[crypto] Expose shifted out bit from big integer shifts

Expose the bit shifted out as a result of shifting a big integer left
or right.

Signed-off-by: Michael Brown <[email protected]>

Author: mcb30

src/arch/arm32/include/bits/bigint.h

@@ -123,16 +123,18 @@ bigint_subtract_raw ( const uint32_t *subtrahend0, uint32_t *value0,
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shl_raw ( uint32_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint32_t *discard_value;
 	uint32_t *discard_end;
 	uint32_t discard_value_i;
+	int carry;
 
-	__asm__ __volatile__ ( "adds %1, %0, %5, lsl #2\n\t" /* clear CF */
+	__asm__ __volatile__ ( "adds %1, %0, %1, lsl #2\n\t" /* clear CF */
 			       "\n1:\n\t"
 			       "ldr %2, [%0]\n\t"
 			       "adcs %2, %2\n\t"
@@ -142,26 +144,29 @@ bigint_shl_raw ( uint32_t *value0, unsigned int size ) {
 			       : "=l" ( discard_value ),
 				 "=l" ( discard_end ),
 				 "=l" ( discard_value_i ),
+				 "=@cccs" ( carry ),
 				 "+m" ( *value )
-			       : "0" ( value0 ), "1" ( size )
-			       : "cc" );
+			       : "0" ( value0 ), "1" ( size ) );
+	return carry;
 }
 
 /**
  * Shift big integer right
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shr_raw ( uint32_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint32_t *discard_value;
 	uint32_t *discard_end;
 	uint32_t discard_value_i;
+	int carry;
 
-	__asm__ __volatile__ ( "adds %1, %0, %5, lsl #2\n\t" /* clear CF */
+	__asm__ __volatile__ ( "adds %1, %0, %1, lsl #2\n\t" /* clear CF */
 			       "\n1:\n\t"
 			       "ldmdb %1!, {%2}\n\t"
 			       "rrxs %2, %2\n\t"
@@ -171,9 +176,10 @@ bigint_shr_raw ( uint32_t *value0, unsigned int size ) {
 			       : "=l" ( discard_value ),
 				 "=l" ( discard_end ),
 				 "=l" ( discard_value_i ),
+				 "=@cccs" ( carry ),
 				 "+m" ( *value )
-			       : "0" ( value0 ), "1" ( size )
-			       : "cc" );
+			       : "0" ( value0 ), "1" ( size ) );
+	return carry;
 }
 
 /**

src/arch/arm64/include/bits/bigint.h

@@ -122,14 +122,16 @@ bigint_subtract_raw ( const uint64_t *subtrahend0, uint64_t *value0,
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shl_raw ( uint64_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint64_t *discard_value;
 	uint64_t discard_value_i;
 	unsigned int discard_size;
+	int carry;
 
 	__asm__ __volatile__ ( "cmn xzr, xzr\n\t" /* clear CF */
 			       "\n1:\n\t"
@@ -141,40 +143,43 @@ bigint_shl_raw ( uint64_t *value0, unsigned int size ) {
 			       : "=r" ( discard_value ),
 				 "=r" ( discard_size ),
 				 "=r" ( discard_value_i ),
+				 "=@cccs" ( carry ),
 				 "+m" ( *value )
-			       : "0" ( value0 ), "1" ( size )
-			       : "cc" );
+			       : "0" ( value0 ), "1" ( size ) );
+	return carry;
 }
 
 /**
  * Shift big integer right
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shr_raw ( uint64_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint64_t *discard_value;
-	uint64_t discard_value_i;
-	uint64_t discard_value_j;
+	uint64_t discard_high;
 	unsigned int discard_size;
+	uint64_t low;
 
-	__asm__ __volatile__ ( "mov %3, #0\n\t"
+	__asm__ __volatile__ ( "mov %2, #0\n\t"
 			       "\n1:\n\t"
 			       "sub %w1, %w1, #1\n\t"
-			       "ldr %2, [%0, %1, lsl #3]\n\t"
-			       "extr %3, %3, %2, #1\n\t"
-			       "str %3, [%0, %1, lsl #3]\n\t"
-			       "mov %3, %2\n\t"
+			       "ldr %3, [%0, %1, lsl #3]\n\t"
+			       "extr %2, %2, %3, #1\n\t"
+			       "str %2, [%0, %1, lsl #3]\n\t"
+			       "mov %2, %3\n\t"
 			       "cbnz %w1, 1b\n\t"
 			       : "=r" ( discard_value ),
 				 "=r" ( discard_size ),
-				 "=r" ( discard_value_i ),
-				 "=r" ( discard_value_j ),
+				 "=r" ( discard_high ),
+				 "=r" ( low ),
 				 "+m" ( *value )
 			       : "0" ( value0 ), "1" ( size ) );
+	return ( low & 1 );
 }
 
