc a
non-heap pointer, etc.. There also could be a problem with the system’s allocations: you’ve run out of memory, some
other function in your program is using the heap allocation functions mmap or sbrk, etc.. However, it is
most likely that some code that has been executed was naughty.
To get more information about the problem, first print via the debugger the dmalloc_errno variable to get the library’s internal error code. You can suspend your debugger and run dmalloc -e value-returned-from-print to get an English translation of the error. A number of the error messages are designed to indicate specific problems with the library administrative structures and may not be user-friendly.
If the problem was due to the arguments or system allocations then the source of the problem has been found. However,
if some code did something wrong, you may have some more work to do to locate the actual problem. The check-heap
token should be enabled and the interval setting disabled or set to a low value so that the library can find the problem
as close as possible to its source. The code that was execute right before the library halted, can then be examined
closely for irregularities. See section Description of the Debugging Tokens, See section Dmalloc Utility Program.
You may also want to put calls to dmalloc_verify(0) in your code before the section which generated the error.
This should locate the problem faster by checking the library’s structures at that point. See section Additional Non-standard Routines.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.9.2 Tracking Down Non-Freed Memory
So you’ve run your program, examined the log-file and discovered (to your horror) some un-freed memory. Memory leaks can become large problems since even the smallest and most insignificant leak can starve the program given the right circumstances.
not freed: '0x45008' (12 bytes) from 'ra=0x1f8f4' not freed: '0x45028' (12 bytes) from 'unknown' not freed: '0x45048' (10 bytes) from 'argv.c:1077' known memory not freed: 1 pointer, 10 bytes unknown memory not freed: 2 pointers, 24 bytes |
Above you will see a sample of some non-freed memory messages from the logfile. In the first line the ‘0x45008’ is the pointer that was not freed, the ‘12 bytes’ is the size of the unfreed block, and the ‘ra=0x1f8f4’ or return-address shows where the allocation originated from. See section Translating Return Addresses into Code Locations.
The systems which cannot provide return-address information show ‘unknown’ instead, as in the 2nd line in the sample above.
The ‘argv.c:1077’ information from the 3rd line shows the file and line number which allocated the memory which was not freed. This information comes from the calls from C files which included ‘dmalloc.h’. See section Macros Providing File and Line Information.
At the bottom of the sample it totals the memory for you and breaks it down to known memory (those calls which supplied the file/line information) and unknown (the rest).
Often, you may allocate memory in via strdup() or another routine, so the logfile listing where in the
strdup routine the memory was allocated does not help locate the true source of the memory leak – the routine
that called strdup. Without a mechanism to trace the calling stack, there is no way for the library to see who
the caller of the caller (so to speak) was.
However, there is a way to track down unfreed memory in this circumstance. You need to compile the library with
STORE_SEEN_COUNT defined in ‘conf.h’. The library will then record how many times a pointer has been
allocated or freed. It will display the unfreed memory as:
not freed: '0x45008|s3' (12 bytes) from 'ra=0x1f8f4' |
The STORE_SEEN_COUNT option adds a ‘|s#’ qualifier to the address. This means that the address in question
was seen ‘#’ many times. In the above example, the address ‘0x45008’ was seen ‘3’ times. The last time
it was allocated, it was not freed.
How can a pointer be “seen” 3 times? Let say you strdup a string of 12 characters and get address
‘0x45008’ – this is #1 time the pointer is seen. You then free the pointer (seen #2) but later strdup
another 12 character string and it gets the ‘0x45008’ address from the free list (seen #3).
So to find out who is allocating this particular 12 bytes the 3rd time, try dmalloc -a 0x45008:3. The library
will stop the program the third time it sees the ‘0x45008’ address. You then enter a debugger and put a break
point at dmalloc_error. Run the program and when the breakpoint is reached you can examine the stack frame to
determine who called strdup to allocate the pointer.
To not bother with the STORE_SEEN_COUNT feature, you can also run your program with the never-reuse token
enabled. This token will cause the library to never reuse memory that has been freed. Unique addresses are always
generated. This should be used with caution since it may cause your program to run out of memory.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.9.3 Diagnosing Fence-Post Overwritten Memory
For a definition of fence-posts please see the “Features” section. See section General Features of the Library.
To detect fence-post overruns, you need to enable the ‘check-fence’ token. See section Description of the Debugging Tokens. This pads your allocations with some extra bytes at the front and the end and watches the space to make sure that they don’t get overwritten. NOTE: The library cannot detect if this space gets read, only written.
If you have encountered a fence-post memory error, the logfile should be able to tell you the offending address.
free: failed UNDER picket-fence magic-number checking: pointer '0x1d008' from 'dmalloc_t.c:427' Dump of proper fence-bottom bytes: '\e\253\300\300\e\253\300\300' Dump of '0x1d008'-8: '\e\253\300\300WOW!\003\001pforger\023\001\123' |
The above sample shows that the pointer ‘0x1d008’ has had its lower fence-post area overwritten. This means that the code wrote below the bottom of the address or above the address right below this one. In the sample, the string that did it was ‘WOW!’.
The library first shows you what the proper fence-post information should look like, and then shows what the pointer’s bad information was. If it cannot print the character, it will display the value as ‘\ddd’ where ddd are three octal digits.
By enabling the check-heap debugging token and assigning the interval setting to a low number, you should be able
to locate approximately when this problem happened. See section Description of the Debugging Tokens, See section Dmalloc Utility Program.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.9.4 Translating Return Addresses into Code Locations
The following gdb commands help you translate the return-addresses (ra=) entries in the logfile into locations in your code. I’ve provided a ‘ra_info.pl’ perl script in the ‘contrib/’ directory with the dmalloc sources which seems to work well with gdb. But, if you need to do it manually, here are the commands in gdb to use.
