and bn_add_words to avoid using fake bignums to window other bignums that
can lead to corruption. This change allows all bignum tests to pass with
BN_DEBUG and BN_DEBUG_RAND debugging and valgrind. NB: This should be
tested on a few different architectures and configuration targets, as the
bignum code this deals with is quite preprocessor (and assembly) sensitive.
Submitted by: Nils Narsch
Reviewed by: Geoff Thorpe, Ulf Moeller
generally a more efficient comparison than comparing two integers, and the
first of these two loops was off-by-one (copying one too many values). This
change also removes a superfluous assignment that would set an unused word
to zero (and potentially allow an overrun in some cases).
Submitted by: Nils Larsch
Reviewed by: Geoff Thorpe
constant BIGNUMs. It turns out that this trips up different but equally
useful compiler warnings to -Wcast-qual, and so wasn't worth the ugliness
it created. (Thanks to Ulf for the forehead-slap.)
and structures as constant without having to cast away const at any point.
There is still plenty of other code that makes gcc's "-Wcast-qual" unhappy,
but crypto/bn/ is now ok. Purists are welcome to suggest alternatives.
constructing BIGNUM structures with pointers offset into other bignums
(among other things). This corrects some of it that is too plainly insane,
and tries to ensure that bignums are normalised when passed to other
functions.
structures being passed in to or out of API functions, and this corrects a
couple of cases found so far.
Also, lop off a couple of bytes of white-space.
I have tried to convert 'len' type variable declarations to unsigned as a
means to address these warnings when appropriate, but when in doubt I have
used casts in the comparisons instead. The better solution (that would get
us all lynched by API users) would be to go through and convert all the
function prototypes and structure definitions to use unsigned variables
except when signed is necessary. The proliferation of (signed) "int" for
strictly non-negative uses is unfortunate.
except internally to the allocator BN_CTX_new(), as such this deprecates
the use of BN_CTX_init() in the API. Moreover, the structure definition of
BN_CTX is taken out of bn_lcl.h and moved into bn_ctx.c itself.
NDEBUG should probably only be "forced" in the top-level configuration, but
until it is I will avoid removing it from bn_ctx.c which might surprise
people with massive slow-downs in their keygens. So I've left it in
bn_ctx.c but tidied up the preprocessor logic a touch and made it more
tolerant of debugging efforts.
be) precompiled out in the API headers. This change is to ensure that if
it is defined when compiling openssl, the deprecated functions aren't
implemented either.
sure the loop does correctly stop and breaking ("division by zero")
modulus operations are not performed. The (pre-generated) prime
table crypto/bn/bn_prime.h was already correct, but it could not be
re-generated on some platforms because of the "division by zero"
situation in the script.
replaced #if logic around bn_sub_part_words in bn_mul.c. I rely upon
OPENSSL_BN_ASM_PART_WORDS being added by ./Configure script. Would it
still work on non-Unix platforms?
- a patch to fix a memory leak in rsa_gen.c
- a note about compiler warnings with unions
- a note about improving structure element names
This applies his patch and implements a solution to the notes.
key-generation and prime-checking functions. Rather than explicitly passing
callback functions and caller-defined context data for the callbacks, a new
structure BN_GENCB is defined that encapsulates this; a pointer to the
structure is passed to all such functions instead.
This wrapper structure allows the encapsulation of "old" and "new" style
callbacks - "new" callbacks return a boolean result on the understanding
that returning FALSE should terminate keygen/primality processing. The
BN_GENCB abstraction will allow future callback modifications without
needing to break binary compatibility nor change the API function
prototypes. The new API functions have been given names ending in "_ex" and
the old functions are implemented as wrappers to the new ones. The
OPENSSL_NO_DEPRECATED symbol has been introduced so that, if defined,
declaration of the older functions will be skipped. NB: Some
openssl-internal code will stick with the older callbacks for now, so
appropriate "#undef" logic will be put in place - this is in case the user
is *building* openssl (rather than *including* its headers) with this
symbol defined.
