pool/openssl-3
SHA256
2
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Accepting request 1120051 from security:tls:unstable

Browse Source 
OBS-URL: https://build.opensuse.org/request/show/1120051
OBS-URL: https://build.opensuse.org/package/show/security:tls/openssl-3?expand=0&rev=76
This commit is contained in:
Otto Hollmann 2023-10-24 16:01:48 +00:00 committed by Git OBS Bridge
parent 6fa380a834
commit 69a932c2a0
15 changed files with 3554 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
openssl-3.1.3.tar.gz
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:f0316a2ebd89e7f2352976445458689f80302093788c466692fb2a188b2eacf6
size 15561739

16
openssl-3.1.3.tar.gz.asc
View File

@ -1,16 +0,0 @@
-----BEGIN PGP SIGNATURE-----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=S8rI
-----END PGP SIGNATURE-----

BIN
openssl-3.1.4.tar.gz (Stored with Git LFS) Normal file
View File

Binary file not shown.

16
openssl-3.1.4.tar.gz.asc Normal file
View File

@ -0,0 +1,16 @@
-----BEGIN PGP SIGNATURE-----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=EH33
-----END PGP SIGNATURE-----

30
openssl-3.changes
View File

@ -1,3 +1,33 @@
-------------------------------------------------------------------
Tue Oct 24 14:53:41 UTC 2023 - Otto Hollmann <otto.hollmann@suse.com>
- Update to 3.1.4:
* Fix incorrect key and IV resizing issues when calling
EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() or EVP_CipherInit_ex2()
with OSSL_PARAM parameters that alter the key or IV length
[bsc#1216163, CVE-2023-5363].
-------------------------------------------------------------------
Thu Oct 19 15:03:14 UTC 2023 - Otto Hollmann <otto.hollmann@suse.com>
- Performance enhancements for cryptography from OpenSSL 3.2
[jsc#PED-5086, jsc#PED-3514]
* Add patches:
- openssl-ec-Use-static-linkage-on-nistp521-felem_-square-mul-.patch
- openssl-ec-56-bit-Limb-Solinas-Strategy-for-secp384r1.patch
- openssl-ec-powerpc64le-Add-asm-implementation-of-felem_-squa.patch
- openssl-ecc-Remove-extraneous-parentheses-in-secp384r1.patch
- openssl-powerpc-ecc-Fix-stack-allocation-secp384r1-asm.patch
- openssl-Improve-performance-for-6x-unrolling-with-vpermxor-i.patch
-------------------------------------------------------------------
Thu Oct 19 11:53:29 UTC 2023 - Pedro Monreal <pmonreal@suse.com>
- FIPS: Add the FIPS_mode() compatibility macro and flag support.
* Add patches:
- openssl-Add-FIPS_mode-compatibility-macro.patch
- openssl-Add-Kernel-FIPS-mode-flag-support.patch
-------------------------------------------------------------------
Thu Oct 12 09:44:19 UTC 2023 - <jengelh@inai.de>

13
openssl-3.spec
View File

@ -22,7 +22,7 @@
%define man_suffix 3ssl
Name: openssl-3
# Don't forget to update the version in the "openssl" meta-package!
Version: 3.1.3
Version: 3.1.4
Release: 0
Summary: Secure Sockets and Transport Layer Security
License: Apache-2.0
@ -48,6 +48,17 @@ Patch7: openssl-Add-support-for-PROFILE-SYSTEM-system-default-cipher.pat
Patch8: openssl-Override-default-paths-for-the-CA-directory-tree.patch
# PATCH-FIX-UPSTREAM: bsc#1209430 Upgrade OpenSSL from 3.0.8 to 3.1.0 in TW
Patch9: openssl-Add_support_for_Windows_CA_certificate_store.patch
# PATCH-FIX-FEDORA Add FIPS_mode compatibility macro and flag support
Patch10: openssl-Add-FIPS_mode-compatibility-macro.patch
Patch11: openssl-Add-Kernel-FIPS-mode-flag-support.patch
# PATCH-FIX-UPSTREAM jsc#PED-5086, jsc#PED-3514
# POWER10 performance enhancements for cryptography
Patch12: openssl-ec-Use-static-linkage-on-nistp521-felem_-square-mul-.patch
Patch13: openssl-ec-56-bit-Limb-Solinas-Strategy-for-secp384r1.patch
Patch14: openssl-ec-powerpc64le-Add-asm-implementation-of-felem_-squa.patch
Patch15: openssl-ecc-Remove-extraneous-parentheses-in-secp384r1.patch
Patch16: openssl-powerpc-ecc-Fix-stack-allocation-secp384r1-asm.patch
Patch17: openssl-Improve-performance-for-6x-unrolling-with-vpermxor-i.patch
BuildRequires: pkgconfig
BuildRequires: pkgconfig(zlib)
Requires: libopenssl3 = %{version}-%{release}

83
openssl-Add-FIPS_mode-compatibility-macro.patch Normal file
View File

@ -0,0 +1,83 @@
From 8e29a10b39a649d751870eb1fd1b8c388e66acc3 Mon Sep 17 00:00:00 2001
From: rpm-build <rpm-build>
Date: Mon, 31 Jul 2023 09:41:27 +0200
Subject: [PATCH 08/35] 0008-Add-FIPS_mode-compatibility-macro.patch
Patch-name: 0008-Add-FIPS_mode-compatibility-macro.patch
Patch-id: 8
Patch-status: |
# Add FIPS_mode() compatibility macro
From-dist-git-commit: 9409bc7044cf4b5773639cce20f51399888c45fd
---
include/openssl/fips.h | 26 ++++++++++++++++++++++++++
test/property_test.c | 14 ++++++++++++++
2 files changed, 40 insertions(+)
create mode 100644 include/openssl/fips.h
diff --git a/include/openssl/fips.h b/include/openssl/fips.h
new file mode 100644
index 0000000000..4162cbf88e
--- /dev/null
+++ b/include/openssl/fips.h
@@ -0,0 +1,26 @@
+/*
+ * Copyright 2016-2020 The OpenSSL Project Authors. All Rights Reserved.
+ *
+ * Licensed under the Apache License 2.0 (the "License"). You may not use
+ * this file except in compliance with the License. You can obtain a copy
+ * in the file LICENSE in the source distribution or at
+ * https://www.openssl.org/source/license.html
+ */
+
+#ifndef OPENSSL_FIPS_H
+# define OPENSSL_FIPS_H
+# pragma once
+
+# include <openssl/evp.h>
+# include <openssl/macros.h>
+
+# ifdef __cplusplus
+extern "C" {
+# endif
+
+# define FIPS_mode() EVP_default_properties_is_fips_enabled(NULL)
+
+# ifdef __cplusplus
+}
+# endif
+#endif
diff --git a/test/property_test.c b/test/property_test.c
index 45b1db3e85..8894c1c1cb 100644
--- a/test/property_test.c
+++ b/test/property_test.c
@@ -677,6 +677,19 @@ static int test_property_list_to_string(int i)
return ret;
}
+#include <openssl/fips.h>
+static int test_downstream_FIPS_mode(void)
+{
+ int ret = 0;
+
+ ret = TEST_true(EVP_set_default_properties(NULL, "fips=yes"))
+ && TEST_true(FIPS_mode())
+ && TEST_true(EVP_set_default_properties(NULL, "fips=no"))
+ && TEST_false(FIPS_mode());
+
+ return ret;
+}
+
int setup_tests(void)
{
ADD_TEST(test_property_string);
@@ -690,6 +703,7 @@ int setup_tests(void)
ADD_TEST(test_property);
ADD_TEST(test_query_cache_stochastic);
ADD_TEST(test_fips_mode);
+ ADD_TEST(test_downstream_FIPS_mode);
ADD_ALL_TESTS(test_property_list_to_string, OSSL_NELEM(to_string_tests));
return 1;
}
--
2.41.0

