SP: big-endian support

[SparkiDev]

Description

Handle reading and writing from big-endian byte array when compiling for big-endian.

Fixes zd#17810

Testing

Emulator not available on Ubuntu for big-endian ARM.

Checklist

• added tests
• updated/added doxygen
• updated appropriate READMEs
• Updated manual and documentation

[dgarske]

@SparkiDev , what's the state of this PR? Is it ready for review?

[douzzer]

to get this to compile with BIG_ENDIAN_ORDER, these changes are needed:

diff --git a/wolfcrypt/src/sp_arm32.c b/wolfcrypt/src/sp_arm32.c
index d08bf9cd3..20d7f5577 100644
--- a/wolfcrypt/src/sp_arm32.c
+++ b/wolfcrypt/src/sp_arm32.c
@@ -91,7 +91,6 @@ static void sp_2048_from_bin(sp_digit* r, int size, const byte* a, int n)
 {
     int i;
     int j;
-    byte* d;
 
     for (i = n - 1,j = 0; i >= 3; i -= 4) {
         r[j]  = ((sp_digit)a[i - 0] <<  0) |
@@ -111,6 +110,8 @@ static void sp_2048_from_bin(sp_digit* r, int size, const byte* a, int n)
             r[j] |= ((sp_digit)a[i]) << s;
         }
 #else
+        byte* d;
+
         r[j] = 0;
 
         d = (byte*)r;
@@ -18274,7 +18275,6 @@ static void sp_3072_from_bin(sp_digit* r, int size, const byte* a, int n)
 {
     int i;
     int j;
-    byte* d;
 
     for (i = n - 1,j = 0; i >= 3; i -= 4) {
         r[j]  = ((sp_digit)a[i - 0] <<  0) |
@@ -18294,6 +18294,8 @@ static void sp_3072_from_bin(sp_digit* r, int size, const byte* a, int n)
             r[j] |= ((sp_digit)a[i]) << s;
         }
 #else
+        byte* d;
+
         r[j] = 0;
 
         d = (byte*)r;
@@ -76393,7 +76395,6 @@ static void sp_256_from_bin(sp_digit* r, int size, const byte* a, int n)
 {
     int i;
     int j;
-    byte* d;
 
     for (i = n - 1,j = 0; i >= 3; i -= 4) {
         r[j]  = ((sp_digit)a[i - 0] <<  0) |
@@ -76413,6 +76414,8 @@ static void sp_256_from_bin(sp_digit* r, int size, const byte* a, int n)
             r[j] |= ((sp_digit)a[i]) << s;
         }
 #else
+        byte* d;
+
         r[j] = 0;
 
         d = (byte*)r;
@@ -156655,7 +156658,6 @@ static void sp_1024_from_bin(sp_digit* r, int size, const byte* a, int n)
 {
     int i;
     int j;
-    byte* d;
 
     for (i = n - 1,j = 0; i >= 3; i -= 4) {
         r[j]  = ((sp_digit)a[i - 0] <<  0) |
@@ -156675,6 +156677,8 @@ static void sp_1024_from_bin(sp_digit* r, int size, const byte* a, int n)
             r[j] |= ((sp_digit)a[i]) << s;
         }
 #else
+        byte* d;
+
         r[j] = 0;
 
         d = (byte*)r;

once that's done, building on armeb-linux-gnueabihf gives

/tmp/tmp.4346_12509/cc4wYqsG.s:133: Error: selected processor does not support `umaal r11,r12,r0,r5' in ARM mode
/tmp/tmp.4346_12509/cc4wYqsG.s:135: Error: selected processor does not support `umaal r12,r8,r1,r5' in ARM mode
/tmp/tmp.4346_12509/cc4wYqsG.s:136: Error: selected processor does not support `umaal r12,r7,r0,r6' in ARM mode
[and a couple hundred more like this]

not sure if I'm using the wrong target or what.

it builds clean targeting armv7a-unknown-linux-gnueabihf.

[dgarske]

@SparkiDev nice work on this, please spend some more time if possible to finish it up.

[douzzer]

multi-test passes on cross-aarch64_be-all-sp-asm-unittest-sanitizer cross-armeb-all-noasm-testsuite quantum-safe-wolfssl-all-cross-armv7a-armasm-unittest, but still fails on cross-armeb-all-armasm-testsuite (problem is in master).