You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@milagro.apache.org by GitBox <gi...@apache.org> on 2021/03/28 06:02:25 UTC
[GitHub] [incubator-milagro-crypto-c] 1one-w01f opened a new issue #90: Incorrect RSA private key generation and potentials for fault attack
1one-w01f opened a new issue #90:
URL: https://github.com/apache/incubator-milagro-crypto-c/issues/90
Hi,
I recently run into a rather interesting problem. It appears to me that the RSA key generation in milagro might sometimes be incorrect, and I have no ideas to why that would be the case.
Here's the source code I prepared to illustrate the problem. For simplicity I just replaced the source code of `test/test_rsa_2048.c` with this. The build target I used was `LINUX_64BIT_NIST256_RSA2048`.
```C
#include <stdio.h>
#include <time.h>
#include "rsa_2048.h"
#include "randapi.h"
void print_keypair(rsa_private_key_2048 *priv, rsa_public_key_2048 *pub)
{
printf(" priv->p = ");
FF_2048_output(priv->p,FFLEN_2048/2);
printf("\n");
printf(" priv->q = ");
FF_2048_output(priv->q,FFLEN_2048/2);
printf("\n");
printf("\n");
printf(" pub->n = ");
FF_2048_output(pub->n,FFLEN_2048);
printf("\n");
printf("\n");
printf("priv->dp = ");
FF_2048_output(priv->dp,FFLEN_2048/2);
printf("\n");
printf("priv->dq = ");
FF_2048_output(priv->dq,FFLEN_2048/2);
printf("\n");
printf(" priv->c = ");
FF_2048_output(priv->c,FFLEN_2048/2);
printf("\n");
}
int main(int argc, char *argv[])
{
char m[RFS_2048],ml[RFS_2048],c[RFS_2048],s[RFS_2048];
char p[RFS_2048/2], q[RFS_2048/2];
rsa_public_key_2048 pub;
rsa_private_key_2048 priv;
octet M= {0,sizeof(m),m};
octet ML= {0,sizeof(ml),ml};
octet C= {0,sizeof(c),c};
octet S= {0,sizeof(s),s};
octet P= {sizeof(p), sizeof(p), p};
octet Q= {sizeof(q), sizeof(q), q};
// somehow this prime number leads to incorrect dp, which is quite weird
OCT_fromHex(&P, "f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed");
OCT_fromHex(&Q,"f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363");
// use our choice of P and Q
RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
print_keypair(&priv, &pub);
// ======
printf("\n===\n\n");
OCT_fromHex(&P, "db59a11194cb81be97715d9ea436484c7fd39cee1a71dd1e896e08cc2d91d95a2d6d561f48db72acea57450ecfd4dc5399c1f8ce7abbb8b99c92f9c7be7ab2df6990b473c5a464eaa4d34747d022e47c17b549905db305986af5970442ad54714ee7fce2c8c443da7b22f4354e105c009a8d36de733d9fb02bfbcd57803bdcfb");
OCT_fromHex(&Q, "f6457c323b66f43fa7a20f03b01a8c09c257f7fa65c7ce6a01acd816bca4de5da7692fd82ac036ca18bcc1d36e03a9ca6c2c1cf9007c106932c946cb51acb151c5c2912369553a70ddea742bc5673baaa962f8028c5ec7f4699bffd6c7beab3c278424a15548a4fce75946784036d2a9a118698224b3f218e66d980786252c93");
// use our choice of P and Q
RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
print_keypair(&priv, &pub);
// === clean up ===
RSA_2048_PRIVATE_KEY_KILL(&priv);
OCT_clear(&M);
OCT_clear(&ML); /* clean up afterwards */
OCT_clear(&C);
return 0;
}
```
Which produces the following output:
---
priv->p = f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed
priv->q = f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
pub->n = e932ac92252f585b3a80a4dd76a897c8b7652952fe788f6ec8dd640587a1ee5647670a8ad4c2be0f9fa6e49c605adf77b5174230af7bd50e5d6d6d6d28ccf0a886a514cc72e51d209cc772a52ef419f6a953f3135929588ebe9b351fca61ced78f346fe00dbb6306e5c2a4c6dfc3779af85ab417371cf34d8387b9b30ae46d7a 5ff5a655b8d8455f1b94ae736989d60a6f2fd5cadbffbd504c5a756a2e6bb5cecc13bca7503f6df8b52ace5c410997e98809db4dc30d943de4e812a47553dce54844a78e36401d13f77dc650619fed88d8b3926e3d8e319c80c744779ac5d6abe252896950917476ece5e8fc27d5f053d6018d91b502c4787558a002b9283da7
**priv->dp = 4c8b6ace1ba4747563b3d9f6c29e01fb968c96e116707fb4556b811f39b047aac0d7fc303650ff6a0a14e969656094a08dab8231a71c76613ae1063d87aea343b8a2146fc5b850693e1d7167ab97e063b8c383462e8394f47ff0fdb4c6d4e5872c49c2b459841670b808f68173248883f16d95c698b0dd39a5e167e840047571**
priv->dq = a2345ca519cd21e1b59b0b3cfaa39d0832e6e74d729db31de5152da6404903f479814863c9fa28b45418a091b60e6e1890fafe3a29e26baf07ff6f9932e7beb6d06fb767e8e7538bc07e47fb7f6f0b1554f3b7e223eefae63bd6350361591e31b34663ef50c98609a6ea9c7726a89155ab29e80ffc56416ff73957802dd5d797
**priv->c = 6f1e5e6d113bffba6e9fe60cbfd0b959f0b78fb8f71db4ab4f63147abec85beb2a982acde258672046843ce3b9f93a33ff1b25602762dfd8dbddf36e3b9815a1bf14a9dbba3d28df3dfb830fb976662cc1b41026207aa16686f49c23f93f1312ba29ef76a0c13fb03e727247b632dda5a06bf6d385da9831af4ec18d1d23df95**
===