# you may need to add the following commands to load in shared libraries (gdb) sharedlibrary (gdb) add-shared-symbol-files (gdb) x 0x10234d 0x10234d <_findbuf+132>: 0x7fffceb7 (gdb) info line *(0x82cc) Line 1092 of argv.c starts at pc 0x7540 and ends at 0x7550. |
In the above example, gdb was used to find that the two non-freed memory pointers were allocated in _findbuf()
and in file argv.c line 1092 respectively. The ‘x address’ (for examine) can always be used on the
return-addresses but the ‘info line *(address)’ will only work if that file was compiled using the -g option
and has not been stripped. This limitation may not be true in later versions of gdb.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.10 Using the Library with a Thread Package
Threads are special operating system facilities which allow your programs to have multiple threads of execution (hence the name). In effect your program can be doing a number of things “at the same time”. This allows you to take full advantage of modern operating system scheduling and multi-processor hardware. If I’ve already lost you or if any of the terminology below does not make sense, see manuals about POSIX threads (pthreads) before going any further. O’Reilly publishes a pretty good pthreads manual for example.
To use dmalloc with your threaded program, you will first need to make sure that you are linking with ‘libdmallocth.a’ which is the threaded version of the library. The support for threads in dmalloc should be adequate for most if not all testing scenarios. It provides support for mutex locking itself to protect against race conditions that result in multiple simultaneous execution. One of the major problems is that most thread libraries uses malloc themselves. Since all of dmalloc’s initialization happens when a call to malloc is made, we may be attempting to initialize or lock the mutex while the thread library is booting up. A very bad thing since thread libraries don’t expect to recurse.
The solution to this problem is to have the library not initialize or lock its mutex variable until after a certain number of allocation calls have been completed. If the library does not wait before initializing the locks, the thread library will probably core dump. If it waits too long then it can’t protect itself from multiple execution and it will abort or other bad things might happen. You adjust the number of times to wait at runtime with the ‘lock-on’ option to the dmalloc program (for example dmalloc -o 20). See section Dmalloc Utility Program. Times values between 5 and 30 are probably good although operating systems will vary significantly. You know its too low if your program immediately core dumps and too high if the dmalloc library says its gone recursive although with low values, you might get either problem.
An additional complexity is when we are initializing the lock before mutex locking around the library. As mentioned, the initialization itself may generate a malloc call causing the library to go recursive and the pthread library to possibly core dump. With the THREAD_INIT_LOCK setting defined in ‘settings.h’, you can tune how many times before we start locking to try and initialize the mutex lock. It defaults to 2 which seems to work for me. If people need to have this runtime configurable or would like to present an alternative default, please let me know.
So to use dmalloc with a threaded program, follow the following steps carefully.
- Follow the installation instructions on how to configure, make, and install the library but make sure to add the --enable-threads argument to configure. See section How to Install the Library.
- Typing make should be enough to build the threaded versions of the libraries including ‘libdmallocth.a’.
- Link the dmalloc threaded library into your program. The dmalloc library should probably be placed at or near the end of the library list.
- Enable the debugging options that you need by typing dmalloc -l logfile -i 100 low (for example). dmalloc --usage will provide verbose usage info for the dmalloc program. See section Dmalloc Utility Program.
- Enable the “lock-on” option (for example dmalloc -o 20). As explained above, you may have to try different values before getting it right. Values between 5 and 30 are probably good.
- If you get a dmalloc error #13 ‘thread locking has not been configured’ then you have not compiled you program with the threaded version of dmalloc or there was a problem building it.
- If everything works, you should be able to run your program, have it not immediately crash, and the dmalloc library should not complain about recursion.
If you have any specific questions or would like addition information posted in this section, please let me know. Experienced thread programmers only please.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.11 Using the library with Cygwin environment.
The Cygwin environment is a Linux-like environment for Windows. It provides Linux look and feel as well as a programming environment. See URL http://www.cygwin.com/ for more details.
Cygwin uses the GetEnvironmentVariableA function to read in environmental variables instead of getenv.
This functions are used to get the value of the ‘DMALLOC_OPTIONS’ variable which sets the debugging options.
See section Environment Variable Name and Features.
As of right now, dmalloc is not detecting the GetEnvironmentVariableA function correctly so you may need to tune
the ‘conf.h’ file to get it to work. See the sections on HAVE_GETENVIRONMENTVARIABLEA and
GETENV_SAVE settings. Feedback is welcome here.
If you still have problems reading in the environmental variables, you can work around this issue. You can add some
code into the main function in your program to initialize the dmalloc flags yourself. Here is a code sample:
main(int argc, char **argv)
{
#ifdef DMALLOC
/*
* Get environ variable DMALLOC_OPTIONS and pass the settings string
* on to dmalloc_debug_setup to setup the dmalloc debugging flags.
*/
dmalloc_debug_setup(getenv("DMALLOC_OPTIONS"));
#endif
/* rest of code in main starts here */
…
}
|
The #ifdef is just a good idea. I means that when debugging with dmalloc you need to compile your code with
-DDMALLOC. When you are done debugging you can remove the flag and the call to dmalloc_debug_setup will
be removed.
Please let me know if there is a better way to do this.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.12 Debugging Memory in a Server or Cgi-Bin Process
There are some specified challenges when trying to debug allocations in processes which do not startup, run, and then shutdown. Server processes (often called daemons) are those that are started (often at system boot time) and run perpetually. Other processes which are difficult to debug are CGI programs which are spawned by web servers or when you want to start debugging inside of a child process.
- Build your server or cgi-bin program with the dmalloc library like any other program. See section Getting Started with the Library.
- Add code into your program to enable the library flags to perform the memory checks that you require. Since these programs often do not run from the command line, you cannot use the dmalloc utility program and modify the process environment. See section Dmalloc Utility Program. The library provides a couple of functions to set the debugging flags when a program is running.