There is another change in the new _ex functions; the key-generation
functions do not return key structures but operate on structures passed by
the caller, the return value is a boolean. This will allow for a smoother
transition to having key-generation as "virtual function" in the various
***_METHOD tables.
the divisor was a bit more complex than I first saw. The lost bit
can't just be discarded, as there are cases where it is important.
For example, look at dividing 320000 with 80000 vs. 80001 (all
decimals), the difference is crucial. The trick here is to check if
that lost bit was 1, and in that case, do the following:
1. subtract the quotient from the remainder
2. as long as the remainder is negative, add the divisor (the whole
divisor, not the shofted down copy) to it, and decrease the
quotient by one.
There's probably a nice mathematical proof for this already, but I
won't bother with that, unless someone requests it from me.
PR: 338
Here's the description, submitted by Gisle Vanem <giva@bgnett.no>:
1. sock_init() renamed to ssl_sock_init() in ./apps/s_socket.c due
to name-clash with Watt-32.
2. rand() renamed to Rand() in ./crypto/bn/divtest.c due to name-clash
with <stdlib.h>
3. Added calls to dbug_init()/sock_init() in some demo programs.
4. Changed cflags/lflags in configure. Watt-32 install root now taken
from $WATT_ROOT.
the new method names where _GF... suffixes have been removed.
Revert changes to ..._{get/set}_Jprojective_coordinates_...:
The current implementation for ECC over binary fields does not use
projective coordinates, and if it did, it would not use Jacobian
projective coordinates; so it's OK to use the ..._GFp prefix for all
this.
Add author attributions to some files so that it doesn't look
as if Sun wrote all of this :-)
The 'OPENSSL_NO_SUN_DIV' default is still subject to change,
so I didn't bother to finish the CHANGES entry yet.
Submitted by: Douglas Stebila <douglas.stebila@sun.com>, Sheueling Chang <sheueling.chang@sun.com>
(CHANGES entry by Bodo Moeller)
happens reliably, even if the BIGNUM is already sufficiently large.
[Note that the bn_expand()/bn_wexpand() macros call bn_expand2() only
if the BIGNUM actually has to grow, so this change does not add any
new overhead as currently bn_expand2() is never called directly.]
HP-UX in common in ./config). Note that for the moment of this writing
none of 64-bit platforms pass bntest. I'm committing this anyway as it's
too frustrating to patch snapshots over and over while 0.9.6 is known to
work.
details (performance numbers and accompanying discussions:-). Note that
the code is not engaged in ./Configure yet. I'll add it later this week
along with updates for .spec file.
Submitted by:
Reviewed by:
PR:
It does not appear to be faster than the current Montgomery code
except for very small moduli (somewhere between 192 and 224 bits
in a 64-bit Sun environment, and even less than 192 bits
on 32 bit systems).
and make all files the depend on it include it without prefixing it
with openssl/.
This means that all Makefiles will have $(TOP) as one of the include
directories.
sure they are available in opensslconf.h, by giving them names starting
with "OPENSSL_" to avoid conflicts with other packages and by making
sure e_os2.h will cover all platform-specific cases together with
opensslconf.h.
I've checked fairly well that nothing breaks with this (apart from
external software that will adapt if they have used something like
NO_KRB5), but I can't guarantee it completely, so a review of this
change would be a good thing.
Bleichenbacher's DSA attack. With this implementation, the expected
number of iterations never exceeds 2.
New semantics for BN_rand_range():
BN_rand_range(r, min, range) now generates r such that
min <= r < min+range.
(Previously, BN_rand_range(r, min, max) generated r such that
min <= r < max.
It is more convenient to have the range; also the previous
prototype was misleading because max was larger than
the actual maximum.)
for its ASN1 operations as well as the old style function
pointers (i2d, d2i, new, free). Change standard extensions
to support this.
Fix a warning in BN_mul(), bn_mul.c about uninitialised 'j'.
