Merge branch 'master' into bit-fields-type

[openjpeg.git] / src / lib / openjp2 / dwt.c

diff --git a/src/lib/openjp2/dwt.c b/src/lib/openjp2/dwt.c
index 4ad99ed98c3ef99c00fc743f6b5de6688e6f0aa6..4fce8b209c085dcf90742cdb675825376e5c95de 100644 (file)
--- a/src/lib/openjp2/dwt.c
+++ b/src/lib/openjp2/dwt.c
@@ -129,16 +129,16 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_tcd_tilecomp_t* tilec, OPJ_UINT32 i, DWT
 static OPJ_BOOL opj_dwt_encode_procedure(      opj_tcd_tilecomp_t * tilec,
                                                                                    void (*p_function)(OPJ_INT32 *, OPJ_INT32,OPJ_INT32,OPJ_INT32) );
 
-static OPJ_UINT32 opj_dwt_max_resolution(opj_tcd_resolution_t* restrict r, OPJ_UINT32 i);
+static OPJ_UINT32 opj_dwt_max_resolution(opj_tcd_resolution_t* OPJ_RESTRICT r, OPJ_UINT32 i);
 
 /* <summary>                             */
 /* Inverse 9-7 wavelet transform in 1-D. */
 /* </summary>                            */
-static void opj_v4dwt_decode(opj_v4dwt_t* restrict dwt);
+static void opj_v4dwt_decode(opj_v4dwt_t* OPJ_RESTRICT dwt);
 
-static void opj_v4dwt_interleave_h(opj_v4dwt_t* restrict w, OPJ_FLOAT32* restrict a, OPJ_INT32 x, OPJ_INT32 size);
+static void opj_v4dwt_interleave_h(opj_v4dwt_t* OPJ_RESTRICT w, OPJ_FLOAT32* OPJ_RESTRICT a, OPJ_INT32 x, OPJ_INT32 size);
 
-static void opj_v4dwt_interleave_v(opj_v4dwt_t* restrict v , OPJ_FLOAT32* restrict a , OPJ_INT32 x, OPJ_INT32 nb_elts_read);
+static void opj_v4dwt_interleave_v(opj_v4dwt_t* OPJ_RESTRICT v , OPJ_FLOAT32* OPJ_RESTRICT a , OPJ_INT32 x, OPJ_INT32 nb_elts_read);
 
 #ifdef __SSE__
 static void opj_v4dwt_decode_step1_sse(opj_v4_t* w, OPJ_INT32 count, const __m128 c);
@@ -409,7 +409,9 @@ static INLINE OPJ_BOOL opj_dwt_encode_procedure(opj_tcd_tilecomp_t * tilec,void
 
        l_data_size = opj_dwt_max_resolution( tilec->resolutions,tilec->numresolutions) * (OPJ_UINT32)sizeof(OPJ_INT32);
        bj = (OPJ_INT32*)opj_malloc((size_t)l_data_size);
-       if (! bj) {
+       /* l_data_size is equal to 0 when numresolutions == 1 but bj is not used */
+       /* in that case, so do not error out */
+       if (l_data_size != 0 && ! bj) {
                return OPJ_FALSE;
        }
        i = l;
@@ -541,7 +543,7 @@ void opj_dwt_calc_explicit_stepsizes(opj_tccp_t * tccp, OPJ_UINT32 prec) {
 /* <summary>                             */
 /* Determine maximum computed resolution level for inverse wavelet transform */
 /* </summary>                            */
-static OPJ_UINT32 opj_dwt_max_resolution(opj_tcd_resolution_t* restrict r, OPJ_UINT32 i) {
+static OPJ_UINT32 opj_dwt_max_resolution(opj_tcd_resolution_t* OPJ_RESTRICT r, OPJ_UINT32 i) {
        OPJ_UINT32 mr   = 0;
        OPJ_UINT32 w;
        while( --i ) {
@@ -567,9 +569,11 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_tcd_tilecomp_t* tilec, OPJ_UINT32 numres
        OPJ_UINT32 rh = (OPJ_UINT32)(tr->y1 - tr->y0);  /* height of the resolution level computed */
 