86
openssl-Add-Kernel-FIPS-mode-flag-support.patch Normal file
View File

@ -0,0 +1,86 @@
From aa3aebf132959e7e44876042efaf9ff24ffe0f2b Mon Sep 17 00:00:00 2001
From: rpm-build <rpm-build>
Date: Mon, 31 Jul 2023 09:41:27 +0200
Subject: [PATCH 09/35] 0009-Add-Kernel-FIPS-mode-flag-support.patch
Patch-name: 0009-Add-Kernel-FIPS-mode-flag-support.patch
Patch-id: 9
Patch-status: |
# Add check to see if fips flag is enabled in kernel
From-dist-git-commit: 9409bc7044cf4b5773639cce20f51399888c45fd
---
crypto/context.c | 36 ++++++++++++++++++++++++++++++++++++
include/internal/provider.h | 3 +++
2 files changed, 39 insertions(+)
diff --git a/crypto/context.c b/crypto/context.c
index e294ea1512..51002ba79a 100644
--- a/crypto/context.c
+++ b/crypto/context.c
@@ -16,6 +16,41 @@
#include "internal/provider.h"
#include "crypto/context.h"
+# include <sys/types.h>
+# include <sys/stat.h>
+# include <fcntl.h>
+# include <unistd.h>
+# include <openssl/evp.h>
+
+# define FIPS_MODE_SWITCH_FILE "/proc/sys/crypto/fips_enabled"
+
+static int kernel_fips_flag;
+
+static void read_kernel_fips_flag(void)
+{
+ char buf[2] = "0";
+ int fd;
+
+ if (ossl_safe_getenv("OPENSSL_FORCE_FIPS_MODE") != NULL) {
+ buf[0] = '1';
+ } else if ((fd = open(FIPS_MODE_SWITCH_FILE, O_RDONLY)) >= 0) {
+ while (read(fd, buf, sizeof(buf)) < 0 && errno == EINTR) ;
+ close(fd);
+ }
+
+ if (buf[0] == '1') {
+ kernel_fips_flag = 1;
+ }
+
+ return;
+}
+
+int ossl_get_kernel_fips_flag()
+{
+ return kernel_fips_flag;
+}
+
+
struct ossl_lib_ctx_st {
CRYPTO_RWLOCK *lock, *rand_crngt_lock;
OSSL_EX_DATA_GLOBAL global;
@@ -336,6 +371,7 @@ static int default_context_inited = 0;
DEFINE_RUN_ONCE_STATIC(default_context_do_init)
{
+ read_kernel_fips_flag();
if (!CRYPTO_THREAD_init_local(&default_context_thread_local, NULL))
goto err;
diff --git a/include/internal/provider.h b/include/internal/provider.h
index 18937f84c7..1446bf7afb 100644
--- a/include/internal/provider.h
+++ b/include/internal/provider.h
@@ -112,6 +112,9 @@ int ossl_provider_init_as_child(OSSL_LIB_CTX *ctx,
const OSSL_DISPATCH *in);
void ossl_provider_deinit_child(OSSL_LIB_CTX *ctx);
+/* FIPS flag access */
+int ossl_get_kernel_fips_flag(void);
+
# ifdef __cplusplus
}
# endif
--
2.41.0

495
openssl-Improve-performance-for-6x-unrolling-with-vpermxor-i.patch Normal file
View File

