Various changes to allow tile buffers of more than 4giga pixels

Untested though, since that means a tile buffer of at least 16 GB. So
there might be places where uint32 overflow on multiplication still occur...

8 files changed, 119 insertions, 106 deletions

diff --git a/src/lib/openjp2/dwt.c b/src/lib/openjp2/dwt.c
index 81a8150a..d0cd3b4e 100644
--- a/src/lib/openjp2/dwt.c
+++ b/src/lib/openjp2/dwt.c
@@ -279,7 +279,7 @@ static void opj_dwt_deinterleave_v(OPJ_INT32 *a, OPJ_INT32 *b, OPJ_INT32 dn,
         l_src += 2;
     } /* b[i*x]=a[2*i+cas]; */
 
-    l_dest = b + sn * x;
+    l_dest = b + (size_t)sn * (size_t)x;
     l_src = a + 1 - cas;
 
     i = dn;
@@ -325,7 +325,7 @@ static void opj_dwt_interleave_v(const opj_dwt_t* v, OPJ_INT32 *a, OPJ_INT32 x)
         bi += 2;
         ai += x;
     }
-    ai = a + (v->sn * x);
+    ai = a + (v->sn * (size_t)x);
     bi = v->mem + 1 - v->cas;
     i = v->dn ;
     while (i--) {
@@ -616,7 +616,7 @@ static
 void opj_idwt53_v_final_memcpy(OPJ_INT32* tiledp_col,
                                const OPJ_INT32* tmp,
                                OPJ_INT32 len,
-                               OPJ_INT32 stride)
+                               size_t stride)
 {
     OPJ_INT32 i;
     for (i = 0; i < len; ++i) {
@@ -625,9 +625,9 @@ void opj_idwt53_v_final_memcpy(OPJ_INT32* tiledp_col,
                     PARALLEL_COLS_53 * sizeof(OPJ_INT32))
            would do but would be a tiny bit slower.
            We can take here advantage of our knowledge of alignment */
-        STOREU(&tiledp_col[i * stride + 0],
+        STOREU(&tiledp_col[(size_t)i * stride + 0],
                LOAD(&tmp[PARALLEL_COLS_53 * i + 0]));
-        STOREU(&tiledp_col[i * stride + VREG_INT_COUNT],
+        STOREU(&tiledp_col[(size_t)i * stride + VREG_INT_COUNT],
                LOAD(&tmp[PARALLEL_COLS_53 * i + VREG_INT_COUNT]));
     }
 }
@@ -639,12 +639,13 @@ static void opj_idwt53_v_cas0_mcols_SSE2_OR_AVX2(
     const OPJ_INT32 sn,
     const OPJ_INT32 len,
     OPJ_INT32* tiledp_col,
-    const OPJ_INT32 stride)
+    const size_t stride)
 {
     const OPJ_INT32* in_even = &tiledp_col[0];
-    const OPJ_INT32* in_odd = &tiledp_col[sn * stride];
+    const OPJ_INT32* in_odd = &tiledp_col[(size_t)sn * stride];
 
-    OPJ_INT32 i, j;
+    OPJ_INT32 i;
+    size_t j;
     VREG d1c_0, d1n_0, s1n_0, s0c_0, s0n_0;
     VREG d1c_1, d1n_1, s1n_1, s0c_1, s0n_1;
     const VREG two = LOAD_CST(2);
@@ -703,7 +704,7 @@ static void opj_idwt53_v_cas0_mcols_SSE2_OR_AVX2(
 
     if (len & 1) {
         VREG tmp_len_minus_1;
-        s1n_0 = LOADU(in_even + ((len - 1) / 2) * stride);
+        s1n_0 = LOADU(in_even + (size_t)((len - 1) / 2) * stride);
         /* tmp_len_minus_1 = s1n - ((d1n + 1) >> 1); */
         tmp_len_minus_1 = SUB(s1n_0, SAR(ADD3(d1n_0, d1n_0, two), 2));
         STORE(tmp + PARALLEL_COLS_53 * (len - 1), tmp_len_minus_1);
@@ -711,7 +712,7 @@ static void opj_idwt53_v_cas0_mcols_SSE2_OR_AVX2(
         STORE(tmp + PARALLEL_COLS_53 * (len - 2),
               ADD(d1n_0, SAR(ADD(s0n_0, tmp_len_minus_1), 1)));
 
-        s1n_1 = LOADU(in_even + ((len - 1) / 2) * stride + VREG_INT_COUNT);
+        s1n_1 = LOADU(in_even + (size_t)((len - 1) / 2) * stride + VREG_INT_COUNT);
         /* tmp_len_minus_1 = s1n - ((d1n + 1) >> 1); */
         tmp_len_minus_1 = SUB(s1n_1, SAR(ADD3(d1n_1, d1n_1, two), 2));
         STORE(tmp + PARALLEL_COLS_53 * (len - 1) + VREG_INT_COUNT,
@@ -739,15 +740,16 @@ static void opj_idwt53_v_cas1_mcols_SSE2_OR_AVX2(
     const OPJ_INT32 sn,
     const OPJ_INT32 len,
     OPJ_INT32* tiledp_col,
-    const OPJ_INT32 stride)
+    const size_t stride)
 {
-    OPJ_INT32 i, j;
+    OPJ_INT32 i;
+    size_t j;
 