- To set the memory debugging flags, use the
dmalloc_debug_setupfunction which takes a string in the same format of the ‘DMALLOC_OPTIONS’ environmental variable. See section Environment Variable Name and Features. Use the dmalloc utility with the-nno-changes argument to see the appropriate settings for the ‘DMALLOC_OPTIONS’ environmental variable.> dmalloc -n -l logfile high Outputed: DMALLOC_OPTIONS=debug=0x4f4ed03,log=logfile export DMALLOC_OPTIONS
So if you want to turn on high debugging and log to the file ‘logfile’ then you would copy the above ‘DMALLOC_OPTIONS’ value into a call to
dmalloc_debug_setup. Notice that I have surrounded the dmalloc code with an#ifdef DMALLOCso you’ll have to compile using the-DDMALLOCflag.main() { #ifdef DMALLOC /* set the 'high' flags */ dmalloc_debug_setup("debug=0x4f47d03,log=logfile"); #endif … }Please note that the
dmalloc_debug_setupfunction does not know abouthigh,low, or other debug tokens but needs the actual flag values. - For earlier versions of the library (before 5.0.0) without
dmalloc_debug_setup, thedmalloc_debugfunction is available to set the flags directly, but it cannot adjust the logfile name and the other environment settings. You can use the dmalloc utility program to see what the numerical equivalent of the high token.> dmalloc -n high Outputed: DMALLOC_OPTIONS=debug=0x4f4ed03 export DMALLOC_OPTIONS
You can then take the
0x4f4ed03hexadecimal number and calldmalloc_debugwith that number.main() { #ifdef DMALLOC /* set the 'high' flags */ dmalloc_debug(0x4f4ed03); #endif … } - Even with the settings enabled, you may have problems getting the logfile to be written if your program is
running as ‘nobody’ or another user without permissions for security reasons. This is especially true for cgi-bin
programs. In this case you should specify a full path to your malloc logfile in a world writable directory
(ex.
dmalloc_debug_setup("debug=0x4f47d03,log=/var/tmp/malloc");). Watch for programs which change into other directories and which may cause logfiles specified as relative or local paths to be dropped in other locations. You may always want to use a full path logfile. - Once you have your settings enabled and your log is being generated, you may now want to check out how your
process is doing in terms of unfreed memory. Since it is not shutting down, the automatic unfreed log entries are not
being dropped to the logfile. By using the
dmalloc_markanddmalloc_log_changedfunctions, you can set a mark point at a certain place inside of your program, and then later see whether there are any unfreed pointers since the mark.main() { #ifdef DMALLOC /* set the 'high' flags */ dmalloc_debug_setup("debug=0x4f47d03,log=logfile"); #endif while (1) { /* accept a connection from a client */ accept_connection(); while (1) { #ifdef DMALLOC unsigned long mark; /* get the current dmalloc position */ mark = dmalloc_mark() ; #endif /* process the connection */ if (process_connection() != PROCESS_OK) { break; } #ifdef DMALLOC /* * log unfreed pointers that have been added to * the heap since mark */ dmalloc_log_changed(mark, 1 /* log unfreed pointers */, 0 /* do not log freed pointers */, 1 /* log each pnt otherwise summary */); #endif } /* close the connection with the client */ close_connection(); } … }Usually you would set the mark after the initializations and before each transaction is processed. Then for each transaction you can use
dmalloc_log_changedto show the unfreed memory. See section Additional Non-standard Routines. - You can also use the
dmalloc_log_statsfunction to dump general information about the heap. Also, remember that you can use thedmalloc_messageanddmalloc_vmessageroutines to annotate the dmalloc logfile with details to help you debug memory problems. See section Additional Non-standard Routines.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.13 Explanation of the Logfile Output
Most of time you will be using the logfile output from library as the sole information source for diagnosing problems in and getting statistics for your program.
1098918225: 3: Dmalloc version 'Version 5.6.5'
1098918225: 3: flags = 0x4f4e503, logfile '/tmp/dmalloc.log'
1098918225: 3: interval = 500, addr = 0, seen # = 0, limit = 0
1098918225: 3: starting time = 1098918225
1098918225: 3: process pid = 32406
1098918226: 4: WARNING: tried to free(0) from foo.c:708'
1098918228: 20: *** free: at 'unknown' pnt '0xed310080|s2': \
size 12, alloced at 'bar.c:102'
1098918230: 50: ERROR: heap_check: free space was overwritten (err 67)
1098918230: 50: error details: checking free pointer
1098918230: 50: pointer '0x291c5' from 'unknown' prev access 'foo.c:787'
|
Here is a short example of some logfile information. Each of the lines are prefixed by the time (in epoch seconds since
1/1/1970) and the iteration or call count which is the number of times the library has been called from malloc, free,
verify, etc.. In the above example, the first 5 log entries where written at epoch 1098918225 or ‘Wed Oct 27
19:03:45 2004 EST’ and they were generated by the 3rd call to the library. See the ‘settings.dist’ file entries to
tune what elements appear on each line: LOG_TIME_NUMBER, LOG_ITERATION, LOG_PID, etc.. You can convert the epoch
seconds to a date from the command line with the following perl code: perl -e 'print localtime($ARGV[0])."\n";'
epoch-seconds-number
The first 5 lines of the sample logfile contain header information for all logfiles. They show the version number and URL for the library as well as all of the settings that the library is currently using. These settings are tuned using the dmalloc utility program. See section Dmalloc Utility Program. The 5th line of is the process-id that generated the logfile.
The 6th line in the above example is what causes the logfile to be opened and the header to be written. It is a warning that tells you that you tried to free a 0L pointer at a certain location. You can disable these warnings by setting ‘ALLOW_FREE_NULL_MESSAGE’ to 0 in ‘settings.dist’.
Line 7 is an example of a transaction log that you get when you enable the log-trans debug token. See section Description of the Debugging Tokens. This line shows that a call to free was made from an unknown location. It is unknown because the file in
question did not include ‘dmalloc.h’ to get file/line-number information. The call to free was freeing the pointer
address 0xed310080 which we have “seen” 2 times (s2). We saw the pointer when it was allocated and then we are
seeing it again when it was freed. Because the library is reusing pointers (reclaiming freed memory) the seen count
helps to track how many times a pointer was used. The last part of the line shows that the pointer to be freed was
allocated by ‘bar.c’ line 102.
