We have little-endian BIGNUM functions now.
Change-Id: Iffc46a14e75c6bba2e170b824b1a08c69d2e9d18
Reviewed-on: https://boringssl-review.googlesource.com/27594
Reviewed-by: Adam Langley <alangley@gmail.com>
This is adapted from upstream's
eb7916960bf50f436593abe3d5f2e0592d291017.
This gives a 22% win for ECDSA signing. (Upstream cites 30-40%, but they
are unnecessarily using BN_mod_exp_mont_consttime in their generic path.
The exponent is public. I expect part of their 30-40% is just offsetting
this.)
Did 506000 ECDSA P-256 signing operations in 25044595us (20204.0 ops/sec)
Did 170506 ECDSA P-256 verify operations in 25033567us (6811.1 ops/sec)
Did 618000 ECDSA P-256 signing operations in 25031294us (24689.1 ops/sec)
Did 182240 ECDSA P-256 verify operations in 25006918us (7287.6 ops/sec)
Most of the performance win appears to be from the assembly operations
and not the addition chain. I have a CL to graft the addition chain onto
the C implementation, but it did not show measurable improvement in
ECDSA verify. ECDSA sign gets 2-4% faster, but we're more concerned
about ECDSA verify in the OPENSSL_SMALL builds.
Change-Id: Ide166f98b146c025f7f80ed7906336c16818540a
Reviewed-on: https://boringssl-review.googlesource.com/27593
Reviewed-by: Adam Langley <alangley@gmail.com>
This introduces a hook for the OpenSSL assembly.
Change-Id: I35e0588f0ed5bed375b12f738d16c9f46ceedeea
Reviewed-on: https://boringssl-review.googlesource.com/27592
Reviewed-by: Adam Langley <alangley@gmail.com>
Largely random data, but make it easy to add things in the future.
Change-Id: I30bee790bd9671b4d0327c2244fe5cd1a8954f90
Reviewed-on: https://boringssl-review.googlesource.com/27591
Reviewed-by: Adam Langley <alangley@gmail.com>
This imports the assembly portion of
eb7916960bf50f436593abe3d5f2e0592d291017 from upstream. Note the
OPENSSL_ia32cap_P bits were tweaked to be delocate-compatible. Those
should be reviewed against the original file.
Change-Id: I19eef722225bb7928275e3d93890f80aa2f8734d
Reviewed-on: https://boringssl-review.googlesource.com/27589
Reviewed-by: Adam Langley <alangley@gmail.com>
We were still using the allocating scalar inversion for ECDSA verify
because previously it seemed to be faster. It appears to have flipped
now, though probably was always just a wash.
While I'm here, save a multiplication by swapping the inversion and
Montgomery reduction.
Did 200000 ECDSA P-256 signing operations in 10025749us (19948.6 ops/sec)
Did 66234 ECDSA P-256 verify operations in 10061123us (6583.2 ops/sec)
Did 202000 ECDSA P-256 signing operations in 10020846us (20158.0 ops/sec)
Did 68052 ECDSA P-256 verify operations in 10020592us (6791.2 ops/sec)
The actual motivation is to get rid of the unchecked EC_SCALAR function
and align sign/verify in preparation for the assembly scalar ops.
Change-Id: I1bd3a5719a67966dc8edaa43535a3864b69f76d0
Reviewed-on: https://boringssl-review.googlesource.com/27588
Reviewed-by: Adam Langley <alangley@gmail.com>
No sense in adding impossible error cases we need to handle.
Additionally, tighten them a bit and require strong bounds. (I wasn't
sure what we'd need at first and made them unnecessarily general.)
Change-Id: I21a0afde90a55be2e9a0b8d7288f595252844f5f
Reviewed-on: https://boringssl-review.googlesource.com/27586
Reviewed-by: Adam Langley <alangley@gmail.com>
This is so the *_small functions can assume somewhat more uniform
widths, to simplify their error-handling.
Change-Id: I0420cb237084b253e918c64b0c170a5dfd99ab40
Reviewed-on: https://boringssl-review.googlesource.com/27584
Reviewed-by: Adam Langley <alangley@gmail.com>
The FIPS 186-4 algorithm we use includes a limit which hits a 2^-20
failure probability, assuming my math is right. We've observed roughly
2^-23. This is a little large at scale. (See b/77854769.)