     VREG s1_0, s2_0, dc_0, dn_0;
     VREG s1_1, s2_1, dc_1, dn_1;
     const VREG two = LOAD_CST(2);
 
-    const OPJ_INT32* in_even = &tiledp_col[sn * stride];
+    const OPJ_INT32* in_even = &tiledp_col[(size_t)sn * stride];
     const OPJ_INT32* in_odd = &tiledp_col[0];
 
     assert(len > 2);
@@ -807,9 +809,9 @@ static void opj_idwt53_v_cas1_mcols_SSE2_OR_AVX2(
 
     if (!(len & 1)) {
         /*dn = in_odd[(len / 2 - 1) * stride] - ((s1 + 1) >> 1); */
-        dn_0 = SUB(LOADU(in_odd + (len / 2 - 1) * stride),
+        dn_0 = SUB(LOADU(in_odd + (size_t)(len / 2 - 1) * stride),
                    SAR(ADD3(s1_0, s1_0, two), 2));
-        dn_1 = SUB(LOADU(in_odd + (len / 2 - 1) * stride + VREG_INT_COUNT),
+        dn_1 = SUB(LOADU(in_odd + (size_t)(len / 2 - 1) * stride + VREG_INT_COUNT),
                    SAR(ADD3(s1_1, s1_1, two), 2));
 
         /* tmp[len - 2] = s1 + ((dn + dc) >> 1); */
@@ -849,7 +851,7 @@ static void opj_idwt3_v_cas0(OPJ_INT32* tmp,
                              const OPJ_INT32 sn,
                              const OPJ_INT32 len,
                              OPJ_INT32* tiledp_col,
-                             const OPJ_INT32 stride)
+                             const size_t stride)
 {
     OPJ_INT32 i, j;
     OPJ_INT32 d1c, d1n, s1n, s0c, s0n;
@@ -860,15 +862,15 @@ static void opj_idwt3_v_cas0(OPJ_INT32* tmp,
     /* accesses and explicit interleaving. */
 
     s1n = tiledp_col[0];
-    d1n = tiledp_col[sn * stride];
+    d1n = tiledp_col[(size_t)sn * stride];
     s0n = s1n - ((d1n + 1) >> 1);
 
     for (i = 0, j = 0; i < (len - 3); i += 2, j++) {
         d1c = d1n;
         s0c = s0n;
 
-        s1n = tiledp_col[(j + 1) * stride];
-        d1n = tiledp_col[(sn + j + 1) * stride];
+        s1n = tiledp_col[(size_t)(j + 1) * stride];
+        d1n = tiledp_col[(size_t)(sn + j + 1) * stride];
 
         s0n = s1n - ((d1c + d1n + 2) >> 2);
 
@@ -880,7 +882,7 @@ static void opj_idwt3_v_cas0(OPJ_INT32* tmp,
 
     if (len & 1) {
         tmp[len - 1] =
-            tiledp_col[((len - 1) / 2) * stride] -
+            tiledp_col[(size_t)((len - 1) / 2) * stride] -
             ((d1n + 1) >> 1);
         tmp[len - 2] = d1n + ((s0n + tmp[len - 1]) >> 1);
     } else {
@@ -888,7 +890,7 @@ static void opj_idwt3_v_cas0(OPJ_INT32* tmp,
     }
 
     for (i = 0; i < len; ++i) {
-        tiledp_col[i * stride] = tmp[i];
+        tiledp_col[(size_t)i * stride] = tmp[i];
     }
 }
 
@@ -899,11 +901,11 @@ static void opj_idwt3_v_cas1(OPJ_INT32* tmp,
                              const OPJ_INT32 sn,
                              const OPJ_INT32 len,
                              OPJ_INT32* tiledp_col,
-                             const OPJ_INT32 stride)
+                             const size_t stride)
 {
     OPJ_INT32 i, j;
     OPJ_INT32 s1, s2, dc, dn;
-    const OPJ_INT32* in_even = &tiledp_col[sn * stride];
+    const OPJ_INT32* in_even = &tiledp_col[(size_t)sn * stride];
     const OPJ_INT32* in_odd = &tiledp_col[0];
 
     assert(len > 2);
@@ -916,9 +918,9 @@ static void opj_idwt3_v_cas1(OPJ_INT32* tmp,
     tmp[0] = in_even[0] + dc;
     for (i = 1, j = 1; i < (len - 2 - !(len & 1)); i += 2, j++) {
 
-        s2 = in_even[(j + 1) * stride];
+        s2 = in_even[(size_t)(j + 1) * stride];
 
-        dn = in_odd[j * stride] - ((s1 + s2 + 2) >> 2);
+        dn = in_odd[(size_t)j * stride] - ((s1 + s2 + 2) >> 2);
         tmp[i  ] = dc;
         tmp[i + 1] = s1 + ((dn + dc) >> 1);
 