Lines 8-10 is the next problem that the library caught and this one is an error. It happened 5 seconds from the start of the log (1098918230) and at the 50th call into the library. It shows that an allocation that had been freed then was overwritten. This may imply that someone tried to use memory after it was freed or that there was a loose pointer reference. The last two lines give more details about when the error was discovered, the address of the offending pointer, and when the pointer was previous accessed, in this case freed. To discover where this problem is happening, you can use a debugger. See section Using Dmalloc With a Debugger.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
3.14 Various Other Hints That May Help
One of the problems that is often seen is that a program crashes in the libc memory code and you suspect a heap
memory problem but both dmalloc and maybe valgrind don’t show any problems. One of the big problems with debugging is
that it is very difficult to do it without effecting how the program is run. Sometimes errors are due to subtle race
conditions that are only seen when the program is running at full speed – not slowed down by debugging code.
This is especially true with threaded code which is often heavily affected when used with dmalloc and valgrind. Older versions of valgrid (maybe current) forced all threads into a single virtual system by design, which often masks reentrance bugs.
One way to work through these issues is to run with the library with very few debugging flags enabled. Many memory problems are fence-post areas so start with dmalloc checking just the fence post and error logging enabled:
dmalloc -d 0 -l dmalloc.log -p log-stats -p log-non-free -p check-fence -p check-funcs |
This enabled a small number of checks and should cause your program to run at close to full speed. The library has never been optimized for speed so some performance penalties will be felt.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4. Dmalloc Utility Program
The dmalloc program is designed to assist in the setting of the environment variable ‘DMALLOC_OPTIONS’.
See section Environment Variable Name and Features. It is designed to print the shell commands necessary to make the appropriate changes to
the environment. Unfortunately, it cannot make the changes on its own so the output from dmalloc should be sent through
the eval shell command which will do the commands.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4.1 Using a Shell Alias with the Utility
The dmalloc program is designed to assist in the setting of the environment variable ‘DMALLOC_OPTIONS’.
See section Environment Variable Name and Features. It is designed to print the shell commands necessary to make the appropriate changes to
the environment. Unfortunately, it cannot make the changes on its own so the output from dmalloc should be sent through
the eval shell command which will do the commands.
With shells that have aliasing or macro capabilities: csh, bash, ksh, tcsh, zsh, etc., setting up an alias to dmalloc to do the eval call is recommended. Bash, ksh, and zsh users should add the following to their ‘.bashrc’, ‘.profile’, or ‘.zshrc’ file respectively (notice the -b option for bourne shell output):
function dmalloc { eval `command dmalloc -b $*`; }
|
If your shell does not support the command function then try:
function dmalloc { eval `\dmalloc -b $*`; }
|
or use the full path to the dmalloc binary:
function dmalloc { eval `/usr/local/bin/dmalloc -b $*`; }
|
If you are using csh or tcsh, you should add the following to your ‘.cshrc’ file (notice the -C option for c-shell output):
alias dmalloc 'eval `\dmalloc -C \!*`' |
This allows the user to execute the dmalloc command as ‘dmalloc arguments’.
Users of versions of the Bourne shell (usually known as /bin/sh) that don’t have command functions will need to send the output to a temporary file and the read it back in with the “.” command:
$ dmalloc -b arguments … > /tmp/out $ . /tmp/out |
By the way, if you are looking for a shell, I heartily recommend trying out zsh. It is a bourne shell written from scratch with much the same features as tcsh without the csh crap.
NOTE: After you add the alias to the file you need to log out and log back in to have it take effect, or you can execute the above appropriate command on the command line. If you enter dmalloc runtime and see any output with DMALLOC_OPTIONS in it then the alias did not work.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4.2 How to Use the Dmalloc Program
The most basic usage for the program is ‘dmalloc [-bC] tag’. The ‘-b’ or ‘-C’ (either but not both flags
used at a time) are for generating Bourne or C shell type commands respectively. dmalloc will try and use the
SHELL environment variable to determine whether bourne or C shell commands should be generated but you may want
to explicitly specify the correct flag.
The ‘tag’ argument to dmalloc should match a line from the user’s runtime configuration file or should be one of the built-in tags. See section Format of the Runtime Configuration File. If no tag is specified and no other option-commands used, dmalloc will display the current settings of the environment variable. It is useful to specify one of the verbose options when doing this.
To find out the usage for the debug malloc program try ‘dmalloc --usage-long’. The standardized usage message that will be displayed is one of the many features of the argv library included with this package.
It is available on the web at URL http://256stuff.com/sources/argv/. See the documentation there for more information.
Here is a detailed list of the flags that can passed to dmalloc:
-
-a address Set the ‘addr’ part of the ‘DMALLOC_OPTIONS’ variable to address (or alternatively address:number).
-
-b Output Bourne shell type commands. Usually handled automagically.
-
-C Output C shell type commands. Usually handled automagically.
-
-c Clear/unset all of the settings not specified with other arguments. You can do this automatically when you set to a new tag with the -r option.
NOTE: clear will never unset the ‘debug’ setting. Use -d 0 or a tag to ‘none’ to achieve this.
-
-d bitmask Set the ‘debug’ part of the ‘DMALLOC_OPTIONS’ env variable to the bitmask value which should be in hex. This is overridden (and unnecessary) if a tag is specified.
-
-D List all of the debug-tokens. Useful for finding a token to be used with the -p or -m options. Use with -v or -V verbose options.
-
-e errno Print the dmalloc error string that corresponds to the error number errno.
-
-f filename Use this configuration file instead of the RC file ‘$HOME/.dmallocrc’.
-
-g Output gdb type commands for using inside of the gdb debugger.
-
-h (or --help) Output a help message for the utility.