 /**

src/arch/loong64/include/bits/bigint.h

@@ -144,26 +144,27 @@ bigint_subtract_raw ( const uint64_t *subtrahend0, uint64_t *value0,
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shl_raw ( uint64_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint64_t *discard_value;
 	uint64_t discard_value_i;
-	uint64_t discard_carry;
 	uint64_t discard_temp;
 	unsigned int discard_size;
+	uint64_t carry;
 
 	__asm__ __volatile__ ( "\n1:\n\t"
 			       /* Load value[i] */
 			       "ld.d %2, %0, 0\n\t"
 			       /* Shift left */
 			       "rotri.d %2, %2, 63\n\t"
-			       "andi %4, %2, 1\n\t"
-			       "xor %2, %2, %4\n\t"
-			       "or %2, %2, %3\n\t"
-			       "move %3, %4\n\t"
+			       "andi %3, %2, 1\n\t"
+			       "xor %2, %2, %3\n\t"
+			       "or %2, %2, %4\n\t"
+			       "move %4, %3\n\t"
 			       /* Store value[i] */
 			       "st.d %2, %0, 0\n\t"
 			       /* Loop  */
@@ -173,37 +174,39 @@ bigint_shl_raw ( uint64_t *value0, unsigned int size ) {
 			       : "=r" ( discard_value ),
 				 "=r" ( discard_size ),
 				 "=r" ( discard_value_i ),
-				 "=r" ( discard_carry ),
 				 "=r" ( discard_temp ),
+				 "=r" ( carry ),
 				 "+m" ( *value )
-			       : "0" ( value0 ), "1" ( size ), "3" ( 0 )
+			       : "0" ( value0 ), "1" ( size ), "4" ( 0 )
 			       : "cc" );
+	return ( carry & 1 );
 }
 
 /**
  * Shift big integer right
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shr_raw ( uint64_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	uint64_t *discard_value;
 	uint64_t discard_value_i;
-	uint64_t discard_carry;
 	uint64_t discard_temp;
 	unsigned int discard_size;
+	uint64_t carry;
 
 	__asm__ __volatile__ ( "\n1:\n\t"
 			       /* Load value[i] */
 			       "ld.d %2, %0, -8\n\t"
 			       /* Shift right */
-			       "andi %4, %2, 1\n\t"
-			       "xor %2, %2, %4\n\t"
-			       "or %2, %2, %3\n\t"
-			       "move %3, %4\n\t"
+			       "andi %3, %2, 1\n\t"
+			       "xor %2, %2, %3\n\t"
+			       "or %2, %2, %4\n\t"
+			       "move %4, %3\n\t"
 			       "rotri.d %2, %2, 1\n\t"
 			       /* Store value[i] */
 			       "st.d %2, %0, -8\n\t"
@@ -214,11 +217,12 @@ bigint_shr_raw ( uint64_t *value0, unsigned int size ) {
 			       : "=r" ( discard_value ),
 				 "=r" ( discard_size ),
 				 "=r" ( discard_value_i ),
-				 "=r" ( discard_carry ),
 				 "=r" ( discard_temp ),
+				 "=r" ( carry ),
 				 "+m" ( *value )
-			       : "0" ( value0 + size ), "1" ( size ), "3" ( 0 )
+			       : "0" ( value0 + size ), "1" ( size ), "4" ( 0 )
 			       : "cc" );
+	return ( carry & 1 );
 }
 
 /**

src/arch/riscv/include/bits/bigint.h

@@ -143,77 +143,81 @@ bigint_subtract_raw ( const unsigned long *subtrahend0, unsigned long *value0,
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shl_raw ( unsigned long *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	unsigned long *valueN = ( value0 + size );
 	unsigned long *discard_value;
 	unsigned long discard_value_i;
-	unsigned long discard_carry;
 	unsigned long discard_temp;
+	unsigned long carry;
 