@@ -927,7 +929,7 @@ static void opj_idwt3_v_cas1(OPJ_INT32* tmp,
     }
     tmp[i] = dc;
     if (!(len & 1)) {
-        dn = in_odd[(len / 2 - 1) * stride] - ((s1 + 1) >> 1);
+        dn = in_odd[(size_t)(len / 2 - 1) * stride] - ((s1 + 1) >> 1);
         tmp[len - 2] = s1 + ((dn + dc) >> 1);
         tmp[len - 1] = dn;
     } else {
@@ -935,7 +937,7 @@ static void opj_idwt3_v_cas1(OPJ_INT32* tmp,
     }
 
     for (i = 0; i < len; ++i) {
-        tiledp_col[i * stride] = tmp[i];
+        tiledp_col[(size_t)i * stride] = tmp[i];
     }
 }
 #endif /* !defined(STANDARD_SLOW_VERSION) */
@@ -946,7 +948,7 @@ static void opj_idwt3_v_cas1(OPJ_INT32* tmp,
 /* Performs interleave, inverse wavelet transform and copy back to buffer */
 static void opj_idwt53_v(const opj_dwt_t *dwt,
                          OPJ_INT32* tiledp_col,
-                         OPJ_INT32 stride,
+                         size_t stride,
                          OPJ_INT32 nb_cols)
 {
 #ifdef STANDARD_SLOW_VERSION
@@ -994,14 +996,14 @@ static void opj_idwt53_v(const opj_dwt_t *dwt,
             OPJ_INT32* out = dwt->mem;
             for (c = 0; c < nb_cols; c++, tiledp_col++) {
                 OPJ_INT32 i;
-                const OPJ_INT32* in_even = &tiledp_col[sn * stride];
+                const OPJ_INT32* in_even = &tiledp_col[(size_t)sn * stride];
                 const OPJ_INT32* in_odd = &tiledp_col[0];
 
                 out[1] = in_odd[0] - ((in_even[0] + 1) >> 1);
                 out[0] = in_even[0] + out[1];
 
                 for (i = 0; i < len; ++i) {
-                    tiledp_col[i * stride] = out[i];
+                    tiledp_col[(size_t)i * stride] = out[i];
                 }
             }
 
@@ -1341,11 +1343,11 @@ static void opj_dwt_decode_v_func(void* user_data, opj_tls_t* tls)
     job = (opj_dwd_decode_v_job_t*)user_data;
     for (j = job->min_j; j + PARALLEL_COLS_53 <= job->max_j;
             j += PARALLEL_COLS_53) {
-        opj_idwt53_v(&job->v, &job->tiledp[j], (OPJ_INT32)job->w,
+        opj_idwt53_v(&job->v, &job->tiledp[j], (size_t)job->w,
                      PARALLEL_COLS_53);
     }
     if (j < job->max_j)
-        opj_idwt53_v(&job->v, &job->tiledp[j], (OPJ_INT32)job->w,
+        opj_idwt53_v(&job->v, &job->tiledp[j], (size_t)job->w,
                      (OPJ_INT32)(job->max_j - j));
 
     opj_aligned_free(job->v.mem);
@@ -1413,7 +1415,7 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_thread_pool_t* tp,
 
         if (num_threads <= 1 || rh <= 1) {
             for (j = 0; j < rh; ++j) {
-                opj_idwt53_h(&h, &tiledp[j * w]);
+                opj_idwt53_h(&h, &tiledp[(size_t)j * w]);
             }
         } else {
             OPJ_UINT32 num_jobs = (OPJ_UINT32)num_threads;
@@ -1465,10 +1467,10 @@ static OPJ_BOOL opj_dwt_decode_tile(opj_thread_pool_t* tp,
         if (num_threads <= 1 || rw <= 1) {
             for (j = 0; j + PARALLEL_COLS_53 <= rw;
                     j += PARALLEL_COLS_53) {
-                opj_idwt53_v(&v, &tiledp[j], (OPJ_INT32)w, PARALLEL_COLS_53);
+                opj_idwt53_v(&v, &tiledp[j], (size_t)w, PARALLEL_COLS_53);
             }
             if (j < rw) {
-                opj_idwt53_v(&v, &tiledp[j], (OPJ_INT32)w, (OPJ_INT32)(rw - j));
+                opj_idwt53_v(&v, &tiledp[j], (size_t)w, (OPJ_INT32)(rw - j));
             }
         } else {
             OPJ_UINT32 num_jobs = (OPJ_UINT32)num_threads;
@@ -2035,14 +2037,16 @@ static void opj_v4dwt_interleave_v(opj_v4dwt_t* OPJ_RESTRICT dwt,
     OPJ_UINT32 i;
 
     for (i = dwt->win_l_x0; i < dwt->win_l_x1; ++i) {
-        memcpy(&bi[i * 2], &a[i * width], (size_t)nb_elts_read * sizeof(OPJ_FLOAT32));
+        memcpy(&bi[i * 2], &a[i * (size_t)width],
+               (size_t)nb_elts_read * sizeof(OPJ_FLOAT32));
     }
 
-    a += (OPJ_UINT32)dwt->sn * width;
+    a += (OPJ_UINT32)dwt->sn * (size_t)width;
     bi = dwt->wavelet + 1 - dwt->cas;
 
     for (i = dwt->win_h_x0; i < dwt->win_h_x1; ++i) {
-        memcpy(&bi[i * 2], &a[i * width], (size_t)nb_elts_read * sizeof(OPJ_FLOAT32));
+        memcpy(&bi[i * 2], &a[i * (size_t)width],
+               (size_t)nb_elts_read * sizeof(OPJ_FLOAT32));
     }
 }
 
