* POSSIBILITY OF SUCH DAMAGE.
*/
+#define OPJ_SKIP_POISON
#include "opj_includes.h"
+
+#ifdef __SSE__
+#include <xmmintrin.h>
+#endif
+#ifdef __SSE2__
+#include <emmintrin.h>
+#endif
+
+#if defined(__GNUC__)
+#pragma GCC poison malloc calloc realloc free
+#endif
+
#include "t1_luts.h"
/** @defgroup T1 T1 - Implementation of the tier-1 coding */
if (!t1->encoder) {
OPJ_UINT32 datasize = w * h;
- if (datasize > (size_t)t1->datasize) {
+ if (datasize > t1->datasize) {
opj_aligned_free(t1->data);
t1->data = (OPJ_INT32*) opj_aligned_malloc(datasize * sizeof(OPJ_INT32));
if (!t1->data) {
}
typedef struct {
+ OPJ_BOOL whole_tile_decoding;
OPJ_UINT32 resno;
opj_tcd_cblk_dec_t* cblk;
opj_tcd_band_t* band;
OPJ_UINT32 tile_w;
job = (opj_t1_cblk_decode_processing_job_t*) user_data;
- resno = job->resno;
+
cblk = job->cblk;
+
+ if (!job->whole_tile_decoding) {
+ cblk_w = (OPJ_UINT32)(cblk->x1 - cblk->x0);
+ cblk_h = (OPJ_UINT32)(cblk->y1 - cblk->y0);
+
+ cblk->decoded_data = opj_aligned_malloc(cblk_w * cblk_h * sizeof(OPJ_INT32));
+ if (cblk->decoded_data == NULL) {
+ if (job->p_manager_mutex) {
+ opj_mutex_lock(job->p_manager_mutex);
+ }
+ opj_event_msg(job->p_manager, EVT_ERROR,
+ "Cannot allocate cblk->decoded_data\n");
+ if (job->p_manager_mutex) {
+ opj_mutex_unlock(job->p_manager_mutex);
+ }
+ *(job->pret) = OPJ_FALSE;
+ opj_free(job);
+ return;
+ }
+ /* Zero-init required */
+ memset(cblk->decoded_data, 0, cblk_w * cblk_h * sizeof(OPJ_INT32));
+ } else if (cblk->decoded_data) {
+ /* Not sure if that code path can happen, but better be */
+ /* safe than sorry */
+ opj_aligned_free(cblk->decoded_data);
+ cblk->decoded_data = NULL;
+ }
+
+ resno = job->resno;
band = job->band;
tilec = job->tilec;
tccp = job->tccp;
}
}
+ /* Both can be non NULL if for example decoding a full tile and then */
+ /* partially a tile. In which case partial decoding should be the */
+ /* priority */
+ assert((cblk->decoded_data != NULL) || (tilec->data != NULL));
+
if (cblk->decoded_data) {
+ OPJ_UINT32 cblk_size = cblk_w * cblk_h;
if (tccp->qmfbid == 1) {
- for (j = 0; j < cblk_h; ++j) {
- i = 0;
- for (; i < (cblk_w & ~(OPJ_UINT32)3U); i += 4U) {
- OPJ_INT32 tmp0 = datap[(j * cblk_w) + i + 0U];
- 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];
- datap[(j * cblk_w) + i + 0U] = tmp0 / 2;
- datap[(j * cblk_w) + i + 1U] = tmp1 / 2;
- datap[(j * cblk_w) + i + 2U] = tmp2 / 2;
- datap[(j * cblk_w) + i + 3U] = tmp3 / 2;
- }
- for (; i < cblk_w; ++i) {
- datap[(j * cblk_w) + i] /= 2;
- }
+ for (i = 0; i < cblk_size; ++i) {
+ datap[i] /= 2;
}
} else { /* if (tccp->qmfbid == 0) */
- for (j = 0; j < cblk_h; ++j) {
- for (i = 0; i < cblk_w; ++i) {
- OPJ_FLOAT32 tmp = ((OPJ_FLOAT32)(*datap)) * band->stepsize;
- memcpy(datap, &tmp, sizeof(tmp));
- datap++;
+ i = 0;
+#ifdef __SSE2__
+ {
+ const __m128 xmm_stepsize = _mm_set1_ps(band->stepsize);
+ for (; i < (cblk_size & ~15U); i += 16) {
+ __m128 xmm0_data = _mm_cvtepi32_ps(_mm_load_si128((__m128i * const)(