@ -0,0 +1,495 @@
From 3d3a7ecd1ae5ab08d22041f7b3b035c34f12fa02 Mon Sep 17 00:00:00 2001
From: Danny Tsen <dtsen@linux.ibm.com>
Date: Tue, 22 Aug 2023 15:58:53 -0400
Subject: [PATCH] Improve performance for 6x unrolling with vpermxor
instruction
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21812)
---
crypto/aes/asm/aesp8-ppc.pl | 145 +++++++++++++++++++++++-------------
1 file changed, 95 insertions(+), 50 deletions(-)
diff --git a/crypto/aes/asm/aesp8-ppc.pl b/crypto/aes/asm/aesp8-ppc.pl
index 60cf86f52aed2..38b9405a283b7 100755
--- a/crypto/aes/asm/aesp8-ppc.pl
+++ b/crypto/aes/asm/aesp8-ppc.pl
@@ -99,11 +99,12 @@
.long 0x1b000000, 0x1b000000, 0x1b000000, 0x1b000000 ?rev
.long 0x0d0e0f0c, 0x0d0e0f0c, 0x0d0e0f0c, 0x0d0e0f0c ?rev
.long 0,0,0,0 ?asis
+.long 0x0f102132, 0x43546576, 0x8798a9ba, 0xcbdcedfe
Lconsts:
mflr r0
bcl 20,31,\$+4
mflr $ptr #vvvvv "distance between . and rcon
- addi $ptr,$ptr,-0x48
+ addi $ptr,$ptr,-0x58
mtlr r0
blr
.long 0
@@ -2405,7 +2406,7 @@ ()
my $key_=$key2;
my ($x00,$x10,$x20,$x30,$x40,$x50,$x60,$x70)=map("r$_",(0,3,26..31));
$x00=0 if ($flavour =~ /osx/);
-my ($in0, $in1, $in2, $in3, $in4, $in5 )=map("v$_",(0..5));
+my ($in0, $in1, $in2, $in3, $in4, $in5)=map("v$_",(0..5));
my ($out0, $out1, $out2, $out3, $out4, $out5)=map("v$_",(7,12..16));
my ($twk0, $twk1, $twk2, $twk3, $twk4, $twk5)=map("v$_",(17..22));
my $rndkey0="v23"; # v24-v25 rotating buffer for first found keys
@@ -2460,6 +2461,18 @@ ()
li $x70,0x70
mtspr 256,r0
+ # Reverse eighty7 to 0x010101..87
+ xxlor 2, 32+$eighty7, 32+$eighty7
+ vsldoi $eighty7,$tmp,$eighty7,1 # 0x010101..87
+ xxlor 1, 32+$eighty7, 32+$eighty7
+
+ # Load XOR contents. 0xf102132435465768798a9bacbdcedfe
+ mr $x70, r6
+ bl Lconsts
+ lxvw4x 0, $x40, r6 # load XOR contents
+ mr r6, $x70
+ li $x70,0x70
+
subi $rounds,$rounds,3 # -4 in total
lvx $rndkey0,$x00,$key1 # load key schedule
@@ -2502,69 +2515,77 @@ ()
?vperm v31,v31,$twk5,$keyperm
lvx v25,$x10,$key_ # pre-load round[2]
+ # Switch to use the following codes with 0x010101..87 to generate tweak.
+ # eighty7 = 0x010101..87
+ # vsrab tmp, tweak, seven # next tweak value, right shift 7 bits
+ # vand tmp, tmp, eighty7 # last byte with carry
+ # vaddubm tweak, tweak, tweak # left shift 1 bit (x2)
+ # xxlor vsx, 0, 0
+ # vpermxor tweak, tweak, tmp, vsx
+
vperm $in0,$inout,$inptail,$inpperm
subi $inp,$inp,31 # undo "caller"
vxor $twk0,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vand $tmp,$tmp,$eighty7
vxor $out0,$in0,$twk0
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in1, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in1
lvx_u $in1,$x10,$inp
vxor $twk1,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in1,$in1,$in1,$leperm
vand $tmp,$tmp,$eighty7
vxor $out1,$in1,$twk1
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in2, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in2
lvx_u $in2,$x20,$inp
andi. $taillen,$len,15
vxor $twk2,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in2,$in2,$in2,$leperm
vand $tmp,$tmp,$eighty7
vxor $out2,$in2,$twk2
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in3, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in3
lvx_u $in3,$x30,$inp
sub $len,$len,$taillen
vxor $twk3,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in3,$in3,$in3,$leperm
vand $tmp,$tmp,$eighty7
vxor $out3,$in3,$twk3
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in4, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in4
lvx_u $in4,$x40,$inp
subi $len,$len,0x60
vxor $twk4,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in4,$in4,$in4,$leperm
vand $tmp,$tmp,$eighty7
vxor $out4,$in4,$twk4
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in5, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in5
lvx_u $in5,$x50,$inp
addi $inp,$inp,0x60
vxor $twk5,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in5,$in5,$in5,$leperm
vand $tmp,$tmp,$eighty7
vxor $out5,$in5,$twk5
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in0, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in0
vxor v31,v31,$rndkey0
mtctr $rounds
@@ -2590,6 +2611,8 @@ ()
lvx v25,$x10,$key_ # round[4]
bdnz Loop_xts_enc6x
+ xxlor 32+$eighty7, 1, 1 # 0x010101..87
+
subic $len,$len,96 # $len-=96
vxor $in0,$twk0,v31 # xor with last round key
vcipher $out0,$out0,v24
@@ -2599,7 +2622,6 @@ ()
vaddubm $tweak,$tweak,$tweak
vcipher $out2,$out2,v24
vcipher $out3,$out3,v24
- vsldoi $tmp,$tmp,$tmp,15
vcipher $out4,$out4,v24
vcipher $out5,$out5,v24
@@ -2607,7 +2629,8 @@ ()
vand $tmp,$tmp,$eighty7
vcipher $out0,$out0,v25
vcipher $out1,$out1,v25
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in1, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in1
vcipher $out2,$out2,v25
vcipher $out3,$out3,v25
vxor $in1,$twk1,v31
@@ -2618,13 +2641,13 @@ ()
and r0,r0,$len
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vcipher $out0,$out0,v26
vcipher $out1,$out1,v26
vand $tmp,$tmp,$eighty7
vcipher $out2,$out2,v26
vcipher $out3,$out3,v26
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in2, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in2
vcipher $out4,$out4,v26
vcipher $out5,$out5,v26
@@ -2638,7 +2661,6 @@ ()
vaddubm $tweak,$tweak,$tweak
vcipher $out0,$out0,v27
vcipher $out1,$out1,v27
- vsldoi $tmp,$tmp,$tmp,15
vcipher $out2,$out2,v27
vcipher $out3,$out3,v27
vand $tmp,$tmp,$eighty7
@@ -2646,7 +2668,8 @@ ()
vcipher $out5,$out5,v27
addi $key_,$sp,$FRAME+15 # rewind $key_
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in3, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in3
vcipher $out0,$out0,v28
vcipher $out1,$out1,v28
vxor $in3,$twk3,v31
@@ -2655,7 +2678,6 @@ ()
vcipher $out2,$out2,v28
vcipher $out3,$out3,v28
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vcipher $out4,$out4,v28
vcipher $out5,$out5,v28
lvx v24,$x00,$key_ # re-pre-load round[1]
@@ -2663,7 +2685,8 @@ ()
vcipher $out0,$out0,v29
vcipher $out1,$out1,v29
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in4, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in4
vcipher $out2,$out2,v29
vcipher $out3,$out3,v29
vxor $in4,$twk4,v31
@@ -2673,14 +2696,14 @@ ()
vcipher $out5,$out5,v29
lvx v25,$x10,$key_ # re-pre-load round[2]
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vcipher $out0,$out0,v30
vcipher $out1,$out1,v30
vand $tmp,$tmp,$eighty7
vcipher $out2,$out2,v30
vcipher $out3,$out3,v30