To avoid modifying the FIPS algorithm, retry the whole thing four times
to bring the failure rate down to 2^-80. Along the way, now that I have
the derivation on hand, adjust
https://boringssl-review.googlesource.com/22584 to target the same
failure probability.
Along the way, fix an issue with RSA_generate_key where, if callers
don't check for failure, there may be half a key in there.
Change-Id: I0e1da98413ebd4ffa65fb74c67a58a0e0cd570ff
Reviewed-on: https://boringssl-review.googlesource.com/27288
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
https://boringssl-review.googlesource.com/10520 and then later
https://boringssl-review.googlesource.com/25285 made BN_MONT_CTX_set
constant-time, which is necessary for RSA's mont_p and mont_q. However,
due to a typo in the benchmark, they did not correctly measure.
Split BN_MONT_CTX creation into a constant-time and variable-time one.
The constant-time one uses our current algorithm and the latter restores
the original BN_mod codepath.
Should we wish to avoid BN_mod, I have an alternate version lying
around:
First, BN_set_bit + bn_mod_lshift1_consttime as now to count up to 2*R.
Next, observe that 2*R = BN_to_montgomery(2) and R*R =
BN_to_montgomery(R) = BN_to_montgomery(2^r_bits) Also observe that
BN_mod_mul_montgomery only needs n0, not RR. Split the core of
BN_mod_exp_mont into its own function so the caller handles conversion.
Raise 2*R to the r_bits power to get 2^r_bits*R = R*R.
The advantage of that algorithm is that it is still constant-time, so we
only need one BN_MONT_CTX_new. Additionally, it avoids BN_mod which is
otherwise (almost, but the remaining links should be easy to cut) out of
the critical path for correctness. One less operation to worry about.
The disadvantage is that it is gives a 25% (RSA-2048) or 32% (RSA-4096)
slower RSA verification speed. I went with the BN_mod one for the time
being.
Before:
Did 9204 RSA 2048 signing operations in 10052053us (915.6 ops/sec)
Did 326000 RSA 2048 verify (same key) operations in 10028823us (32506.3 ops/sec)
Did 50830 RSA 2048 verify (fresh key) operations in 10033794us (5065.9 ops/sec)
Did 1269 RSA 4096 signing operations in 10019204us (126.7 ops/sec)
Did 88435 RSA 4096 verify (same key) operations in 10031129us (8816.1 ops/sec)
Did 14552 RSA 4096 verify (fresh key) operations in 10053411us (1447.5 ops/sec)
After:
Did 9150 RSA 2048 signing operations in 10022831us (912.9 ops/sec)
Did 322000 RSA 2048 verify (same key) operations in 10028604us (32108.2 ops/sec)
Did 289000 RSA 2048 verify (fresh key) operations in 10017205us (28850.4 ops/sec)
Did 1270 RSA 4096 signing operations in 10072950us (126.1 ops/sec)
Did 87480 RSA 4096 verify (same key) operations in 10036328us (8716.3 ops/sec)
Did 80730 RSA 4096 verify (fresh key) operations in 10073614us (8014.0 ops/sec)
Change-Id: Ie8916d1634ccf8513ceda458fa302f09f3e93c07
Reviewed-on: https://boringssl-review.googlesource.com/27287
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
The first non-zero window (which we can condition on for public
exponents) always multiplies by one. This means we can cut out one
Montgomery multiplication. It also means we never actually need to
initialize r to one, saving another Montgomery multiplication for P-521.
This, in turn, means we don't need the bn_one_to_montgomery optimization
for the public-exponent exponentations, so we can delete
bn_one_to_montgomery_small. (The function does currently promise to
handle p = 0, but this is not actually reachable, so it can just do a
reduction on RR.)
For RSA, where we're not doing many multiplications to begin with,
saving one is noticeable.
