nhyatt/putty-source
1
0
Fork 0
mirror of https://git.tartarus.org/simon/putty.git synced 2025-01-09 17:38:00 +00:00
Code Issues Projects Releases Wiki Activity
pre-0.83
putty-source/crypto/aes.h

161 lines
7.4 KiB
C
Raw Permalink Normal View History

Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
/*
* Definitions likely to be helpful to multiple AES implementations.
*/
/*
* The 'extra' structure used by AES implementations is used to
* include information about how to check if a given implementation is
* available at run time, and whether we've already checked.
*/
struct aes_extra_mutable;
struct aes_extra {
/* Function to check availability. Might be expensive, so we don't
* want to call it more than once. */
bool (*check_available)(void);
/* Point to a writable substructure. */
struct aes_extra_mutable *mut;
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
/* Extra API function specific to AES, to encrypt a single block
* in ECB mode without touching the IV. Used by AES-GCM MAC
* setup. */
void (*encrypt_ecb_block)(ssh_cipher *, void *);
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
};
struct aes_extra_mutable {
bool checked_availability;
bool is_available;
};
static inline bool check_availability(const struct aes_extra *extra)
{
if (!extra->mut->checked_availability) {
extra->mut->is_available = extra->check_available();
extra->mut->checked_availability = true;
}
return extra->mut->is_available;
}
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
/* Shared stub function for all the AES-GCM vtables. */
void aesgcm_cipher_crypt_length(
ssh_cipher *cipher, void *blk, int len, unsigned long seq);
/* External entry point for the encrypt_ecb_block function. */
static inline void aes_encrypt_ecb_block(ssh_cipher *ciph, void *blk)
{
const struct aes_extra *extra = ciph->vt->extra;
extra->encrypt_ecb_block(ciph, blk);
}
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
/*
* Macros to define vtables for AES variants. There are a lot of
* these, because of the cross product between cipher modes, key
* sizes, and assorted HW/SW implementations, so it's worth spending
* some effort here to reduce the boilerplate in the sub-files.
*/
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
#define AES_EXTRA_BITS(impl_c, bits) \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
static struct aes_extra_mutable aes ## impl_c ## _extra_mut; \
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
static const struct aes_extra aes ## bits ## impl_c ## _extra = { \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
.check_available = aes ## impl_c ## _available, \
.mut = &aes ## impl_c ## _extra_mut, \
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
.encrypt_ecb_block = &aes ## bits ## impl_c ## _encrypt_ecb_block, \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
}
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
#define AES_EXTRA(impl_c) \
AES_EXTRA_BITS(impl_c, 128); \
AES_EXTRA_BITS(impl_c, 192); \
AES_EXTRA_BITS(impl_c, 256)
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
#define AES_CBC_VTABLE(impl_c, impl_display, bits) \
const ssh_cipheralg ssh_aes ## bits ## _cbc ## impl_c = { \
.new = aes ## impl_c ## _new, \
.free = aes ## impl_c ## _free, \
.setiv = aes ## impl_c ## _setiv_cbc, \
.setkey = aes ## impl_c ## _setkey, \
.encrypt = aes ## bits ## impl_c ## _cbc_encrypt, \
.decrypt = aes ## bits ## impl_c ## _cbc_decrypt, \
Add 'next_message' methods to cipher and MAC vtables. This provides a convenient hook to be called between SSH messages, for the crypto components to do any per-message processing like incrementing a sequence number.
2022-08-16 17:27:06 +00:00
.next_message = nullcipher_next_message, \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
.ssh2_id = "aes" #bits "-cbc", \
.blksize = 16, \
.real_keybits = bits, \
.padded_keybytes = bits/8, \
.flags = SSH_CIPHER_IS_CBC, \
.text_name = "AES-" #bits " CBC (" impl_display ")", \
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
.extra = &aes ## bits ## impl_c ## _extra, \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
}
#define AES_SDCTR_VTABLE(impl_c, impl_display, bits) \
const ssh_cipheralg ssh_aes ## bits ## _sdctr ## impl_c = { \
.new = aes ## impl_c ## _new, \
.free = aes ## impl_c ## _free, \
.setiv = aes ## impl_c ## _setiv_sdctr, \
.setkey = aes ## impl_c ## _setkey, \
.encrypt = aes ## bits ## impl_c ## _sdctr, \
.decrypt = aes ## bits ## impl_c ## _sdctr, \
Add 'next_message' methods to cipher and MAC vtables. This provides a convenient hook to be called between SSH messages, for the crypto components to do any per-message processing like incrementing a sequence number.
