From c17157e4105bc92f8841083db67a17ac39062749 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Linus=20Unneb=C3=A4ck?= <linus@folkdatorn.se>
Date: Tue, 24 Nov 2020 19:04:42 +0100
Subject: [PATCH 1/4] Add mipsel to Github Actions CI setup

---
 .github/workflows/ci.yml  | 4 ++++
 mk/cargo.sh               | 7 +++++++
 mk/install-build-tools.sh | 6 ++++++
 3 files changed, 17 insertions(+)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 1c59d83fe5..dda5338bb9 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -174,6 +174,7 @@ jobs:
           - i686-pc-windows-msvc
           - i686-unknown-linux-gnu
           - i686-unknown-linux-musl
+          - mipsel-unknown-linux-gnu
           - x86_64-pc-windows-gnu
           - x86_64-pc-windows-msvc
           - x86_64-apple-darwin
@@ -259,6 +260,9 @@ jobs:
           - target: i686-unknown-linux-musl
             host_os: ubuntu-18.04
 
+          - target: mipsel-unknown-linux-gnu
+            host_os: ubuntu-18.04
+
           - target: x86_64-pc-windows-gnu
             host_os: windows-latest
 
diff --git a/mk/cargo.sh b/mk/cargo.sh
index 85a1cbc385..95c0f25614 100755
--- a/mk/cargo.sh
+++ b/mk/cargo.sh
@@ -20,6 +20,7 @@ IFS=$'\n\t'
 rustflags_self_contained="-Clink-self-contained=yes -Clinker=rust-lld"
 qemu_aarch64="qemu-aarch64 -L /usr/aarch64-linux-gnu"
 qemu_arm="qemu-arm -L /usr/arm-linux-gnueabihf"
+qemu_mipsel="qemu-system-mipsel -L /usr/mipsel-linux-gnu"
 
 # Avoid putting the Android tools in `$PATH` because there are tools in this
 # directory like `clang` that would conflict with the same-named tools that may
@@ -87,6 +88,12 @@ case $target in
     export AR_i686_unknown_linux_musl=llvm-ar-$llvm_version
     export CARGO_TARGET_I686_UNKNOWN_LINUX_MUSL_RUSTFLAGS="$rustflags_self_contained"
     ;;
+  mipsel-unknown-linux-gnu)
+    export CC_mipsel_unknown_linux_gnu=mipsel-linux-gnu-gcc
+    export AR_mipsel_unknown_linux_gnu=mipsel-linux-gnu-gcc-ar
+    export CARGO_TARGET_MIPSEL_UNKNOWN_LINUX_GNU_LINKER=mipsel-linux-gnu-gcc
+    export CARGO_TARGET_MIPSEL_UNKNOWN_LINUX_GNU_RUNNER="$qemu_mipsel"
+    ;;
   x86_64-unknown-linux-musl)
     export CC_x86_64_unknown_linux_musl=clang-$llvm_version
     export AR_x86_64_unknown_linux_musl=llvm-ar-$llvm_version
diff --git a/mk/install-build-tools.sh b/mk/install-build-tools.sh
index db50246e8d..459873cbea 100755
--- a/mk/install-build-tools.sh
+++ b/mk/install-build-tools.sh
@@ -65,6 +65,12 @@ case $target in
 --target=i686-unknown-linux-musl|--target=x86_64-unknown-linux-musl)
   use_clang=1
   ;;
+--target=mipsel-unknown-linux-gnu)
+  install_packages \
+    qemu-user \
+    gcc-mipsel-linux-gnu \
+    libc6-dev-mipsel-cross
+  ;;
 --target=wasm32-unknown-unknown)
   # The version of wasm-bindgen-cli must match the wasm-bindgen version.
   wasm_bindgen_version=$(cargo metadata --format-version 1 | jq -r '.packages | map(select( .name == "wasm-bindgen")) | map(.version) | .[0]')

From 53cfcd6ceecad90fde5995b3c073bd8a1af6d3e5 Mon Sep 17 00:00:00 2001
From: Bruno Tavares <connect+github@bltavares.com>
Date: Mon, 3 May 2021 23:19:56 -0300
Subject: [PATCH 2/4] Add mips support

This commit ports the changes from https://github.com/briansmith/ring/pull/1181 into the updated codebase

The following command is working:
> cross build --target mipsel-unknown-linux-musl

But tests are still not working:
> cross test --target mipsel-unknown-linux-musl
---
 Cargo.toml                            |   2 +
 build.rs                              |  37 ++-
 crypto/fipsmodule/bn/asm/mips-mont.pl | 435 ++++++++++++++++++++++++++
 include/ring-core/mips_arch.h         |  40 +++
 mk/cargo.sh                           |   7 +
 mk/install-build-tools.sh             |   6 +
 src/arithmetic/bigint.rs              |  36 ++-
 7 files changed, 548 insertions(+), 15 deletions(-)
 create mode 100644 crypto/fipsmodule/bn/asm/mips-mont.pl
 create mode 100644 include/ring-core/mips_arch.h

diff --git a/Cargo.toml b/Cargo.toml
index 2d5e11c9b9..ae83fac0d6 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -58,6 +58,7 @@ include = [
     "crypto/fipsmodule/bn/asm/x86-mont.pl",
     "crypto/fipsmodule/bn/asm/x86_64-mont.pl",
     "crypto/fipsmodule/bn/asm/x86_64-mont5.pl",
+    "crypto/fipsmodule/bn/asm/mips-mont.pl",
     "crypto/fipsmodule/bn/internal.h",
     "crypto/fipsmodule/bn/montgomery.c",
     "crypto/fipsmodule/bn/montgomery_inv.c",
@@ -105,6 +106,7 @@ include = [
     "examples/checkdigest.rs",
     "include/ring-core/aes.h",
     "include/ring-core/arm_arch.h",
+    "include/ring-core/mips_arch.h",
     "include/ring-core/base.h",
     "include/ring-core/check.h",
     "include/ring-core/cpu.h",
diff --git a/build.rs b/build.rs
index 64d78b8aa9..4663f13779 100644
--- a/build.rs
+++ b/build.rs
@@ -30,6 +30,8 @@ const X86: &str = "x86";
 const X86_64: &str = "x86_64";
 const AARCH64: &str = "aarch64";
 const ARM: &str = "arm";
+const MIPS: &str = "mips";
+const MIPS64: &str = "mips64";
 
 #[rustfmt::skip]
 const RING_SRCS: &[(&[&str], &str)] = &[
@@ -40,12 +42,12 @@ const RING_SRCS: &[(&[&str], &str)] = &[
     (&[], "crypto/mem.c"),
     (&[], "crypto/poly1305/poly1305.c"),
 
-    (&[AARCH64, ARM, X86_64, X86], "crypto/crypto.c"),
-    (&[AARCH64, ARM, X86_64, X86], "crypto/curve25519/curve25519.c"),
-    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/ecp_nistz.c"),
-    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/gfp_p256.c"),
-    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/gfp_p384.c"),
-    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/p256.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/crypto.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/curve25519/curve25519.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/ecp_nistz.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/gfp_p256.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/gfp_p384.c"),
+    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/p256.c"),
 