        OPJ_UINT32 w = (OPJ_UINT32)(tilec->x1 - tilec->x0);
-
-       h.mem = (OPJ_INT32*)
-       opj_aligned_malloc(opj_dwt_max_resolution(tr, numres) * sizeof(OPJ_INT32));
+       
+       if (numres == 1U) {
+               return OPJ_TRUE;
+       }
+       h.mem = (OPJ_INT32*)opj_aligned_malloc(opj_dwt_max_resolution(tr, numres) * sizeof(OPJ_INT32));
        if (! h.mem){
                /* FIXME event manager error callback */
                return OPJ_FALSE;
@@ -578,7 +582,7 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_tcd_tilecomp_t* tilec, OPJ_UINT32 numres
        v.mem = h.mem;
 
        while( --numres) {
-               OPJ_INT32 * restrict tiledp = tilec->data;
+               OPJ_INT32 * OPJ_RESTRICT tiledp = tilec->data;
                OPJ_UINT32 j;
 
                ++tr;
@@ -613,8 +617,8 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_tcd_tilecomp_t* tilec, OPJ_UINT32 numres
        return OPJ_TRUE;
 }
 
-static void opj_v4dwt_interleave_h(opj_v4dwt_t* restrict w, OPJ_FLOAT32* restrict a, OPJ_INT32 x, OPJ_INT32 size){
-       OPJ_FLOAT32* restrict bi = (OPJ_FLOAT32*) (w->wavelet + w->cas);
+static void opj_v4dwt_interleave_h(opj_v4dwt_t* OPJ_RESTRICT w, OPJ_FLOAT32* OPJ_RESTRICT a, OPJ_INT32 x, OPJ_INT32 size){
+       OPJ_FLOAT32* OPJ_RESTRICT bi = (OPJ_FLOAT32*) (w->wavelet + w->cas);
        OPJ_INT32 count = w->sn;
        OPJ_INT32 i, k;
 
@@ -656,8 +660,8 @@ static void opj_v4dwt_interleave_h(opj_v4dwt_t* restrict w, OPJ_FLOAT32* restric
        }
 }
 
-static void opj_v4dwt_interleave_v(opj_v4dwt_t* restrict v , OPJ_FLOAT32* restrict a , OPJ_INT32 x, OPJ_INT32 nb_elts_read){
-       opj_v4_t* restrict bi = v->wavelet + v->cas;
+static void opj_v4dwt_interleave_v(opj_v4dwt_t* OPJ_RESTRICT v , OPJ_FLOAT32* OPJ_RESTRICT a , OPJ_INT32 x, OPJ_INT32 nb_elts_read){
+       opj_v4_t* OPJ_RESTRICT bi = v->wavelet + v->cas;
        OPJ_INT32 i;
 
        for(i = 0; i < v->sn; ++i){
@@ -675,7 +679,7 @@ static void opj_v4dwt_interleave_v(opj_v4dwt_t* restrict v , OPJ_FLOAT32* restri
 #ifdef __SSE__
 
 static void opj_v4dwt_decode_step1_sse(opj_v4_t* w, OPJ_INT32 count, const __m128 c){
-       __m128* restrict vw = (__m128*) w;
+       __m128* OPJ_RESTRICT vw = (__m128*) w;
        OPJ_INT32 i;
        /* 4x unrolled loop */
        for(i = 0; i < count >> 2; ++i){
@@ -696,8 +700,8 @@ static void opj_v4dwt_decode_step1_sse(opj_v4_t* w, OPJ_INT32 count, const __m12
 }
 