Before:
Did 92000 RSA 2048 verify (same key) operations in 3002557us (30640.6 ops/sec)
Did 25165 RSA 4096 verify (same key) operations in 3045046us (8264.2 ops/sec)
After:
Did 100000 RSA 2048 verify (same key) operations in 3002483us (33305.8 ops/sec)
Did 26603 RSA 4096 verify (same key) operations in 3010942us (8835.4 ops/sec)
(Not looking at the fresh key number yet as that still needs to be
fixed.)
Change-Id: I81a025a68d9b0f8eb0f9c6c04ec4eedf0995a345
Reviewed-on: https://boringssl-review.googlesource.com/27286
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
It's defined to return one in Montgomery form, not a normal one.
(Not that this matters. This function is only used to Fermat's Little
Theorem. Probably it should have been less general, though we'd need to
make new test vectors first.)
Change-Id: Ia8d7588e6a413b25f01280af9aacef0192283771
Reviewed-on: https://boringssl-review.googlesource.com/27285
Reviewed-by: Adam Langley <agl@google.com>
BN_mod_exp_mont is intended to protect the base, but not the exponent.
Accordingly, it shouldn't treat a base of zero as special.
Change-Id: Ib053e8ce65ab1741973a9f9bfeff8c353567439c
Reviewed-on: https://boringssl-review.googlesource.com/27284
Reviewed-by: Adam Langley <agl@google.com>
Our technique to perform the reduction only works for balanced key
sizes. For unbalanced keys, we fall back to variable-time logic.
Instead, fall back earlier to the non-CRT codepath, which is still
secure, just slower. This also aligns with the advice here:
https://github.com/HACS-workshop/spectre-mitigations/blob/master/crypto_guidelines.md#1-do-not-conditionally-choose-between-constant-and-non-constant-time
Update-Note: This is a performance hit (some keys will run 3x slower),
but only for keys with different-sized primes. I believe the Windows
crypto APIs will not accept such keys at all. There are two scenarios to
be concerned with for RSA performance:
1. Performance of reasonably-generated keys. Keys that BoringSSL or
anyone else reasonable generates will all be balanced, so this change
does not affect them.
2. Worst-case performance for DoS purposes. This CL does not change the
worst-case performance for RSA at a given bit size. In fact, it improves
it slightly. A sufficiently unbalanced RSA key is as slow as not doing
CRT at all.
In both cases, this change does not affect performance. The affected
keys are pathologically-generated ones that were not quite pathological
enough.
Bug: 235
Change-Id: Ie298dabb549ab9108fa9374aa86ebffe8b6c6c88
Reviewed-on: https://boringssl-review.googlesource.com/27504
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This is helpful at smaller sizes because the benefits of an unlikely hit
by trival-division are smaller.
The full set of kPrimes eliminates about 94.3% of random numbers. The
first quarter eliminates about 93.2% of them. But the little extra power
of the full set seems to be borderline for RSA 3072 and clearly positive
for RSA 4096.
Did 316 RSA 2048 key-gen operations in 30035598us (10.5 ops/sec)
min: 19423us, median: 80448us, max: 394265us
Change-Id: Iee53f721329674ae7a08fabd85b4f645c24e119d
Reviewed-on: https://boringssl-review.googlesource.com/26944
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: David Benjamin <davidben@google.com>
The generic code special-cases affine points, but this leaks
information. (Of course, the generic code also doesn't have a
constant-time multiply and other problems, but one thing at a time.)
The optimization in point doubling is not useful. Point multiplication
more-or-less never doubles an affine point. The optimization in point
addition *is* useful because the wNAF code converts the tables to
affine. Accordingly, align with the P-256 code which adds a 'mixed'
parameter.
(I haven't aligned the formally-verified point formulas themselves yet;
initial testing suggests that the large number of temporaries take a
perf hit with BIGNUM. I'll check the results in EC_FELEM, which will be
stack-allocated, to see if we still need to help the compiler out.)
Strangly, it actually got a bit faster with this change. I'm guessing
because now it doesn't need to bother with unnecessary comparisons and
maybe was kinder to the branch predictor?