     (&[X86_64, X86], "crypto/cpu-intel.c"),
 
@@ -87,6 +89,8 @@ const RING_SRCS: &[(&[&str], &str)] = &[
     (&[AARCH64], "crypto/chacha/asm/chacha-armv8.pl"),
     (&[AARCH64], "crypto/fipsmodule/modes/asm/ghash-neon-armv8.pl"),
     (&[AARCH64], SHA512_ARMV8),
+
+    (&[MIPS, MIPS64], "crypto/fipsmodule/bn/asm/mips-mont.pl"),
 ];
 
 const SHA256_X86_64: &str = "crypto/fipsmodule/sha/asm/sha256-x86_64.pl";
@@ -200,6 +204,20 @@ const ASM_TARGETS: &[AsmTarget] = &[
         asm_extension: "S",
         preassemble: false,
     },
+    AsmTarget {
+        oss: LINUX_ABI,
+        arch: "mips",
+        perlasm_format: "elf",
+        asm_extension: "S",
+        preassemble: false,
+    },
+    AsmTarget {
+        oss: LINUX_ABI,
+        arch: "mips64",
+        perlasm_format: "elf",
+        asm_extension: "S",
+        preassemble: false,
+    },
     AsmTarget {
         oss: MACOS_ABI,
         arch: "aarch64",
@@ -309,6 +327,7 @@ fn ring_build_rs_main() {
     let arch = env::var("CARGO_CFG_TARGET_ARCH").unwrap();
     let os = env::var("CARGO_CFG_TARGET_OS").unwrap();
     let env = env::var("CARGO_CFG_TARGET_ENV").unwrap();
+    let endian = env::var("CARGO_CFG_TARGET_ENDIAN").unwrap();
     let (obj_ext, obj_opt) = if env == MSVC {
         (MSVC_OBJ_EXT, MSVC_OBJ_OPT)
     } else {
@@ -322,6 +341,7 @@ fn ring_build_rs_main() {
 
     let target = Target {
         arch,
+        endian,
         os,
         env,
         obj_ext,
@@ -373,6 +393,7 @@ fn pregenerate_asm_main() {
 
 struct Target {
     arch: String,
+    endian: String,
     os: String,
     env: String,
     obj_ext: &'static str,
@@ -391,6 +412,10 @@ fn build_c_code(target: &Target, pregenerated: PathBuf, out_dir: &Path, ring_cor
         }
     }
 