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in5, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in5
vcipher $out4,$out4,v30
vcipher $out5,$out5,v30
vxor $in5,$twk5,v31
@@ -2690,7 +2713,6 @@ ()
vcipherlast $out0,$out0,$in0
lvx_u $in0,$x00,$inp # load next input block
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vcipherlast $out1,$out1,$in1
lvx_u $in1,$x10,$inp
vcipherlast $out2,$out2,$in2
@@ -2703,7 +2725,10 @@ ()
vcipherlast $out4,$out4,$in4
le?vperm $in2,$in2,$in2,$leperm
lvx_u $in4,$x40,$inp
- vxor $tweak,$tweak,$tmp
+ xxlor 10, 32+$in0, 32+$in0
+ xxlor 32+$in0, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in0
+ xxlor 32+$in0, 10, 10
vcipherlast $tmp,$out5,$in5 # last block might be needed
# in stealing mode
le?vperm $in3,$in3,$in3,$leperm
@@ -2736,6 +2761,8 @@ ()
mtctr $rounds
beq Loop_xts_enc6x # did $len-=96 borrow?
+ xxlor 32+$eighty7, 2, 2 # 0x870101..01
+
addic. $len,$len,0x60
beq Lxts_enc6x_zero
cmpwi $len,0x20
@@ -3112,6 +3139,18 @@ ()
li $x70,0x70
mtspr 256,r0
+ # Reverse eighty7 to 0x010101..87
+ xxlor 2, 32+$eighty7, 32+$eighty7
+ vsldoi $eighty7,$tmp,$eighty7,1 # 0x010101..87
+ xxlor 1, 32+$eighty7, 32+$eighty7
+
+ # Load XOR contents. 0xf102132435465768798a9bacbdcedfe
+ mr $x70, r6
+ bl Lconsts
+ lxvw4x 0, $x40, r6 # load XOR contents
+ mr r6, $x70
+ li $x70,0x70
+
subi $rounds,$rounds,3 # -4 in total
lvx $rndkey0,$x00,$key1 # load key schedule
@@ -3159,64 +3198,64 @@ ()
vxor $twk0,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vand $tmp,$tmp,$eighty7
vxor $out0,$in0,$twk0
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in1, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in1
lvx_u $in1,$x10,$inp
vxor $twk1,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in1,$in1,$in1,$leperm
vand $tmp,$tmp,$eighty7
vxor $out1,$in1,$twk1
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in2, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in2
lvx_u $in2,$x20,$inp
andi. $taillen,$len,15
vxor $twk2,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in2,$in2,$in2,$leperm
vand $tmp,$tmp,$eighty7
vxor $out2,$in2,$twk2
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in3, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in3
lvx_u $in3,$x30,$inp
sub $len,$len,$taillen
vxor $twk3,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in3,$in3,$in3,$leperm
vand $tmp,$tmp,$eighty7
vxor $out3,$in3,$twk3
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in4, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in4
lvx_u $in4,$x40,$inp
subi $len,$len,0x60
vxor $twk4,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in4,$in4,$in4,$leperm
vand $tmp,$tmp,$eighty7
vxor $out4,$in4,$twk4
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in5, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in5
lvx_u $in5,$x50,$inp
addi $inp,$inp,0x60
vxor $twk5,$tweak,$rndkey0
vsrab $tmp,$tweak,$seven # next tweak value
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
le?vperm $in5,$in5,$in5,$leperm
vand $tmp,$tmp,$eighty7
vxor $out5,$in5,$twk5
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in0, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in0
vxor v31,v31,$rndkey0
mtctr $rounds
@@ -3242,6 +3281,8 @@ ()
lvx v25,$x10,$key_ # round[4]
bdnz Loop_xts_dec6x
+ xxlor 32+$eighty7, 1, 1
+
subic $len,$len,96 # $len-=96
vxor $in0,$twk0,v31 # xor with last round key
vncipher $out0,$out0,v24
@@ -3251,7 +3292,6 @@ ()
vaddubm $tweak,$tweak,$tweak
vncipher $out2,$out2,v24
vncipher $out3,$out3,v24
- vsldoi $tmp,$tmp,$tmp,15
vncipher $out4,$out4,v24
vncipher $out5,$out5,v24
@@ -3259,7 +3299,8 @@ ()
vand $tmp,$tmp,$eighty7
vncipher $out0,$out0,v25
vncipher $out1,$out1,v25
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in1, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in1
vncipher $out2,$out2,v25
vncipher $out3,$out3,v25
vxor $in1,$twk1,v31
@@ -3270,13 +3311,13 @@ ()
and r0,r0,$len
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vncipher $out0,$out0,v26
vncipher $out1,$out1,v26
vand $tmp,$tmp,$eighty7
vncipher $out2,$out2,v26
vncipher $out3,$out3,v26
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in2, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in2
vncipher $out4,$out4,v26
vncipher $out5,$out5,v26
@@ -3290,7 +3331,6 @@ ()
vaddubm $tweak,$tweak,$tweak
vncipher $out0,$out0,v27
vncipher $out1,$out1,v27
- vsldoi $tmp,$tmp,$tmp,15
vncipher $out2,$out2,v27
vncipher $out3,$out3,v27
vand $tmp,$tmp,$eighty7
@@ -3298,7 +3338,8 @@ ()
vncipher $out5,$out5,v27
addi $key_,$sp,$FRAME+15 # rewind $key_
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in3, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in3
vncipher $out0,$out0,v28
vncipher $out1,$out1,v28
vxor $in3,$twk3,v31
@@ -3307,7 +3348,6 @@ ()
vncipher $out2,$out2,v28
vncipher $out3,$out3,v28
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vncipher $out4,$out4,v28
vncipher $out5,$out5,v28
lvx v24,$x00,$key_ # re-pre-load round[1]
@@ -3315,7 +3355,8 @@ ()
vncipher $out0,$out0,v29
vncipher $out1,$out1,v29
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in4, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in4
vncipher $out2,$out2,v29
vncipher $out3,$out3,v29
vxor $in4,$twk4,v31
@@ -3325,14 +3366,14 @@ ()
vncipher $out5,$out5,v29
lvx v25,$x10,$key_ # re-pre-load round[2]
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vncipher $out0,$out0,v30
vncipher $out1,$out1,v30
vand $tmp,$tmp,$eighty7
vncipher $out2,$out2,v30
vncipher $out3,$out3,v30
- vxor $tweak,$tweak,$tmp
+ xxlor 32+$in5, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in5
vncipher $out4,$out4,v30
vncipher $out5,$out5,v30
vxor $in5,$twk5,v31
@@ -3342,7 +3383,6 @@ ()
vncipherlast $out0,$out0,$in0
lvx_u $in0,$x00,$inp # load next input block
vaddubm $tweak,$tweak,$tweak
- vsldoi $tmp,$tmp,$tmp,15
vncipherlast $out1,$out1,$in1
lvx_u $in1,$x10,$inp
vncipherlast $out2,$out2,$in2
@@ -3355,7 +3395,10 @@ ()
vncipherlast $out4,$out4,$in4
le?vperm $in2,$in2,$in2,$leperm
lvx_u $in4,$x40,$inp
- vxor $tweak,$tweak,$tmp
+ xxlor 10, 32+$in0, 32+$in0
+ xxlor 32+$in0, 0, 0
+ vpermxor $tweak, $tweak, $tmp, $in0
+ xxlor 32+$in0, 10, 10
vncipherlast $out5,$out5,$in5
le?vperm $in3,$in3,$in3,$leperm
lvx_u $in5,$x50,$inp
@@ -3386,6 +3429,8 @@ ()
mtctr $rounds
beq Loop_xts_dec6x # did $len-=96 borrow?
+ xxlor 32+$eighty7, 2, 2
+
addic. $len,$len,0x60
beq Lxts_dec6x_zero
cmpwi $len,0x20