Before:
Did 2201 ECDH P-384 operations in 3068341us (717.3 ops/sec)
Did 4092 ECDSA P-384 signing operations in 3076981us (1329.9 ops/sec)
Did 3503 ECDSA P-384 verify operations in 3024753us (1158.1 ops/sec)
Did 992 ECDH P-521 operations in 3017884us (328.7 ops/sec)
Did 1798 ECDSA P-521 signing operations in 3059000us (587.8 ops/sec)
Did 1581 ECDSA P-521 verify operations in 3033142us (521.2 ops/sec)
After:
Did 2310 ECDH P-384 operations in 3092648us (746.9 ops/sec)
Did 4080 ECDSA P-384 signing operations in 3044588us (1340.1 ops/sec)
Did 3520 ECDSA P-384 verify operations in 3056070us (1151.8 ops/sec)
Did 992 ECDH P-521 operations in 3012779us (329.3 ops/sec)
Did 1792 ECDSA P-521 signing operations in 3019459us (593.5 ops/sec)
Did 1600 ECDSA P-521 verify operations in 3047749us (525.0 ops/sec)
Bug: 239
Change-Id: If5d13825fc98e4c58bdd1580cf0245bf7ce93a82
Reviewed-on: https://boringssl-review.googlesource.com/27004
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This used to work, but I broke it on accident in the recent rewrite.
Change-Id: I06ab5e06eb0c0a6b67ecc97919654e386f3c2198
Reviewed-on: https://boringssl-review.googlesource.com/26984
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Martin Kreichgauer <martinkr@google.com>
Reviewed-by: Martin Kreichgauer <martinkr@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
This is in preparation for representing field elements with
stack-allocated types in the generic code. While there is likely little
benefit in threading all the turned field arithmetic through all the
generic code, and the P-224 logic, in particular, does not have a tight
enough abstraction for this, the current implementations depend on
BN_div, which is not compatible with stack-allocating things and avoiding
malloc.
This also speeds things up slightly, now that benchmarks cover point
validation.
Before:
Did 82786 ECDH P-224 operations in 10024326us (8258.5 ops/sec)
After:
Did 89991 ECDH P-224 operations in 10012429us (8987.9 ops/sec)
Change-Id: I468483b49f5dc69187aebd62834365ce5caab795
Reviewed-on: https://boringssl-review.googlesource.com/26971
Reviewed-by: Adam Langley <agl@google.com>
Alas, it is reachable by way of the legacy custom curves API. Add a
basic test to ensure those codepaths work.
Change-Id: If631110045a664001133a0d07fdac4c67971a15f
Reviewed-on: https://boringssl-review.googlesource.com/26970
Reviewed-by: Adam Langley <agl@google.com>
ECDSA converts digests to scalars by taking the leftmost n bits, where n
is the number of bits in the group order. This does not necessarily
produce a fully-reduced scalar.
Montgomery multiplication actually tolerates this slightly looser bound,
so we did not bother with the conditional subtraction. However, this
subtraction is free compared to the multiplication, inversion, and base
point multiplication. Simplify things by keeping it fully-reduced.
Change-Id: If49dffefccc21510f40418dc52ea4da7e3ff198f
Reviewed-on: https://boringssl-review.googlesource.com/26968
Reviewed-by: Adam Langley <agl@google.com>
ECDSA's logic for converting digests to scalars sometimes produces
slightly unreduced values. Test these cases.
Change-Id: I67a5078db684ee82c286f41e71b13b57c3ee707b
Reviewed-on: https://boringssl-review.googlesource.com/26967
Reviewed-by: Adam Langley <agl@google.com>
May as well use it. Also avoid an overflow with digest_len if someone
asks to sign a truly enormous digest.
Change-Id: Ia0a53007a496f9c7cadd44b1020ec2774b310936
Reviewed-on: https://boringssl-review.googlesource.com/26966
Reviewed-by: Adam Langley <agl@google.com>
For non-custom curves, this only comes up with P-521 and, even then,
only with excessively large hashes. Still, we should have test coverage
for this.
Change-Id: Id17a6f47d59d6dd4a43a93857fd3df490f9fa965
Reviewed-on: https://boringssl-review.googlesource.com/26965
Reviewed-by: Adam Langley <agl@google.com>
We do this in four different places, with the same long comment, and I'm
about to add yet another one.