+    if target.arch == "mips" && target.endian == "big" {
+        panic!("MIPS Big-Endian detected. Stoping compilation as BoringSSL code are not available for this platform");
+    }
+
     let asm_target = ASM_TARGETS.iter().find(|asm_target| {
         asm_target.arch == target.arch && asm_target.oss.contains(&target.os.as_ref())
     });
diff --git a/crypto/fipsmodule/bn/asm/mips-mont.pl b/crypto/fipsmodule/bn/asm/mips-mont.pl
new file mode 100644
index 0000000000..7158c3a8bc
--- /dev/null
+++ b/crypto/fipsmodule/bn/asm/mips-mont.pl
@@ -0,0 +1,435 @@
+#! /usr/bin/env perl
+# Copyright 2010-2018 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 Andy Polyakov <appro@openssl.org> for the OpenSSL
+# project. The module is, however, dual licensed under OpenSSL and
+# CRYPTOGAMS licenses depending on where you obtain it. For further
+# details see http://www.openssl.org/~appro/cryptogams/.
+# ====================================================================
+
+# This module doesn't present direct interest for OpenSSL, because it
+# doesn't provide better performance for longer keys, at least not on
+# in-order-execution cores. While 512-bit RSA sign operations can be
+# 65% faster in 64-bit mode, 1024-bit ones are only 15% faster, and
+# 4096-bit ones are up to 15% slower. In 32-bit mode it varies from
+# 16% improvement for 512-bit RSA sign to -33% for 4096-bit RSA
+# verify:-( All comparisons are against bn_mul_mont-free assembler.
+# The module might be of interest to embedded system developers, as
+# the code is smaller than 1KB, yet offers >3x improvement on MIPS64
+# and 75-30% [less for longer keys] on MIPS32 over compiler-generated
+# code.
+
+######################################################################
+# There is a number of MIPS ABI in use, O32 and N32/64 are most
+# widely used. Then there is a new contender: NUBI. It appears that if
+# one picks the latter, it's possible to arrange code in ABI neutral
+# manner. Therefore let's stick to NUBI register layout:
+#
+($zero,$at,$t0,$t1,$t2)=map("\$$_",(0..2,24,25));
+($a0,$a1,$a2,$a3,$a4,$a5,$a6,$a7)=map("\$$_",(4..11));
+($s0,$s1,$s2,$s3,$s4,$s5,$s6,$s7,$s8,$s9,$s10,$s11)=map("\$$_",(12..23));
+($gp,$tp,$sp,$fp,$ra)=map("\$$_",(3,28..31));
+#
+# The return value is placed in $a0. Following coding rules facilitate
+# interoperability:
+#
+# - never ever touch $tp, "thread pointer", former $gp;
+# - copy return value to $t0, former $v0 [or to $a0 if you're adapting
+#   old code];
+# - on O32 populate $a4-$a7 with 'lw $aN,4*N($sp)' if necessary;
+#
+# For reference here is register layout for N32/64 MIPS ABIs:
+#
+# ($zero,$at,$v0,$v1)=map("\$$_",(0..3));
+# ($a0,$a1,$a2,$a3,$a4,$a5,$a6,$a7)=map("\$$_",(4..11));
+# ($t0,$t1,$t2,$t3,$t8,$t9)=map("\$$_",(12..15,24,25));
+# ($s0,$s1,$s2,$s3,$s4,$s5,$s6,$s7)=map("\$$_",(16..23));
+# ($gp,$sp,$fp,$ra)=map("\$$_",(28..31));
+
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+# supported flavours are o32,n32,64,nubi32,nubi64, default is o32
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : "o32";
+
+if ($flavour =~ /64|n32/i) {
+	$PTR_ADD="daddu";	# incidentally works even on n32
+	$PTR_SUB="dsubu";	# incidentally works even on n32
+	$REG_S="sd";
+	$REG_L="ld";
+	$SZREG=8;
+} else {
+	$PTR_ADD="addu";
+	$PTR_SUB="subu";
+	$REG_S="sw";
+	$REG_L="lw";
+	$SZREG=4;
+}
+$SAVED_REGS_MASK = ($flavour =~ /nubi/i) ? 0x00fff000 : 0x00ff0000;
+#
+# <appro@openssl.org>
+#
+######################################################################
+
+$output and open STDOUT,">$output";
+
+if ($flavour =~ /64|n32/i) {
+	$LD="ld";
+	$ST="sd";
+	$MULTU="dmultu";
+	$ADDU="daddu";
+	$SUBU="dsubu";
+	$BNSZ=8;
+} else {
+	$LD="lw";
+	$ST="sw";
+	$MULTU="multu";
+	$ADDU="addu";
+	$SUBU="subu";
+	$BNSZ=4;
+}
+
+# int bn_mul_mont(
+$rp=$a0;	# BN_ULONG *rp,
+$ap=$a1;	# const BN_ULONG *ap,
+$bp=$a2;	# const BN_ULONG *bp,
+$np=$a3;	# const BN_ULONG *np,
+$n0=$a4;	# const BN_ULONG *n0,
+$num=$a5;	# int num);
+
+$lo0=$a6;
+$hi0=$a7;
+$lo1=$t1;
+$hi1=$t2;
+$aj=$s0;
+$bi=$s1;
+$nj=$s2;
+$tp=$s3;
+$alo=$s4;
+$ahi=$s5;
+$nlo=$s6;
+$nhi=$s7;
+$tj=$s8;
+$i=$s9;
+$j=$s10;
+$m1=$s11;
+
+$FRAMESIZE=14;
+
+$code=<<___;
+#include <ring-core/mips_arch.h>
+
+.text
+
+.set	noat
+.set	noreorder
+
+.align	5
+.globl	bn_mul_mont
+.ent	bn_mul_mont
+bn_mul_mont:
+___
+$code.=<<___ if ($flavour =~ /o32/i);
+	lw	$n0,16($sp)
+	lw	$num,20($sp)
+___
+$code.=<<___;
+	slt	$at,$num,4
+	bnez	$at,1f
+	li	$t0,0
+	b	bn_mul_mont_internal
+	nop
+1:	jr	$ra
+	li	$a0,0
+.end	bn_mul_mont
+
+.align	5
+.ent	bn_mul_mont_internal
+bn_mul_mont_internal:
+	.frame	$fp,$FRAMESIZE*$SZREG,$ra
+	.mask	0x40000000|$SAVED_REGS_MASK,-$SZREG
+	$PTR_SUB $sp,$FRAMESIZE*$SZREG
+	$REG_S	$fp,($FRAMESIZE-1)*$SZREG($sp)
+	$REG_S	$s11,($FRAMESIZE-2)*$SZREG($sp)
+	$REG_S	$s10,($FRAMESIZE-3)*$SZREG($sp)
+	$REG_S	$s9,($FRAMESIZE-4)*$SZREG($sp)
+	$REG_S	$s8,($FRAMESIZE-5)*$SZREG($sp)
+	$REG_S	$s7,($FRAMESIZE-6)*$SZREG($sp)
+	$REG_S	$s6,($FRAMESIZE-7)*$SZREG($sp)
+	$REG_S	$s5,($FRAMESIZE-8)*$SZREG($sp)
+	$REG_S	$s4,($FRAMESIZE-9)*$SZREG($sp)
+___
+$code.=<<___ if ($flavour =~ /nubi/i);
+	$REG_S	$s3,($FRAMESIZE-10)*$SZREG($sp)
+	$REG_S	$s2,($FRAMESIZE-11)*$SZREG($sp)
+	$REG_S	$s1,($FRAMESIZE-12)*$SZREG($sp)
+	$REG_S	$s0,($FRAMESIZE-13)*$SZREG($sp)
+___
+$code.=<<___;
+	move	$fp,$sp
+
+	.set	reorder
+	$LD	$n0,0($n0)
+	$LD	$bi,0($bp)	# bp[0]
+	$LD	$aj,0($ap)	# ap[0]
+	$LD	$nj,0($np)	# np[0]
+
+	$PTR_SUB $sp,2*$BNSZ	# place for two extra words
+	sll	$num,`log($BNSZ)/log(2)`
+	li	$at,-4096
+	$PTR_SUB $sp,$num
+	and	$sp,$at
+
+	$MULTU	($aj,$bi)
+	$LD	$ahi,$BNSZ($ap)
+	$LD	$nhi,$BNSZ($np)
+	mflo	($lo0,$aj,$bi)
+	mfhi	($hi0,$aj,$bi)
+	$MULTU	($lo0,$n0)
+	mflo	($m1,$lo0,$n0)
+
+	$MULTU	($ahi,$bi)
+	mflo	($alo,$ahi,$bi)
+	mfhi	($ahi,$ahi,$bi)
+
+	$MULTU	($nj,$m1)
+	mflo	($lo1,$nj,$m1)
+	mfhi	($hi1,$nj,$m1)
+	$MULTU	($nhi,$m1)
+	$ADDU	$lo1,$lo0
+	sltu	$at,$lo1,$lo0
+	$ADDU	$hi1,$at
+	mflo	($nlo,$nhi,$m1)
+	mfhi	($nhi,$nhi,$m1)
+
+	move	$tp,$sp
+	li	$j,2*$BNSZ
+.align	4
+.L1st:
+	.set	noreorder
+	$PTR_ADD $aj,$ap,$j
+	$PTR_ADD $nj,$np,$j
+	$LD	$aj,($aj)
+	$LD	$nj,($nj)
+
+	$MULTU	($aj,$bi)
+	$ADDU	$lo0,$alo,$hi0
+	$ADDU	$lo1,$nlo,$hi1
+	sltu	$at,$lo0,$hi0
+	sltu	$t0,$lo1,$hi1
+	$ADDU	$hi0,$ahi,$at
+	$ADDU	$hi1,$nhi,$t0
+	mflo	($alo,$aj,$bi)
+	mfhi	($ahi,$aj,$bi)
+
+	$ADDU	$lo1,$lo0
+	sltu	$at,$lo1,$lo0
+	$MULTU	($nj,$m1)
+	$ADDU	$hi1,$at
+	addu	$j,$BNSZ
+	$ST	$lo1,($tp)
+	sltu	$t0,$j,$num
+	mflo	($nlo,$nj,$m1)
+	mfhi	($nhi,$nj,$m1)
+
+	bnez	$t0,.L1st
+	$PTR_ADD $tp,$BNSZ
+	.set	reorder
+
+	$ADDU	$lo0,$alo,$hi0
+	sltu	$at,$lo0,$hi0
+	$ADDU	$hi0,$ahi,$at
+
+	$ADDU	$lo1,$nlo,$hi1
+	sltu	$t0,$lo1,$hi1
+	$ADDU	$hi1,$nhi,$t0
+	$ADDU	$lo1,$lo0
+	sltu	$at,$lo1,$lo0
+	$ADDU	$hi1,$at
+
+	$ST	$lo1,($tp)
+
+	$ADDU	$hi1,$hi0
+	sltu	$at,$hi1,$hi0
+	$ST	$hi1,$BNSZ($tp)
+	$ST	$at,2*$BNSZ($tp)
+
+	li	$i,$BNSZ
+.align	4
+.Louter:
+	$PTR_ADD $bi,$bp,$i
+	$LD	$bi,($bi)
+	$LD	$aj,($ap)
+	$LD	$ahi,$BNSZ($ap)
+	$LD	$tj,($sp)
+
+	$MULTU	($aj,$bi)
+	$LD	$nj,($np)
+	$LD	$nhi,$BNSZ($np)
+	mflo	($lo0,$aj,$bi)
+	mfhi	($hi0,$aj,$bi)
+	$ADDU	$lo0,$tj
+	$MULTU	($lo0,$n0)
+	sltu	$at,$lo0,$tj
+	$ADDU	$hi0,$at
+	mflo	($m1,$lo0,$n0)
+
+	$MULTU	($ahi,$bi)
+	mflo	($alo,$ahi,$bi)
+	mfhi	($ahi,$ahi,$bi)
+
+	$MULTU	($nj,$m1)
+	mflo	($lo1,$nj,$m1)
+	mfhi	($hi1,$nj,$m1)
+
+	$MULTU	($nhi,$m1)
+	$ADDU	$lo1,$lo0
+	sltu	$at,$lo1,$lo0
+	$ADDU	$hi1,$at
+	mflo	($nlo,$nhi,$m1)
+	mfhi	($nhi,$nhi,$m1)
+
+	move	$tp,$sp
+	li	$j,2*$BNSZ
+	$LD	$tj,$BNSZ($tp)
+.align	4
+.Linner:
+	.set	noreorder
+	$PTR_ADD $aj,$ap,$j
+	$PTR_ADD $nj,$np,$j
+	$LD	$aj,($aj)
+	$LD	$nj,($nj)
+
+	$MULTU	($aj,$bi)
+	$ADDU	$lo0,$alo,$hi0
+	$ADDU	$lo1,$nlo,$hi1
+	sltu	$at,$lo0,$hi0
+	sltu	$t0,$lo1,$hi1
+	$ADDU	$hi0,$ahi,$at
+	$ADDU	$hi1,$nhi,$t0
+	mflo	($alo,$aj,$bi)
+	mfhi	($ahi,$aj,$bi)
+
+	$ADDU	$lo0,$tj
+	addu	$j,$BNSZ
+	$MULTU	($nj,$m1)
+	sltu	$at,$lo0,$tj
+	$ADDU	$lo1,$lo0
+	$ADDU	$hi0,$at
+	sltu	$t0,$lo1,$lo0
+	$LD	$tj,2*$BNSZ($tp)
+	$ADDU	$hi1,$t0
+	sltu	$at,$j,$num
+	mflo	($nlo,$nj,$m1)
+	mfhi	($nhi,$nj,$m1)
+	$ST	$lo1,($tp)
+	bnez	$at,.Linner
+	$PTR_ADD $tp,$BNSZ
+	.set	reorder
+
+	$ADDU	$lo0,$alo,$hi0
+	sltu	$at,$lo0,$hi0
+	$ADDU	$hi0,$ahi,$at
+	$ADDU	$lo0,$tj
+	sltu	$t0,$lo0,$tj
+	$ADDU	$hi0,$t0
+
+	$LD	$tj,2*$BNSZ($tp)
+	$ADDU	$lo1,$nlo,$hi1
+	sltu	$at,$lo1,$hi1
+	$ADDU	$hi1,$nhi,$at
+	$ADDU	$lo1,$lo0
+	sltu	$t0,$lo1,$lo0
+	$ADDU	$hi1,$t0
+	$ST	$lo1,($tp)
+
+	$ADDU	$lo1,$hi1,$hi0
+	sltu	$hi1,$lo1,$hi0
+	$ADDU	$lo1,$tj
+	sltu	$at,$lo1,$tj
+	$ADDU	$hi1,$at
+	$ST	$lo1,$BNSZ($tp)
+	$ST	$hi1,2*$BNSZ($tp)
+
+	addu	$i,$BNSZ
+	sltu	$t0,$i,$num
+	bnez	$t0,.Louter
+
+	.set	noreorder
+	$PTR_ADD $tj,$sp,$num	# &tp[num]
+	move	$tp,$sp
+	move	$ap,$sp
+	li	$hi0,0		# clear borrow bit
+
+.align	4
+.Lsub:	$LD	$lo0,($tp)
+	$LD	$lo1,($np)
+	$PTR_ADD $tp,$BNSZ
+	$PTR_ADD $np,$BNSZ
+	$SUBU	$lo1,$lo0,$lo1	# tp[i]-np[i]
+	sgtu	$at,$lo1,$lo0
+	$SUBU	$lo0,$lo1,$hi0
+	sgtu	$hi0,$lo0,$lo1
+	$ST	$lo0,($rp)
+	or	$hi0,$at
+	sltu	$at,$tp,$tj
+	bnez	$at,.Lsub
+	$PTR_ADD $rp,$BNSZ
+
+	$SUBU	$hi0,$hi1,$hi0	# handle upmost overflow bit
+	move	$tp,$sp
+	$PTR_SUB $rp,$num	# restore rp
+	not	$hi1,$hi0
+
+.Lcopy:	$LD	$nj,($tp)	# conditional move
+	$LD	$aj,($rp)
+	$ST	$zero,($tp)
+	$PTR_ADD $tp,$BNSZ
+	and	$nj,$hi0
+	and	$aj,$hi1
+	or	$aj,$nj
+	sltu	$at,$tp,$tj
+	$ST	$aj,($rp)
+	bnez	$at,.Lcopy
+	$PTR_ADD $rp,$BNSZ
+
+	li	$a0,1
+	li	$t0,1
+
+	.set	noreorder
+	move	$sp,$fp
+	$REG_L	$fp,($FRAMESIZE-1)*$SZREG($sp)
+	$REG_L	$s11,($FRAMESIZE-2)*$SZREG($sp)
+	$REG_L	$s10,($FRAMESIZE-3)*$SZREG($sp)
+	$REG_L	$s9,($FRAMESIZE-4)*$SZREG($sp)
+	$REG_L	$s8,($FRAMESIZE-5)*$SZREG($sp)
+	$REG_L	$s7,($FRAMESIZE-6)*$SZREG($sp)
+	$REG_L	$s6,($FRAMESIZE-7)*$SZREG($sp)
+	$REG_L	$s5,($FRAMESIZE-8)*$SZREG($sp)
+	$REG_L	$s4,($FRAMESIZE-9)*$SZREG($sp)
+___
+$code.=<<___ if ($flavour =~ /nubi/i);
+	$REG_L	$s3,($FRAMESIZE-10)*$SZREG($sp)
+	$REG_L	$s2,($FRAMESIZE-11)*$SZREG($sp)
+	$REG_L	$s1,($FRAMESIZE-12)*$SZREG($sp)
+	$REG_L	$s0,($FRAMESIZE-13)*$SZREG($sp)
+___
+$code.=<<___;
+	jr	$ra
+	$PTR_ADD $sp,$FRAMESIZE*$SZREG
+.end	bn_mul_mont_internal
+.rdata
+.asciiz	"Montgomery Multiplication for MIPS, CRYPTOGAMS by <appro\@openssl.org>"
+___
+
+$code =~ s/\`([^\`]*)\`/eval $1/gem;
+
+print $code;
+close STDOUT or die "error closing STDOUT";
\ No newline at end of file
diff --git a/include/ring-core/mips_arch.h b/include/ring-core/mips_arch.h
new file mode 100644
index 0000000000..e7562b8d8b
--- /dev/null
+++ b/include/ring-core/mips_arch.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2011-2016 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 OSSL_CRYPTO_MIPS_ARCH_H
+# define OSSL_CRYPTO_MIPS_ARCH_H
+
+# if (defined(__mips_smartmips) || defined(_MIPS_ARCH_MIPS32R3) || \
+      defined(_MIPS_ARCH_MIPS32R5) || defined(_MIPS_ARCH_MIPS32R6)) \
+      && !defined(_MIPS_ARCH_MIPS32R2)
+#  define _MIPS_ARCH_MIPS32R2
+# endif
+
+# if (defined(_MIPS_ARCH_MIPS64R3) || defined(_MIPS_ARCH_MIPS64R5) || \
+      defined(_MIPS_ARCH_MIPS64R6)) \
+      && !defined(_MIPS_ARCH_MIPS64R2)
+#  define _MIPS_ARCH_MIPS64R2
+# endif
+
+# if defined(_MIPS_ARCH_MIPS64R6)
+#  define dmultu(rs,rt)
+#  define mflo(rd,rs,rt)	dmulu	rd,rs,rt
+#  define mfhi(rd,rs,rt)	dmuhu	rd,rs,rt
+# elif defined(_MIPS_ARCH_MIPS32R6)
+#  define multu(rs,rt)
+#  define mflo(rd,rs,rt)	mulu	rd,rs,rt
+#  define mfhi(rd,rs,rt)	muhu	rd,rs,rt
+# else
+#  define dmultu(rs,rt)		dmultu	rs,rt
+#  define multu(rs,rt)		multu	rs,rt
+#  define mflo(rd,rs,rt)	mflo	rd
+#  define mfhi(rd,rs,rt)	mfhi	rd
+# endif
+
+#endif
\ No newline at end of file
diff --git a/mk/cargo.sh b/mk/cargo.sh
index 95c0f25614..9a1b16d1af 100755
--- a/mk/cargo.sh
+++ b/mk/cargo.sh
@@ -21,6 +21,7 @@ rustflags_self_contained="-Clink-self-contained=yes -Clinker=rust-lld"
 qemu_aarch64="qemu-aarch64 -L /usr/aarch64-linux-gnu"
 qemu_arm="qemu-arm -L /usr/arm-linux-gnueabihf"
 qemu_mipsel="qemu-system-mipsel -L /usr/mipsel-linux-gnu"
+qemu_mips64el="qemu-system-mips64el -L /usr/mips64el-linux-gnu"
 
