Fix compiler warnings

- celt/modes.c:430:14: warning: cast from 'const unsigned char *' to
  'opus_int16 *' increases required alignment from 1 to 2 [-Wcast-align]
- 'C[0][1]' may be used uninitialized [-Wmaybe-uninitialized]
- Unused variable/parameter
- Value stored is never read
- MSVC warnings about "possible loss of data" due to type conversions
- MSVC warning C4146: unary minus operator applied to unsigned type
- silk/NLSF_del_dec_quant.c:137:20: warning: array subscript is above
  array bounds [-Warray-bounds] (gcc -O3 false positive)
- src/mlp_train.h:39:20: warning: function declaration isn't a prototype
  [-Wstrict-prototypes]
- Remove SMALL_FOOTPRINT code from SSE 4.1 FIR implementation, matching
  the C implementation.

The clang -Wcast-align warnings with SSE intrinsics are a known
clang issue: https://llvm.org/bugs/show_bug.cgi?id=20670

6 files changed, 19 insertions, 17 deletions

diff --git a/silk/NLSF_del_dec_quant.c b/silk/NLSF_del_dec_quant.c
index 5155caef..44a16acd 100644
--- a/silk/NLSF_del_dec_quant.c
+++ b/silk/NLSF_del_dec_quant.c
@@ -131,7 +131,7 @@ opus_int32 silk_NLSF_del_dec_quant(                             /* O    Returns
             RD_Q25[ j + nStates ] = silk_SMLABB( silk_MLA( RD_tmp_Q25, silk_SMULBB( diff_Q10, diff_Q10 ), w_Q5[ i ] ), mu_Q20, rate1_Q5 );
         }
 
-        if( silk_LSHIFT( nStates, 1 ) <= NLSF_QUANT_DEL_DEC_STATES ) {
+        if( nStates <= NLSF_QUANT_DEL_DEC_STATES/2 ) {
             /* double number of states and copy */
             for( j = 0; j < nStates; j++ ) {
                 ind[ j + nStates ][ i ] = ind[ j ][ i ] + 1;
diff --git a/silk/enc_API.c b/silk/enc_API.c
index ba3db060..701c2905 100644
--- a/silk/enc_API.c
+++ b/silk/enc_API.c
@@ -233,7 +233,6 @@ opus_int silk_Encode(                                   /* O    Returns error co
         }
     }
 
-    TargetRate_bps = silk_RSHIFT32( encControl->bitRate, encControl->nChannelsInternal - 1 );
     for( n = 0; n < encControl->nChannelsInternal; n++ ) {
         /* Force the side channel to the same rate as the mid */
         opus_int force_fs_kHz = (n==1) ? psEnc->state_Fxx[0].sCmn.fs_kHz : 0;
diff --git a/silk/fixed/schur64_FIX.c b/silk/fixed/schur64_FIX.c
index 4d3b0932..b2cb12d9 100644
--- a/silk/fixed/schur64_FIX.c
+++ b/silk/fixed/schur64_FIX.c
@@ -43,7 +43,7 @@ opus_int32 silk_schur64(                            /* O    returns residual ene
     opus_int32 C[ SILK_MAX_ORDER_LPC + 1 ][ 2 ];
     opus_int32 Ctmp1_Q30, Ctmp2_Q30, rc_tmp_Q31;
 
-    silk_assert( order <= SILK_MAX_ORDER_LPC );
+    silk_assert( order >= 0 && order <= SILK_MAX_ORDER_LPC );
 
     /* Check for invalid input */
     if( c[ 0 ] <= 0 ) {
@@ -51,9 +51,10 @@ opus_int32 silk_schur64(                            /* O    returns residual ene
         return 0;
     }
 