Change-Id: If28e3f87ea71020d9b07b92e8947f3848473d99d
Reviewed-on: https://boringssl-review.googlesource.com/26964
Reviewed-by: Adam Langley <agl@google.com>
RSA keygen uses this to pick primes. May as well avoid bouncing on
malloc. (The BIGNUM internally allocates, of course, but that allocation
will be absorbed by BN_CTX in RSA keygen.)
Change-Id: Ie2243a6e48b9c55f777153cbf67ba5c06688c2f1
Reviewed-on: https://boringssl-review.googlesource.com/26887
Reviewed-by: Adam Langley <agl@google.com>
With this, in 0.02% of 1024-bit primes (which is what's used with an RSA
2048 generation), we'll leak that we struggled to generate values less
than the prime. I.e. that there's a greater likelihood of zero bits
after the leading 1 bit in the prime.
But this recovers all the speed loss from making key generation
constant-time, and then some.
Did 273 RSA 2048 key-gen operations in 30023223us (9.1 ops/sec)
min: 23867us, median: 93688us, max: 421466us
Did 66 RSA 3072 key-gen operations in 30041763us (2.2 ops/sec)
min: 117044us, median: 402095us, max: 1096538us
Did 31 RSA 4096 key-gen operations in 31673405us (1.0 ops/sec)
min: 245109us, median: 769480us, max: 2659386us
Change-Id: Id82dedde35f5fbb36b278189c0685a13c7824590
Reviewed-on: https://boringssl-review.googlesource.com/26924
Reviewed-by: Adam Langley <alangley@gmail.com>
Windows CryptoAPI and Go bound public exponents at 2^32-1, so don't
generate keys which would violate that.
https://github.com/golang/go/issues/3161https://msdn.microsoft.com/en-us/library/aa387685(VS.85).aspx
BoringSSL itself also enforces a 33-bit limit.
I don't currently have plans to take much advantage of it, but the
modular inverse step and one of the GCDs in RSA key generation are
helped by small public exponents[0]. In case someone feels inspired
later, get this limit enforced now. Use 32-bits as that's a more
convenient limit, and there's no requirement to produce e=2^32+1 keys.
(Is there still a requirement to accept them?)
[0] This isn't too bad, but it's only worth it if it produces simpler or
smaller code. RSA keygen is not performance-critical.
1. Make bn_mod_u16_consttime work for uint32_t. It only barely doesn't
work. Maybe only accept 3 and 65537 and pre-compute, maybe call into
bn_div_rem_words and friends, maybe just tighten the bound a hair
longer.
2. Implement bn_div_u32_consttime by incorporating 32-bit chunks much
like bn_mod_u32_consttime.
3. Perform one normal Euclidean algorithm iteration rather than using the
binary version. u, v, B, and D are now single words, while A and C
are full-width.
4. Continue with binary Euclidean algorithm (u and v are still secret),
taking advantage of most values being small.
Update-Note: RSA_generate_key_ex will no longer generate keys with
public exponents larger than 2^32-1. Everyone uses 65537, save some
folks who use 3, so this shouldn't matter.
Change-Id: I0d28a29a30d9ff73bff282e34dd98e2b64c35c79
Reviewed-on: https://boringssl-review.googlesource.com/26365
Reviewed-by: Adam Langley <alangley@gmail.com>
We don't check it is fully reduced because different implementations use
Carmichael vs Euler totients, but if d exceeds n, something is wrong.
Note the fixed-width BIGNUM changes already fail operations with
oversized d.
Update-Note: Some blatantly invalid RSA private keys will be rejected at
RSA_check_key time. Note that most of those keys already are not
usable with BoringSSL anyway. This CL moves the failure from
sign/decrypt to RSA_check_key.
Change-Id: I468dbba74a148aa58c5994cc27f549e7ae1486a2
Reviewed-on: https://boringssl-review.googlesource.com/26374
Reviewed-by: Adam Langley <alangley@gmail.com>
Rather than recompute values the same as in key generation, where
possible, we check differently. In particular, most RSA values are
modular inverses of some value. Check each of them by multiplying and
using our naive constant-time division function.