-
-i number -
Set the checking interval to number. If the
check-heaptoken is enabled, this causes the library to only check the heap every Nth time which can significantly increase the running speed of your program. If a problem is found, however, this limits your ability to determine when the problem occurred. Try values of 50 or 100 initially. -
-k Do not reset all of the settings when a tag is specified. This specifically overrides the -r option and is provided here to override -r if it has been added to the dmalloc alias.
-
-l filename Write the debugging output and other log-file information to the filename. Filename can include some of the following patterns which get expanded into strings:
-
%h Gets expanded into the hostname if the
gethostname()function is available.-
%i Gets expanded into the thread-id if the library has been configure to be used with threads. See section Using the Library with a Thread Package. See the end of the ‘settings.dist’ file for settings which return the thread-id and convert it into a string.
-
%p Gets expanded into the process-id if the
getpid()function is available.-
%t Gets expanded into the time value in seconds if the
time()function is available.-
%u Gets expanded into the user-id number if the
getuid()function is available.
Some examples:
# logfile produced with pid extension: # logfile.8412 or logfile.31451 dmalloc -l logfile.%p # hostname and time extensions: # dmalloc-box1.foo.com-1055213240 dmalloc -l dmalloc-%h-%t # if threads enabled, have thread-id extension: log.thread32 dmalloc -l log.thread%i
-
-L Write the debug-value into the environment not in hex but by individual debug-tokens in long form.
-
-m token(s) Remove (minus) the debug capabilities of token(s) from the current debug setting or from the selected tag (or -d value). Multiple -m options can be specified.
-
-M limit Set the memory allocation limit which will abort the program if the total memory allocations exceed this number of bytes. The limit can be a number with a k, m, or g at the end to indicate kilobyte, megabyte, and gigabyte respectively. Ex: 100k, 200m, 1g. If the limit is exceeded, this will generate an
ERROR_OVER_LIMITerror. See section Description of the Internal Error Codes.-
-n Without changing the environment, output the commands resulting from the supplied options.
-
-o times Set the “lock-on” period which dictates to the threaded version of the library to not initialize or lock the mutex lock around the library until after a certain number of allocation calls have been made. Some number between 2 and 30 is probably good. See the “Using With Threads” section for more information about the operation of the library with threads. See section Using the Library with a Thread Package.
-
-p token(s) Add (plus) the debug capabilities of token(s) to the current debug setting or to the selected tag (or -d value). Multiple -p options can be specified.
-
-r Remove (unset) all settings when using a tag. This is useful when you are returning to a standard development tag and want the logfile, address, and interval settings to be cleared automatically. If you want this behavior by default, this can be put into the dmalloc alias.
-
-R Output rc shell type commands. This is not for the runtime configuration file but for the rc shell program.
-
-s file:line Set the ‘start’ part of the ‘DMALLOC_OPTIONS’ env variable to a file-name and line-number location in the source where the library should begin more extensive heap checking. The file and line numbers for heap transactions must be working for this option to be obeyed. This is used if you are trying to locate a problem and you want the extensive checking to not happen initially because it’s too slow.
-
-S number Set the ‘start’ part of the ‘DMALLOC_OPTIONS’ env variable to an dmalloc mark number. The library will begin more extensive heap checking after this number of memory transactions. If you
LOG_ITERATIONenabled in your ‘settings.h’ file then the entries in the log file will be prepended with the number of memory transactions that the library has handled so far. This number can be used to delay the start of the fine grained heap checking which can be very slow.-
--start-size size Set the ‘start’ part of the ‘DMALLOC_OPTIONS’ env variable to a number of bytes. The library will begin more extensive heap checking after this amount of memory has been allocated by the library. This allows you to start the slow and detailed checking of the library later in the program execution. You can use patterns like 250m, 1g, or 102k to mean 250 megabytes, 1 gigabyte, and 102 kilobytes respectively.
-
-t List all of the tags in the rc-file. Use with -v or -V verbose options.
-
-u (or --usage) Output the usage information for the utility.
-
-v Give verbose output. Especially useful when dumping current settings or listing all of the tags.
-
-V Give very verbose output for outputting even more details about settings.
-
--version Output the version string for the utility. Please note that the version of the library that is installed or has been linked into your application may be different from the utility version.
If no arguments are specified, dmalloc dumps out the current settings that you have for the environment variable. For example:
Debug-Flags '0x40005c7' (runtime) Address 0x1f008, count = 3 Interval 100 Logpath 'malloc' Start-File not-set |
With a -v option and no arguments, dmalloc dumps out the current settings in a verbose manner. For example:
Debug-Flags '0x40005c7' (runtime) log-stats, log-non-free, log-bad-space, check-fence, catch-null Address 0x1f008, count = 10 Interval 100 Logpath 'malloc' Start-File not-set |
Here are some examples of dmalloc usage:
# start tough debugging, check the heap every 100 times, # send the log information to file 'logfile' dmalloc high -i 100 -l logfile # find out what error code 20 is (from the logfile) dmalloc -e 20 # cause the library to halt itself when it sees the address 0x34238 # for the 6th time. dmalloc -a 0x34238:6 # send the log information to file 'logfile' with the time in seconds # as an extension. dmalloc -l logfile.%t # return to the normal 'runtime' settings and clear out all # other settings dmalloc -c runtime # enable basic 'low' settings plus (-p) the logging of # transactions (log-trans) to file 'logfile' dmalloc low -p log-trans -l logfile # print out the current settings with Very-verbose output dmalloc -V # list the available debug malloc tokens with Very-verbose output dmalloc -DV # list the available tags from the rc file with verbose output dmalloc -tv |
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4.3 Environment Variable Name and Features
An environment variable is a variable that is part of the user’s working environment and is shared by all the
programs. The ‘DMALLOC_OPTIONS’ variable is used by the dmalloc library to enable or disable the memory debugging
features, at runtime. NOTE: you can also use the dmalloc_debug_setup function to set the option string.
It can be set either by hand or with the help of the dmalloc program. See section Dmalloc Utility Program.
To set it by hand, Bourne shell (sh, bash, ksh, or zsh) users should use:
DMALLOC_OPTIONS=value export DMALLOC_OPTIONS |
C shell (csh or tcsh) users need to invoke:
setenv DMALLOC_OPTIONS value |
The value in the above examples is a comma separated list of tokens each having a corresponding value. The tokens are described below:
-
debug -
This should be set to a value in hexadecimal which corresponds to the functionality token values added together. See section Description of the Debugging Tokens. For instance, if the user wanted to enable the logging of memory transactions (value ‘0x008’) and wanted to check fence-post memory (value ‘0x400’) then ‘debug’ should be set to ‘0x408’ (‘0x008’ + ‘0x400’).
NOTE: You don’t have to worry about remembering all the hex values of the tokens because the dmalloc program automates the setting of this variable especially.
NOTE: You can also specify the debug tokens directly, separated by commas. See section Description of the Debugging Tokens. If ‘debug’ and the tokens are both used, the token values will be added to the debug value.
-
lockon -
Set this to a number which is the “lock-on” period. This dictates to the threaded version of the library to not initialize or lock the mutex lock around the library until after a certain number of allocation calls have been made. See the “Using With Threads” section for more information about the operation of the library with threads. See section Using the Library with a Thread Package.
-
log -
Set this to a filename so that if ‘debug’ has logging enabled, the library can log transactions, administration information, and/or errors to the file so memory problems and usage can be tracked.
To get different logfiles for different processes, you can assign ‘log’ to a string with
%din it (for instance ‘logfile.%d’). This will be replaced with the pid of the running process (for instance ‘logfile.2451’).WARNING: it is easy to core dump any program with dmalloc, if you send in a format with arguments other than the one
%d. -
addr -
When this is set to a hex address (taken from the dmalloc log-file for instance) dmalloc will abort when it finds itself either allocating or freeing that address.
The address can also have an ‘:number’ argument. For instance, if it was set it to ‘0x3e45:10’, the library will kill itself the 10th time it sees address ‘0x3e45’. By setting the number argument to 0, the program will never stop when it sees the address. This is useful for logging all activity on the address and makes it easier to track down specific addresses not being freed.
This works well in conjunction with the
STORE_SEEN_COUNToption. See section Tracking Down Non-Freed Memory.NOTE: dmalloc will also log all activity on this address along with a count.
-
inter -
By setting this to a number X, dmalloc will only check the heap every X times. This means a number of debugging features can be enabled while still running the program within a finite amount of time.
A setting of ‘100’ works well with reasonably memory intensive programs. This of course means that the library will not catch errors exactly when they happen but possibly 100 library calls later.
-
start -
Set this to a number X and dmalloc will begin checking the heap after X times. This means the intensive debugging can be started after a certain point in a program.
‘start’ also has the format ‘file:line’. For instance, if it is set to ‘dmalloc_t.c:126’ dmalloc will start checking the heap after it sees a dmalloc call from the ‘dmalloc_t.c’ file, line number 126. If you use ‘dmalloc_t.c:0’, with a 0 line number, then dmalloc will start checking the heap after it sees a call from anywhere in the ‘dmalloc_t.c’ file.
This allows the intensive debugging to be started after a certain routine or file has been reached in the program.
Some examples are:
# turn on transaction and stats logging and set # 'logfile' as the log-file setenv DMALLOC_OPTIONS log-trans,log-stats,log=logfile # enable debug flags 0x1f as well as heap-checking and # set the interval to be 100 setenv DMALLOC_OPTIONS debug=0x1f,check-heap,inter=100 # enable 'logfile' as the log-file, watch for # address '0x1234', and start checking when we see # file.c line 123 setenv DMALLOC_OPTIONS log=logfile,addr=0x1234,start=file.c:123 |
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4.4 Description of the Debugging Tokens
The below tokens and their corresponding descriptions are for the setting of the debug library setting in the environment variable. See section Environment Variable Name and Features. They should be specified in the user’s ‘.dmallocrc’ file. See section Format of the Runtime Configuration File.
Each token, when specified, enables a specific debugging feature. For instance, if you have the log-stats token
enabled, the library will log general statistics to the logfile.
To get this information on the fly, use dmalloc -DV. This will print out the Debug tokens in Very-verbose mode. See section Dmalloc Utility Program.
-
none No debugging functionality
-
log-stats Log general statistics when dmalloc_shutdown or dmalloc_log_stats is called.
-
log-non-free Log non-freed memory pointers when dmalloc_shutdown or dmalloc_log_unfreed is called.
-
log-known Log only known memory pointers that have not been freed. Pointers which do not have file/line or return-address information will not be logged.
-
log-trans Log general memory transactions (quite verbose).
-
log-admin Log administrative information (quite verbose).
-
log-bad-space Log actual bytes in and around bad pointers.
-
log-nonfree-space Log actual bytes in non-freed pointers.
-
log-elapsed-time Log elapsed-time for allocated pointers (see ‘conf.h’).
-
log-current-time Log current-time for allocated pointers (see ‘conf.h’).
-
check-fence Check fence-post memory areas.
-
check-heap Verify heap administrative structure.
-
check-blank Check to see if space that was blanked when a pointer was allocated or when it was freed has been overwritten. If this is enabled then it will enable
free-blankandalloc-blankautomatically.-
check-funcs Check the arguments of some functions (mostly string operations) looking for bad pointers.
-
check-shutdown Check all of the pointers in the heap when the program exits.
-
catch-signals Shutdown the library automatically on SIGHUP, SIGINT, or SIGTERM. This will cause the library to dump its statistics (if requested) when you press control-c on the program (for example).
-
realloc-copy Always copy data to a new pointer when realloc.