 # Avoid putting the Android tools in `$PATH` because there are tools in this
 # directory like `clang` that would conflict with the same-named tools that may
@@ -94,6 +95,12 @@ case $target in
     export CARGO_TARGET_MIPSEL_UNKNOWN_LINUX_GNU_LINKER=mipsel-linux-gnu-gcc
     export CARGO_TARGET_MIPSEL_UNKNOWN_LINUX_GNU_RUNNER="$qemu_mipsel"
     ;;
+  mips64el-unknown-linux-gnuabi64)
+    export CC_mips64el_unknown_linux_gnuabi64=mips64el-linux-gnu-gcc
+    export AR_mips64el_unknown_linux_gnuabi64=mips64el-linux-gnu-gcc-ar
+    export CARGO_TARGET_MIPS64EL_UNKNOWN_LINUX_GNUABI64_LINKER=mips64el-linux-gnu-gcc
+    export CARGO_TARGET_MIPS64EL_UNKNOWN_LINUX_GNUABI64_RUNNER="$qemu_mips64el"
+    ;;
   x86_64-unknown-linux-musl)
     export CC_x86_64_unknown_linux_musl=clang-$llvm_version
     export AR_x86_64_unknown_linux_musl=llvm-ar-$llvm_version
diff --git a/mk/install-build-tools.sh b/mk/install-build-tools.sh
index 459873cbea..87f2dd1921 100755
--- a/mk/install-build-tools.sh
+++ b/mk/install-build-tools.sh
@@ -71,6 +71,12 @@ case $target in
     gcc-mipsel-linux-gnu \
     libc6-dev-mipsel-cross
   ;;
+--target=mips64el-unknown-linux-gnu)
+  install_packages \
+    qemu-user \
+    gcc-mips64el-linux-gnu \
+    libc6-dev-mips64el-cross
+  ;;
 --target=wasm32-unknown-unknown)
   # The version of wasm-bindgen-cli must match the wasm-bindgen version.
   wasm_bindgen_version=$(cargo metadata --format-version 1 | jq -r '.packages | map(select( .name == "wasm-bindgen")) | map(.version) | .[0]')
diff --git a/src/arithmetic/bigint.rs b/src/arithmetic/bigint.rs
index 028be1a4bf..44a1cf0695 100644
--- a/src/arithmetic/bigint.rs
+++ b/src/arithmetic/bigint.rs
@@ -1194,7 +1194,9 @@ fn limbs_mont_mul(r: &mut [Limb], a: &[Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     ))]
     unsafe {
         bn_mul_mont(
@@ -1211,7 +1213,9 @@ fn limbs_mont_mul(r: &mut [Limb], a: &[Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1251,7 +1255,9 @@ fn limbs_from_mont_in_place(r: &mut [Limb], tmp: &mut [Limb], m: &[Limb], n0: &N
     target_arch = "aarch64",
     target_arch = "arm",
     target_arch = "x86_64",
-    target_arch = "x86"
+    target_arch = "x86",
+    target_arch = "mips",
+    target_arch = "mips64",
 )))]
 fn limbs_mul(r: &mut [Limb], a: &[Limb], b: &[Limb]) {
     debug_assert_eq!(r.len(), 2 * a.len());
@@ -1282,7 +1288,9 @@ fn limbs_mont_product(r: &mut [Limb], a: &[Limb], b: &[Limb], m: &[Limb], n0: &N
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     ))]
     unsafe {
         bn_mul_mont(
@@ -1299,7 +1307,9 @@ fn limbs_mont_product(r: &mut [Limb], a: &[Limb], b: &[Limb], m: &[Limb], n0: &N
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1316,7 +1326,9 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     ))]
     unsafe {
         bn_mul_mont(
@@ -1333,7 +1345,9 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1347,7 +1361,9 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
     target_arch = "aarch64",
     target_arch = "arm",
     target_arch = "x86_64",
-    target_arch = "x86"
+    target_arch = "x86",
+    target_arch = "mips",
+    target_arch = "mips64",
 ))]
 prefixed_extern! {
     // `r` and/or 'a' and/or 'b' may alias.
@@ -1367,7 +1383,9 @@ prefixed_extern! {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86"
+        target_arch = "x86",
+        target_arch = "mips",
+        target_arch = "mips64",
     ))
 ))]
 prefixed_extern! {