 	__asm__ __volatile__ ( "\n1:\n\t"
 			       /* Load value[i] */
 			       LOADN " %1, (%0)\n\t"
 			       /* Shift left */
-			       "slli %3, %1, 1\n\t"
-			       "or %3, %3, %2\n\t"
-			       "srli %2, %1, %7\n\t"
+			       "slli %2, %1, 1\n\t"
+			       "or %2, %2, %3\n\t"
+			       "srli %3, %1, %7\n\t"
 			       /* Store value[i] */
-			       STOREN " %3, (%0)\n\t"
+			       STOREN " %2, (%0)\n\t"
 			       /* Loop  */
 			       "addi %0, %0, %6\n\t"
 			       "bne %0, %5, 1b\n\t"
 			       : "=&r" ( discard_value ),
 				 "=&r" ( discard_value_i ),
-				 "=&r" ( discard_carry ),
 				 "=&r" ( discard_temp ),
+				 "=&r" ( carry ),
 				 "+m" ( *value )
 			       : "r" ( valueN ),
 				 "i" ( sizeof ( unsigned long ) ),
 				 "i" ( ( 8 * sizeof ( unsigned long ) - 1 ) ),
-				 "0" ( value0 ), "2" ( 0 ) );
+				 "0" ( value0 ), "3" ( 0 ) );
+	return carry;
 }
 
 /**
  * Shift big integer right
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shr_raw ( unsigned long *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	unsigned long *valueN = ( value0 + size );
 	unsigned long *discard_value;
 	unsigned long discard_value_i;
-	unsigned long discard_carry;
 	unsigned long discard_temp;
+	unsigned long carry;
 
 	__asm__ __volatile__ ( "\n1:\n\t"
 			       /* Load value[i] */
 			       LOADN " %1, %6(%0)\n\t"
 			       /* Shift right */
-			       "srli %3, %1, 1\n\t"
-			       "or %3, %3, %2\n\t"
-			       "slli %2, %1, %7\n\t"
+			       "srli %2, %1, 1\n\t"
+			       "or %2, %2, %3\n\t"
+			       "slli %3, %1, %7\n\t"
 			       /* Store value[i] */
-			       STOREN " %3, %6(%0)\n\t"
+			       STOREN " %2, %6(%0)\n\t"
 			       /* Loop  */
 			       "addi %0, %0, %6\n\t"
 			       "bne %0, %5, 1b\n\t"
 			       : "=&r" ( discard_value ),
 				 "=&r" ( discard_value_i ),
-				 "=&r" ( discard_carry ),
 				 "=&r" ( discard_temp ),
+				 "=&r" ( carry ),
 				 "+m" ( *value )
 			       : "r" ( value0 ),
 				 "i" ( -( sizeof ( unsigned long ) ) ),
 				 "i" ( ( 8 * sizeof ( unsigned long ) - 1 ) ),
-				 "0" ( valueN ), "2" ( 0 ) );
+				 "0" ( valueN ), "3" ( 0 ) );
+	return ( !! carry );
 }
 
 /**

src/arch/x86/include/bits/bigint.h

@@ -116,42 +116,49 @@ bigint_subtract_raw ( const uint32_t *subtrahend0, uint32_t *value0,
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shl_raw ( uint32_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	long index;
 	long discard_c;
+	int out;
 
 	__asm__ __volatile__ ( "xor %0, %0\n\t" /* Zero %0 and clear CF */
 			       "\n1:\n\t"
-			       "rcll $1, (%3,%0,4)\n\t"
+			       "rcll $1, (%4,%0,4)\n\t"
 			       "inc %0\n\t" /* Does not affect CF */
 			       "loop 1b\n\t"
 			       : "=&r" ( index ), "=&c" ( discard_c ),
-				 "+m" ( *value )
+				 "=@ccc" ( out ), "+m" ( *value )
 			       : "r" ( value0 ), "1" ( size ) );
+	return out;
 }
 
 /**
  * Shift big integer right
  *
  * @v value0		Element 0 of big integer
  * @v size		Number of elements
+ * @ret out		Bit shifted out
  */
-static inline __attribute__ (( always_inline )) void
+static inline __attribute__ (( always_inline )) int
 bigint_shr_raw ( uint32_t *value0, unsigned int size ) {
 	bigint_t ( size ) __attribute__ (( may_alias )) *value =
 		( ( void * ) value0 );
 	long discard_c;
+	int out;
 
 	__asm__ __volatile__ ( "clc\n\t"
 			       "\n1:\n\t"
-			       "rcrl $1, -4(%2,%0,4)\n\t"
+			       "rcrl $1, -4(%3,%0,4)\n\t"
 			       "loop 1b\n\t"
-			       : "=&c" ( discard_c ), "+m" ( *value )
+			       : "=&c" ( discard_c ), "=@ccc" ( out ),
+				 "+m" ( *value )
 			       : "r" ( value0 ), "0" ( size ) );
+	return out;
 }
 
 /**