priv->p = db59a11194cb81be97715d9ea436484c7fd39cee1a71dd1e896e08cc2d91d95a2d6d561f48db72acea57450ecfd4dc5399c1f8ce7abbb8b99c92f9c7be7ab2df6990b473c5a464eaa4d34747d022e47c17b549905db305986af5970442ad54714ee7fce2c8c443da7b22f4354e105c009a8d36de733d9fb02bfbcd57803bdcfb
priv->q = f6457c323b66f43fa7a20f03b01a8c09c257f7fa65c7ce6a01acd816bca4de5da7692fd82ac036ca18bcc1d36e03a9ca6c2c1cf9007c106932c946cb51acb151c5c2912369553a70ddea742bc5673baaa962f8028c5ec7f4699bffd6c7beab3c278424a15548a4fce75946784036d2a9a118698224b3f218e66d980786252c93
pub->n = d303aa59c3248225f2ff52e32e33cdcb797c3b501fed5e51c0189f27ef7cd614c7c4cbf2079e663b24f7ae2e9683b4878ce662d6c944196874f4a48422353303c9ad8ffd2824b4f70996ee558a89e1e98679305209a124aaf62eecd3187d2d8ee2cf5eb1b5a1b7347958aa53205b1f726856903d6a953ed72a74673c92f46f52 92b4f1c59cdb645a6afc4d0381dec80444775658c54995bf0cba58e5f90f152a557172bc040c432fe63ede5b1cf14f2f1a11d30d3a96baa24c3a3982461292be8e9db35fd088a8cbaf0e26326e630ee690368cd0a8868502e9ddb17dde45179ce7bb22b10f9deb2e3d8ddcd30ca97c9bae84ca7c3c9a13afa95de4783ea20821
priv->dp = 923bc0b66332567f0fa0e9146d7985885537bdf411a13e145b9eb0881e613b91739e396a30924c7346e4d8b48a8de837bbd6a5defc7d25d1130ca68529a721ea4660784d2e6d989c6de22f85356c9852ba78dbb593ccae659ca3ba02d71e384b89effdec85d82d3c52174d78deb592ab11b379e9a22915201d5288e50027e8a7**
priv->dq = a42e52cc2799f82a6fc15f57cabc5d5bd6e54ffc43da899c011de5647dc33ee91a461fe571d579dc107dd68cf40271319d72bdfb55a80af0cc862f32367320e12e81b6179b8e26f5e946f81d2e44d271c641faac5d94854d9bbd5539da7f1cd2c502c31638db18a89a3b84502acf371bc0baf1016dcd4c10999e655a596e1db7
priv->c = b163cc8298e63ea294f20288143304adc9a555ca170e55942954773801405652d8f998da549541782bb565ff49b32ae5f6d814bf38a878c598c6307fc294a0151f2681b8a39d6ace0a732feb6cb514a03de8d92290a8f1499231a945d698c8461cde13156d9bfc5b09ca3bf0d27a770f3ab3fedb375bd4135c8dcb57f9d60881
---
The primes `P` and `Q` were all copied from RSA private keys generated by OpenSSL. If we use OpenSSL to check the numbers of the 2 private keys:
```bash
$ openssl rsa -in privkey.pem -noout -text
RSA Private-Key: (2048 bit, 2 primes)
modulus:
00:e9:32:ac:92:25:2f:58:5b:3a:80:a4:dd:76:a8:
97:c8:b7:65:29:52:fe:78:8f:6e:c8:dd:64:05:87:
a1:ee:56:47:67:0a:8a:d4:c2:be:0f:9f:a6:e4:9c:
60:5a:df:77:b5:17:42:30:af:7b:d5:0e:5d:6d:6d:
6d:28:cc:f0:a8:86:a5:14:cc:72:e5:1d:20:9c:c7:
72:a5:2e:f4:19:f6:a9:53:f3:13:59:29:58:8e:be:
9b:35:1f:ca:61:ce:d7:8f:34:6f:e0:0d:bb:63:06:
e5:c2:a4:c6:df:c3:77:9a:f8:5a:b4:17:37:1c:f3:
4d:83:87:b9:b3:0a:e4:6d:7a:5f:f5:a6:55:b8:d8:
45:5f:1b:94:ae:73:69:89:d6:0a:6f:2f:d5:ca:db:
ff:bd:50:4c:5a:75:6a:2e:6b:b5:ce:cc:13:bc:a7:
50:3f:6d:f8:b5:2a:ce:5c:41:09:97:e9:88:09:db:
4d:c3:0d:94:3d:e4:e8:12:a4:75:53:dc:e5:48:44:
a7:8e:36:40:1d:13:f7:7d:c6:50:61:9f:ed:88:d8:
b3:92:6e:3d:8e:31:9c:80:c7:44:77:9a:c5:d6:ab:
e2:52:89:69:50:91:74:76:ec:e5:e8:fc:27:d5:f0:
53:d6:01:8d:91:b5:02:c4:78:75:58:a0:02:b9:28:
3d:a7
publicExponent: 3 (0x3)
privateExponent:
00:9b:77:1d:b6:c3:74:e5:92:27:00:6d:e8:f9:c5:
ba:85:cf:98:c6:37:54:50:5f:9f:30:93:98:03:af:
c1:49:8e:da:44:b1:b1:e3:2c:7e:b5:15:19:ed:bd:
95:91:ea:4f:ce:0f:81:75:ca:52:8e:09:93:9e:48:
f3:70:88:a0:70:59:c3:63:32:f7:43:68:c0:68:84:
f7:18:c9:f8:11:4f:1b:8d:4c:b7:90:c6:3b:09:d4:
67:78:bf:dc:41:34:8f:b4:cd:9f:ea:b3:d2:42:04:
99:2c:6d:d9:ea:82:4f:bc:a5:91:cd:64:cf:68:a2:
33:ad:05:26:77:5c:98:48:fa:fa:31:52:81:77:e1:
f8:df:91:81:a8:b9:45:08:11:06:fd:58:bd:3d:73:
79:9b:22:95:75:c4:f3:b2:91:01:a0:3e:e1:f0:54:
72:b3:61:57:84:d9:24:4c:e0:ed:63:9c:77:e8:e2:
12:ab:52:ab:dd:f4:a9:28:22:4b:6b:6f:74:b7:11:
47:86:dd:60:71:bd:91:13:d7:87:0c:6b:52:c0:bc:
8b:9c:10:2c:fe:32:1d:ac:35:7e:03:0e:d6:c5:80:
04:0c:a4:1c:13:d6:b4:96:78:11:80:7e:f2:a2:25:
98:3e:a9:f8:8d:67:fa:a4:26:20:f4:2a:4f:5b:db:
e0:3b
prime1:
00:f5:5d:1f:9b:de:51:9d:3d:30:e9:a0:82:0a:08:
50:f3:a6:d0:5e:fa:82:dc:c7:ef:94:8a:09:83:42:
24:ad:6f:b7:7e:e7:92:f5:3b:1e:e6:ef:c0:27:0e:
5e:8f:dd:59:fa:b4:0a:da:83:3d:f0:ea:69:eb:2f:
0b:37:d7:17:a3:fd:03:29:28:0c:d4:75:25:fd:22:
97:0c:8f:d8:60:c7:be:74:94:82:c4:2a:6e:16:a4:
b5:b7:ed:f2:7e:01:21:4f:55:fd:58:39:a1:1c:86:
be:6b:bd:8b:01:e5:dd:ee:f7:43:bc:a5:9e:89:6c:
17:51:14:5d:4b:6a:9d:a9:ed
prime2:
00:f3:4e:8a:f7:a6:b3:b2:d2:90:68:90:db:77:f5:
6b:8c:4c:5a:5a:f4:2b:ec:8c:ac:d7:9f:c4:79:60:
6d:85:ee:b6:41:ec:95:ae:f7:3d:0e:7e:24:f0:da:
91:15:a5:24:d9:78:7d:57:3e:d3:a1:86:8b:ff:27:
65:cc:5b:9e:12:38:a7:93:1b:dd:5a:fd:51:a0:bd:
6b:f9:3f:26:90:9f:ff:6d:93:d3:35:e6:78:59:59:
c1:4f:85:12:05:ad:4a:8c:e9:95:e6:f9:2e:49:0e:
7a:5f:ea:b2:b9:fc:da:00:80:be:dc:17:fa:81:62:
27:f2:d6:03:40:44:c0:c3:63
exponent1:
00:a3:93:6a:67:e9:8b:be:28:cb:46:6b:01:5c:05:
8b:4d:19:e0:3f:51:ac:93:2f:f5:0d:b1:5b:ac:d6:
c3:1e:4a:7a:54:9a:61:f8:d2:14:99:f5:2a:c4:b4:
3f:0a:93:91:51:cd:5c:91:ac:d3:f5:f1:9b:f2:1f:
5c:cf:e4:ba:6d:53:57:70:c5:5d:e2:f8:c3:fe:17:
0f:5d:b5:3a:eb:2f:d4:4d:b8:57:2d:71:9e:b9:c3:
23:cf:f3:f6:fe:ab:6b:8a:39:53:90:26:6b:68:59:
d4:47:d3:b2:01:43:e9:49:fa:2d:28:6e:69:b0:f2:
ba:36:0d:93:87:9c:69:1b:f3
exponent2:
00:a2:34:5c:a5:19:cd:21:e1:b5:9b:0b:3c:fa:a3:
9d:08:32:e6:e7:4d:72:9d:b3:1d:e5:15:2d:a6:40:
49:03:f4:79:81:48:63:c9:fa:28:b4:54:18:a0:91:
b6:0e:6e:18:90:fa:fe:3a:29:e2:6b:af:07:ff:6f:
99:32:e7:be:b6:d0:6f:b7:67:e8:e7:53:8b:c0:7e:
47:fb:7f:6f:0b:15:54:f3:b7:e2:23:ee:fa:e6:3b:
d6:35:03:61:59:1e:31:b3:46:63:ef:50:c9:86:09:
a6:ea:9c:77:26:a8:91:55:ab:29:e8:0f:fc:56:41:
6f:f7:39:57:80:2d:d5:d7:97
coefficient:
00:85:4e:43:ad:27:2d:f5:81:12:c9:84:ca:73:eb:
bf:9a:57:8a:56:53:ff:eb:79:dc:43:a4:20:27:b5:
5a:69:27:d7:74:65:c4:2d:83:e9:d8:ee:9e:7c:bc:
91:6f:ed:76:2a:7b:90:24:a5:bd:ee:8a:f3:1d:38:
d5:c1:84:0d:80:78:13:9a:3c:b0:04:7c:c7:36:cf:
22:4d:99:25:f6:1a:42:fd:0f:48:87:09:e7:24:a5:
c6:f3:21:40:a0:5f:22:e4:9b:37:2d:f7:f4:39:c6:
ad:86:07:89:6b:d6:b3:8a:d5:9e:bd:be:1b:8d:e9:
32:54:35:25:7e:4c:a4:b0:9d
```
```bash
$ openssl rsa -in privkey.pem -noout -text
RSA Private-Key: (2048 bit, 2 primes)
modulus:
00:d3:03:aa:59:c3:24:82:25:f2:ff:52:e3:2e:33:
cd:cb:79:7c:3b:50:1f:ed:5e:51:c0:18:9f:27:ef:
7c:d6:14:c7:c4:cb:f2:07:9e:66:3b:24:f7:ae:2e:
96:83:b4:87:8c:e6:62:d6:c9:44:19:68:74:f4:a4:
84:22:35:33:03:c9:ad:8f:fd:28:24:b4:f7:09:96:
ee:55:8a:89:e1:e9:86:79:30:52:09:a1:24:aa:f6:
2e:ec:d3:18:7d:2d:8e:e2:cf:5e:b1:b5:a1:b7:34:
79:58:aa:53:20:5b:1f:72:68:56:90:3d:6a:95:3e:
d7:2a:74:67:3c:92:f4:6f:52:92:b4:f1:c5:9c:db:
64:5a:6a:fc:4d:03:81:de:c8:04:44:77:56:58:c5:
49:95:bf:0c:ba:58:e5:f9:0f:15:2a:55:71:72:bc:
04:0c:43:2f:e6:3e:de:5b:1c:f1:4f:2f:1a:11:d3:
0d:3a:96:ba:a2:4c:3a:39:82:46:12:92:be:8e:9d:
b3:5f:d0:88:a8:cb:af:0e:26:32:6e:63:0e:e6:90:
36:8c:d0:a8:86:85:02:e9:dd:b1:7d:de:45:17:9c:
e7:bb:22:b1:0f:9d:eb:2e:3d:8d:dc:d3:0c:a9:7c:
9b:ae:84:ca:7c:3c:9a:13:af:a9:5d:e4:78:3e:a2:
08:21
publicExponent: 3 (0x3)
privateExponent:
00:8c:ad:1c:3b:d7:6d:ac:19:4c:aa:37:42:1e:cd:
33:dc:fb:a8:27:8a:bf:f3:94:36:80:10:6a:1a:9f:
a8:8e:b8:85:2d:dd:4c:05:14:44:27:6d:fa:74:1f:
0f:02:78:5a:5d:ee:ec:8f:30:d8:10:f0:4d:f8:6d:
ad:6c:23:77:57:db:c9:0a:a8:c5:6d:cd:fa:06:64:
9e:e3:b1:b1:41:46:59:a6:20:36:b1:16:18:71:f9:
74:9d:e2:10:53:73:b4:97:34:e9:cb:ce:6b:cf:78:
50:e5:c6:e2:15:92:14:f6:f0:39:b5:7e:47:0e:29:
e4:c6:f8:44:d3:0c:a2:f4:e0:80:b9:38:56:88:70:
9e:e8:1d:45:ea:eb:73:b3:f7:c9:56:dd:2b:a0:2e:
0a:9c:24:56:6a:50:02:09:e5:93:a1:ab:11:f3:2d:
b5:a0:66:7b:42:1c:8f:a5:ea:10:86:0b:62:c2:7e:
2e:7f:94:a0:ff:a8:93:fb:4a:23:f2:1f:08:ea:31:
9e:85:c1:0a:06:4a:c8:35:9c:7f:3b:3b:49:d5:34:
be:dc:d3:d4:4d:cf:a4:0e:32:bc:6c:8d:3b:65:4a:
4b:8a:00:c8:a1:0b:56:e4:92:0b:c1:6e:54:41:89:
4b:a1:ea:1c:12:6d:c5:ab:ef:0f:4d:aa:10:d0:2b:
54:63
prime1:
00:f6:45:7c:32:3b:66:f4:3f:a7:a2:0f:03:b0:1a:
8c:09:c2:57:f7:fa:65:c7:ce:6a:01:ac:d8:16:bc:
a4:de:5d:a7:69:2f:d8:2a:c0:36:ca:18:bc:c1:d3:
6e:03:a9:ca:6c:2c:1c:f9:00:7c:10:69:32:c9:46:
cb:51:ac:b1:51:c5:c2:91:23:69:55:3a:70:dd:ea:
74:2b:c5:67:3b:aa:a9:62:f8:02:8c:5e:c7:f4:69:
9b:ff:d6:c7:be:ab:3c:27:84:24:a1:55:48:a4:fc:
e7:59:46:78:40:36:d2:a9:a1:18:69:82:24:b3:f2:
18:e6:6d:98:07:86:25:2c:93
prime2:
00:db:59:a1:11:94:cb:81:be:97:71:5d:9e:a4:36:
48:4c:7f:d3:9c:ee:1a:71:dd:1e:89:6e:08:cc:2d:
91:d9:5a:2d:6d:56:1f:48:db:72:ac:ea:57:45:0e:
cf:d4:dc:53:99:c1:f8:ce:7a:bb:b8:b9:9c:92:f9:
c7:be:7a:b2:df:69:90:b4:73:c5:a4:64:ea:a4:d3:
47:47:d0:22:e4:7c:17:b5:49:90:5d:b3:05:98:6a:
f5:97:04:42:ad:54:71:4e:e7:fc:e2:c8:c4:43:da:
7b:22:f4:35:4e:10:5c:00:9a:8d:36:de:73:3d:9f:
b0:2b:fb:cd:57:80:3b:dc:fb
exponent1:
00:a4:2e:52:cc:27:99:f8:2a:6f:c1:5f:57:ca:bc:
5d:5b:d6:e5:4f:fc:43:da:89:9c:01:1d:e5:64:7d:
c3:3e:e9:1a:46:1f:e5:71:d5:79:dc:10:7d:d6:8c:
f4:02:71:31:9d:72:bd:fb:55:a8:0a:f0:cc:86:2f:
32:36:73:20:e1:2e:81:b6:17:9b:8e:26:f5:e9:46:
f8:1d:2e:44:d2:71:c6:41:fa:ac:5d:94:85:4d:9b:
bd:55:39:da:7f:1c:d2:c5:02:c3:16:38:db:18:a8:
9a:3b:84:50:2a:cf:37:1b:c0:ba:f1:01:6d:cd:4c:
10:99:9e:65:5a:59:6e:1d:b7
exponent2:
00:92:3b:c0:b6:63:32:56:7f:0f:a0:e9:14:6d:79:
85:88:55:37:bd:f4:11:a1:3e:14:5b:9e:b0:88:1e:
61:3b:91:73:9e:39:6a:30:92:4c:73:46:e4:d8:b4:
8a:8d:e8:37:bb:d6:a5:de:fc:7d:25:d1:13:0c:a6:
85:29:a7:21:ea:46:60:78:4d:2e:6d:98:9c:6d:e2:
2f:85:35:6c:98:52:ba:78:db:b5:93:cc:ae:65:9c:
a3:ba:02:d7:1e:38:4b:89:ef:fd:ec:85:d8:2d:3c:
52:17:4d:78:de:b5:92:ab:11:b3:79:e9:a2:29:15:
20:1d:52:88:e5:00:27:e8:a7
coefficient:
00:b1:63:cc:82:98:e6:3e:a2:94:f2:02:88:14:33:
04:ad:c9:a5:55:ca:17:0e:55:94:29:54:77:38:01:
40:56:52:d8:f9:98:da:54:95:41:78:2b:b5:65:ff:
49:b3:2a:e5:f6:d8:14:bf:38:a8:78:c5:98:c6:30:
7f:c2:94:a0:15:1f:26:81:b8:a3:9d:6a:ce:0a:73:
2f:eb:6c:b5:14:a0:3d:e8:d9:22:90:a8:f1:49:92:
31:a9:45:d6:98:c8:46:1c:de:13:15:6d:9b:fc:5b:
09:ca:3b:f0:d2:7a:77:0f:3a:b3:fe:db:37:5b:d4:
13:5c:8d:cb:57:f9:d6:08:81
```
It can be seen that for the first RSA private key, the numbers `dp` and `c` (both styled in bold) are not right. This could be potentially dangerous when signatures are involved.
Because of the Chinese Remainder Theorem, if one uses the incorrectly computed private key to sign a message and sends the resulting wrong signature directly without first verifying it, this can lead to the so-called "fault attack" (e.g., explained [here](https://www.cryptologie.net/article/371/fault-attacks-on-rsas-signatures/)).
Continuing with the example above, if we sign the message `Hello World\n` using the first RSA private key:
```C
// somehow this prime number leads to incorrect dp, which is quite weird
OCT_fromHex(&P, "f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed");
OCT_fromHex(&Q,"f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363");
// use our choice of P and Q
RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
print_keypair(&priv, &pub);
printf("Signing message\n");
OCT_jstring(&M,(char *)"Hello World\n");
PKCS15(HASH_TYPE_RSA_2048,&M,&C);
RSA_2048_DECRYPT(&priv,&C,&S); /* create signature in S */
printf("Signature= ");
OCT_output(&S);
```
We get a faulty signature: `s = 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`.
If an attacker obtains this signature, and knows the choice of hash algorithm as well as the message of `Hello World\n`, then he/she can also compute the hash value and prepare the PKCS1v1.5 structure accordingly: `r = 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`
And now the attacker can factorize the modulus `n` by computing `GCD(s^e - r, n)`. In Python3:
```python
import math
# faulty signature
s = 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
# precomputed PKCS1v1.5 structure
r = 0x0001ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff003031300d060960864801650304020105000420d2a84f4b8b650937ec8f73cd8be2c74add5a911ba64df27458ed8229da804a26
# modulus
n = 0xf55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed * 0xf34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
# now factorize n
p1 = math.gcd(pow(s,3,n)-r, n)
p2 = n // p1
print("p=",hex(max(p1,p2)))
print("q=",hex(min(p1,p2)))
```
Which outputs
---
p= 0xf55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed
q= 0xf34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
---
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@milagro.apache.org
For additional commands, e-mail: issues-help@milagro.apache.org
[GitHub] [incubator-milagro-crypto-c] mcarrickscott commented on issue #90: Incorrect RSA private key generation and potentials for fault attack
Posted by GitBox <gi...@apache.org>.
mcarrickscott commented on issue #90:
URL: https://github.com/apache/incubator-milagro-crypto-c/issues/90#issuecomment-808887175
The problem here is that p=1 mod 4 and q=3 mod 4
The Milagro code only works for primes that are 3 mod 4,and the primes it
generates internally are all of this form. In general 3 mod 4 primes are
easier to handle,and since half of all primes are of this form,there is no
loss of generality in restricting to such primes.
The specific problem is that the function invmodp(r,a,n) only calculates
the correct inverse if n is odd. Now p-1 is even, but in p=3 mod 4 case
(given that e is odd) we can calculate inverse(r,a,(p-1)/2) and this will
be correct. But the same trick won't work if p is 1 mod 4, as in that case
(p-1)/2 will still be even. It would not be hard to fix, but probably
simpler just to restrict to 3 mod 4 primes
Mike
On Sun, Mar 28, 2021 at 7:02 AM 1one.w01f ***@***.***> wrote:
> Hi,
>
> I recently run into a rather interesting problem. It appears to me that
> the RSA key generation in milagro might sometimes be incorrect, and I have
> no ideas to why that would be the case.
>
> Here's the source code I prepared to illustrate the problem. For
> simplicity I just replaced the source code of test/test_rsa_2048.c with
> this. The build target I used was LINUX_64BIT_NIST256_RSA2048.