From bd27ae2c87237b61e3fd73dfbbc020e21f0cbf9c Mon Sep 17 00:00:00 2001
From: Bruno Tavares <connect+github@bltavares.com>
Date: Mon, 3 May 2021 23:57:50 -0300
Subject: [PATCH 3/4] Remove changes based on GitHub feedback

---
 Cargo.toml                            |   2 -
 build.rs                              |  41 +--
 crypto/fipsmodule/bn/asm/mips-mont.pl | 435 --------------------------
 include/ring-core/mips_arch.h         |  40 ---
 src/arithmetic/bigint.rs              |  36 +--
 5 files changed, 19 insertions(+), 535 deletions(-)
 delete mode 100644 crypto/fipsmodule/bn/asm/mips-mont.pl
 delete mode 100644 include/ring-core/mips_arch.h

diff --git a/Cargo.toml b/Cargo.toml
index ae83fac0d6..2d5e11c9b9 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -58,7 +58,6 @@ include = [
     "crypto/fipsmodule/bn/asm/x86-mont.pl",
     "crypto/fipsmodule/bn/asm/x86_64-mont.pl",
     "crypto/fipsmodule/bn/asm/x86_64-mont5.pl",
-    "crypto/fipsmodule/bn/asm/mips-mont.pl",
     "crypto/fipsmodule/bn/internal.h",
     "crypto/fipsmodule/bn/montgomery.c",
     "crypto/fipsmodule/bn/montgomery_inv.c",
@@ -106,7 +105,6 @@ include = [
     "examples/checkdigest.rs",
     "include/ring-core/aes.h",
     "include/ring-core/arm_arch.h",
-    "include/ring-core/mips_arch.h",
     "include/ring-core/base.h",
     "include/ring-core/check.h",
     "include/ring-core/cpu.h",
diff --git a/build.rs b/build.rs
index 4663f13779..ed4b212759 100644
--- a/build.rs
+++ b/build.rs
@@ -30,8 +30,6 @@ const X86: &str = "x86";
 const X86_64: &str = "x86_64";
 const AARCH64: &str = "aarch64";
 const ARM: &str = "arm";
-const MIPS: &str = "mips";
-const MIPS64: &str = "mips64";
 
 #[rustfmt::skip]
 const RING_SRCS: &[(&[&str], &str)] = &[
@@ -41,13 +39,13 @@ const RING_SRCS: &[(&[&str], &str)] = &[
     (&[], "crypto/limbs/limbs.c"),
     (&[], "crypto/mem.c"),
     (&[], "crypto/poly1305/poly1305.c"),
+    (&[], "crypto/curve25519/curve25519.c"),
+    (&[], "crypto/fipsmodule/ec/ecp_nistz.c"),
+    (&[], "crypto/fipsmodule/ec/gfp_p256.c"),
+    (&[], "crypto/fipsmodule/ec/gfp_p384.c"),
+    (&[], "crypto/fipsmodule/ec/p256.c"),
 
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/crypto.c"),
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/curve25519/curve25519.c"),
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/ecp_nistz.c"),
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/gfp_p256.c"),
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/gfp_p384.c"),
-    (&[AARCH64, ARM, MIPS, MIPS64, X86_64, X86], "crypto/fipsmodule/ec/p256.c"),
+    (&[AARCH64, X86_64, X86], "crypto/crypto.c"),
 
     (&[X86_64, X86], "crypto/cpu-intel.c"),
 
@@ -89,8 +87,6 @@ const RING_SRCS: &[(&[&str], &str)] = &[
     (&[AARCH64], "crypto/chacha/asm/chacha-armv8.pl"),
     (&[AARCH64], "crypto/fipsmodule/modes/asm/ghash-neon-armv8.pl"),
     (&[AARCH64], SHA512_ARMV8),
-
-    (&[MIPS, MIPS64], "crypto/fipsmodule/bn/asm/mips-mont.pl"),
 ];
 
 const SHA256_X86_64: &str = "crypto/fipsmodule/sha/asm/sha256-x86_64.pl";
@@ -204,20 +200,6 @@ const ASM_TARGETS: &[AsmTarget] = &[
         asm_extension: "S",
         preassemble: false,
     },
-    AsmTarget {
-        oss: LINUX_ABI,
-        arch: "mips",
-        perlasm_format: "elf",
-        asm_extension: "S",
-        preassemble: false,
-    },
-    AsmTarget {
-        oss: LINUX_ABI,
-        arch: "mips64",
-        perlasm_format: "elf",
-        asm_extension: "S",
-        preassemble: false,
-    },
     AsmTarget {
         oss: MACOS_ABI,
         arch: "aarch64",
@@ -321,13 +303,16 @@ fn main() {
 fn ring_build_rs_main() {
     use std::env;
 
+    if env::var("CARGO_CFG_TARGET_ENDIAN").unwrap() == "big" {
+        panic!("Big-endian targets are not supported yet");
+    }
+
     let out_dir = env::var("OUT_DIR").unwrap();
     let out_dir = PathBuf::from(out_dir);
 
     let arch = env::var("CARGO_CFG_TARGET_ARCH").unwrap();
     let os = env::var("CARGO_CFG_TARGET_OS").unwrap();
     let env = env::var("CARGO_CFG_TARGET_ENV").unwrap();
-    let endian = env::var("CARGO_CFG_TARGET_ENDIAN").unwrap();
     let (obj_ext, obj_opt) = if env == MSVC {
         (MSVC_OBJ_EXT, MSVC_OBJ_OPT)
     } else {
@@ -341,7 +326,6 @@ fn ring_build_rs_main() {
 
     let target = Target {
         arch,
-        endian,
         os,
         env,
         obj_ext,
@@ -393,7 +377,6 @@ fn pregenerate_asm_main() {
 
 struct Target {
     arch: String,
-    endian: String,
     os: String,
     env: String,
     obj_ext: &'static str,
@@ -412,10 +395,6 @@ fn build_c_code(target: &Target, pregenerated: PathBuf, out_dir: &Path, ring_cor
         }
     }
 