2159
openssl-ec-56-bit-Limb-Solinas-Strategy-for-secp384r1.patch Normal file
View File

File diff suppressed because it is too large Load Diff

65
openssl-ec-Use-static-linkage-on-nistp521-felem_-square-mul-.patch Normal file
View File

@ -0,0 +1,65 @@
From 3e47a286dc3274bda72a196c3a4030a1fc8302f1 Mon Sep 17 00:00:00 2001
From: Rohan McLure <rohanmclure@linux.ibm.com>
Date: Fri, 23 Jun 2023 16:41:48 +1000
Subject: [PATCH] ec: Use static linkage on nistp521 felem_{square,mul}
wrappers
Runtime selection of implementations for felem_{square,mul} depends on
felem_{square,mul}_wrapper functions, which overwrite function points in
a similar design to that of .plt.got sections used by program loaders
during dynamic linking.
There's no reason why these functions need to have external linkage.
Mark static.
Signed-off-by: Rohan McLure <rohanmclure@linux.ibm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Todd Short <todd.short@me.com>
(Merged from https://github.com/openssl/openssl/pull/21471)
---
crypto/ec/ecp_nistp521.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/crypto/ec/ecp_nistp521.c b/crypto/ec/ecp_nistp521.c
index 97815cac1f13..32a9268ecf17 100644
--- a/crypto/ec/ecp_nistp521.c
+++ b/crypto/ec/ecp_nistp521.c
@@ -676,8 +676,8 @@ static void felem_reduce(felem out, const largefelem in)
}
#if defined(ECP_NISTP521_ASM)
-void felem_square_wrapper(largefelem out, const felem in);
-void felem_mul_wrapper(largefelem out, const felem in1, const felem in2);
+static void felem_square_wrapper(largefelem out, const felem in);
+static void felem_mul_wrapper(largefelem out, const felem in1, const felem in2);
static void (*felem_square_p)(largefelem out, const felem in) =
felem_square_wrapper;
@@ -691,7 +691,7 @@ void p521_felem_mul(largefelem out, const felem in1, const felem in2);
# include "crypto/ppc_arch.h"
# endif
-void felem_select(void)
+static void felem_select(void)
{
# if defined(_ARCH_PPC64)
if ((OPENSSL_ppccap_P & PPC_MADD300) && (OPENSSL_ppccap_P & PPC_ALTIVEC)) {
@@ -707,13 +707,13 @@ void felem_select(void)
felem_mul_p = felem_mul_ref;
}
-void felem_square_wrapper(largefelem out, const felem in)
+static void felem_square_wrapper(largefelem out, const felem in)
{
felem_select();
felem_square_p(out, in);
}
-void felem_mul_wrapper(largefelem out, const felem in1, const felem in2)
+static void felem_mul_wrapper(largefelem out, const felem in1, const felem in2)
{
felem_select();
felem_mul_p(out, in1, in2);

428
openssl-ec-powerpc64le-Add-asm-implementation-of-felem_-squa.patch Normal file
View File