>
> #include <stdio.h>
> #include <time.h>
> #include "rsa_2048.h"
> #include "randapi.h"
> void print_keypair(rsa_private_key_2048 *priv, rsa_public_key_2048 *pub)
> {
> printf(" priv->p = ");
> FF_2048_output(priv->p,FFLEN_2048/2);
> printf("\n");
> printf(" priv->q = ");
> FF_2048_output(priv->q,FFLEN_2048/2);
> printf("\n");
>
> printf("\n");
>
> printf(" pub->n = ");
> FF_2048_output(pub->n,FFLEN_2048);
> printf("\n");
>
> printf("\n");
>
> printf("priv->dp = ");
> FF_2048_output(priv->dp,FFLEN_2048/2);
> printf("\n");
> printf("priv->dq = ");
> FF_2048_output(priv->dq,FFLEN_2048/2);
> printf("\n");
> printf(" priv->c = ");
> FF_2048_output(priv->c,FFLEN_2048/2);
> printf("\n");
> }
> int main(int argc, char *argv[])
> {
> char m[RFS_2048],ml[RFS_2048],c[RFS_2048],s[RFS_2048];
> char p[RFS_2048/2], q[RFS_2048/2];
>
> rsa_public_key_2048 pub;
> rsa_private_key_2048 priv;
>
> octet M= {0,sizeof(m),m};
> octet ML= {0,sizeof(ml),ml};
> octet C= {0,sizeof(c),c};
> octet S= {0,sizeof(s),s};
> octet P= {sizeof(p), sizeof(p), p};
> octet Q= {sizeof(q), sizeof(q), q};
>
> // somehow this prime number leads to incorrect dp, which is quite weird
> OCT_fromHex(&P, "f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed");
>
> OCT_fromHex(&Q,"f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363");
>
> // use our choice of P and Q
> RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
>
> print_keypair(&priv, &pub);
> // ======
> printf("\n===\n\n");
>
> OCT_fromHex(&P, "db59a11194cb81be97715d9ea436484c7fd39cee1a71dd1e896e08cc2d91d95a2d6d561f48db72acea57450ecfd4dc5399c1f8ce7abbb8b99c92f9c7be7ab2df6990b473c5a464eaa4d34747d022e47c17b549905db305986af5970442ad54714ee7fce2c8c443da7b22f4354e105c009a8d36de733d9fb02bfbcd57803bdcfb");
>
> OCT_fromHex(&Q, "f6457c323b66f43fa7a20f03b01a8c09c257f7fa65c7ce6a01acd816bca4de5da7692fd82ac036ca18bcc1d36e03a9ca6c2c1cf9007c106932c946cb51acb151c5c2912369553a70ddea742bc5673baaa962f8028c5ec7f4699bffd6c7beab3c278424a15548a4fce75946784036d2a9a118698224b3f218e66d980786252c93");
>
> // use our choice of P and Q
> RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
>
> print_keypair(&priv, &pub);
> // === clean up ===
>
> RSA_2048_PRIVATE_KEY_KILL(&priv);
>
> OCT_clear(&M);
> OCT_clear(&ML); /* clean up afterwards */
> OCT_clear(&C);
>
> return 0;
> }
>
> Which produces the following output:
> ------------------------------
>
> priv->p =
> f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed
> priv->q =
> f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
>
> pub->n =
> e932ac92252f585b3a80a4dd76a897c8b7652952fe788f6ec8dd640587a1ee5647670a8ad4c2be0f9fa6e49c605adf77b5174230af7bd50e5d6d6d6d28ccf0a886a514cc72e51d209cc772a52ef419f6a953f3135929588ebe9b351fca61ced78f346fe00dbb6306e5c2a4c6dfc3779af85ab417371cf34d8387b9b30ae46d7a
> 5ff5a655b8d8455f1b94ae736989d60a6f2fd5cadbffbd504c5a756a2e6bb5cecc13bca7503f6df8b52ace5c410997e98809db4dc30d943de4e812a47553dce54844a78e36401d13f77dc650619fed88d8b3926e3d8e319c80c744779ac5d6abe252896950917476ece5e8fc27d5f053d6018d91b502c4787558a002b9283da7
>
> *priv->dp =
> 4c8b6ace1ba4747563b3d9f6c29e01fb968c96e116707fb4556b811f39b047aac0d7fc303650ff6a0a14e969656094a08dab8231a71c76613ae1063d87aea343b8a2146fc5b850693e1d7167ab97e063b8c383462e8394f47ff0fdb4c6d4e5872c49c2b459841670b808f68173248883f16d95c698b0dd39a5e167e840047571*
> priv->dq =
> a2345ca519cd21e1b59b0b3cfaa39d0832e6e74d729db31de5152da6404903f479814863c9fa28b45418a091b60e6e1890fafe3a29e26baf07ff6f9932e7beb6d06fb767e8e7538bc07e47fb7f6f0b1554f3b7e223eefae63bd6350361591e31b34663ef50c98609a6ea9c7726a89155ab29e80ffc56416ff73957802dd5d797
> *priv->c =
> 6f1e5e6d113bffba6e9fe60cbfd0b959f0b78fb8f71db4ab4f63147abec85beb2a982acde258672046843ce3b9f93a33ff1b25602762dfd8dbddf36e3b9815a1bf14a9dbba3d28df3dfb830fb976662cc1b41026207aa16686f49c23f93f1312ba29ef76a0c13fb03e727247b632dda5a06bf6d385da9831af4ec18d1d23df95*
>
> ===
>
> priv->p =
> db59a11194cb81be97715d9ea436484c7fd39cee1a71dd1e896e08cc2d91d95a2d6d561f48db72acea57450ecfd4dc5399c1f8ce7abbb8b99c92f9c7be7ab2df6990b473c5a464eaa4d34747d022e47c17b549905db305986af5970442ad54714ee7fce2c8c443da7b22f4354e105c009a8d36de733d9fb02bfbcd57803bdcfb
> priv->q =
> f6457c323b66f43fa7a20f03b01a8c09c257f7fa65c7ce6a01acd816bca4de5da7692fd82ac036ca18bcc1d36e03a9ca6c2c1cf9007c106932c946cb51acb151c5c2912369553a70ddea742bc5673baaa962f8028c5ec7f4699bffd6c7beab3c278424a15548a4fce75946784036d2a9a118698224b3f218e66d980786252c93
>
> pub->n =
> d303aa59c3248225f2ff52e32e33cdcb797c3b501fed5e51c0189f27ef7cd614c7c4cbf2079e663b24f7ae2e9683b4878ce662d6c944196874f4a48422353303c9ad8ffd2824b4f70996ee558a89e1e98679305209a124aaf62eecd3187d2d8ee2cf5eb1b5a1b7347958aa53205b1f726856903d6a953ed72a74673c92f46f52