-    if target.arch == "mips" && target.endian == "big" {
-        panic!("MIPS Big-Endian detected. Stoping compilation as BoringSSL code are not available for this platform");
-    }
-
     let asm_target = ASM_TARGETS.iter().find(|asm_target| {
         asm_target.arch == target.arch && asm_target.oss.contains(&target.os.as_ref())
     });
diff --git a/crypto/fipsmodule/bn/asm/mips-mont.pl b/crypto/fipsmodule/bn/asm/mips-mont.pl
deleted file mode 100644
index 7158c3a8bc..0000000000
--- a/crypto/fipsmodule/bn/asm/mips-mont.pl
+++ /dev/null
@@ -1,435 +0,0 @@
-#! /usr/bin/env perl
-# Copyright 2010-2018 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 Andy Polyakov <appro@openssl.org> for the OpenSSL
-# project. The module is, however, dual licensed under OpenSSL and
-# CRYPTOGAMS licenses depending on where you obtain it. For further
-# details see http://www.openssl.org/~appro/cryptogams/.
-# ====================================================================
-
-# This module doesn't present direct interest for OpenSSL, because it
-# doesn't provide better performance for longer keys, at least not on
-# in-order-execution cores. While 512-bit RSA sign operations can be
-# 65% faster in 64-bit mode, 1024-bit ones are only 15% faster, and
-# 4096-bit ones are up to 15% slower. In 32-bit mode it varies from
-# 16% improvement for 512-bit RSA sign to -33% for 4096-bit RSA
-# verify:-( All comparisons are against bn_mul_mont-free assembler.
-# The module might be of interest to embedded system developers, as
-# the code is smaller than 1KB, yet offers >3x improvement on MIPS64
-# and 75-30% [less for longer keys] on MIPS32 over compiler-generated
-# code.
-
-######################################################################
-# There is a number of MIPS ABI in use, O32 and N32/64 are most
-# widely used. Then there is a new contender: NUBI. It appears that if
-# one picks the latter, it's possible to arrange code in ABI neutral
-# manner. Therefore let's stick to NUBI register layout:
-#
-($zero,$at,$t0,$t1,$t2)=map("\$$_",(0..2,24,25));
-($a0,$a1,$a2,$a3,$a4,$a5,$a6,$a7)=map("\$$_",(4..11));
-($s0,$s1,$s2,$s3,$s4,$s5,$s6,$s7,$s8,$s9,$s10,$s11)=map("\$$_",(12..23));
-($gp,$tp,$sp,$fp,$ra)=map("\$$_",(3,28..31));
-#
-# The return value is placed in $a0. Following coding rules facilitate
-# interoperability:
-#
-# - never ever touch $tp, "thread pointer", former $gp;
-# - copy return value to $t0, former $v0 [or to $a0 if you're adapting
-#   old code];
-# - on O32 populate $a4-$a7 with 'lw $aN,4*N($sp)' if necessary;
-#
-# For reference here is register layout for N32/64 MIPS ABIs:
-#
-# ($zero,$at,$v0,$v1)=map("\$$_",(0..3));
-# ($a0,$a1,$a2,$a3,$a4,$a5,$a6,$a7)=map("\$$_",(4..11));
-# ($t0,$t1,$t2,$t3,$t8,$t9)=map("\$$_",(12..15,24,25));
-# ($s0,$s1,$s2,$s3,$s4,$s5,$s6,$s7)=map("\$$_",(16..23));
-# ($gp,$sp,$fp,$ra)=map("\$$_",(28..31));
-
-# $output is the last argument if it looks like a file (it has an extension)
-# $flavour is the first argument if it doesn't look like a file
-$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
-# supported flavours are o32,n32,64,nubi32,nubi64, default is o32
-$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : "o32";
-
-if ($flavour =~ /64|n32/i) {
-	$PTR_ADD="daddu";	# incidentally works even on n32
-	$PTR_SUB="dsubu";	# incidentally works even on n32
-	$REG_S="sd";
-	$REG_L="ld";
-	$SZREG=8;
-} else {
-	$PTR_ADD="addu";
-	$PTR_SUB="subu";
-	$REG_S="sw";
-	$REG_L="lw";
-	$SZREG=4;
-}
-$SAVED_REGS_MASK = ($flavour =~ /nubi/i) ? 0x00fff000 : 0x00ff0000;
-#
-# <appro@openssl.org>
-#
-######################################################################
-
-$output and open STDOUT,">$output";
-
-if ($flavour =~ /64|n32/i) {
-	$LD="ld";
-	$ST="sd";
-	$MULTU="dmultu";
-	$ADDU="daddu";
-	$SUBU="dsubu";
-	$BNSZ=8;
-} else {
-	$LD="lw";
-	$ST="sw";
-	$MULTU="multu";
-	$ADDU="addu";
-	$SUBU="subu";
-	$BNSZ=4;
-}
-
-# int bn_mul_mont(
-$rp=$a0;	# BN_ULONG *rp,
-$ap=$a1;	# const BN_ULONG *ap,
-$bp=$a2;	# const BN_ULONG *bp,
-$np=$a3;	# const BN_ULONG *np,
-$n0=$a4;	# const BN_ULONG *n0,
-$num=$a5;	# int num);
-
-$lo0=$a6;
-$hi0=$a7;
-$lo1=$t1;
-$hi1=$t2;
-$aj=$s0;
-$bi=$s1;
-$nj=$s2;
-$tp=$s3;
-$alo=$s4;
-$ahi=$s5;
-$nlo=$s6;
-$nhi=$s7;
-$tj=$s8;
-$i=$s9;
-$j=$s10;
-$m1=$s11;
-
-$FRAMESIZE=14;
-
-$code=<<___;
-#include <ring-core/mips_arch.h>
-
-.text
-
-.set	noat
-.set	noreorder
-
-.align	5
-.globl	bn_mul_mont
-.ent	bn_mul_mont
-bn_mul_mont:
-___
-$code.=<<___ if ($flavour =~ /o32/i);
-	lw	$n0,16($sp)
-	lw	$num,20($sp)
-___
-$code.=<<___;
-	slt	$at,$num,4
-	bnez	$at,1f
-	li	$t0,0
-	b	bn_mul_mont_internal
-	nop
-1:	jr	$ra
-	li	$a0,0
-.end	bn_mul_mont
-
-.align	5
-.ent	bn_mul_mont_internal
-bn_mul_mont_internal:
-	.frame	$fp,$FRAMESIZE*$SZREG,$ra
-	.mask	0x40000000|$SAVED_REGS_MASK,-$SZREG
-	$PTR_SUB $sp,$FRAMESIZE*$SZREG
-	$REG_S	$fp,($FRAMESIZE-1)*$SZREG($sp)
-	$REG_S	$s11,($FRAMESIZE-2)*$SZREG($sp)
-	$REG_S	$s10,($FRAMESIZE-3)*$SZREG($sp)
-	$REG_S	$s9,($FRAMESIZE-4)*$SZREG($sp)
-	$REG_S	$s8,($FRAMESIZE-5)*$SZREG($sp)
-	$REG_S	$s7,($FRAMESIZE-6)*$SZREG($sp)
-	$REG_S	$s6,($FRAMESIZE-7)*$SZREG($sp)
-	$REG_S	$s5,($FRAMESIZE-8)*$SZREG($sp)
-	$REG_S	$s4,($FRAMESIZE-9)*$SZREG($sp)
-___
-$code.=<<___ if ($flavour =~ /nubi/i);
-	$REG_S	$s3,($FRAMESIZE-10)*$SZREG($sp)
-	$REG_S	$s2,($FRAMESIZE-11)*$SZREG($sp)
-	$REG_S	$s1,($FRAMESIZE-12)*$SZREG($sp)
-	$REG_S	$s0,($FRAMESIZE-13)*$SZREG($sp)
-___
-$code.=<<___;
-	move	$fp,$sp
-
-	.set	reorder
-	$LD	$n0,0($n0)
-	$LD	$bi,0($bp)	# bp[0]
-	$LD	$aj,0($ap)	# ap[0]
-	$LD	$nj,0($np)	# np[0]
-
-	$PTR_SUB $sp,2*$BNSZ	# place for two extra words
-	sll	$num,`log($BNSZ)/log(2)`
-	li	$at,-4096
-	$PTR_SUB $sp,$num
-	and	$sp,$at
-
-	$MULTU	($aj,$bi)
-	$LD	$ahi,$BNSZ($ap)
-	$LD	$nhi,$BNSZ($np)
-	mflo	($lo0,$aj,$bi)
-	mfhi	($hi0,$aj,$bi)
-	$MULTU	($lo0,$n0)
-	mflo	($m1,$lo0,$n0)
-
-	$MULTU	($ahi,$bi)
-	mflo	($alo,$ahi,$bi)
-	mfhi	($ahi,$ahi,$bi)
-
-	$MULTU	($nj,$m1)
-	mflo	($lo1,$nj,$m1)
-	mfhi	($hi1,$nj,$m1)
-	$MULTU	($nhi,$m1)
-	$ADDU	$lo1,$lo0
-	sltu	$at,$lo1,$lo0
-	$ADDU	$hi1,$at
-	mflo	($nlo,$nhi,$m1)
-	mfhi	($nhi,$nhi,$m1)
-
-	move	$tp,$sp
-	li	$j,2*$BNSZ
-.align	4
-.L1st:
-	.set	noreorder
-	$PTR_ADD $aj,$ap,$j
-	$PTR_ADD $nj,$np,$j
-	$LD	$aj,($aj)
-	$LD	$nj,($nj)
-
-	$MULTU	($aj,$bi)
-	$ADDU	$lo0,$alo,$hi0
-	$ADDU	$lo1,$nlo,$hi1
-	sltu	$at,$lo0,$hi0
-	sltu	$t0,$lo1,$hi1
-	$ADDU	$hi0,$ahi,$at
-	$ADDU	$hi1,$nhi,$t0
-	mflo	($alo,$aj,$bi)
-	mfhi	($ahi,$aj,$bi)
-
-	$ADDU	$lo1,$lo0
-	sltu	$at,$lo1,$lo0
-	$MULTU	($nj,$m1)
-	$ADDU	$hi1,$at
-	addu	$j,$BNSZ
-	$ST	$lo1,($tp)
-	sltu	$t0,$j,$num
-	mflo	($nlo,$nj,$m1)
-	mfhi	($nhi,$nj,$m1)
-
-	bnez	$t0,.L1st
-	$PTR_ADD $tp,$BNSZ
-	.set	reorder
-
-	$ADDU	$lo0,$alo,$hi0
-	sltu	$at,$lo0,$hi0
-	$ADDU	$hi0,$ahi,$at
-
-	$ADDU	$lo1,$nlo,$hi1
-	sltu	$t0,$lo1,$hi1
-	$ADDU	$hi1,$nhi,$t0
-	$ADDU	$lo1,$lo0
-	sltu	$at,$lo1,$lo0
