diff --git a/include/polarssl/aesni.h b/include/polarssl/aesni.h
index b53c0c0..f3f9fc3 100644
--- a/include/polarssl/aesni.h
+++ b/include/polarssl/aesni.h
@@ -58,13 +58,29 @@
  * \param input    16-byte input block
  * \param output   16-byte output block
  *
- * \return         0 if success, 1 if operation failed
+ * \return         0 on success (cannot fail)
  */
 int aesni_crypt_ecb( aes_context *ctx,
                      int mode,
                      const unsigned char input[16],
                      unsigned char output[16] );
 
+/**
+ * \brief          GCM multiplication: c = a * b in GF(2^128)
+ *
+ * \param c        Result
+ * \param a        First operand
+ * \param b        Second operand
+ *
+ * \note           Both operands and result are bit strings interpreted as
+ *                 elements of GF(2^128) as per the GCM spec.
+ *
+ * \return         0 on success (cannot fail)
+ */
+int aesni_gcm_mult( unsigned char c[16],
+                    const unsigned char a[16],
+                    const unsigned char b[16] );
+
 #endif /* POLARSSL_HAVE_X86_64 */
 
 #endif /* POLARSSL_AESNI_H */
diff --git a/library/aesni.c b/library/aesni.c
index 9b41c36..0431f78 100644
--- a/library/aesni.c
+++ b/library/aesni.c
@@ -25,6 +25,7 @@
 
 /*
  * [AES-WP] http://software.intel.com/en-us/articles/intel-advanced-encryption-standard-aes-instructions-set
+ * [CLMUL-WP] http://software.intel.com/en-us/articles/intel-carry-less-multiplication-instruction-and-its-usage-for-computing-the-gcm-mode/
  */
 