@@ -2325,9 +2329,9 @@ OPJ_BOOL opj_dwt_decode_tile_97(opj_tcd_tilecomp_t* OPJ_RESTRICT tilec,
 
             for (k = 0; k < rw; k++) {
                 aj[k      ] = h.wavelet[k].f[0];
-                aj[k + w  ] = h.wavelet[k].f[1];
-                aj[k + w * 2] = h.wavelet[k].f[2];
-                aj[k + w * 3] = h.wavelet[k].f[3];
+                aj[k + (size_t)w  ] = h.wavelet[k].f[1];
+                aj[k + (size_t)w * 2] = h.wavelet[k].f[2];
+                aj[k + (size_t)w * 3] = h.wavelet[k].f[3];
             }
 
             aj += w * 4;
@@ -2340,10 +2344,10 @@ OPJ_BOOL opj_dwt_decode_tile_97(opj_tcd_tilecomp_t* OPJ_RESTRICT tilec,
             for (k = 0; k < rw; k++) {
                 switch (rh - j) {
                 case 3:
-                    aj[k + w * 2] = h.wavelet[k].f[2];
+                    aj[k + (size_t)w * 2] = h.wavelet[k].f[2];
                 /* FALLTHRU */
                 case 2:
-                    aj[k + w  ] = h.wavelet[k].f[1];
+                    aj[k + (size_t)w  ] = h.wavelet[k].f[1];
                 /* FALLTHRU */
                 case 1:
                     aj[k] = h.wavelet[k].f[0];
@@ -2366,7 +2370,7 @@ OPJ_BOOL opj_dwt_decode_tile_97(opj_tcd_tilecomp_t* OPJ_RESTRICT tilec,
             opj_v4dwt_decode(&v);
 
             for (k = 0; k < rh; ++k) {
-                memcpy(&aj[k * w], &v.wavelet[k], 4 * sizeof(OPJ_FLOAT32));
+                memcpy(&aj[k * (size_t)w], &v.wavelet[k], 4 * sizeof(OPJ_FLOAT32));
             }
             aj += 4;
         }
@@ -2380,7 +2384,7 @@ OPJ_BOOL opj_dwt_decode_tile_97(opj_tcd_tilecomp_t* OPJ_RESTRICT tilec,
             opj_v4dwt_decode(&v);
 
             for (k = 0; k < rh; ++k) {
-                memcpy(&aj[k * w], &v.wavelet[k], (size_t)j * sizeof(OPJ_FLOAT32));
+                memcpy(&aj[k * (size_t)w], &v.wavelet[k], (size_t)j * sizeof(OPJ_FLOAT32));
             }
         }
     }
diff --git a/src/lib/openjp2/j2k.c b/src/lib/openjp2/j2k.c
index 400fcaa4..aef0c1c0 100644
--- a/src/lib/openjp2/j2k.c
+++ b/src/lib/openjp2/j2k.c
@@ -10923,7 +10923,7 @@ OPJ_BOOL opj_j2k_encode(opj_j2k_t * p_j2k,
 {
     OPJ_UINT32 i, j;
     OPJ_UINT32 l_nb_tiles;
-    OPJ_UINT32 l_max_tile_size = 0, l_current_tile_size;
+    OPJ_SIZE_T l_max_tile_size = 0, l_current_tile_size;
     OPJ_BYTE * l_current_data = 00;
     OPJ_BOOL l_reuse_data = OPJ_FALSE;
     opj_tcd_t* p_tcd = 00;
diff --git a/src/lib/openjp2/mct.c b/src/lib/openjp2/mct.c
index 8c82ee20..b79d4b87 100644
--- a/src/lib/openjp2/mct.c
+++ b/src/lib/openjp2/mct.c
@@ -77,7 +77,7 @@ void opj_mct_encode(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_SIZE_T i;
     const OPJ_SIZE_T len = n;
@@ -119,7 +119,7 @@ void opj_mct_encode(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_SIZE_T i;
     const OPJ_SIZE_T len = n;
@@ -146,7 +146,7 @@ void opj_mct_decode(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_SIZE_T i;
     const OPJ_SIZE_T len = n;
@@ -181,7 +181,7 @@ void opj_mct_decode(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_UINT32 i;
     for (i = 0; i < n; ++i) {
@@ -214,7 +214,7 @@ void opj_mct_encode_real(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_SIZE_T i;
     const OPJ_SIZE_T len = n;
@@ -359,7 +359,7 @@ void opj_mct_encode_real(
     OPJ_INT32* OPJ_RESTRICT c0,
     OPJ_INT32* OPJ_RESTRICT c1,
     OPJ_INT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_UINT32 i;
     for (i = 0; i < n; ++i) {
@@ -386,7 +386,7 @@ void opj_mct_decode_real(
     OPJ_FLOAT32* OPJ_RESTRICT c0,
     OPJ_FLOAT32* OPJ_RESTRICT c1,
     OPJ_FLOAT32* OPJ_RESTRICT c2,
-    OPJ_UINT32 n)
+    OPJ_SIZE_T n)
 {
     OPJ_UINT32 i;
 #ifdef __SSE__
@@ -451,13 +451,13 @@ OPJ_FLOAT64 opj_mct_getnorm_real(OPJ_UINT32 compno)
 