@ -0,0 +1,428 @@
From 966047ee13188e8634af25af348940acceb9316d Mon Sep 17 00:00:00 2001
From: Rohan McLure <rohanmclure@linux.ibm.com>
Date: Wed, 31 May 2023 14:32:26 +1000
Subject: [PATCH] ec: powerpc64le: Add asm implementation of felem_{square,mul}
Add an assembly implementation of felem_{square,mul}, which will be
implemented whenever Altivec support is present and the core implements
ISA 3.0 (Power 9) or greater.
Signed-off-by: Rohan McLure <rohanmclure@linux.ibm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Todd Short <todd.short@me.com>
(Merged from https://github.com/openssl/openssl/pull/21471)
---
crypto/ec/asm/ecp_nistp384-ppc64.pl | 355 ++++++++++++++++++++++++++++
crypto/ec/build.info | 6 +-
crypto/ec/ecp_nistp384.c | 9 +
3 files changed, 368 insertions(+), 2 deletions(-)
create mode 100755 crypto/ec/asm/ecp_nistp384-ppc64.pl
diff --git a/crypto/ec/asm/ecp_nistp384-ppc64.pl b/crypto/ec/asm/ecp_nistp384-ppc64.pl
new file mode 100755
index 000000000000..3f86b391af69
--- /dev/null
+++ b/crypto/ec/asm/ecp_nistp384-ppc64.pl
@@ -0,0 +1,355 @@
+#! /usr/bin/env perl
+# Copyright 2023 The OpenSSL Project Authors. All Rights Reserved.
+#
+# Licensed under the Apache License 2.0 (the "License"). You may not use
+# this file except in compliance with the License. You can obtain a copy
+# in the file LICENSE in the source distribution or at
+# https://www.openssl.org/source/license.html
+#
+# ====================================================================
+# Written by Rohan McLure <rmclure@linux.ibm.com> for the OpenSSL
+# project.
+# ====================================================================
+#
+# p384 lower-level primitives for PPC64 using vector instructions.
+#
+
+use strict;
+use warnings;
+
+my $flavour = shift;
+my $output = "";
+while (($output=shift) && ($output!~/\w[\w\-]*\.\w+$/)) {}
+if (!$output) {
+ $output = "-";
+}
+
+my ($xlate, $dir);
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}ppc-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/ppc-xlate.pl" and -f $xlate) or
+die "can't locate ppc-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+my $code = "";
+
+my ($sp, $outp, $savelr, $savesp) = ("r1", "r3", "r10", "r12");
+
+my $vzero = "v32";
+
+sub startproc($)
+{
+ my ($name) = @_;
+
+ $code.=<<___;
+ .globl ${name}
+ .align 5
+${name}:
+
+___
+}
+
+sub endproc($)
+{
+ my ($name) = @_;
+
+ $code.=<<___;
+ blr
+ .size ${name},.-${name}
+
+___
+}
+
+
+sub push_vrs($$)
+{
+ my ($min, $max) = @_;
+
+ my $count = $max - $min + 1;
+
+ $code.=<<___;
+ mr $savesp,$sp
+ stdu $sp,-16*`$count+1`($sp)
+
+___
+ for (my $i = $min; $i <= $max; $i++) {
+ my $mult = $max - $i + 1;
+ $code.=<<___;
+ stxv $i,-16*$mult($savesp)
+___
+
+ }
+
+ $code.=<<___;
+
+___
+}
+
+sub pop_vrs($$)
+{
+ my ($min, $max) = @_;
+
+ $code.=<<___;
+ ld $savesp,0($sp)
+___
+ for (my $i = $min; $i <= $max; $i++) {
+ my $mult = $max - $i + 1;
+ $code.=<<___;
+ lxv $i,-16*$mult($savesp)
+___
+ }
+
+ $code.=<<___;
+ mr $sp,$savesp
+
+___
+}
+
+sub load_vrs($$)
+{
+ my ($pointer, $reg_list) = @_;
+
+ for (my $i = 0; $i <= 6; $i++) {
+ my $offset = $i * 8;
+ $code.=<<___;
+ lxsd $reg_list->[$i],$offset($pointer)
+___
+ }
+
+ $code.=<<___;
+
+___
+}
+
+sub store_vrs($$)
+{
+ my ($pointer, $reg_list) = @_;
+
+ for (my $i = 0; $i <= 12; $i++) {
+ my $offset = $i * 16;
+ $code.=<<___;
+ stxv $reg_list->[$i],$offset($pointer)
+___
+ }
+
+ $code.=<<___;
+
+___
+}
+
+$code.=<<___;
+.machine "any"
+.text
+
+___
+
+{
+ # mul/square common
+ my ($t1, $t2, $t3, $t4) = ("v33", "v34", "v42", "v43");
+ my ($zero, $one) = ("r8", "r9");
+ my $out = "v51";
+
+ {
+ #
+ # p384_felem_mul
+ #
+
+ my ($in1p, $in2p) = ("r4", "r5");
+ my @in1 = map("v$_",(44..50));
+ my @in2 = map("v$_",(35..41));
+
+ startproc("p384_felem_mul");
+
+ push_vrs(52, 63);
+
+ $code.=<<___;
+ vspltisw $vzero,0
+
+___
+
+ load_vrs($in1p, \@in1);
+ load_vrs($in2p, \@in2);
+
+ $code.=<<___;
+ vmsumudm $out,$in1[0],$in2[0],$vzero
+ stxv $out,0($outp)
+
+ xxpermdi $t1,$in1[0],$in1[1],0b00
+ xxpermdi $t2,$in2[1],$in2[0],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ stxv $out,16($outp)
+
+ xxpermdi $t2,$in2[2],$in2[1],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$in1[2],$in2[0],$out
+ stxv $out,32($outp)
+
+ xxpermdi $t2,$in2[1],$in2[0],0b00
+ xxpermdi $t3,$in1[2],$in1[3],0b00
+ xxpermdi $t4,$in2[3],$in2[2],0b00
+ vmsumudm $out,$t1,$t4,$vzero
+ vmsumudm $out,$t3,$t2,$out
+ stxv $out,48($outp)
+
+ xxpermdi $t2,$in2[4],$in2[3],0b00
+ xxpermdi $t4,$in2[2],$in2[1],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$t3,$t4,$out
+ vmsumudm $out,$in1[4],$in2[0],$out
+ stxv $out,64($outp)
+
+ xxpermdi $t2,$in2[5],$in2[4],0b00
+ xxpermdi $t4,$in2[3],$in2[2],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$t3,$t4,$out
+ xxpermdi $t4,$in2[1],$in2[0],0b00