> 92b4f1c59cdb645a6afc4d0381dec80444775658c54995bf0cba58e5f90f152a557172bc040c432fe63ede5b1cf14f2f1a11d30d3a96baa24c3a3982461292be8e9db35fd088a8cbaf0e26326e630ee690368cd0a8868502e9ddb17dde45179ce7bb22b10f9deb2e3d8ddcd30ca97c9bae84ca7c3c9a13afa95de4783ea20821
>
> priv->dp =
> 923bc0b66332567f0fa0e9146d7985885537bdf411a13e145b9eb0881e613b91739e396a30924c7346e4d8b48a8de837bbd6a5defc7d25d1130ca68529a721ea4660784d2e6d989c6de22f85356c9852ba78dbb593ccae659ca3ba02d71e384b89effdec85d82d3c52174d78deb592ab11b379e9a22915201d5288e50027e8a7**
> priv->dq =
> a42e52cc2799f82a6fc15f57cabc5d5bd6e54ffc43da899c011de5647dc33ee91a461fe571d579dc107dd68cf40271319d72bdfb55a80af0cc862f32367320e12e81b6179b8e26f5e946f81d2e44d271c641faac5d94854d9bbd5539da7f1cd2c502c31638db18a89a3b84502acf371bc0baf1016dcd4c10999e655a596e1db7
> priv->c =
> b163cc8298e63ea294f20288143304adc9a555ca170e55942954773801405652d8f998da549541782bb565ff49b32ae5f6d814bf38a878c598c6307fc294a0151f2681b8a39d6ace0a732feb6cb514a03de8d92290a8f1499231a945d698c8461cde13156d9bfc5b09ca3bf0d27a770f3ab3fedb375bd4135c8dcb57f9d60881
> ------------------------------
>
> The primes P and Q were all copied from RSA private keys generated by
> OpenSSL. If we use OpenSSL to check the numbers of the 2 private keys:
>
> $ openssl rsa -in privkey.pem -noout -text
> RSA Private-Key: (2048 bit, 2 primes)
> modulus:
> 00:e9:32:ac:92:25:2f:58:5b:3a:80:a4:dd:76:a8:
> 97:c8:b7:65:29:52:fe:78:8f:6e:c8:dd:64:05:87:
> a1:ee:56:47:67:0a:8a:d4:c2:be:0f:9f:a6:e4:9c:
> 60:5a:df:77:b5:17:42:30:af:7b:d5:0e:5d:6d:6d:
> 6d:28:cc:f0:a8:86:a5:14:cc:72:e5:1d:20:9c:c7:
> 72:a5:2e:f4:19:f6:a9:53:f3:13:59:29:58:8e:be:
> 9b:35:1f:ca:61:ce:d7:8f:34:6f:e0:0d:bb:63:06:
> e5:c2:a4:c6:df:c3:77:9a:f8:5a:b4:17:37:1c:f3:
> 4d:83:87:b9:b3:0a:e4:6d:7a:5f:f5:a6:55:b8:d8:
> 45:5f:1b:94:ae:73:69:89:d6:0a:6f:2f:d5:ca:db:
> ff:bd:50:4c:5a:75:6a:2e:6b:b5:ce:cc:13:bc:a7:
> 50:3f:6d:f8:b5:2a:ce:5c:41:09:97:e9:88:09:db:
> 4d:c3:0d:94:3d:e4:e8:12:a4:75:53:dc:e5:48:44:
> a7:8e:36:40:1d:13:f7:7d:c6:50:61:9f:ed:88:d8:
> b3:92:6e:3d:8e:31:9c:80:c7:44:77:9a:c5:d6:ab:
> e2:52:89:69:50:91:74:76:ec:e5:e8:fc:27:d5:f0:
> 53:d6:01:8d:91:b5:02:c4:78:75:58:a0:02:b9:28:
> 3d:a7
> publicExponent: 3 (0x3)
> privateExponent:
> 00:9b:77:1d:b6:c3:74:e5:92:27:00:6d:e8:f9:c5:
> ba:85:cf:98:c6:37:54:50:5f:9f:30:93:98:03:af:
> c1:49:8e:da:44:b1:b1:e3:2c:7e:b5:15:19:ed:bd:
> 95:91:ea:4f:ce:0f:81:75:ca:52:8e:09:93:9e:48:
> f3:70:88:a0:70:59:c3:63:32:f7:43:68:c0:68:84:
> f7:18:c9:f8:11:4f:1b:8d:4c:b7:90:c6:3b:09:d4:
> 67:78:bf:dc:41:34:8f:b4:cd:9f:ea:b3:d2:42:04:
> 99:2c:6d:d9:ea:82:4f:bc:a5:91:cd:64:cf:68:a2:
> 33:ad:05:26:77:5c:98:48:fa:fa:31:52:81:77:e1:
> f8:df:91:81:a8:b9:45:08:11:06:fd:58:bd:3d:73:
> 79:9b:22:95:75:c4:f3:b2:91:01:a0:3e:e1:f0:54:
> 72:b3:61:57:84:d9:24:4c:e0:ed:63:9c:77:e8:e2:
> 12:ab:52:ab:dd:f4:a9:28:22:4b:6b:6f:74:b7:11:
> 47:86:dd:60:71:bd:91:13:d7:87:0c:6b:52:c0:bc:
> 8b:9c:10:2c:fe:32:1d:ac:35:7e:03:0e:d6:c5:80:
> 04:0c:a4:1c:13:d6:b4:96:78:11:80:7e:f2:a2:25:
> 98:3e:a9:f8:8d:67:fa:a4:26:20:f4:2a:4f:5b:db:
> e0:3b
> prime1:
> 00:f5:5d:1f:9b:de:51:9d:3d:30:e9:a0:82:0a:08:
> 50:f3:a6:d0:5e:fa:82:dc:c7:ef:94:8a:09:83:42:
> 24:ad:6f:b7:7e:e7:92:f5:3b:1e:e6:ef:c0:27:0e:
> 5e:8f:dd:59:fa:b4:0a:da:83:3d:f0:ea:69:eb:2f:
> 0b:37:d7:17:a3:fd:03:29:28:0c:d4:75:25:fd:22:
> 97:0c:8f:d8:60:c7:be:74:94:82:c4:2a:6e:16:a4:
> b5:b7:ed:f2:7e:01:21:4f:55:fd:58:39:a1:1c:86:
> be:6b:bd:8b:01:e5:dd:ee:f7:43:bc:a5:9e:89:6c:
> 17:51:14:5d:4b:6a:9d:a9:ed
> prime2:
> 00:f3:4e:8a:f7:a6:b3:b2:d2:90:68:90:db:77:f5:
> 6b:8c:4c:5a:5a:f4:2b:ec:8c:ac:d7:9f:c4:79:60:
> 6d:85:ee:b6:41:ec:95:ae:f7:3d:0e:7e:24:f0:da:
> 91:15:a5:24:d9:78:7d:57:3e:d3:a1:86:8b:ff:27:
> 65:cc:5b:9e:12:38:a7:93:1b:dd:5a:fd:51:a0:bd:
> 6b:f9:3f:26:90:9f:ff:6d:93:d3:35:e6:78:59:59:
> c1:4f:85:12:05:ad:4a:8c:e9:95:e6:f9:2e:49:0e:
> 7a:5f:ea:b2:b9:fc:da:00:80:be:dc:17:fa:81:62:
> 27:f2:d6:03:40:44:c0:c3:63
> exponent1:
> 00:a3:93:6a:67:e9:8b:be:28:cb:46:6b:01:5c:05:
> 8b:4d:19:e0:3f:51:ac:93:2f:f5:0d:b1:5b:ac:d6:
> c3:1e:4a:7a:54:9a:61:f8:d2:14:99:f5:2a:c4:b4:
> 3f:0a:93:91:51:cd:5c:91:ac:d3:f5:f1:9b:f2:1f:
> 5c:cf:e4:ba:6d:53:57:70:c5:5d:e2:f8:c3:fe:17:
> 0f:5d:b5:3a:eb:2f:d4:4d:b8:57:2d:71:9e:b9:c3:
> 23:cf:f3:f6:fe:ab:6b:8a:39:53:90:26:6b:68:59:
> d4:47:d3:b2:01:43:e9:49:fa:2d:28:6e:69:b0:f2:
> ba:36:0d:93:87:9c:69:1b:f3
> exponent2:
> 00:a2:34:5c:a5:19:cd:21:e1:b5:9b:0b:3c:fa:a3:
> 9d:08:32:e6:e7:4d:72:9d:b3:1d:e5:15:2d:a6:40:
> 49:03:f4:79:81:48:63:c9:fa:28:b4:54:18:a0:91:
> b6:0e:6e:18:90:fa:fe:3a:29:e2:6b:af:07:ff:6f:
> 99:32:e7:be:b6:d0:6f:b7:67:e8:e7:53:8b:c0:7e:
> 47:fb:7f:6f:0b:15:54:f3:b7:e2:23:ee:fa:e6:3b:
> d6:35:03:61:59:1e:31:b3:46:63:ef:50:c9:86:09:
> a6:ea:9c:77:26:a8:91:55:ab:29:e8:0f:fc:56:41:
> 6f:f7:39:57:80:2d:d5:d7:97
> coefficient:
> 00:85:4e:43:ad:27:2d:f5:81:12:c9:84:ca:73:eb:
> bf:9a:57:8a:56:53:ff:eb:79:dc:43:a4:20:27:b5:
> 5a:69:27:d7:74:65:c4:2d:83:e9:d8:ee:9e:7c:bc:
> 91:6f:ed:76:2a:7b:90:24:a5:bd:ee:8a:f3:1d:38:
> d5:c1:84:0d:80:78:13:9a:3c:b0:04:7c:c7:36:cf:
> 22:4d:99:25:f6:1a:42:fd:0f:48:87:09:e7:24:a5:
> c6:f3:21:40:a0:5f:22:e4:9b:37:2d:f7:f4:39:c6:
> ad:86:07:89:6b:d6:b3:8a:d5:9e:bd:be:1b:8d:e9:
> 32:54:35:25:7e:4c:a4:b0:9d
>
> $ openssl rsa -in privkey.pem -noout -text
> RSA Private-Key: (2048 bit, 2 primes)
> modulus:
> 00:d3:03:aa:59:c3:24:82:25:f2:ff:52:e3:2e:33:
> cd:cb:79:7c:3b:50:1f:ed:5e:51:c0:18:9f:27:ef:
> 7c:d6:14:c7:c4:cb:f2:07:9e:66:3b:24:f7:ae:2e:
> 96:83:b4:87:8c:e6:62:d6:c9:44:19:68:74:f4:a4:
> 84:22:35:33:03:c9:ad:8f:fd:28:24:b4:f7:09:96:
> ee:55:8a:89:e1:e9:86:79:30:52:09:a1:24:aa:f6:
> 2e:ec:d3:18:7d:2d:8e:e2:cf:5e:b1:b5:a1:b7:34:
> 79:58:aa:53:20:5b:1f:72:68:56:90:3d:6a:95:3e:
> d7:2a:74:67:3c:92:f4:6f:52:92:b4:f1:c5:9c:db:
> 64:5a:6a:fc:4d:03:81:de:c8:04:44:77:56:58:c5:
> 49:95:bf:0c:ba:58:e5:f9:0f:15:2a:55:71:72:bc:
> 04:0c:43:2f:e6:3e:de:5b:1c:f1:4f:2f:1a:11:d3:
> 0d:3a:96:ba:a2:4c:3a:39:82:46:12:92:be:8e:9d:
> b3:5f:d0:88:a8:cb:af:0e:26:32:6e:63:0e:e6:90:
> 36:8c:d0:a8:86:85:02:e9:dd:b1:7d:de:45:17:9c:
> e7:bb:22:b1:0f:9d:eb:2e:3d:8d:dc:d3:0c:a9:7c:
> 9b:ae:84:ca:7c:3c:9a:13:af:a9:5d:e4:78:3e:a2:
> 08:21
> publicExponent: 3 (0x3)
> privateExponent:
> 00:8c:ad:1c:3b:d7:6d:ac:19:4c:aa:37:42:1e:cd:
> 33:dc:fb:a8:27:8a:bf:f3:94:36:80:10:6a:1a:9f:
> a8:8e:b8:85:2d:dd:4c:05:14:44:27:6d:fa:74:1f:
> 0f:02:78:5a:5d:ee:ec:8f:30:d8:10:f0:4d:f8:6d:
> ad:6c:23:77:57:db:c9:0a:a8:c5:6d:cd:fa:06:64:
> 9e:e3:b1:b1:41:46:59:a6:20:36:b1:16:18:71:f9:
> 74:9d:e2:10:53:73:b4:97:34:e9:cb:ce:6b:cf:78:
> 50:e5:c6:e2:15:92:14:f6:f0:39:b5:7e:47:0e:29:
> e4:c6:f8:44:d3:0c:a2:f4:e0:80:b9:38:56:88:70:
> 9e:e8:1d:45:ea:eb:73:b3:f7:c9:56:dd:2b:a0:2e:
> 0a:9c:24:56:6a:50:02:09:e5:93:a1:ab:11:f3:2d:
> b5:a0:66:7b:42:1c:8f:a5:ea:10:86:0b:62:c2:7e:
> 2e:7f:94:a0:ff:a8:93:fb:4a:23:f2:1f:08:ea:31:
> 9e:85:c1:0a:06:4a:c8:35:9c:7f:3b:3b:49:d5:34:
> be:dc:d3:d4:4d:cf:a4:0e:32:bc:6c:8d:3b:65:4a:
> 4b:8a:00:c8:a1:0b:56:e4:92:0b:c1:6e:54:41:89:
> 4b:a1:ea:1c:12:6d:c5:ab:ef:0f:4d:aa:10:d0:2b:
> 54:63
> prime1:
> 00:f6:45:7c:32:3b:66:f4:3f:a7:a2:0f:03:b0:1a:
> 8c:09:c2:57:f7:fa:65:c7:ce:6a:01:ac:d8:16:bc:
> a4:de:5d:a7:69:2f:d8:2a:c0:36:ca:18:bc:c1:d3:
> 6e:03:a9:ca:6c:2c:1c:f9:00:7c:10:69:32:c9:46:
> cb:51:ac:b1:51:c5:c2:91:23:69:55:3a:70:dd:ea:
> 74:2b:c5:67:3b:aa:a9:62:f8:02:8c:5e:c7:f4:69:
> 9b:ff:d6:c7:be:ab:3c:27:84:24:a1:55:48:a4:fc:
> e7:59:46:78:40:36:d2:a9:a1:18:69:82:24:b3:f2:
> 18:e6:6d:98:07:86:25:2c:93
> prime2:
> 00:db:59:a1:11:94:cb:81:be:97:71:5d:9e:a4:36:
> 48:4c:7f:d3:9c:ee:1a:71:dd:1e:89:6e:08:cc:2d:
> 91:d9:5a:2d:6d:56:1f:48:db:72:ac:ea:57:45:0e:
> cf:d4:dc:53:99:c1:f8:ce:7a:bb:b8:b9:9c:92:f9:
> c7:be:7a:b2:df:69:90:b4:73:c5:a4:64:ea:a4:d3:
> 47:47:d0:22:e4:7c:17:b5:49:90:5d:b3:05:98:6a:
> f5:97:04:42:ad:54:71:4e:e7:fc:e2:c8:c4:43:da:
> 7b:22:f4:35:4e:10:5c:00:9a:8d:36:de:73:3d:9f:
> b0:2b:fb:cd:57:80:3b:dc:fb
> exponent1:
> 00:a4:2e:52:cc:27:99:f8:2a:6f:c1:5f:57:ca:bc:
> 5d:5b:d6:e5:4f:fc:43:da:89:9c:01:1d:e5:64:7d:
> c3:3e:e9:1a:46:1f:e5:71:d5:79:dc:10:7d:d6:8c:
> f4:02:71:31:9d:72:bd:fb:55:a8:0a:f0:cc:86:2f:
> 32:36:73:20:e1:2e:81:b6:17:9b:8e:26:f5:e9:46:
> f8:1d:2e:44:d2:71:c6:41:fa:ac:5d:94:85:4d:9b:
> bd:55:39:da:7f:1c:d2:c5:02:c3:16:38:db:18:a8:
> 9a:3b:84:50:2a:cf:37:1b:c0:ba:f1:01:6d:cd:4c:
> 10:99:9e:65:5a:59:6e:1d:b7
> exponent2:
> 00:92:3b:c0:b6:63:32:56:7f:0f:a0:e9:14:6d:79:
> 85:88:55:37:bd:f4:11:a1:3e:14:5b:9e:b0:88:1e:
> 61:3b:91:73:9e:39:6a:30:92:4c:73:46:e4:d8:b4:
> 8a:8d:e8:37:bb:d6:a5:de:fc:7d:25:d1:13:0c:a6:
> 85:29:a7:21:ea:46:60:78:4d:2e:6d:98:9c:6d:e2:
> 2f:85:35:6c:98:52:ba:78:db:b5:93:cc:ae:65:9c:
> a3:ba:02:d7:1e:38:4b:89:ef:fd:ec:85:d8:2d:3c:
> 52:17:4d:78:de:b5:92:ab:11:b3:79:e9:a2:29:15:
> 20:1d:52:88:e5:00:27:e8:a7
> coefficient:
> 00:b1:63:cc:82:98:e6:3e:a2:94:f2:02:88:14:33:
> 04:ad:c9:a5:55:ca:17:0e:55:94:29:54:77:38:01:
> 40:56:52:d8:f9:98:da:54:95:41:78:2b:b5:65:ff:
> 49:b3:2a:e5:f6:d8:14:bf:38:a8:78:c5:98:c6:30:
> 7f:c2:94:a0:15:1f:26:81:b8:a3:9d:6a:ce:0a:73:
> 2f:eb:6c:b5:14:a0:3d:e8:d9:22:90:a8:f1:49:92:
> 31:a9:45:d6:98:c8:46:1c:de:13:15:6d:9b:fc:5b:
> 09:ca:3b:f0:d2:7a:77:0f:3a:b3:fe:db:37:5b:d4:
> 13:5c:8d:cb:57:f9:d6:08:81
>
> It can be seen that for the first RSA private key, the numbers dp and c
> (both styled in bold) are not right. This could be potentially dangerous
> when signatures are involved.