Median of 29 RSA keygens: 0m0.218s -> 0m0.205s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: Iaca19f12c045457013def844a17bf502ed09136e
Reviewed-on: https://boringssl-review.googlesource.com/26373
Reviewed-by: Adam Langley <alangley@gmail.com>
This leaves RSA_check_key, which will be fixed in subsequent commits.
Median of 29 RSA keygens: 0m0.220s -> 0m0.209s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I325f23fcc59302e68570908e5427b65471b799f6
Reviewed-on: https://boringssl-review.googlesource.com/26371
Reviewed-by: Adam Langley <alangley@gmail.com>
This uses the full binary GCD algorithm, where all four of A, B, C, and
D must be retained. (BN_mod_inverse_odd implements the odd number
version which only needs A and C.) It is patterned after the version
in the Handbook of Applied Cryptography, but tweaked so the coefficients
are non-negative and bounded.
Median of 29 RSA keygens: 0m0.225s -> 0m0.220s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I6dc13524ea7c8ac1072592857880ddf141d87526
Reviewed-on: https://boringssl-review.googlesource.com/26370
Reviewed-by: Adam Langley <alangley@gmail.com>
RSA key generation requires computing a GCD (p-1 and q-1 are relatively
prime with e) and an LCM (the Carmichael totient). I haven't made BN_gcd
itself constant-time here to save having to implement
bn_lshift_secret_shift, since the two necessary operations can be served
by bn_rshift_secret_shift, already added for Rabin-Miller. However, the
guts of BN_gcd are replaced. Otherwise, the new functions are only
connected to tests for now, they'll be used in subsequent CLs.
To support LCM, there is also now a constant-time division function.
This does not replace BN_div because bn_div_consttime is some 40x slower
than BN_div. That penalty is fine for RSA keygen because that operation
is not bottlenecked on division, so we prefer simplicity over
performance.
Median of 29 RSA keygens: 0m0.212s -> 0m0.225s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: Idbfbfa6e7f5a3b8782ce227fa130417b3702cf97
Reviewed-on: https://boringssl-review.googlesource.com/26369
Reviewed-by: Adam Langley <alangley@gmail.com>
Expose the constant-time abs_sub functions from the fixed Karatsuba code
in BIGNUM form for RSA to call into. RSA key generation involves
checking if |p - q| is above some lower bound.
BN_sub internally branches on which of p or q is bigger. For any given
iteration, this is not secret---one of p or q is necessarily the larger,
and whether we happened to pick the larger or smaller first is
irrelevant. Accordingly, there is no need to perform the p/q swap at the
end in constant-time.
However, this stage of the algorithm picks p first, sticks with it, and
then computes |p - q| for various q candidates. The distribution of
comparisons leaks information about p. The leak is unlikely to be
problematic, but plug it anyway.
Median of 29 RSA keygens: 0m0.210s -> 0m0.212s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I024b4e51b364f5ca2bcb419a0393e7be13249aec
Reviewed-on: https://boringssl-review.googlesource.com/26368
Reviewed-by: Adam Langley <alangley@gmail.com>
It costs us a malloc, but it's one less function to test and implement
in constant time, now that BN_cmp and BIGNUM are okay.
Median of 29 RSA keygens: 0m0.207s -> 0m0.210s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: Ic56f92f0dcf04da1f542290a7e8cdab8036699ed
Reviewed-on: https://boringssl-review.googlesource.com/26367
Reviewed-by: Adam Langley <alangley@gmail.com>
RSA key generation currently does the GCD check before the primality
test, in hopes of discarding things invalid by other means before
running the expensive primality check.
However, GCD is about to get a bit more expensive to clear the timing
leak, and the trial division part of primality testing is quite fast.
Thus, split that portion out via a new bn_is_obviously_composite and
call it before GCD.
Median of 29 RSA keygens: 0m0.252s -> 0m0.207s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I3999771fb73cca16797cab9332d14c4ebeb02046
Reviewed-on: https://boringssl-review.googlesource.com/26366
Reviewed-by: Adam Langley <alangley@gmail.com>
I'm not sure why I separated "fixed" and "quick_ctx" names. That's
annoying and doesn't generalize well to, say, adding a bn_div_consttime
function for RSA keygen.