+ xxpermdi $t1,$in1[4],$in1[5],0b00
+ vmsumudm $out,$t1,$t4,$out
+ stxv $out,80($outp)
+
+ xxpermdi $t1,$in1[0],$in1[1],0b00
+ xxpermdi $t2,$in2[6],$in2[5],0b00
+ xxpermdi $t4,$in2[4],$in2[3],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$t3,$t4,$out
+ xxpermdi $t2,$in2[2],$in2[1],0b00
+ xxpermdi $t1,$in1[4],$in1[5],0b00
+ vmsumudm $out,$t1,$t2,$out
+ vmsumudm $out,$in1[6],$in2[0],$out
+ stxv $out,96($outp)
+
+ xxpermdi $t1,$in1[1],$in1[2],0b00
+ xxpermdi $t2,$in2[6],$in2[5],0b00
+ xxpermdi $t3,$in1[3],$in1[4],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$t3,$t4,$out
+ xxpermdi $t3,$in2[2],$in2[1],0b00
+ xxpermdi $t1,$in1[5],$in1[6],0b00
+ vmsumudm $out,$t1,$t3,$out
+ stxv $out,112($outp)
+
+ xxpermdi $t1,$in1[2],$in1[3],0b00
+ xxpermdi $t3,$in1[4],$in1[5],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$t3,$t4,$out
+ vmsumudm $out,$in1[6],$in2[2],$out
+ stxv $out,128($outp)
+
+ xxpermdi $t1,$in1[3],$in1[4],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ xxpermdi $t1,$in1[5],$in1[6],0b00
+ vmsumudm $out,$t1,$t4,$out
+ stxv $out,144($outp)
+
+ vmsumudm $out,$t3,$t2,$vzero
+ vmsumudm $out,$in1[6],$in2[4],$out
+ stxv $out,160($outp)
+
+ vmsumudm $out,$t1,$t2,$vzero
+ stxv $out,176($outp)
+
+ vmsumudm $out,$in1[6],$in2[6],$vzero
+ stxv $out,192($outp)
+___
+
+ endproc("p384_felem_mul");
+ }
+
+ {
+ #
+ # p384_felem_square
+ #
+
+ my ($inp) = ("r4");
+ my @in = map("v$_",(44..50));
+ my @inx2 = map("v$_",(35..41));
+
+ startproc("p384_felem_square");
+
+ push_vrs(52, 63);
+
+ $code.=<<___;
+ vspltisw $vzero,0
+
+___
+
+ load_vrs($inp, \@in);
+
+ $code.=<<___;
+ li $zero,0
+ li $one,1
+ mtvsrdd $t1,$one,$zero
+___
+
+ for (my $i = 0; $i <= 6; $i++) {
+ $code.=<<___;
+ vsld $inx2[$i],$in[$i],$t1
+___
+ }
+
+ $code.=<<___;
+ vmsumudm $out,$in[0],$in[0],$vzero
+ stxv $out,0($outp)
+
+ vmsumudm $out,$in[0],$inx2[1],$vzero
+ stxv $out,16($outp)
+
+ vmsumudm $out,$in[0],$inx2[2],$vzero
+ vmsumudm $out,$in[1],$in[1],$out
+ stxv $out,32($outp)
+
+ xxpermdi $t1,$in[0],$in[1],0b00
+ xxpermdi $t2,$inx2[3],$inx2[2],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ stxv $out,48($outp)
+
+ xxpermdi $t4,$inx2[4],$inx2[3],0b00
+ vmsumudm $out,$t1,$t4,$vzero
+ vmsumudm $out,$in[2],$in[2],$out
+ stxv $out,64($outp)
+
+ xxpermdi $t2,$inx2[5],$inx2[4],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$in[2],$inx2[3],$out
+ stxv $out,80($outp)
+
+ xxpermdi $t2,$inx2[6],$inx2[5],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$in[2],$inx2[4],$out
+ vmsumudm $out,$in[3],$in[3],$out
+ stxv $out,96($outp)
+
+ xxpermdi $t3,$in[1],$in[2],0b00
+ vmsumudm $out,$t3,$t2,$vzero
+ vmsumudm $out,$in[3],$inx2[4],$out
+ stxv $out,112($outp)
+
+ xxpermdi $t1,$in[2],$in[3],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ vmsumudm $out,$in[4],$in[4],$out
+ stxv $out,128($outp)
+
+ xxpermdi $t1,$in[3],$in[4],0b00
+ vmsumudm $out,$t1,$t2,$vzero
+ stxv $out,144($outp)
+
+ vmsumudm $out,$in[4],$inx2[6],$vzero
+ vmsumudm $out,$in[5],$in[5],$out
+ stxv $out,160($outp)
+
+ vmsumudm $out,$in[5],$inx2[6],$vzero
+ stxv $out,176($outp)
+
+ vmsumudm $out,$in[6],$in[6],$vzero
+ stxv $out,192($outp)
+___
+
+ endproc("p384_felem_square");
+ }
+}
+
+$code =~ s/\`([^\`]*)\`/eval $1/gem;
+print $code;
+close STDOUT or die "error closing STDOUT: $!";
diff --git a/crypto/ec/build.info b/crypto/ec/build.info
index 1fa60a1deddd..4077bead7bdb 100644
--- a/crypto/ec/build.info
+++ b/crypto/ec/build.info
@@ -39,8 +39,9 @@ IF[{- !$disabled{asm} -}]
$ECASM_ppc64=ecp_nistz256.c ecp_ppc.c ecp_nistz256-ppc64.s x25519-ppc64.s
$ECDEF_ppc64=ECP_NISTZ256_ASM X25519_ASM
IF[{- !$disabled{'ec_nistp_64_gcc_128'} -}]
- $ECASM_ppc64=$ECASM_ppc64 ecp_nistp521-ppc64.s
- $ECDEF_ppc64=$ECDEF_ppc64 ECP_NISTP521_ASM
+ $ECASM_ppc64=$ECASM_ppc64 ecp_nistp384-ppc64.s ecp_nistp521-ppc64.s
+ $ECDEF_ppc64=$ECDEF_ppc64 ECP_NISTP384_ASM ECP_NISTP521_ASM
+ INCLUDE[ecp_nistp384.o]=..
INCLUDE[ecp_nistp521.o]=..
ENDIF
@@ -119,6 +120,7 @@ GENERATE[ecp_nistz256-armv8.S]=asm/ecp_nistz256-armv8.pl
INCLUDE[ecp_nistz256-armv8.o]=..
GENERATE[ecp_nistz256-ppc64.s]=asm/ecp_nistz256-ppc64.pl
+GENERATE[ecp_nistp384-ppc64.s]=asm/ecp_nistp384-ppc64.pl
GENERATE[ecp_nistp521-ppc64.s]=asm/ecp_nistp521-ppc64.pl
GENERATE[x25519-x86_64.s]=asm/x25519-x86_64.pl
diff --git a/crypto/ec/ecp_nistp384.c b/crypto/ec/ecp_nistp384.c
index a0559487ed4e..14f9530d07c6 100644
--- a/crypto/ec/ecp_nistp384.c
+++ b/crypto/ec/ecp_nistp384.c
@@ -691,6 +691,15 @@ void p384_felem_mul(widefelem out, const felem in1, const felem in2);
static void felem_select(void)
{
+# if defined(_ARCH_PPC64)
+ if ((OPENSSL_ppccap_P & PPC_MADD300) && (OPENSSL_ppccap_P & PPC_ALTIVEC)) {
+ felem_square_p = p384_felem_square;
+ felem_mul_p = p384_felem_mul;
+
+ return;
+ }
+# endif
+
/* Default */
felem_square_p = felem_square_ref;
felem_mul_p = felem_mul_ref;