-    for( k = 0; k < order + 1; k++ ) {
+    k = 0;
+    do {
         C[ k ][ 0 ] = C[ k ][ 1 ] = c[ k ];
-    }
+    } while( ++k <= order );
 
     for( k = 0; k < order; k++ ) {
         /* Check that we won't be getting an unstable rc, otherwise stop here. */
diff --git a/silk/fixed/schur_FIX.c b/silk/fixed/schur_FIX.c
index 9fe7f419..59d44a6f 100644
--- a/silk/fixed/schur_FIX.c
+++ b/silk/fixed/schur_FIX.c
@@ -43,28 +43,29 @@ opus_int32 silk_schur(                              /* O    Returns residual ene
     opus_int32    C[ SILK_MAX_ORDER_LPC + 1 ][ 2 ];
     opus_int32    Ctmp1, Ctmp2, rc_tmp_Q15;
 
-    silk_assert( order <= SILK_MAX_ORDER_LPC );
+    silk_assert( order >= 0 && order <= SILK_MAX_ORDER_LPC );
 
     /* Get number of leading zeros */
     lz = silk_CLZ32( c[ 0 ] );
 
     /* Copy correlations and adjust level to Q30 */
+    k = 0;
     if( lz < 2 ) {
         /* lz must be 1, so shift one to the right */
-        for( k = 0; k < order + 1; k++ ) {
+        do {
             C[ k ][ 0 ] = C[ k ][ 1 ] = silk_RSHIFT( c[ k ], 1 );
-        }
+        } while( ++k <= order );
     } else if( lz > 2 ) {
         /* Shift to the left */
         lz -= 2;
-        for( k = 0; k < order + 1; k++ ) {
+        do {
             C[ k ][ 0 ] = C[ k ][ 1 ] = silk_LSHIFT( c[ k ], lz );
-        }
+        } while( ++k <= order );
     } else {
         /* No need to shift */
-        for( k = 0; k < order + 1; k++ ) {
+        do {
             C[ k ][ 0 ] = C[ k ][ 1 ] = c[ k ];
-        }
+        } while( ++k <= order );
     }
 
     for( k = 0; k < order; k++ ) {
diff --git a/silk/fixed/x86/prefilter_FIX_sse.c b/silk/fixed/x86/prefilter_FIX_sse.c
index 488a603f..d8c9c2f5 100644
--- a/silk/fixed/x86/prefilter_FIX_sse.c
+++ b/silk/fixed/x86/prefilter_FIX_sse.c
@@ -107,8 +107,8 @@ void silk_warped_LPC_analysis_filter_FIX_sse4_1(
                 xmm_tempb = _mm_add_epi32( xmm_tempb, xmm_product2 );
                 xmm_tempa = _mm_add_epi32( xmm_tempa, xmm_tempb );
 
-                sum  = (coef_Q13_8 * state_8) >> 16;
-                sum += (coef_Q13_9 * state_9) >> 16;
+                sum  = (opus_int32)((coef_Q13_8 * state_8) >> 16);
+                sum += (opus_int32)((coef_Q13_9 * state_9) >> 16);
 
                 xmm_tempa = _mm_add_epi32( xmm_tempa, _mm_shuffle_epi32( xmm_tempa, _MM_SHUFFLE( 0, 0, 0, 2 ) ) );
                 sum += _mm_cvtsi128_si32( xmm_tempa);
diff --git a/silk/float/schur_FLP.c b/silk/float/schur_FLP.c
index f4b4072f..c1e0bbb5 100644
--- a/silk/float/schur_FLP.c
+++ b/silk/float/schur_FLP.c
@@ -41,12 +41,13 @@ silk_float silk_schur_FLP(                  /* O    returns residual energy
     double C[ SILK_MAX_ORDER_LPC + 1 ][ 2 ];
     double Ctmp1, Ctmp2, rc_tmp;
 
-    silk_assert( order <= SILK_MAX_ORDER_LPC );
+    silk_assert( order >= 0 && order <= SILK_MAX_ORDER_LPC );
 
     /* Copy correlations */
-    for( k = 0; k < order+1; k++ ) {
+    k = 0;
+    do {
         C[ k ][ 0 ] = C[ k ][ 1 ] = auto_corr[ k ];
-    }
+    } while( ++k <= order );
 
     for( k = 0; k < order; k++ ) {
         /* Get reflection coefficient */