Change-Id: I751d52b30e079de2f0d37a952de380fbf2c1e6b7
Reviewed-on: https://boringssl-review.googlesource.com/26364
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
Rabin-Miller requires selecting a random number from 2 to |w|-1.
This is done by picking an N-bit number and discarding out-of-range
values. This leaks information about |w|, so apply blinding. Rather than
discard bad values, adjust them to be in range.
Though not uniformly selected, these adjusted values
are still usable as Rabin-Miller checks.
Rabin-Miller is already probabilistic, so we could reach the desired
confidence levels by just suitably increasing the iteration count.
However, to align with FIPS 186-4, we use a more pessimal analysis: we
do not count the non-uniform values towards the iteration count. As a
result, this function is more complex and has more timing risk than
necessary.
We count both total iterations and uniform ones and iterate until we've
reached at least |BN_PRIME_CHECKS_BLINDED| and |iterations|,
respectively. If the latter is large enough, it will be the limiting
factor with high probability and we won't leak information.
Note this blinding does not impact most calls when picking primes
because composites are rejected early. Only the two secret primes see
extra work. So while this does make the BNTest.PrimeChecking test take
about 2x longer to run on debug mode, RSA key generation time is fine.
Another, perhaps simpler, option here would have to run
bn_rand_range_words to the full 100 count, select an arbitrary
successful try, and declare failure of the entire keygen process (as we
do already) if all tries failed. I went with the option in this CL
because I happened to come up with it first, and because the failure
probability decreases much faster. Additionally, the option in this CL
does not affect composite numbers, while the alternate would. This gives
a smaller multiplier on our entropy draw. We also continue to use the
"wasted" work for stronger assurance on primality. FIPS' numbers are
remarkably low, considering the increase has negligible cost.
Thanks to Nathan Benjamin for helping me explore the failure rate as the
target count and blinding count change.
Now we're down to the rest of RSA keygen, which will require all the
operations we've traditionally just avoided in constant-time code!
Median of 29 RSA keygens: 0m0.169s -> 0m0.298s
(Accuracy beyond 0.1s is questionable. The runs at subsequent test- and
rename-only CLs were 0m0.217s, 0m0.245s, 0m0.244s, 0m0.247s.)
Bug: 238
Change-Id: Id6406c3020f2585b86946eb17df64ac42f30ebab
Reviewed-on: https://boringssl-review.googlesource.com/25890
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Reviewed-by: Adam Langley <agl@google.com>
(This is actually slightly silly as |a|'s probability distribution falls
off exponentially, but it's easy enough to do right.)
Instead, we run the loop to the end. This is still performant because we
can, as before, return early on composite numbers. Only two calls
actually run to the end. Moreover, running to the end has comparable
cost to BN_mod_exp_mont_consttime.
Median time goes from 0.140s to 0.231s. That cost some, but we're still
faster than the original implementation.
We're down to one more leak, which is that the BN_rand_range_ex call
does not hide |w1|. That one may only be solved probabilistically...
Median of 29 RSA keygens: 0m0.123s -> 0m0.145s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I4847cb0053118c572d2dd5f855388b5199fa6ce2
Reviewed-on: https://boringssl-review.googlesource.com/25888
Reviewed-by: Adam Langley <agl@google.com>
Compilers use a variant of Barrett reduction to divide by constants,
which conveniently also avoids problematic operations on the secret
numerator. Implement the variant as described here:
http://ridiculousfish.com/blog/posts/labor-of-division-episode-i.html
Repurpose this to implement a constant-time BN_mod_word replacement.
It's even much faster! I've gone ahead and replaced the other
BN_mod_word calls on the primes table.
That should give plenty of budget for the other changes. (I am assuming
that a regression is okay, as RSA keygen is not performance-sensitive,
but that I should avoid anything too dramatic.)
Proof of correctness: https://github.com/davidben/fiat-crypto/blob/barrett/src/Arithmetic/BarrettReduction/RidiculousFish.v
Median of 29 RSA keygens: 0m0.621s -> 0m0.123s
(Accuracy beyond 0.1s is questionable, though this particular
improvement is quite solid.)