76
openssl-ecc-Remove-extraneous-parentheses-in-secp384r1.patch Normal file
View File

@ -0,0 +1,76 @@
From 670e73d9084465384b11ef24802ca4a313e1d2f4 Mon Sep 17 00:00:00 2001
From: Rohan McLure <rohanmclure@linux.ibm.com>
Date: Tue, 15 Aug 2023 15:20:20 +1000
Subject: [PATCH] ecc: Remove extraneous parentheses in secp384r1
Substitutions in the felem_reduce() method feature unecessary
parentheses, remove them.
Signed-off-by: Rohan McLure <rohan.mclure@linux.ibm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21749)
---
crypto/ec/ecp_nistp384.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/crypto/ec/ecp_nistp384.c b/crypto/ec/ecp_nistp384.c
index 14f9530d07c6..ff68f9cc7ad0 100644
--- a/crypto/ec/ecp_nistp384.c
+++ b/crypto/ec/ecp_nistp384.c
@@ -540,7 +540,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[7] += in[12] >> 8;
acc[6] += (in[12] & 0xff) << 48;
acc[6] -= in[12] >> 16;
- acc[5] -= ((in[12] & 0xffff) << 40);
+ acc[5] -= (in[12] & 0xffff) << 40;
acc[6] += in[12] >> 48;
acc[5] += (in[12] & 0xffffffffffff) << 8;
@@ -549,7 +549,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[6] += in[11] >> 8;
acc[5] += (in[11] & 0xff) << 48;
acc[5] -= in[11] >> 16;
- acc[4] -= ((in[11] & 0xffff) << 40);
+ acc[4] -= (in[11] & 0xffff) << 40;
acc[5] += in[11] >> 48;
acc[4] += (in[11] & 0xffffffffffff) << 8;
@@ -558,7 +558,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[5] += in[10] >> 8;
acc[4] += (in[10] & 0xff) << 48;
acc[4] -= in[10] >> 16;
- acc[3] -= ((in[10] & 0xffff) << 40);
+ acc[3] -= (in[10] & 0xffff) << 40;
acc[4] += in[10] >> 48;
acc[3] += (in[10] & 0xffffffffffff) << 8;
@@ -567,7 +567,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[4] += in[9] >> 8;
acc[3] += (in[9] & 0xff) << 48;
acc[3] -= in[9] >> 16;
- acc[2] -= ((in[9] & 0xffff) << 40);
+ acc[2] -= (in[9] & 0xffff) << 40;
acc[3] += in[9] >> 48;
acc[2] += (in[9] & 0xffffffffffff) << 8;
@@ -582,7 +582,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[3] += acc[8] >> 8;
acc[2] += (acc[8] & 0xff) << 48;
acc[2] -= acc[8] >> 16;
- acc[1] -= ((acc[8] & 0xffff) << 40);
+ acc[1] -= (acc[8] & 0xffff) << 40;
acc[2] += acc[8] >> 48;
acc[1] += (acc[8] & 0xffffffffffff) << 8;
@@ -591,7 +591,7 @@ static void felem_reduce(felem out, const widefelem in)
acc[2] += acc[7] >> 8;
acc[1] += (acc[7] & 0xff) << 48;
acc[1] -= acc[7] >> 16;
- acc[0] -= ((acc[7] & 0xffff) << 40);
+ acc[0] -= (acc[7] & 0xffff) << 40;
acc[1] += acc[7] >> 48;
acc[0] += (acc[7] & 0xffffffffffff) << 8;

10
openssl-no-html-docs.patch
View File

@ -1,8 +1,8 @@
Index: openssl-3.0.0-alpha1/Configurations/unix-Makefile.tmpl
Index: openssl-3.1.4/Configurations/unix-Makefile.tmpl
===================================================================
--- openssl-3.0.0-alpha1.orig/Configurations/unix-Makefile.tmpl 2020-04-23 22:56:27.365853133 +0200
+++ openssl-3.0.0-alpha1/Configurations/unix-Makefile.tmpl 2020-04-23 22:56:52.474004636 +0200
@@ -544,7 +544,7 @@ install_sw: install_dev install_engines
--- openssl-3.1.4.orig/Configurations/unix-Makefile.tmpl
+++ openssl-3.1.4/Configurations/unix-Makefile.tmpl
@@ -611,7 +611,7 @@ install_sw: install_dev install_engines
uninstall_sw: uninstall_runtime uninstall_modules uninstall_engines uninstall_dev
@ -10,4 +10,4 @@ Index: openssl-3.0.0-alpha1/Configurations/unix-Makefile.tmpl
+install_docs: install_man_docs
uninstall_docs: uninstall_man_docs uninstall_html_docs
$(RM) -r $(DESTDIR)$(DOCDIR)
$(RM) -r "$(DESTDIR)$(DOCDIR)"

96
openssl-powerpc-ecc-Fix-stack-allocation-secp384r1-asm.patch Normal file
View File

@ -0,0 +1,96 @@
From 50f8b936b00dc18ce1f622a7a6aa46daf03da48b Mon Sep 17 00:00:00 2001
From: Rohan McLure <rohanmclure@linux.ibm.com>
Date: Wed, 16 Aug 2023 16:52:47 +1000
Subject: [PATCH] powerpc: ecc: Fix stack allocation secp384r1 asm
Assembly acceleration secp384r1 opts to not use any callee-save VSRs, as
VSX enabled systems make extensive use of renaming, and so writebacks in
felem_{mul,square}() can be reordered for best cache effects.
Remove stack allocations. This in turn fixes unmatched push/pops in
felem_{mul,square}().
Signed-off-by: Rohan McLure <rohan.mclure@linux.ibm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21749)
---
crypto/ec/asm/ecp_nistp384-ppc64.pl | 49 -----------------------------
1 file changed, 49 deletions(-)
diff --git a/crypto/ec/asm/ecp_nistp384-ppc64.pl b/crypto/ec/asm/ecp_nistp384-ppc64.pl
index 3f86b391af69..28f4168e5218 100755
--- a/crypto/ec/asm/ecp_nistp384-ppc64.pl
+++ b/crypto/ec/asm/ecp_nistp384-ppc64.pl
@@ -62,51 +62,6 @@ ($)
___
}
-
-sub push_vrs($$)
-{
- my ($min, $max) = @_;
-
- my $count = $max - $min + 1;
-
- $code.=<<___;
- mr $savesp,$sp
- stdu $sp,-16*`$count+1`($sp)
-
-___
- for (my $i = $min; $i <= $max; $i++) {
- my $mult = $max - $i + 1;
- $code.=<<___;
- stxv $i,-16*$mult($savesp)
-___
-
- }
-
- $code.=<<___;
-
-___
-}
-
-sub pop_vrs($$)
-{
- my ($min, $max) = @_;
-
- $code.=<<___;
- ld $savesp,0($sp)
-___
- for (my $i = $min; $i <= $max; $i++) {
- my $mult = $max - $i + 1;
- $code.=<<___;
- lxv $i,-16*$mult($savesp)
-___
- }
-
- $code.=<<___;
- mr $sp,$savesp
-
-___
-}
-
sub load_vrs($$)
{
my ($pointer, $reg_list) = @_;
@@ -162,8 +117,6 @@ ($$)
startproc("p384_felem_mul");
- push_vrs(52, 63);
-
$code.=<<___;
vspltisw $vzero,0
@@ -268,8 +221,6 @@ ($$)
startproc("p384_felem_square");
- push_vrs(52, 63);
-
$code.=<<___;
vspltisw $vzero,0