2022-08-16 17:27:06 +00:00
.next_message = nullcipher_next_message, \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
.ssh2_id = "aes" #bits "-ctr", \
.blksize = 16, \
.real_keybits = bits, \
.padded_keybytes = bits/8, \
.flags = 0, \
.text_name = "AES-" #bits " SDCTR (" impl_display ")", \
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
.extra = &aes ## bits ## impl_c ## _extra, \
}
#define AES_GCM_VTABLE(impl_c, impl_display, bits) \
const ssh_cipheralg ssh_aes ## bits ## _gcm ## impl_c = { \
.new = aes ## impl_c ## _new, \
.free = aes ## impl_c ## _free, \
.setiv = aes ## impl_c ## _setiv_gcm, \
.setkey = aes ## impl_c ## _setkey, \
.encrypt = aes ## bits ## impl_c ## _gcm, \
.decrypt = aes ## bits ## impl_c ## _gcm, \
.encrypt_length = aesgcm_cipher_crypt_length, \
.decrypt_length = aesgcm_cipher_crypt_length, \
.next_message = aes ## impl_c ## _next_message_gcm, \
/* 192-bit AES-GCM is included only so that testcrypt can run \
* standard test vectors against it. OpenSSH doesn't define a \
* protocol id for it. So we set its ssh2_id to NULL. */ \
.ssh2_id = bits==192 ? NULL : "aes" #bits "-gcm@openssh.com", \
.blksize = 16, \
.real_keybits = bits, \
.padded_keybytes = bits/8, \
.flags = SSH_CIPHER_SEPARATE_LENGTH, \
.text_name = "AES-" #bits " GCM (" impl_display ")", \
.required_mac = &ssh2_aesgcm_mac, \
.extra = &aes ## bits ## impl_c ## _extra, \
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
}
#define AES_ALL_VTABLES(impl_c, impl_display) \
AES_CBC_VTABLE(impl_c, impl_display, 128); \
AES_CBC_VTABLE(impl_c, impl_display, 192); \
AES_CBC_VTABLE(impl_c, impl_display, 256); \
AES_SDCTR_VTABLE(impl_c, impl_display, 128); \
AES_SDCTR_VTABLE(impl_c, impl_display, 192); \
Implement AES-GCM using the @openssh.com protocol IDs. I only recently found out that OpenSSH defined their own protocol IDs for AES-GCM, defined to work the same as the standard ones except that they fixed the semantics for how you select the linked cipher+MAC pair during key exchange. (RFC 5647 defines protocol ids for AES-GCM in both the cipher and MAC namespaces, and requires that you MUST select both or neither - but this contradicts the selection policy set out in the base SSH RFCs, and there's no discussion of how you resolve a conflict between them! OpenSSH's answer is to do it the same way ChaCha20-Poly1305 works, because that will ensure the two suites don't fight.) People do occasionally ask us for this linked cipher/MAC pair, and now I know it's actually feasible, I've implemented it, including a pair of vector implementations for x86 and Arm using their respective architecture extensions for multiplying polynomials over GF(2). Unlike ChaCha20-Poly1305, I've kept the cipher and MAC implementations in separate objects, with an arm's-length link between them that the MAC uses when it needs to encrypt single cipher blocks to use as the inputs to the MAC algorithm. That enables the cipher and the MAC to be independently selected from their hardware-accelerated versions, just in case someone runs on a system that has polynomial multiplication instructions but not AES acceleration, or vice versa. There's a fourth implementation of the GCM MAC, which is a pure software implementation of the same algorithm used in the vectorised versions. It's too slow to use live, but I've kept it in the code for future testing needs, and because it's a convenient place to dump my design comments. The vectorised implementations are fairly crude as far as optimisation goes. I'm sure serious x86 _or_ Arm optimisation engineers would look at them and laugh. But GCM is a fast MAC compared to HMAC-SHA-256 (indeed compared to HMAC-anything-at-all), so it should at least be good enough to use. And we've got a working version with some tests now, so if someone else wants to improve them, they can.
2022-08-16 17:36:58 +00:00
AES_SDCTR_VTABLE(impl_c, impl_display, 256); \
AES_GCM_VTABLE(impl_c, impl_display, 128); \
AES_GCM_VTABLE(impl_c, impl_display, 192); \
AES_GCM_VTABLE(impl_c, impl_display, 256)
Break up crypto modules containing HW acceleration. This applies to all of AES, SHA-1, SHA-256 and SHA-512. All those source files previously contained multiple implementations of the algorithm, enabled or disabled by ifdefs detecting whether they would work on a given compiler. And in order to get advanced machine instructions like AES-NI or NEON crypto into the output file when the compile flags hadn't enabled them, we had to do nasty stuff with compiler-specific pragmas or attributes. Now we can do the detection at cmake time, and enable advanced instructions in the more sensible way, by compile-time flags. So I've broken up each of these modules into lots of sub-pieces: a file called (e.g.) 'foo-common.c' containing common definitions across all implementations (such as round constants), one called 'foo-select.c' containing the top-level vtable(s), and a separate file for each implementation exporting just the vtable(s) for that implementation. One advantage of this is that it depends a lot less on compiler- specific bodgery. My particular least favourite part of the previous setup was the part where I had to _manually_ define some Arm ACLE feature macros before including <arm_neon.h>, so that it would define the intrinsics I wanted. Now I'm enabling interesting architecture features in the normal way, on the compiler command line, there's no need for that kind of trick: the right feature macros are already defined and <arm_neon.h> does the right thing. Another change in this reorganisation is that I've stopped assuming there's just one hardware implementation per platform. Previously, the accelerated vtables were called things like sha256_hw, and varied between FOO-NI and NEON depending on platform; and the selection code would simply ask 'is hw available? if so, use hw, else sw'. Now, each HW acceleration strategy names its vtable its own way, and the selection vtable has a whole list of possibilities to iterate over looking for a supported one. So if someone feels like writing a second accelerated implementation of something for a given platform - for example, I've heard you can use plain NEON to speed up AES somewhat even without the crypto extension - then it will now have somewhere to drop in alongside the existing ones.
2021-04-19 05:42:12 +00:00
/*
* Macros to repeat a piece of code particular numbers of times that
* correspond to 1 fewer than the number of AES rounds. (Because the
* last round is different.)
*/
#define REP2(x) x x
#define REP4(x) REP2(REP2(x))
#define REP8(x) REP2(REP4(x))
#define REP9(x) REP8(x) x
#define REP11(x) REP8(x) REP2(x) x
#define REP13(x) REP8(x) REP4(x) x
/*
* The round constants used in key schedule expansion.
*/
extern const uint8_t aes_key_setup_round_constants[10];
/*
* The largest number of round keys ever needed.
*/
#define MAXROUNDKEYS 15
Reference in New Issue Copy Permalink