 OPJ_BOOL opj_mct_encode_custom(
     OPJ_BYTE * pCodingdata,
-    OPJ_UINT32 n,
+    OPJ_SIZE_T n,
     OPJ_BYTE ** pData,
     OPJ_UINT32 pNbComp,
     OPJ_UINT32 isSigned)
 {
     OPJ_FLOAT32 * lMct = (OPJ_FLOAT32 *) pCodingdata;
-    OPJ_UINT32 i;
+    OPJ_SIZE_T i;
     OPJ_UINT32 j;
     OPJ_UINT32 k;
     OPJ_UINT32 lNbMatCoeff = pNbComp * pNbComp;
@@ -505,13 +505,13 @@ OPJ_BOOL opj_mct_encode_custom(
 
 OPJ_BOOL opj_mct_decode_custom(
     OPJ_BYTE * pDecodingData,
-    OPJ_UINT32 n,
+    OPJ_SIZE_T n,
     OPJ_BYTE ** pData,
     OPJ_UINT32 pNbComp,
     OPJ_UINT32 isSigned)
 {
     OPJ_FLOAT32 * lMct;
-    OPJ_UINT32 i;
+    OPJ_SIZE_T i;
     OPJ_UINT32 j;
     OPJ_UINT32 k;
 
diff --git a/src/lib/openjp2/mct.h b/src/lib/openjp2/mct.h
index 0ed980e8..2e37ce73 100644
--- a/src/lib/openjp2/mct.h
+++ b/src/lib/openjp2/mct.h
@@ -61,7 +61,7 @@ Apply a reversible multi-component transform to an image
 @param n Number of samples for each component
 */
 void opj_mct_encode(OPJ_INT32* OPJ_RESTRICT c0, OPJ_INT32* OPJ_RESTRICT c1,
-                    OPJ_INT32* OPJ_RESTRICT c2, OPJ_UINT32 n);
+                    OPJ_INT32* OPJ_RESTRICT c2, OPJ_SIZE_T n);
 /**
 Apply a reversible multi-component inverse transform to an image
 @param c0 Samples for luminance component
@@ -70,7 +70,7 @@ Apply a reversible multi-component inverse transform to an image
 @param n Number of samples for each component
 */
 void opj_mct_decode(OPJ_INT32* OPJ_RESTRICT c0, OPJ_INT32* OPJ_RESTRICT c1,
-                    OPJ_INT32* OPJ_RESTRICT c2, OPJ_UINT32 n);
+                    OPJ_INT32* OPJ_RESTRICT c2, OPJ_SIZE_T n);
 /**
 Get norm of the basis function used for the reversible multi-component transform
 @param compno Number of the component (0->Y, 1->U, 2->V)
@@ -86,7 +86,7 @@ Apply an irreversible multi-component transform to an image
 @param n Number of samples for each component
 */
 void opj_mct_encode_real(OPJ_INT32* OPJ_RESTRICT c0, OPJ_INT32* OPJ_RESTRICT c1,
-                         OPJ_INT32* OPJ_RESTRICT c2, OPJ_UINT32 n);
+                         OPJ_INT32* OPJ_RESTRICT c2, OPJ_SIZE_T n);
 /**
 Apply an irreversible multi-component inverse transform to an image
 @param c0 Samples for luminance component
@@ -95,7 +95,7 @@ Apply an irreversible multi-component inverse transform to an image
 @param n Number of samples for each component
 */
 void opj_mct_decode_real(OPJ_FLOAT32* OPJ_RESTRICT c0,
-                         OPJ_FLOAT32* OPJ_RESTRICT c1, OPJ_FLOAT32* OPJ_RESTRICT c2, OPJ_UINT32 n);
+                         OPJ_FLOAT32* OPJ_RESTRICT c1, OPJ_FLOAT32* OPJ_RESTRICT c2, OPJ_SIZE_T n);
 /**
 Get norm of the basis function used for the irreversible multi-component transform
 @param compno Number of the component (0->Y, 1->U, 2->V)
@@ -114,7 +114,7 @@ FIXME DOC
 */
 OPJ_BOOL opj_mct_encode_custom(
     OPJ_BYTE * p_coding_data,
-    OPJ_UINT32 n,
+    OPJ_SIZE_T n,
     OPJ_BYTE ** p_data,
     OPJ_UINT32 p_nb_comp,
     OPJ_UINT32 is_signed);
@@ -129,7 +129,7 @@ FIXME DOC
 */
 OPJ_BOOL opj_mct_decode_custom(
     OPJ_BYTE * pDecodingData,
-    OPJ_UINT32 n,
+    OPJ_SIZE_T n,
     OPJ_BYTE ** pData,
     OPJ_UINT32 pNbComp,
     OPJ_UINT32 isSigned);
diff --git a/src/lib/openjp2/sparse_array.c b/src/lib/openjp2/sparse_array.c
index fb552f8b..3402dca2 100644
--- a/src/lib/openjp2/sparse_array.c
+++ b/src/lib/openjp2/sparse_array.c
@@ -141,27 +141,30 @@ static OPJ_BOOL opj_sparse_array_int32_read_or_write(
                 if (src_block == NULL) {
                     for (j = 0; j < y_incr; j++) {
                         if (buf_col_stride == 1) {
-                            memset(buf + (y - y0 + j) * buf_line_stride + (x - x0) * buf_col_stride,
+                            memset(buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0) * buf_col_stride,
                                    0,
                                    sizeof(OPJ_INT32) * x_incr);
                         } else {
                             OPJ_UINT32 k;
                             for (k = 0; k < x_incr; k++) {
-                                *(buf + (y - y0 + j) * buf_line_stride + (x - x0 + k) * buf_col_stride) = 0;
+                                *(buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0 + k) * buf_col_stride)
+                                    = 0;
                             }
                         }
                     }
                 } else {
                     for (j = 0; j < y_incr; j++) {
                         if (buf_col_stride == 1) {
-                            memcpy(buf + (y - y0 + j) * buf_line_stride + (x - x0) * buf_col_stride,
-                                   src_block + (block_y_offset + j) * sa->block_width + block_x_offset,
+                            memcpy(buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0) * buf_col_stride,
+                                   src_block + (block_y_offset + j) * (size_t)sa->block_width + block_x_offset,
                                    sizeof(OPJ_INT32) * x_incr);
                         } else {
                             OPJ_UINT32 k;
                             for (k = 0; k < x_incr; k++) {
-                                *(buf + (y - y0 + j) * buf_line_stride + (x - x0 + k) * buf_col_stride) =
-                                    *(src_block + (block_y_offset + j) * sa->block_width + block_x_offset + k);
+                                *(buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0 + k) * buf_col_stride)
+                                    =
+                                        *(src_block + (block_y_offset + j) * (size_t)sa->block_width + block_x_offset +
+                                          k);
                             }
                         }
                     }
@@ -178,14 +181,16 @@ static OPJ_BOOL opj_sparse_array_int32_read_or_write(
 