-
free-blank Write special “dmalloc-free” byte (hexadecimal
0xdf, octal0337, decimal223) into space when it is freed. You can set this to be something else in the ‘settings.dist’ file. This ensures that your program is not using memory after it has been freed. You can check to see if areas have been improperly overwritten with thecheck-blanktoken. If the free space has been overwritten, thenERROR_FREE_OVERWRITTENis triggered. See section Description of the Internal Error Codes.-
error-abort Abort the program (and dump core) on errors. See
error-dumpbelow. See section Generating a Core File on Errors.-
alloc-blank Write special “dmalloc-alloc” byte (hexadecimal
0xda, octal0332, decimal218) into space when it is allocated. You can set this to be something else in the ‘settings.dist’ file. If you are not usingcallocthis will overwrite the user space with the special bytes ensuring that your program is initializing its dynamic memory appropriately. Also, if you ask for 35 bytes and the library has to give you a block of 64 because of rounding issues, it will overwrite the extra memory with the special byte. You can then check to see if the extra areas have been improperly overwritten by enabling thecheck-blanktoken.-
print-messages Log any errors and messages to the screen via standard-error.
-
catch-null Abort the program immediately if the library fails to get more heap space from the heap allocation routine
mmaporsbrk.-
never-reuse Have the heap never use space that has been used before and freed. See section Tracking Down Non-Freed Memory. WARNING: This should be used with caution since you may run out of heap space.
-
error-dump Dump core on error and then continue. Later core dumps overwrite earlier ones if the program encounters more than one error. See
error-abortabove. See section Generating a Core File on Errors.NOTE: This will only work if your system supports the
forksystem call and the configuration utility was able to fork without going recursive.-
error-free-null By default the library will not generate an error when a program tries to free a NULL pointer. By enabling this token, you can change this behavior so an error is reported. See also the ALLOW_FREE_NULL and ALLOW_FREE_NULL_MESSAGE settings in the ‘settings.h’ file to change the default behavior.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
4.5 Format of the Runtime Configuration File
By using a RC File (or runtime configuration file) you can alias tags to combinations of debug tokens. See section Description of the Debugging Tokens.
NOTE: For beginning users, the dmalloc program has a couple of tags built into it so it is not necessary for you to setup a RC file:
-
runtime Enables basic runtime tests including fence-post checking, null handling, and logging of any errors.
-
low Runtime settings plus minimal checking of heap structures and overwriting of allocated and freed space.
-
medium Low settings plus checking of all heap structures on each memory call, always relocates block on realloc, and aborts on errors. You may want to use -i option to the dmalloc utility. See section Dmalloc Utility Program.
-
high Medium settings plus checking of overwritten freed and allocated memory and checking of arguments to a number of common functions. You may want to use -i option to the dmalloc utility. See section Dmalloc Utility Program.
For expert users, a sample ‘dmallocrc’ file has been provided but you are encouraged to roll your own combinations. The name of default rc-file is ‘$HOME/.dmallocrc’. The ‘$HOME’ environment variable should be set by the system to point to your home-directory.
The file should contain lines in the general form of:
‘tag’ is to be matched with the tag argument passed to the dmalloc program, while ‘token1, token2, …’ are debug capability tokens. See section Dmalloc Utility Program, Description of the Debugging Tokens.
A line can be finished with a ‘\’ meaning it continues onto the next line. Lines beginning with ‘#’ are treated as comments and are ignored along with empty lines.
Here is an example of a ‘.dmallocrc’ file:
#
# Dmalloc runtime configuration file for the debug malloc library
#
# no debugging
none none
# basic debugging
debug1 log-stats, log-non-free, check-fence
# more logging and some heap checking
debug2 log-stats, log-non-free, log-trans, \
check-fence, check-heap, error-abort
# good utilities
debug3 log-stats, log-non-free, log-trans, \
log-admin, check-fence, check-heap, realloc-copy, \
free-blank, error-abort
…
|
For example, with the above file installed, you can type dmalloc debug1 after setting up your shell alias.
See section Dmalloc Utility Program. This enables the logging of statistics, the logging of non-freed memory, and the checking of
fence-post memory areas.
Enter dmalloc none to disable all memory debugging features.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
5. Information on the Source Code
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
5.1 General Compatibility Concerns
- Realloc() backwards compatibility with being able to realloc from the last freed block is not supported. The author is interested to know who is using this (cough, cough) feature and for what reason.
-
Realloc() of a NULL pointer is supported in which case the library will just make a call to malloc(). This can be
disabled with the help of the
ALLOW_REALLOC_NULLmanual compilation option in the ‘settings.h’ file to adjust the library’s default behavior. -
Some systems allow free(0) to not be an error for some reason. Since 0 is not a valid address returned by the malloc
call, it is debatable that this should be allowed. See ‘settings.h’ for the
ALLOW_FREE_NULLmanual compilation option to adjust the library’s default behavior. - Aside from possibly being slower than the system’s memory allocation functions, the library should be fully compatible with the standard memory routines. If this is not the case, please bring this to my attention.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
5.2 Issues Important for Porting the Library
General portability issues center around:
-
mmap, sbrk, or compatible function usages. The library does support a preallocated memory chunk heap. See the
INTERNAL_MEMORY_SPACEdefine in the ‘settings.dist’ file. -
The locating of the caller’s address from the dmalloc functions. This is useful in locating problems from dmalloc
functions called from C files which did not include ‘dmalloc.h’: C library calls for instance.
See ‘return.h’ for the available architecture/compiler combinations. You may want to examine the assembly code from gcc (GNUs superior c-compiler) version 2+ being run on the following code. It should give you a good start on building a hack for your box.
static char * x; a() { x = __builtin_return_address(0); } main() { a(); }
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
6. Some Solutions to Common Problems
This section provides some answers to some common problems and questions. Please send me mail with any additions to this list – either problems you are still having or tips that you would like to pass on.
When diagnosing a problem, if possible, always make sure you are running the most up to date version of Dmalloc available from the home page at URL http://dmalloc.com/. Problems are often fixed and a new release can be published before people encounter them.
