* sparc/machine.m4 (TMP3): A third temporary register.

(AES_FINAL_ROUND): Prepared for scheduling.

Rev: src/nettle/sparc/machine.m4:1.10

diff --git a/sparc/machine.m4 b/sparc/machine.m4
index 0fbdffcaeb3c2fd96a29d19e85499729297a26d7..adc205e22e7e106b0176828a1428a73613eafc13 100644 (file)
--- a/sparc/machine.m4
+++ b/sparc/machine.m4
@@ -7,6 +7,7 @@ C eval.
 C Used as temporaries by the AES macros
 define(<TMP1>, <%g1>)
 define(<TMP2>, <%g2>)
+define(<TMP3>, <%g3>)
 
 C Loop invariants used by AES_ROUND
 define(<T0>,   <%o0>)
@@ -62,31 +63,32 @@ C Compute one word in the final round function. Output is converted to
 C octets and stored at dst. Relies on AES_SBOX being zero.
 define(<AES_FINAL_ROUND>, <
        C       Load subkey
-       ld      [$7 + eval(4*$1)], TMP1
+       ld      [$7 + eval(4*$1)], TMP3
 
-       and     $3, 0xff, TMP2
-       ldub    [T + TMP2], TMP2
+       and     $3, 0xff, TMP1          C  0
+       ldub    [T + TMP1], TMP1        C  0
        nop
-       xor     TMP1, TMP2, TMP2
-       stb     TMP2, [$8 + eval(4*$1)]
+       xor     TMP3, TMP1, TMP1        C  0
+       stb     TMP1, [$8 + eval(4*$1)] C  0
        
-       srl     $4, 8, TMP2
-       and     TMP2, 0xff, TMP2
-       ldub    [T + TMP2], TMP2
-       srl     TMP1, 8, TMP1
-       xor     TMP1, TMP2, TMP2
-       stb     TMP2, [$8 + eval(4*$1 + 1)]
+       srl     $4, 8, TMP2             C  1
+       and     TMP2, 0xff, TMP2        C  1
+       ldub    [T + TMP2], TMP2        C  1
+       srl     TMP3, 8, TMP3           C  1
+       xor     TMP3, TMP2, TMP2        C  1
+       stb     TMP2, [$8 + eval(4*$1 + 1)]     C  1
 
-       srl     $5, 16, TMP2
-       and     TMP2, 0xff, TMP2
-       ldub    [T + TMP2], TMP2
-       srl     TMP1, 8, TMP1
-       xor     TMP1, TMP2, TMP2
-       stb     TMP2, [$8 + eval(4*$1 + 2)]
+       srl     $5, 16, TMP1            C  2
+       and     TMP1, 0xff, TMP1        C  2
+       ldub    [T + TMP1], TMP1        C  2
+       srl     TMP3, 8, TMP3           C  2
+       xor     TMP3, TMP1, TMP1        C  2
+       stb     TMP1, [$8 + eval(4*$1 + 2)]     C  2
 
-       srl     $6, 24, TMP2
-       ldub    [T + TMP2], TMP2
-       srl     TMP1, 8, TMP1
-       xor     TMP1, TMP2, TMP2
-       stb     TMP2, [$8 + eval(4*$1 + 3)]>)
+       srl     $6, 24, TMP2            C  3
+       ldub    [T + TMP2], TMP2        C  3
+       srl     TMP3, 8, TMP3           C  3
+       xor     TMP3, TMP2, TMP2        C  3
+       stb     TMP2, [$8 + eval(4*$1 + 3)]     C  3
+>)