aef84bb4ef
If OPENSSL_init_crypto() hasn't been called yet when ERR_get_state()
is called, it need to be called early, so the base initialization is
done. On some platforms (those who support DSO functionality and
don't define OPENSSL_USE_NODELETE), that includes a call of
ERR_set_mark(), which calls this function again.
Furthermore, we know that ossl_init_thread_start(), which is called
later in ERR_get_state(), calls OPENSSL_init_crypto(0, NULL), except
that's too late.
Here's what happens without an early call of OPENSSL_init_crypto():
=> ERR_get_state():
=> CRYPTO_THREAD_get_local():
<= NULL;
# no state is found, so it gets allocated.
=> ossl_init_thread_start():
=> OPENSSL_init_crypto():
# Here, base_inited is set to 1
# before ERR_set_mark() call
=> ERR_set_mark():
=> ERR_get_state():
=> CRYPTO_THREAD_get_local():
<= NULL;
# no state is found, so it gets allocated!!!!!
=> ossl_init_thread_start():
=> OPENSSL_init_crypto():
# base_inited is 1,
# so no more init to be done
<= 1
<=
=> CRYPTO_thread_set_local():
<=
<=
<=
<= 1
<=
=> CRYPTO_thread_set_local() # previous value removed!
<=
Result: double allocation, and we have a leak.
By calling the base OPENSSL_init_crypto() early, we get this instead:
=> ERR_get_state():
=> OPENSSL_init_crypto():
# Here, base_inited is set to 1
# before ERR_set_mark() call
=> ERR_set_mark():
=> ERR_get_state():
=> OPENSSL_init_crypto():
# base_inited is 1,
# so no more init to be done
<= 1
=> CRYPTO_THREAD_get_local():
<= NULL;
# no state is found, so it gets allocated
# let's assume we got 0xDEADBEEF
=> ossl_init_thread_start():
=> OPENSSL_init_crypto():
# base_inited is 1,
# so no more init to be done
<= 1
<= 1
=> CRYPTO_thread_set_local():
<=
<=
<=
<= 1
=> CRYPTO_THREAD_get_local():
<= 0xDEADBEEF
<= 0xDEADBEEF
Result: no leak.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4913)
|
||
|---|---|---|
| .. | ||
| build.info | ||
| err.c | ||
| err_all.c | ||
| err_prn.c | ||
| openssl.ec | ||
| openssl.txt | ||
| README | ||
Adding new libraries
--------------------
When adding a new sub-library to OpenSSL, assign it a library number
ERR_LIB_XXX, define a macro XXXerr() (both in err.h), add its
name to ERR_str_libraries[] (in crypto/err/err.c), and add
ERR_load_XXX_strings() to the ERR_load_crypto_strings() function
(in crypto/err/err_all.c). Finally, add an entry:
L XXX xxx.h xxx_err.c
to crypto/err/openssl.ec, and add xxx_err.c to the Makefile.
Running make errors will then generate a file xxx_err.c, and
add all error codes used in the library to xxx.h.
Additionally the library include file must have a certain form.
Typically it will initially look like this:
#ifndef HEADER_XXX_H
#define HEADER_XXX_H
#ifdef __cplusplus
extern "C" {
#endif
/* Include files */
#include <openssl/bio.h>
#include <openssl/x509.h>
/* Macros, structures and function prototypes */
/* BEGIN ERROR CODES */
The BEGIN ERROR CODES sequence is used by the error code
generation script as the point to place new error codes, any text
after this point will be overwritten when make errors is run.
The closing #endif etc will be automatically added by the script.
The generated C error code file xxx_err.c will load the header
files stdio.h, openssl/err.h and openssl/xxx.h so the
header file must load any additional header files containing any
definitions it uses.