 void opj_v4dwt_decode_step2_sse(opj_v4_t* l, opj_v4_t* w, OPJ_INT32 k, OPJ_INT32 m, __m128 c){
-       __m128* restrict vl = (__m128*) l;
-       __m128* restrict vw = (__m128*) w;
+       __m128* OPJ_RESTRICT vl = (__m128*) l;
+       __m128* OPJ_RESTRICT vw = (__m128*) w;
        OPJ_INT32 i;
        __m128 tmp1, tmp2, tmp3;
        tmp1 = vl[0];
@@ -725,7 +729,7 @@ void opj_v4dwt_decode_step2_sse(opj_v4_t* l, opj_v4_t* w, OPJ_INT32 k, OPJ_INT32
 
 static void opj_v4dwt_decode_step1(opj_v4_t* w, OPJ_INT32 count, const OPJ_FLOAT32 c)
 {
-       OPJ_FLOAT32* restrict fw = (OPJ_FLOAT32*) w;
+       OPJ_FLOAT32* OPJ_RESTRICT fw = (OPJ_FLOAT32*) w;
        OPJ_INT32 i;
        for(i = 0; i < count; ++i){
                OPJ_FLOAT32 tmp1 = fw[i*8    ];
@@ -741,8 +745,8 @@ static void opj_v4dwt_decode_step1(opj_v4_t* w, OPJ_INT32 count, const OPJ_FLOAT
 
 static void opj_v4dwt_decode_step2(opj_v4_t* l, opj_v4_t* w, OPJ_INT32 k, OPJ_INT32 m, OPJ_FLOAT32 c)
 {
-       OPJ_FLOAT32* restrict fl = (OPJ_FLOAT32*) l;
-       OPJ_FLOAT32* restrict fw = (OPJ_FLOAT32*) w;
+       OPJ_FLOAT32* fl = (OPJ_FLOAT32*) l;
+       OPJ_FLOAT32* fw = (OPJ_FLOAT32*) w;
        OPJ_INT32 i;
        for(i = 0; i < m; ++i){
                OPJ_FLOAT32 tmp1_1 = fl[0];
@@ -793,7 +797,7 @@ static void opj_v4dwt_decode_step2(opj_v4_t* l, opj_v4_t* w, OPJ_INT32 k, OPJ_IN
 /* <summary>                             */
 /* Inverse 9-7 wavelet transform in 1-D. */
 /* </summary>                            */
-void opj_v4dwt_decode(opj_v4dwt_t* restrict dwt)
+static void opj_v4dwt_decode(opj_v4dwt_t* OPJ_RESTRICT dwt)
 {
        OPJ_INT32 a, b;
        if(dwt->cas == 0) {
@@ -830,7 +834,7 @@ void opj_v4dwt_decode(opj_v4dwt_t* restrict dwt)
 /* <summary>                             */
 /* Inverse 9-7 wavelet transform in 2-D. */
 /* </summary>                            */
-OPJ_BOOL opj_dwt_decode_real(opj_tcd_tilecomp_t* restrict tilec, OPJ_UINT32 numres)
+OPJ_BOOL opj_dwt_decode_real(opj_tcd_tilecomp_t* OPJ_RESTRICT tilec, OPJ_UINT32 numres)
 {
        opj_v4dwt_t h;
        opj_v4dwt_t v;
@@ -850,7 +854,7 @@ OPJ_BOOL opj_dwt_decode_real(opj_tcd_tilecomp_t* restrict tilec, OPJ_UINT32 numr
        v.wavelet = h.wavelet;
 
        while( --numres) {
-               OPJ_FLOAT32 * restrict aj = (OPJ_FLOAT32*) tilec->data;
+               OPJ_FLOAT32 * OPJ_RESTRICT aj = (OPJ_FLOAT32*) tilec->data;
                OPJ_UINT32 bufsize = (OPJ_UINT32)((tilec->x1 - tilec->x0) * (tilec->y1 - tilec->y0));
                OPJ_INT32 j;