>
> Because of the Chinese Remainder Theorem, if one uses the incorrectly
> computed private key to sign a message and sends the resulting wrong
> signature directly without first verifying it, this can lead to the
> so-called "fault attack" (e.g., explained here
> <https://www.cryptologie.net/article/371/fault-attacks-on-rsas-signatures/>
> ).
>
> Continuing with the example above, if we sign the message Hello World\n
> using the first RSA private key:
>
> // somehow this prime number leads to incorrect dp, which is quite weird
> OCT_fromHex(&P, "f55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed");
>
> OCT_fromHex(&Q,"f34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363");
>
> // use our choice of P and Q
> RSA_2048_KEY_PAIR(NULL,3,&priv,&pub,&P,&Q);
>
> print_keypair(&priv, &pub);
>
> printf("Signing message\n");
> OCT_jstring(&M,(char *)"Hello World\n");
> PKCS15(HASH_TYPE_RSA_2048,&M,&C);
>
> RSA_2048_DECRYPT(&priv,&C,&S); /* create signature in S */
>
> printf("Signature= ");
> OCT_output(&S);
>
> We get a faulty signature: s =
> 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
> .
>
> If an attacker obtains this signature, and knows the choice of hash
> algorithm as well as the message of Hello World\n, then he/she can also
> compute the hash value and prepare the PKCS1v1.5 structure accordingly: r
> =
> 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
>
> And now the attacker can factorize the modulus n by computing GCD(s^e -
> r, n). In Python3:
>
> import math
> # faulty signatures = 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
> # precomputed PKCS1v1.5 structurer = 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
> # modulusn = 0xf55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed * 0xf34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
> # now factorize np1 = math.gcd(pow(s,3,n)-r, n)p2 = n // p1
> print("p=",hex(max(p1,p2)))print("q=",hex(min(p1,p2)))
>
> Which outputs
> ------------------------------
>
> p=
> 0xf55d1f9bde519d3d30e9a0820a0850f3a6d05efa82dcc7ef948a09834224ad6fb77ee792f53b1ee6efc0270e5e8fdd59fab40ada833df0ea69eb2f0b37d717a3fd0329280cd47525fd22970c8fd860c7be749482c42a6e16a4b5b7edf27e01214f55fd5839a11c86be6bbd8b01e5ddeef743bca59e896c1751145d4b6a9da9ed
> q=
> 0xf34e8af7a6b3b2d2906890db77f56b8c4c5a5af42bec8cacd79fc479606d85eeb641ec95aef73d0e7e24f0da9115a524d9787d573ed3a1868bff2765cc5b9e1238a7931bdd5afd51a0bd6bf93f26909fff6d93d335e6785959c14f851205ad4a8ce995e6f92e490e7a5feab2b9fcda0080bedc17fa816227f2d6034044c0c363
> ------------------------------
>
> —
> You are receiving this because you are subscribed to this thread.
> Reply to this email directly, view it on GitHub
> <https://github.com/apache/incubator-milagro-crypto-c/issues/90>, or
> unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AAU3ZDS5XX6G5MYXVIHHFA3TF3A7LANCNFSM4Z5WGOQA>
> .
>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@milagro.apache.org
For additional commands, e-mail: issues-help@milagro.apache.org
[GitHub] [incubator-milagro-crypto-c] 1one-w01f commented on issue #90: Incorrect RSA private key generation and potentials for fault attack
Posted by GitBox <gi...@apache.org>.
1one-w01f commented on issue #90:
URL: https://github.com/apache/incubator-milagro-crypto-c/issues/90#issuecomment-808910920
Thanks for the explanation, that is rather interesting. I didn't think of the reason why when I first posted the question, but it makes sense to me now.
If one plans to stick with milagro alone then it's probably going to be fine, but if one needs to integrate/interoperate with other systems/crypto libraries as I did in my case, then this might not be ideal, because the API call doesn't seem to block/warn about 1 mod 4 primes, which might increase the chance of fault attack due to oversight in API usage, as explained in my original question.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@milagro.apache.org
For additional commands, e-mail: issues-help@milagro.apache.org
[GitHub] [incubator-milagro-crypto-c] mcarrickscott commented on issue #90: Incorrect RSA private key generation and potentials for fault attack
Posted by GitBox <gi...@apache.org>.
mcarrickscott commented on issue #90:
URL: https://github.com/apache/incubator-milagro-crypto-c/issues/90#issuecomment-808917572
You make a fair point.
A particular solution in the context of interoperation with OpenSSL would
be to copy p,q,dp, dq and c directly into the Milagro rsa_private_key
structure. Then everything should work fine (and even a little faster)
(But note that since c= 1/p mod q in Milagro and 1/q mod p in OpenSSL, p
and q (and dp and dq) would need to be swapped.)
Mike
On Sun, Mar 28, 2021 at 4:19 PM 1one.w01f ***@***.***> wrote:
> Thanks for the explanation, that is rather interesting. I didn't think of
> the reason why when I first posted the question, but it makes sense to me
> now.
>
> If one plans to stick with milagro alone then it's probably going to be
> fine, but if one needs to integrate/interoperate with other systems/crypto
> libraries as I did in my case, then this might not be ideal, because the
> API call doesn't seem to block/warn about 1 mod 4 primes, which might
> increase the chance of fault attack due to oversight in API usage, as
> explained in my original question.
>
> —
> You are receiving this because you commented.
> Reply to this email directly, view it on GitHub
> <https://github.com/apache/incubator-milagro-crypto-c/issues/90#issuecomment-808910920>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AAU3ZDTHZ6TMHXB2TRAYPKTTF5CGLANCNFSM4Z5WGOQA>
> .
>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@milagro.apache.org
For additional commands, e-mail: issues-help@milagro.apache.org