                 for (j = 0; j < y_incr; j++) {
                     if (buf_col_stride == 1) {
-                        memcpy(src_block + (block_y_offset + j) * sa->block_width + block_x_offset,
-                               buf + (y - y0 + j) * buf_line_stride + (x - x0) * buf_col_stride,
+                        memcpy(src_block + (block_y_offset + j) * (size_t)sa->block_width +
+                               block_x_offset,
+                               buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0) * buf_col_stride,
                                sizeof(OPJ_INT32) * x_incr);
                     } else {
                         OPJ_UINT32 k;
                         for (k = 0; k < x_incr; k++) {
-                            *(src_block + (block_y_offset + j) * sa->block_width + block_x_offset + k) =
-                                *(buf + (y - y0 + j) * buf_line_stride + (x - x0 + k) * buf_col_stride);
+                            *(src_block + (block_y_offset + j) * (size_t)sa->block_width + block_x_offset +
+                              k) =
+                                  *(buf + (y - y0 + j) * (size_t)buf_line_stride + (x - x0 + k) * buf_col_stride);
                         }
                     }
                 }
diff --git a/src/lib/openjp2/t1.c b/src/lib/openjp2/t1.c
index 953c7ab1..44a2f243 100644
--- a/src/lib/openjp2/t1.c
+++ b/src/lib/openjp2/t1.c
@@ -1696,8 +1696,8 @@ static void opj_t1_clbl_decode_processor(void* user_data, opj_tls_t* tls)
             }
         }
     } else if (tccp->qmfbid == 1) {
-        OPJ_INT32* OPJ_RESTRICT tiledp = &tilec->data[(OPJ_UINT32)y * tile_w +
-                                                       (OPJ_UINT32)x];
+        OPJ_INT32* OPJ_RESTRICT tiledp = &tilec->data[(size_t)y * tile_w +
+                                                   (size_t)x];
         for (j = 0; j < cblk_h; ++j) {
             i = 0;
             for (; i < (cblk_w & ~(OPJ_UINT32)3U); i += 4U) {
@@ -1705,19 +1705,19 @@ static void opj_t1_clbl_decode_processor(void* user_data, opj_tls_t* tls)
                 OPJ_INT32 tmp1 = datap[(j * cblk_w) + i + 1U];
                 OPJ_INT32 tmp2 = datap[(j * cblk_w) + i + 2U];
                 OPJ_INT32 tmp3 = datap[(j * cblk_w) + i + 3U];
-                ((OPJ_INT32*)tiledp)[(j * tile_w) + i + 0U] = tmp0 / 2;
-                ((OPJ_INT32*)tiledp)[(j * tile_w) + i + 1U] = tmp1 / 2;
-                ((OPJ_INT32*)tiledp)[(j * tile_w) + i + 2U] = tmp2 / 2;
-                ((OPJ_INT32*)tiledp)[(j * tile_w) + i + 3U] = tmp3 / 2;
+                ((OPJ_INT32*)tiledp)[(j * (size_t)tile_w) + i + 0U] = tmp0 / 2;
+                ((OPJ_INT32*)tiledp)[(j * (size_t)tile_w) + i + 1U] = tmp1 / 2;
+                ((OPJ_INT32*)tiledp)[(j * (size_t)tile_w) + i + 2U] = tmp2 / 2;
+                ((OPJ_INT32*)tiledp)[(j * (size_t)tile_w) + i + 3U] = tmp3 / 2;
             }
             for (; i < cblk_w; ++i) {
                 OPJ_INT32 tmp = datap[(j * cblk_w) + i];
-                ((OPJ_INT32*)tiledp)[(j * tile_w) + i] = tmp / 2;
+                ((OPJ_INT32*)tiledp)[(j * (size_t)tile_w) + i] = tmp / 2;
             }
         }
     } else {        /* if (tccp->qmfbid == 0) */
-        OPJ_FLOAT32* OPJ_RESTRICT tiledp = (OPJ_FLOAT32*) &tilec->data[(OPJ_UINT32)y *
-                                                         tile_w + (OPJ_UINT32)x];
+        OPJ_FLOAT32* OPJ_RESTRICT tiledp = (OPJ_FLOAT32*) &tilec->data[(size_t)y *
+                                                     tile_w + (size_t)x];
         for (j = 0; j < cblk_h; ++j) {
             OPJ_FLOAT32* OPJ_RESTRICT tiledp2 = tiledp;
             for (i = 0; i < cblk_w; ++i) {
@@ -2052,7 +2052,8 @@ OPJ_BOOL opj_t1_encode_cblks(opj_t1_t *t1,
                         OPJ_INT32* OPJ_RESTRICT tiledp;
                         OPJ_UINT32 cblk_w;
                         OPJ_UINT32 cblk_h;
-                        OPJ_UINT32 i, j, tileIndex = 0, tileLineAdvance;
+                        OPJ_UINT32 i, j, tileLineAdvance;
+                        size_t tileIndex = 0;
 