Bug: 238
Change-Id: I67fa36ffe522365b13feb503c687b20d91e72932
Reviewed-on: https://boringssl-review.googlesource.com/25887
Reviewed-by: Adam Langley <agl@google.com>
The extra details in Enhanced Rabin-Miller are only used in
RSA_check_key_fips, on the public RSA modulus, which the static linker
will drop in most of our consumers anyway. Implement normal Rabin-Miller
for RSA keygen and use Montgomery reduction so it runs in constant-time.
Note that we only need to avoid leaking information about the input if
it's a large prime. If the number ends up composite, or we find it in
our table of small primes, we can return immediately.
The leaks not addressed by this CL are:
- The difficulty of selecting |b| leaks information about |w|.
- The distribution of whether step 4.4 runs leaks information about w.
- We leak |a| (the largest power of two which divides w) everywhere.
- BN_mod_word in the trial division is not constant-time.
These will be resolved in follow-up changes.
Median of 29 RSA keygens: 0m0.521 -> 0m0.621s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I0cf0ff22079732a0a3ababfe352bb4327e95b879
Reviewed-on: https://boringssl-review.googlesource.com/25886
Reviewed-by: Adam Langley <agl@google.com>
Probably worth having actual test vectors for these, rather than
checking our code against itself. Additionally, small negative numbers
have, in the past been valuable test vectors (see long comment in
point_add from OpenSSL's ecp_nistp521.c).
Change-Id: Ia5aa8a80eb5b6d0089c3601c5fec2364e699794d
Reviewed-on: https://boringssl-review.googlesource.com/26848
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
p224_felem_neg does not produce an output within the tight bounds
suitable for p224_felem_contract. This was found by inspection of the
code.
This only affects the final y-coordinate output of arbitrary-point
multiplication, so it is a no-op for ECDH and ECDSA.
Change-Id: I1d929458d1f21d02cd8e745d2f0f7040a6bb0627
Reviewed-on: https://boringssl-review.googlesource.com/26847
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Primality testing checks for small words in random places.
Median of 29 RSA keygens: 0m0.811s -> 0m0.521s
(Accuracy beyond 0.1s is questionable, and this "speed up" is certainly
noise.)
Bug: 238
Change-Id: Ie5efab7291302a42ac6e283d25da0c094d8577e7
Reviewed-on: https://boringssl-review.googlesource.com/25885
Reviewed-by: Adam Langley <agl@google.com>
There are a number of random subtractions in RSA key generation. Add a
fixed-width version.
Median of 29 RSA keygens: 0m0.859s -> 0m0.811s
(Accuracy beyond 0.1s is questionable.)
Bug: 238
Change-Id: I9fa0771b95a438fd7d2635fd77a332146ccc96d9
Reviewed-on: https://boringssl-review.googlesource.com/25884
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>
No semantic change: the table is the same as before, but now with less
magic.
Change-Id: I351c2446e9765f25b7dfb901c9e98f12099a325c
Reviewed-on: https://boringssl-review.googlesource.com/26744
Reviewed-by: Adam Langley <agl@google.com>
Reviewed-by: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Rather than writing the answer into the output, it wrote it into some
awkwardly-named temporaries. Thanks to Daniel Hirche for reporting this
issue!
Bug: chromium:825273
Change-Id: I5def4be045cd1925453c9873218e5449bf25e3f5
Reviewed-on: https://boringssl-review.googlesource.com/26785
Reviewed-by: Adam Langley <agl@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
These are composite numbers whose composite witnesses aren't in the
first however many prime numbers, so deterministically checking small
numbers may not work.
We don't check composite witnesses deterministically but these are
probably decent tests. (Not sure how else to find composites with
scarce witnesses, but these seemed decent candidates.)
Change-Id: I23dcb7ba603a64c1f7d1e9a16942e7c29c76da51
Reviewed-on: https://boringssl-review.googlesource.com/26645
Commit-Queue: Steven Valdez <svaldez@google.com>
Reviewed-by: Steven Valdez <svaldez@google.com>
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