One problem that looked like a problem in bn_recp.c at first turned
out to be a BN_mul bug. An example is given in bn_recp.c; finding
the bug responsible for this is left as an exercise.
course, that means we need to handle the cases where the two arrays to
bn_mul_recursive() and bn_mul_part_recursive() differ in size.
I haven't yet changed the comments that describe bn_mul_recursive()
and bn_mul_part_recursive(). I want this to be tested by more people
before I consider this change final. Please test away!
The bn_cmp_part_words bug was only caught in the BN_mod_mul() test,
not in the BN_mul() test, so apparently the choice of parameters in
some cases is bad.
But even if this is avoided, there are still segmentation violations
(during one of the BN_free()s at the end of test_kron
in some cases, in other cases during BN_kronecker, or
later in BN_sqrt; choosing a different exponentiation
algorithm in bntest.c appears to influence when the SIGSEGV
takes place).
so we have to reduce the random numbers used in test_mont.
Before this change, test_mont failed in [debug-]solaris-sparcv9-gcc
configurations ("Montgomery multiplication test failed!" because
the multiplication result obtained with Montgomery multiplication
differed from the result obtained by BN_mod_mul).
Substituing the old version of bn_gcd.c (BN_mod_inverse) did not avoid
the problem.
The strange thing is that it I did not observe any problems
when using debug-solaris-sparcv8-gcc and solaris-sparcv9-cc,
as well as when compiling OpenSSL 0.9.6 in the solaric-sparcv9-gcc
configuration on the same system.
This caused a segmentation fault in calls to malloc, so I cleaned up
bn_lib.c a little so that it is easier to see what is going on.
The bug turned out to be an off-by-one error in BN_bin2bn.
These new files will not be included literally in OpenSSL, but I intend
to integrate most of their contents. Most file names will change,
and when the integration is done, the superfluous files will be deleted.
Submitted by: Lenka Fibikova <fibikova@exp-math.uni-essen.de>
I'm a little bit nervous about bn_div_words, as I don't know what it's
supposed to return on overflow. For now, I trust the rest of the
system to give it numbers that will not cause any overflow...
BN_mul() correctly constified, avoids two realloc()'s that aren't
really necessary and saves memory to boot. This required a small
change in bn_mul_part_recursive() and the addition of variants of
bn_cmp_words(), bn_add_words() and bn_sub_words() that can take arrays
with differing sizes.
The test results show a performance that very closely matches the
original code from before my constification. This may seem like a
very small win from a performance point of view, but if one remembers
that the variants of bn_cmp_words(), bn_add_words() and bn_sub_words()
are not at all optimized for the moment (and there's no corresponding
assembler code), and that their use may be just as non-optimal, I'm
pretty confident there are possibilities...
This code needs reviewing!
two functions that did expansion on in parameters (BN_mul() and
BN_sqr()). The problem was solved by making bn_dup_expand() which is
a mix of bn_expand2() and BN_dup().
BN_mod_mul_montgomery, which calls bn_sqr_recursive
without much preparation.
bn_sqr_recursive requires the length of its argument to be
a power of 2, which is not always the case here.
There's no reason for not using BN_sqr -- if a simpler
approach to squaring made sense, then why not change
BN_sqr? (Using BN_sqr should also speed up DH where g is chosen
such that it becomes small [e.g., 2] when converted
to Montgomery representation.)
Case closed :-)
Also, "make update" has added some missing functions to libeay.num,
updated the TABLE for the alpha changes, and updated thousands of
dependancies that have changed from recent commits.
because we're only handling words anyway) in BN_mod_exp_mont_word
making it a little faster for very small exponents,
and adjust the performance gain estimate in CHANGES according
to slightly more thorough measurements.
(15% faster than BN_mod_exp_mont for "large" base,
20% faster than BN_mod_exp_mont for small base.)
like Malloc, Realloc and especially Free conflict with already existing names
on some operating systems or other packages. That is reason enough to change
the names of the OpenSSL memory allocation macros to something that has a
better chance of being unique, like prepending them with OPENSSL_.