                         OPJ_INT32 x = cblk->x0 - band->x0;
                         OPJ_INT32 y = cblk->y0 - band->y0;
@@ -2076,7 +2077,7 @@ OPJ_BOOL opj_t1_encode_cblks(opj_t1_t *t1,
                         cblk_h = t1->h;
                         tileLineAdvance = tile_w - cblk_w;
 
-                        tiledp = &tilec->data[(OPJ_UINT32)y * tile_w + (OPJ_UINT32)x];
+                        tiledp = &tilec->data[(size_t)y * tile_w + (size_t)x];
                         t1->data = tiledp;
                         t1->data_stride = tile_w;
                         if (tccp->qmfbid == 1) {
diff --git a/src/lib/openjp2/tcd.c b/src/lib/openjp2/tcd.c
index 1213f757..485eb9e2 100644
--- a/src/lib/openjp2/tcd.c
+++ b/src/lib/openjp2/tcd.c
@@ -1974,17 +1974,17 @@ static OPJ_BOOL opj_tcd_mct_decode(opj_tcd_t *p_tcd, opj_event_mgr_t *p_manager)
                                           p_tcd->image->comps[1].resno_decoded;
         opj_tcd_resolution_t* res_comp2 = l_tile->comps[2].resolutions +
                                           p_tcd->image->comps[2].resno_decoded;
-        OPJ_INT32 l_res_samples = (OPJ_INT32)(res_comp0->x1 - res_comp0->x0) *
-                                  (res_comp0->y1 - res_comp0->y0);
+        size_t l_res_samples = (size_t)(res_comp0->x1 - res_comp0->x0) *
+                               (size_t)(res_comp0->y1 - res_comp0->y0);
         /* testcase 1336.pdf.asan.47.376 */
         if (p_tcd->image->comps[0].resno_decoded !=
                 p_tcd->image->comps[1].resno_decoded ||
                 p_tcd->image->comps[0].resno_decoded !=
                 p_tcd->image->comps[2].resno_decoded ||
-                (res_comp1->x1 - res_comp1->x0) * (res_comp1->y1 -
-                        res_comp1->y0) != l_res_samples ||
-                (res_comp2->x1 - res_comp2->x0) * (res_comp2->y1 -
-                        res_comp2->y0) != l_res_samples) {
+                (size_t)(res_comp1->x1 - res_comp1->x0) *
+                (size_t)(res_comp1->y1 - res_comp1->y0) != l_res_samples ||
+                (size_t)(res_comp2->x1 - res_comp2->x0) *
+                (size_t)(res_comp2->y1 - res_comp2->y0) != l_res_samples) {
             opj_event_msg(p_manager, EVT_ERROR,
                           "Tiles don't all have the same dimension. Skip the MCT step.\n");
             return OPJ_FALSE;
@@ -2225,9 +2225,10 @@ static void opj_tcd_code_block_enc_deallocate(opj_tcd_precinct_t * p_precinct)
     }
 }
 
-OPJ_UINT32 opj_tcd_get_encoded_tile_size(opj_tcd_t *p_tcd)
+OPJ_SIZE_T opj_tcd_get_encoded_tile_size(opj_tcd_t *p_tcd)
 {
-    OPJ_UINT32 i, l_data_size = 0;
+    OPJ_UINT32 i;
+    OPJ_SIZE_T l_data_size = 0;
     opj_image_comp_t * l_img_comp = 00;
     opj_tcd_tilecomp_t * l_tilec = 00;
     OPJ_UINT32 l_size_comp, l_remaining;
@@ -2246,8 +2247,8 @@ OPJ_UINT32 opj_tcd_get_encoded_tile_size(opj_tcd_t *p_tcd)
             l_size_comp = 4;
         }
 
-        l_data_size += l_size_comp * (OPJ_UINT32)((l_tilec->x1 - l_tilec->x0) *
-                       (l_tilec->y1 - l_tilec->y0));
+        l_data_size += l_size_comp * ((OPJ_SIZE_T)(l_tilec->x1 - l_tilec->x0) *
+                                      (OPJ_SIZE_T)(l_tilec->y1 - l_tilec->y0));
         ++l_img_comp;
         ++l_tilec;
     }
@@ -2262,7 +2263,7 @@ static OPJ_BOOL opj_tcd_dc_level_shift_encode(opj_tcd_t *p_tcd)
     opj_tccp_t * l_tccp = 00;
     opj_image_comp_t * l_img_comp = 00;
     opj_tcd_tile_t * l_tile;
-    OPJ_UINT32 l_nb_elem, i;
+    size_t l_nb_elem, i;
     OPJ_INT32 * l_current_ptr;
 
     l_tile = p_tcd->tcd_image->tiles;
@@ -2272,8 +2273,8 @@ static OPJ_BOOL opj_tcd_dc_level_shift_encode(opj_tcd_t *p_tcd)
 
     for (compno = 0; compno < l_tile->numcomps; compno++) {
         l_current_ptr = l_tile_comp->data;
-        l_nb_elem = (OPJ_UINT32)((l_tile_comp->x1 - l_tile_comp->x0) *
-                                 (l_tile_comp->y1 - l_tile_comp->y0));
+        l_nb_elem = (size_t)(l_tile_comp->x1 - l_tile_comp->x0) *
+                    (size_t)(l_tile_comp->y1 - l_tile_comp->y0);
 
         if (l_tccp->qmfbid == 1) {
             for (i = 0; i < l_nb_elem; ++i) {
@@ -2299,8 +2300,8 @@ static OPJ_BOOL opj_tcd_mct_encode(opj_tcd_t *p_tcd)
 {
     opj_tcd_tile_t * l_tile = p_tcd->tcd_image->tiles;
     opj_tcd_tilecomp_t * l_tile_comp = p_tcd->tcd_image->tiles->comps;
-    OPJ_UINT32 samples = (OPJ_UINT32)((l_tile_comp->x1 - l_tile_comp->x0) *
-                                      (l_tile_comp->y1 - l_tile_comp->y0));
+    size_t samples = (size_t)(l_tile_comp->x1 - l_tile_comp->x0) *
+                     (size_t)(l_tile_comp->y1 - l_tile_comp->y0);
     OPJ_UINT32 i;
     OPJ_BYTE ** l_data = 00;
     opj_tcp_t * l_tcp = p_tcd->tcp;
@@ -2482,13 +2483,15 @@ static OPJ_BOOL opj_tcd_rate_allocate_encode(opj_tcd_t *p_tcd,
 
 OPJ_BOOL opj_tcd_copy_tile_data(opj_tcd_t *p_tcd,
                                 OPJ_BYTE * p_src,
-                                OPJ_UINT32 p_src_length)
+                                OPJ_SIZE_T p_src_length)
 {
-    OPJ_UINT32 i, j, l_data_size = 0;
+    OPJ_UINT32 i;
+    OPJ_SIZE_T j;
+    OPJ_SIZE_T l_data_size = 0;
     opj_image_comp_t * l_img_comp = 00;
     opj_tcd_tilecomp_t * l_tilec = 00;
     OPJ_UINT32 l_size_comp, l_remaining;
-    OPJ_UINT32 l_nb_elem;
+    OPJ_SIZE_T l_nb_elem;
 
     l_data_size = opj_tcd_get_encoded_tile_size(p_tcd);
     if (l_data_size != p_src_length) {
@@ -2500,8 +2503,8 @@ OPJ_BOOL opj_tcd_copy_tile_data(opj_tcd_t *p_tcd,
     for (i = 0; i < p_tcd->image->numcomps; ++i) {
         l_size_comp = l_img_comp->prec >> 3; /*(/ 8)*/
         l_remaining = l_img_comp->prec & 7;  /* (%8) */
-        l_nb_elem = (OPJ_UINT32)((l_tilec->x1 - l_tilec->x0) * (l_tilec->y1 -
-                                 l_tilec->y0));
+        l_nb_elem = (size_t)(l_tilec->x1 - l_tilec->x0) *
+                    (size_t)(l_tilec->y1 - l_tilec->y0);
 
         if (l_remaining) {
             ++l_size_comp;
diff --git a/src/lib/openjp2/tcd.h b/src/lib/openjp2/tcd.h
index 7c974c53..63c22c45 100644
--- a/src/lib/openjp2/tcd.h
+++ b/src/lib/openjp2/tcd.h
@@ -409,7 +409,7 @@ OPJ_BOOL opj_tcd_update_tile_data(opj_tcd_t *p_tcd,
 /**
  *
  */
-OPJ_UINT32 opj_tcd_get_encoded_tile_size(opj_tcd_t *p_tcd);
+OPJ_SIZE_T opj_tcd_get_encoded_tile_size(opj_tcd_t *p_tcd);
 
 /**
  * Initialize the tile coder and may reuse some meory.
@@ -428,7 +428,7 @@ OPJ_BOOL opj_tcd_init_encode_tile(opj_tcd_t *p_tcd,
  */
 OPJ_BOOL opj_tcd_copy_tile_data(opj_tcd_t *p_tcd,
                                 OPJ_BYTE * p_src,
-                                OPJ_UINT32 p_src_length);
+                                OPJ_SIZE_T p_src_length);
 
 /**
  * Allocates tile component data