 #include "polarssl/config.h"
@@ -101,6 +102,119 @@
 
     return( 0 );
 }
+
+/*
+ * GCM multiplication: c = a times b in GF(2^128)
+ * Based on [CLMUL-WP] algorithms 1 (with equation 27) and 5.
+ */
+int aesni_gcm_mult( unsigned char c[16],
+                    const unsigned char a[16],
+                    const unsigned char b[16] )
+{
+    unsigned char aa[16], bb[16], cc[16];
+    size_t i;
+
+    /* The inputs are in big-endian order, so byte-reverse them */
+    for( i = 0; i < 16; i++ )
+    {
+        aa[i] = a[15 - i];
+        bb[i] = b[15 - i];
+    }
+
+    asm( "movdqu (%0), %%xmm0               \n" // a1:a0
+         "movdqu (%1), %%xmm1               \n" // b1:b0
+
+         /*
+          * Caryless multiplication xmm2:xmm1 = xmm0 * xmm1
+          * using [CLMUL-WP] algorithm 1 (p. 13).
+          */
+         "movdqa %%xmm1, %%xmm2             \n" // copy of b1:b0
+         "movdqa %%xmm1, %%xmm3             \n" // same
+         "movdqa %%xmm1, %%xmm4             \n" // same
+         "pclmulqdq $0x00, %%xmm0, %%xmm1   \n" // a0*b0 = c1:c0
+         "pclmulqdq $0x11, %%xmm0, %%xmm2   \n" // a1*b1 = d1:d0
+         "pclmulqdq $0x10, %%xmm0, %%xmm3   \n" // a0*b1 = e1:e0
+         "pclmulqdq $0x01, %%xmm0, %%xmm4   \n" // a1*b0 = f1:f0
+         "pxor %%xmm3, %%xmm4               \n" // e1+f1:e0+f0
+         "movdqa %%xmm4, %%xmm3             \n" // same
+         "psrldq $8, %%xmm4                 \n" // 0:e1+f1
+         "pslldq $8, %%xmm3                 \n" // e0+f0:0
+         "pxor %%xmm4, %%xmm2               \n" // d1:d0+e1+f1
+         "pxor %%xmm3, %%xmm1               \n" // c1+e0+f1:c0
+
+         /*
+          * Now shift the result one bit to the left,
+          * taking advantage of [CLMUL-WP] eq 27 (p. 20)
+          */
+         "movdqa %%xmm1, %%xmm3             \n" // r1:r0
+         "movdqa %%xmm2, %%xmm4             \n" // r3:r2
+         "psllq $1, %%xmm1                  \n" // r1<<1:r0<<1
+         "psllq $1, %%xmm2                  \n" // r3<<1:r2<<1
+         "psrlq $63, %%xmm3                 \n" // r1>>63:r0>>63
+         "psrlq $63, %%xmm4                 \n" // r3>>63:r2>>63
+         "movdqa %%xmm3, %%xmm5             \n" // r1>>63:r0>>63
+         "pslldq $8, %%xmm3                 \n" // r0>>63:0
+         "pslldq $8, %%xmm4                 \n" // r2>>63:0
+         "psrldq $8, %%xmm5                 \n" // 0:r1>>63
+         "por %%xmm3, %%xmm1                \n" // r1<<1|r0>>63:r0<<1
+         "por %%xmm4, %%xmm2                \n" // r3<<1|r2>>62:r2<<1
+         "por %%xmm5, %%xmm2                \n" // r3<<1|r2>>62:r2<<1|r1>>63
+
+         /*
+          * Now reduce modulo the GCM polynomial x^128 + x^7 + x^2 + x + 1
+          * using [CLMUL-WP] algorithm 5 (p. 20).
+          * Currently xmm2:xmm1 holds x3:x2:x1:x0 (already shifted).
+          */
+         /* Step 2 (1) */
+         "movdqa %%xmm1, %%xmm3             \n" // x1:x0
+         "movdqa %%xmm1, %%xmm4             \n" // same
+         "movdqa %%xmm1, %%xmm5             \n" // same
+         "psllq $63, %%xmm3                 \n" // x1<<63:x0<<63 = stuff:a
+         "psllq $62, %%xmm4                 \n" // x1<<62:x0<<62 = stuff:b
+         "psllq $57, %%xmm5                 \n" // x1<<57:x0<<57 = stuff:c
+
+         /* Step 2 (2) */
+         "pxor %%xmm4, %%xmm3               \n" // stuff:a+b
+         "pxor %%xmm5, %%xmm3               \n" // stuff:a+b+c
+         "pslldq $8, %%xmm3                 \n" // a+b+c:0
+         "pxor %%xmm3, %%xmm1               \n" // x1+a+b+c:x0 = d:x0
+
+         /* Steps 3 and 4 */
+         "movdqa %%xmm1,%%xmm0              \n" // d:x0
+         "movdqa %%xmm1,%%xmm4              \n" // same
+         "movdqa %%xmm1,%%xmm5              \n" // same
+         "psrlq $1, %%xmm0                  \n" // e1:x0>>1 = e1:e0'
+         "psrlq $2, %%xmm4                  \n" // f1:x0>>2 = f1:f0'
+         "psrlq $7, %%xmm5                  \n" // g1:x0>>7 = g1:g0'
+         "pxor %%xmm4, %%xmm0               \n" // e1+f1:e0'+f0'
+         "pxor %%xmm5, %%xmm0               \n" // e1+f1+g1:e0'+f0'+g0'
+         // e0'+f0'+g0' is almost e0+f0+g0, except for some missing
+         // bits carried from d. Now get those bits back in.
+         "movdqa %%xmm1,%%xmm3              \n" // d:x0
+         "movdqa %%xmm1,%%xmm4              \n" // same
+         "movdqa %%xmm1,%%xmm5              \n" // same
+         "psllq $63, %%xmm3                 \n" // d<<63:stuff
+         "psllq $62, %%xmm4                 \n" // d<<62:stuff
+         "psllq $57, %%xmm5                 \n" // d<<57:stuff
+         "pxor %%xmm4, %%xmm3               \n" // d<<63+d<<62:stuff
+         "pxor %%xmm5, %%xmm3               \n" // missing bits of d:stuff
+         "psrldq $8, %%xmm3                 \n" // 0:missing bits of d
+         "pxor %%xmm3, %%xmm0               \n" // e1+f1+g1:e0+f0+g0
+         "pxor %%xmm1, %%xmm0               \n" // h1:h0
+         "pxor %%xmm2, %%xmm0               \n" // x3+h1:x2+h0
+
+         "movdqu %%xmm0, (%2)               \n" // done
+         :
+         : "r" (aa), "r" (bb), "r" (cc)
+         : "memory", "cc", "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5" );
+
+    /* Now byte-reverse the outputs */
+    for( i = 0; i < 16; i++ )
+        c[i] = cc[15 - i];
+
+    return( 0 );
+}
+
 #endif /* POLARSSL_HAVE_X86_64 */
 
 #endif /* POLARSSL_AESNI_C */