- ‘Why does my program run so slow?’
-
This library has never been (and maybe never will be) optimized for space nor speed. Some of its features make it unable to use some of the organizational methods of other more efficient heap libraries.
If you have the
check-heaptoken enabled, your program might run slow or seem to hang. This is because by default, the library will run a full check of the heap with every memory allocation or free. You can have the library check itself less frequently by using the -i option to the dmalloc utility. See section Dmalloc Utility Program. If you are using the high token and you need your program to run faster, try the medium or low tokens which don’t check as many heap features and so run faster although they will not catch as many problems. See section Format of the Runtime Configuration File. - ‘Why was a log-file not produced after I ran my program?’
-
This could be caused by a number of different problems.
- Are you sure you followed all of the items in the “Getting Started” section? Please review them if there is any doubt. See section Getting Started with the Library.
- Use the env or printenv commands to make sure that the ‘DMALLOC_OPTIONS’ variable is set in your exported environment. See section Environment Variable Name and Features.
- Make sure that your program has been compiled correctly with the dmalloc library. The ident program should show chunk.c and other dmalloc files compiled into your program. You can also do strings -a your-program | grep chunk.c and look for something like ‘$Id: chunk.c,v 1.152 1999/08/25 12:37:01 gray Exp $’ with different versions and date information. If this doesn’t show up then chances are dmalloc was not linked into your program.
- If your program changes its working directory, it may write the dmalloc log-file somewhere else in the filesystem. You will need to check both where the program was started and to where it might change directory.
- The logfile is only produced when
dmalloc_shutdown()is called. By default it will be called whenexit()gets called. If you are running your program and press Control-C under Unix the program will stop immediately anddmalloc_shutdown()will not get called. You can either setup a signal handler forSIGINTRand call exit yourself, or you can enable thecatch-signalstoken. See section Description of the Debugging Tokens. - If your program is segfaulting or otherwise crashing when it exits, the
exit()routine may not being called. You will have to resolve these issues so the dmalloc library can gracefully exit and write its log file. - You may want to call
dmalloc_log_stats()anddmalloc_log_unfreed()(ordmalloc_log_changed()) directly to have the library write its log file. Some system modules may not have shutdown if you call this beforeexit()so extra unfreed memory may be reported.
- ‘I don't see any information about my non-freed (leaked) memory?’
-
The library will not (by default) report on “unknown” non-freed memory. Unknown means memory that does not have associated file and line information.
This will be necessary if you are not including ‘dmalloc.h’ in all of your C files or if you are interested in tracking leaks in system functions.
- ‘Dmalloc is returning the error "malloc library has gone recursive"’
-
This most likely indicates that you are using the Dmalloc library within a threaded application and two threads are trying to use the dmalloc library at once. Please see the section of the manual about threads for more information about properly configuring the library. See section Using the Library with a Thread Package.
If you are not using threads, then your program could have caught a signal while within Dmalloc, which then in turn called a memory allocation routine. It is unwise to allocate memory on the heap in most signal handlers. Lastly, some functions called by the library may call memory routines that it does not anticipate. If you think this the case, please report the problem and include a stack trace, operating system version/type, and the version of Dmalloc you are using.
| [ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
Index of Concepts
| Jump to: | %
-
.
0
1
2
3
4
6
7
A B C D E F G H I J K L M N O P Q R S T U V W Z |
|---|
| Jump to: | %
-
.
0
1
2
3
4
6
7
A B C D E F G H I J K L M N O P Q R S T U V W Z |
|---|
| [Top] | [Contents] | [Index] | [ ? ] |
Table of Contents
- 1. Library License and Copying Information
- 2. Description of Features and How to Get Started
- 3. How to Program with the Library
- 3.1 Macros Providing File and Line Information
- 3.2 Getting Caller Address Information
- 3.3 Checking of Function Arguments
- 3.4 Generating a Core File on Errors
- 3.5 Additional Non-standard Routines
- 3.6 Description of the Internal Error Codes
- 3.7 How to Disable the library
- 3.8 Using the Library with C++
- 3.9 Using Dmalloc With a Debugger
- 3.10 Using the Library with a Thread Package
- 3.11 Using the library with Cygwin environment.
- 3.12 Debugging Memory in a Server or Cgi-Bin Process
- 3.13 Explanation of the Logfile Output
- 3.14 Various Other Hints That May Help
- 4. Dmalloc Utility Program
- 5. Information on the Source Code
- 6. Some Solutions to Common Problems
- Index of Concepts
| [Top] | [Contents] | [Index] | [ ? ] |
About This Document
This document was generated by Gray Watson on December 28, 2020 using texi2html 1.82.
The buttons in the navigation panels have the following meaning:
| Button | Name | Go to | From 1.2.3 go to |
|---|---|---|---|
| [ < ] | Back | Previous section in reading order | 1.2.2 |
| [ > ] | Forward | Next section in reading order | 1.2.4 |
| [ << ] | FastBack | Beginning of this chapter or previous chapter | 1 |
| [ Up ] | Up | Up section | 1.2 |
| [ >> ] | FastForward | Next chapter | 2 |
| [Top] | Top | Cover (top) of document | |
| [Contents] | Contents | Table of contents | |
| [Index] | Index | Index | |
| [ ? ] | About | About (help) |
where the Example assumes that the current position is at Subsubsection One-Two-Three of a document of the following structure:
- 1. Section One
- 1.1 Subsection One-One
- ...
- 1.2 Subsection One-Two
- 1.2.1 Subsubsection One-Two-One
- 1.2.2 Subsubsection One-Two-Two
- 1.2.3 Subsubsection One-Two-Three <== Current Position
- 1.2.4 Subsubsection One-Two-Four
- 1.3 Subsection One-Three
- ...
- 1.4 Subsection One-Four
- 1.1 Subsection One-One
This document was generated by Gray Watson on December 28, 2020 using texi2html 1.82.