-	$ADDU	$hi1,$at
-
-	$ST	$lo1,($tp)
-
-	$ADDU	$hi1,$hi0
-	sltu	$at,$hi1,$hi0
-	$ST	$hi1,$BNSZ($tp)
-	$ST	$at,2*$BNSZ($tp)
-
-	li	$i,$BNSZ
-.align	4
-.Louter:
-	$PTR_ADD $bi,$bp,$i
-	$LD	$bi,($bi)
-	$LD	$aj,($ap)
-	$LD	$ahi,$BNSZ($ap)
-	$LD	$tj,($sp)
-
-	$MULTU	($aj,$bi)
-	$LD	$nj,($np)
-	$LD	$nhi,$BNSZ($np)
-	mflo	($lo0,$aj,$bi)
-	mfhi	($hi0,$aj,$bi)
-	$ADDU	$lo0,$tj
-	$MULTU	($lo0,$n0)
-	sltu	$at,$lo0,$tj
-	$ADDU	$hi0,$at
-	mflo	($m1,$lo0,$n0)
-
-	$MULTU	($ahi,$bi)
-	mflo	($alo,$ahi,$bi)
-	mfhi	($ahi,$ahi,$bi)
-
-	$MULTU	($nj,$m1)
-	mflo	($lo1,$nj,$m1)
-	mfhi	($hi1,$nj,$m1)
-
-	$MULTU	($nhi,$m1)
-	$ADDU	$lo1,$lo0
-	sltu	$at,$lo1,$lo0
-	$ADDU	$hi1,$at
-	mflo	($nlo,$nhi,$m1)
-	mfhi	($nhi,$nhi,$m1)
-
-	move	$tp,$sp
-	li	$j,2*$BNSZ
-	$LD	$tj,$BNSZ($tp)
-.align	4
-.Linner:
-	.set	noreorder
-	$PTR_ADD $aj,$ap,$j
-	$PTR_ADD $nj,$np,$j
-	$LD	$aj,($aj)
-	$LD	$nj,($nj)
-
-	$MULTU	($aj,$bi)
-	$ADDU	$lo0,$alo,$hi0
-	$ADDU	$lo1,$nlo,$hi1
-	sltu	$at,$lo0,$hi0
-	sltu	$t0,$lo1,$hi1
-	$ADDU	$hi0,$ahi,$at
-	$ADDU	$hi1,$nhi,$t0
-	mflo	($alo,$aj,$bi)
-	mfhi	($ahi,$aj,$bi)
-
-	$ADDU	$lo0,$tj
-	addu	$j,$BNSZ
-	$MULTU	($nj,$m1)
-	sltu	$at,$lo0,$tj
-	$ADDU	$lo1,$lo0
-	$ADDU	$hi0,$at
-	sltu	$t0,$lo1,$lo0
-	$LD	$tj,2*$BNSZ($tp)
-	$ADDU	$hi1,$t0
-	sltu	$at,$j,$num
-	mflo	($nlo,$nj,$m1)
-	mfhi	($nhi,$nj,$m1)
-	$ST	$lo1,($tp)
-	bnez	$at,.Linner
-	$PTR_ADD $tp,$BNSZ
-	.set	reorder
-
-	$ADDU	$lo0,$alo,$hi0
-	sltu	$at,$lo0,$hi0
-	$ADDU	$hi0,$ahi,$at
-	$ADDU	$lo0,$tj
-	sltu	$t0,$lo0,$tj
-	$ADDU	$hi0,$t0
-
-	$LD	$tj,2*$BNSZ($tp)
-	$ADDU	$lo1,$nlo,$hi1
-	sltu	$at,$lo1,$hi1
-	$ADDU	$hi1,$nhi,$at
-	$ADDU	$lo1,$lo0
-	sltu	$t0,$lo1,$lo0
-	$ADDU	$hi1,$t0
-	$ST	$lo1,($tp)
-
-	$ADDU	$lo1,$hi1,$hi0
-	sltu	$hi1,$lo1,$hi0
-	$ADDU	$lo1,$tj
-	sltu	$at,$lo1,$tj
-	$ADDU	$hi1,$at
-	$ST	$lo1,$BNSZ($tp)
-	$ST	$hi1,2*$BNSZ($tp)
-
-	addu	$i,$BNSZ
-	sltu	$t0,$i,$num
-	bnez	$t0,.Louter
-
-	.set	noreorder
-	$PTR_ADD $tj,$sp,$num	# &tp[num]
-	move	$tp,$sp
-	move	$ap,$sp
-	li	$hi0,0		# clear borrow bit
-
-.align	4
-.Lsub:	$LD	$lo0,($tp)
-	$LD	$lo1,($np)
-	$PTR_ADD $tp,$BNSZ
-	$PTR_ADD $np,$BNSZ
-	$SUBU	$lo1,$lo0,$lo1	# tp[i]-np[i]
-	sgtu	$at,$lo1,$lo0
-	$SUBU	$lo0,$lo1,$hi0
-	sgtu	$hi0,$lo0,$lo1
-	$ST	$lo0,($rp)
-	or	$hi0,$at
-	sltu	$at,$tp,$tj
-	bnez	$at,.Lsub
-	$PTR_ADD $rp,$BNSZ
-
-	$SUBU	$hi0,$hi1,$hi0	# handle upmost overflow bit
-	move	$tp,$sp
-	$PTR_SUB $rp,$num	# restore rp
-	not	$hi1,$hi0
-
-.Lcopy:	$LD	$nj,($tp)	# conditional move
-	$LD	$aj,($rp)
-	$ST	$zero,($tp)
-	$PTR_ADD $tp,$BNSZ
-	and	$nj,$hi0
-	and	$aj,$hi1
-	or	$aj,$nj
-	sltu	$at,$tp,$tj
-	$ST	$aj,($rp)
-	bnez	$at,.Lcopy
-	$PTR_ADD $rp,$BNSZ
-
-	li	$a0,1
-	li	$t0,1
-
-	.set	noreorder
-	move	$sp,$fp
-	$REG_L	$fp,($FRAMESIZE-1)*$SZREG($sp)
-	$REG_L	$s11,($FRAMESIZE-2)*$SZREG($sp)
-	$REG_L	$s10,($FRAMESIZE-3)*$SZREG($sp)
-	$REG_L	$s9,($FRAMESIZE-4)*$SZREG($sp)
-	$REG_L	$s8,($FRAMESIZE-5)*$SZREG($sp)
-	$REG_L	$s7,($FRAMESIZE-6)*$SZREG($sp)
-	$REG_L	$s6,($FRAMESIZE-7)*$SZREG($sp)
-	$REG_L	$s5,($FRAMESIZE-8)*$SZREG($sp)
-	$REG_L	$s4,($FRAMESIZE-9)*$SZREG($sp)
-___
-$code.=<<___ if ($flavour =~ /nubi/i);
-	$REG_L	$s3,($FRAMESIZE-10)*$SZREG($sp)
-	$REG_L	$s2,($FRAMESIZE-11)*$SZREG($sp)
-	$REG_L	$s1,($FRAMESIZE-12)*$SZREG($sp)
-	$REG_L	$s0,($FRAMESIZE-13)*$SZREG($sp)
-___
-$code.=<<___;
-	jr	$ra
-	$PTR_ADD $sp,$FRAMESIZE*$SZREG
-.end	bn_mul_mont_internal
-.rdata
-.asciiz	"Montgomery Multiplication for MIPS, CRYPTOGAMS by <appro\@openssl.org>"
-___
-
-$code =~ s/\`([^\`]*)\`/eval $1/gem;
-
-print $code;
-close STDOUT or die "error closing STDOUT";
\ No newline at end of file
diff --git a/include/ring-core/mips_arch.h b/include/ring-core/mips_arch.h
deleted file mode 100644
index e7562b8d8b..0000000000
--- a/include/ring-core/mips_arch.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/*
- * Copyright 2011-2016 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 OSSL_CRYPTO_MIPS_ARCH_H
-# define OSSL_CRYPTO_MIPS_ARCH_H
-
-# if (defined(__mips_smartmips) || defined(_MIPS_ARCH_MIPS32R3) || \
-      defined(_MIPS_ARCH_MIPS32R5) || defined(_MIPS_ARCH_MIPS32R6)) \
-      && !defined(_MIPS_ARCH_MIPS32R2)
-#  define _MIPS_ARCH_MIPS32R2
-# endif
-
-# if (defined(_MIPS_ARCH_MIPS64R3) || defined(_MIPS_ARCH_MIPS64R5) || \
-      defined(_MIPS_ARCH_MIPS64R6)) \
-      && !defined(_MIPS_ARCH_MIPS64R2)
-#  define _MIPS_ARCH_MIPS64R2
-# endif
-
-# if defined(_MIPS_ARCH_MIPS64R6)
-#  define dmultu(rs,rt)
-#  define mflo(rd,rs,rt)	dmulu	rd,rs,rt
-#  define mfhi(rd,rs,rt)	dmuhu	rd,rs,rt
-# elif defined(_MIPS_ARCH_MIPS32R6)
-#  define multu(rs,rt)
-#  define mflo(rd,rs,rt)	mulu	rd,rs,rt
-#  define mfhi(rd,rs,rt)	muhu	rd,rs,rt
-# else
-#  define dmultu(rs,rt)		dmultu	rs,rt
-#  define multu(rs,rt)		multu	rs,rt
-#  define mflo(rd,rs,rt)	mflo	rd
-#  define mfhi(rd,rs,rt)	mfhi	rd
-# endif
-
-#endif
\ No newline at end of file
diff --git a/src/arithmetic/bigint.rs b/src/arithmetic/bigint.rs
index 44a1cf0695..028be1a4bf 100644
--- a/src/arithmetic/bigint.rs
+++ b/src/arithmetic/bigint.rs
@@ -1194,9 +1194,7 @@ fn limbs_mont_mul(r: &mut [Limb], a: &[Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     ))]
     unsafe {
         bn_mul_mont(
@@ -1213,9 +1211,7 @@ fn limbs_mont_mul(r: &mut [Limb], a: &[Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1255,9 +1251,7 @@ fn limbs_from_mont_in_place(r: &mut [Limb], tmp: &mut [Limb], m: &[Limb], n0: &N
     target_arch = "aarch64",
     target_arch = "arm",
     target_arch = "x86_64",
-    target_arch = "x86",
-    target_arch = "mips",
-    target_arch = "mips64",
+    target_arch = "x86"
 )))]
 fn limbs_mul(r: &mut [Limb], a: &[Limb], b: &[Limb]) {
     debug_assert_eq!(r.len(), 2 * a.len());
@@ -1288,9 +1282,7 @@ fn limbs_mont_product(r: &mut [Limb], a: &[Limb], b: &[Limb], m: &[Limb], n0: &N
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     ))]
     unsafe {
         bn_mul_mont(
@@ -1307,9 +1299,7 @@ fn limbs_mont_product(r: &mut [Limb], a: &[Limb], b: &[Limb], m: &[Limb], n0: &N
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1326,9 +1316,7 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     ))]
     unsafe {
         bn_mul_mont(
@@ -1345,9 +1333,7 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     )))]
     {
         let mut tmp = [0; 2 * MODULUS_MAX_LIMBS];
@@ -1361,9 +1347,7 @@ fn limbs_mont_square(r: &mut [Limb], m: &[Limb], n0: &N0) {
     target_arch = "aarch64",
     target_arch = "arm",
     target_arch = "x86_64",
-    target_arch = "x86",
-    target_arch = "mips",
-    target_arch = "mips64",
+    target_arch = "x86"
 ))]
 prefixed_extern! {
     // `r` and/or 'a' and/or 'b' may alias.
@@ -1383,9 +1367,7 @@ prefixed_extern! {
         target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "x86_64",
-        target_arch = "x86",
-        target_arch = "mips",
-        target_arch = "mips64",
+        target_arch = "x86"
     ))
 ))]
 prefixed_extern! {