This change includes all the name changes needed throughout all C files.
Remove one ampersand so the compiler may complain less.
Make rand() static so it will not conflict with the C RTL.
Make bug() static too, for good measure.
Apparently BN_div_recp reports an error for small divisors
(1,2,4,8,40).
I haven't got mismatches so far. If you can, please run the test
program for a few days (nohup divtest >out& or something), and if it
reports a mismatch, post the output.
X is 5120 on 32-bit and 151552 on 64-bit architectures and I varies
from 0 to 4. As result the test was *unreasonably* slow and virtually
impossible to complete on 64-bit architectures (e.g. IRIX bc couldn't
even swallow such long lines).
crypto/bn/bn_lcl.h for further details. It should be noted that for
the moment of this writing the code was tested only on Alpha. If
compiled with DEC C the C implementation exhibits 12% performance
improvement over the crypto/bn/asm/alpha.s (on EV56 box running
AlphaLinux). GNU C is (unfortunately) 8% behind the assembler
implementation. But it's OpenVMS Alpha users who *may* benefit most
as 'apps/openssl speed rsa' exhibits 6 (six) times performance
improvement over the original VMS bignum implementation. Where "*may*"
means "as soon as code is enabled though #define SIXTY_FOUR_BIT and
crypto/bn/asm/vms.mar is skipped."
because otherwise BN_rand will fail unless DEVRANDOM works,
which causes the programs to dump core because they
don't check the return value of BN_rand (and if they
did, we still couldn't test anything).
- add comment to some files that appear not to be used at all.
returns int (1 = ok, 0 = not seeded). New function RAND_add() is the
same as RAND_seed() but takes an estimate of the entropy as an additional
argument.
problem was that one of the replacement routines had not been working since
SSLeay releases. For now the offending routine has been replaced with
non-optimised assembler. Even so, this now gives around 95% performance
improvement for 1024 bit RSA signs.
the remainder left in %edx. Here is the resulting performance improvement
matrix (improvement as a result of this *and* previous tune-up committed
two days ago). The results were obtained by profiling the "div" part of
the crypto/bn/bnspeed.c.
CPU BN_div bn_div_words overall comment
------------------------------------------------------------------------
PII +16% accumulated by +2-3% PII multiplies damn fast! Taking
inlining multiplication out of the loop
didn't make too much difference.
Eliminating of the multiplication
involved in remainder calculation
is the major factor.
Pentium +45% accumulated by +7-9% mull isn't that fast and replacing
inlining multiplications with additions in
the loop has more visible effect:-)
MIPS +75% +12% +20-25% In addition to the taking mults
R10000 out of the loop (giving 12% in the
asm/mips3.s) three mults were
eliminated in BN_div.
Alpha +30% +50% +10-15% Same as above. But remember that
EV4 bn_div_words is a C implementation.
It takes 4 Alpha mults in C to do
the same thing as 1 MIPS mult in
assembler does. So the effect (50%)
is more impressive. But not the
overall one... Well, if Alpha
bn_mul_add would be implemented
in assembler overall improvement
would be closer to MIPS...
in cryptlib.h (which is often included as "../cryptlib.h"), then the
question remains relative to which directory this is to be interpreted.
gcc went one further directory up, as intended; but makedepend thinks
differently, and so probably do some C compilers. So the ../ must go away;
thus e_os.h goes back into include/openssl (but I now use
#include "openssl/e_os.h" instead of <openssl/e_os.h> to make the point) --
and we have another huge bunch of dependency changes. Argh.
There were problems with putting e_os.h just into the top directory,
because the test programs are compiled within test/ in the "standard"
case in in their original directories in the makefile.one case;
and in the latter symlinks may not be available.
used with negative char values, so I've added casts to unsigned char.
Maybe what really should be done is change all those arrays and
pointers to type unsigned char [] or unsigned char *, respectively;
but using plain char with those predicates is just wrong, so something
had to be done.
Submitted by:
Reviewed by:
PR:
