Apache 2.0 Thread Safety Issues

Global and static variables

When writing your module or when trying to determine if a module or 3rd party library is thread safe there are some common things to keep in mind.

First, you need to recognize that in a threaded model each individual thread has its own program counter, stack and registers. Local variables live on the stack, so those are fine. You need to watch out for any static or global variables. This doesn't mean that you are absolutely not allowed to use static or global variables. There are times when you actually want something to affect all threads, but generally you need to avoid using them if you want your code to be thread safe.

In the case where you have a global variable that needs to be global and accessed by all threads, be very careful when you update it. If, for example, it is an incrementing counter, you need to atomically increment it to avoid race conditions with other threads. You do this using a mutex (mutual exclusion). Lock the mutex, read the current value, increment it and write it back and then unlock the mutex. Any other thread that wants to modify the value has to first check the mutex and block until it is cleared.

If you are using APR, have a look at the apr_atomic_* functions and the apr_thread_mutex_* functions.

errno

This is a common global variable that holds the error number of the last error that occurred. If one thread calls a low-level function that sets errno and then another thread checks it, we are bleeding error numbers from one thread into another. To solve this, make sure your module or library defines _REENTRANT or is compiled with -D_REENTRANT. This will make errno a per-thread variable and should hopefully be transparent to the code. It does this by doing something like this:

which means that accessing errno will call __errno_location() which is provided by the libc. Setting _REENTRANT also forces redefinition of some other functions to their *_r equivalents and sometimes changes the common getc/putc macros into safer function calls. Check your libc documentation for specifics. Instead of, or in addition to _REENTRANT the symbols that may affect this are _POSIX_C_SOURCE, _THREAD_SAFE, _SVID_SOURCE, and _BSD_SOURCE.

Common standard troublesome functions

Not only do things have to be thread safe, but they also have to be reentrant. strtok() is an obvious one. You call it the first time with your delimiter which it then remembers and on each subsequent call it returns the next token. Obviously if multiple threads are calling it you will have a problem. Most systems have a reentrant version of of the function called strtok_r() where you pass in an extra argument which contains an allocated char * which the function will use instead of its own static storage for maintaining the tokenizing state. If you are using APR you can use apr_strtok().

crypt() is another function that tends to not be reentrant, so if you run across calls to that function in a library, watch out. On some systems it is reentrant though, so it is not always a problem. If your system has crypt_r() chances are you should be using that, or if possible simply avoid the whole mess by using md5 instead.

Common 3rd Party Libraries

The following is a list of common libraries that are used by 3rd party Apache modules. You can check to see if your module is using a potentially unsafe library by using tools such as ldd(1) and nm(1). For PHP, for example, try this:

In addition to these libraries you will need to have a look at any libraries linked statically into the module. You can use nm(1) to look for individual symbols in the module.

Library List

Please drop a note to [email protected] if you have additions or corrections to this list.

Library	Version	Thread Safe?	Notes
ASpell/PSpell		?
Berkeley DB	3.x, 4.x	Yes	Be careful about sharing a connection across threads.
bzip2		Yes	Both low-level and high-level APIs are thread-safe. However, high-level API requires thread-safe access to errno.
cdb		?
C-Client		Perhaps	c-client uses strtok() and gethostbyname() which are not thread-safe on most C library implementations. c-client's static data is meant to be shared across threads. If strtok() and gethostbyname() are thread-safe on your OS, c-client may be thread-safe.
cpdflib		?
libcrypt		?
Expat		Yes	Need a separate parser instance per thread
FreeTDS		?
FreeType		?
GD 1.8.x		?
GD 2.0.x		?
gdbm		No	Errors returned via a static gdbm_error variable
ImageMagick	5.2.2	Yes	ImageMagick docs claim it is thread safe since version 5.2.2 (see Change log).
Imlib2		?
libjpeg	v6b	?
libmysqlclient		Yes	Use mysqlclient_r library variant to ensure thread-safety. For more information, please read http://www.mysql.com/doc/en/Threaded_clients.html.
Ming	0.2a	?
Net-SNMP	5.0.x	?
OpenLDAP	2.1.x	Yes	Use ldap_r library variant to ensure thread-safety.
OpenSSL	0.9.6g	Yes	Requires proper usage of CRYPTO_num_locks, CRYPTO_set_locking_callback, CRYPTO_set_id_callback
liboci8 (Oracle 8+)	8.x,9.x	?
pdflib	5.0.x	Yes	PDFLib docs claim it is thread safe; changes.txt indicates it has been partially thread-safe since V1.91: http://www.pdflib.com/products/pdflib/index.html.
libpng	1.0.x	?
libpng	1.2.x	?
libpq (PostgreSQL)	7.x	Yes	Don't share connections across threads and watch out for crypt() calls
Sablotron	0.95	?
zlib	1.1.4	Yes	Relies upon thread-safe zalloc and zfree functions Default is to use libc's calloc/free which are thread-safe.