From d80f17814262258b4a2c8e3b71de7329dd366d5b Mon Sep 17 00:00:00 2001
From: Bruno Tavares <connect+github@bltavares.com>
Date: Tue, 4 May 2021 00:13:22 -0300
Subject: [PATCH 4/4] Keep files only for the big 4

---
 build.rs | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/build.rs b/build.rs
index ed4b212759..84010f43a0 100644
--- a/build.rs
+++ b/build.rs
@@ -41,11 +41,11 @@ const RING_SRCS: &[(&[&str], &str)] = &[
     (&[], "crypto/poly1305/poly1305.c"),
     (&[], "crypto/curve25519/curve25519.c"),
     (&[], "crypto/fipsmodule/ec/ecp_nistz.c"),
-    (&[], "crypto/fipsmodule/ec/gfp_p256.c"),
-    (&[], "crypto/fipsmodule/ec/gfp_p384.c"),
     (&[], "crypto/fipsmodule/ec/p256.c"),
 
-    (&[AARCH64, X86_64, X86], "crypto/crypto.c"),
+    (&[AARCH64, ARM, X86_64, X86], "crypto/crypto.c"),
+    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/gfp_p256.c"),
+    (&[AARCH64, ARM, X86_64, X86], "crypto/fipsmodule/ec/gfp_p384.c"),
 
     (&[X86_64, X86], "crypto/cpu-intel.c"),