+ datap + 0)));
+ __m128 xmm1_data = _mm_cvtepi32_ps(_mm_load_si128((__m128i * const)(
+ datap + 4)));
+ __m128 xmm2_data = _mm_cvtepi32_ps(_mm_load_si128((__m128i * const)(
+ datap + 8)));
+ __m128 xmm3_data = _mm_cvtepi32_ps(_mm_load_si128((__m128i * const)(
+ datap + 12)));
+ _mm_store_ps((float*)(datap + 0), _mm_mul_ps(xmm0_data, xmm_stepsize));
+ _mm_store_ps((float*)(datap + 4), _mm_mul_ps(xmm1_data, xmm_stepsize));
+ _mm_store_ps((float*)(datap + 8), _mm_mul_ps(xmm2_data, xmm_stepsize));
+ _mm_store_ps((float*)(datap + 12), _mm_mul_ps(xmm3_data, xmm_stepsize));
+ datap += 16;
}
}
+#endif
+ for (; i < cblk_size; ++i) {
+ OPJ_FLOAT32 tmp = ((OPJ_FLOAT32)(*datap)) * band->stepsize;
+ memcpy(datap, &tmp, sizeof(tmp));
+ datap++;
+ }
}
} 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[(OPJ_SIZE_T)y * tile_w +
+ (OPJ_SIZE_T)x];
for (j = 0; j < cblk_h; ++j) {
i = 0;
for (; i < (cblk_w & ~(OPJ_UINT32)3U); i += 4U) {
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 * (OPJ_SIZE_T)tile_w) + i + 0U] = tmp0 / 2;
+ ((OPJ_INT32*)tiledp)[(j * (OPJ_SIZE_T)tile_w) + i + 1U] = tmp1 / 2;
+ ((OPJ_INT32*)tiledp)[(j * (OPJ_SIZE_T)tile_w) + i + 2U] = tmp2 / 2;
+ ((OPJ_INT32*)tiledp)[(j * (OPJ_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 * (OPJ_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[(OPJ_SIZE_T)y *
+ tile_w + (OPJ_SIZE_T)x];
for (j = 0; j < cblk_h; ++j) {
OPJ_FLOAT32* OPJ_RESTRICT tiledp2 = tiledp;
for (i = 0; i < cblk_w; ++i) {
opj_thread_pool_t* tp = tcd->thread_pool;
OPJ_UINT32 resno, bandno, precno, cblkno;
+#ifdef DEBUG_VERBOSE
+ OPJ_UINT32 codeblocks_decoded = 0;
+ printf("Enter opj_t1_decode_cblks()\n");
+#endif
+
for (resno = 0; resno < tilec->minimum_num_resolutions; ++resno) {
opj_tcd_resolution_t* res = &tilec->resolutions[resno];
(OPJ_UINT32)precinct->y0,
(OPJ_UINT32)precinct->x1,
(OPJ_UINT32)precinct->y1)) {
+ for (cblkno = 0; cblkno < precinct->cw * precinct->ch; ++cblkno) {
+ opj_tcd_cblk_dec_t* cblk = &precinct->cblks.dec[cblkno];
+ if (cblk->decoded_data) {
+#ifdef DEBUG_VERBOSE
+ printf("Discarding codeblock %d,%d at resno=%d, bandno=%d\n",
+ cblk->x0, cblk->y0, resno, bandno);
+#endif
+ opj_aligned_free(cblk->decoded_data);
+ cblk->decoded_data = NULL;
+ }
+ }
continue;
}
opj_tcd_cblk_dec_t* cblk = &precinct->cblks.dec[cblkno];
opj_t1_cblk_decode_processing_job_t* job;
- assert(cblk->decoded_data == NULL);
-
if (!opj_tcd_is_subband_area_of_interest(tcd,
tilec->compno,
resno,
(OPJ_UINT32)cblk->y0,
(OPJ_UINT32)cblk->x1,
(OPJ_UINT32)cblk->y1)) {
+ if (cblk->decoded_data) {
+#ifdef DEBUG_VERBOSE
+ printf("Discarding codeblock %d,%d at resno=%d, bandno=%d\n",
+ cblk->x0, cblk->y0, resno, bandno);
+#endif
+ opj_aligned_free(cblk->decoded_data);
+ cblk->decoded_data = NULL;
+ }
continue;
}
if (!tcd->whole_tile_decoding) {
OPJ_UINT32 cblk_w = (OPJ_UINT32)(cblk->x1 - cblk->x0);
OPJ_UINT32 cblk_h = (OPJ_UINT32)(cblk->y1 - cblk->y0);
- if (cblk_w == 0 || cblk_h == 0) {
+ if (cblk->decoded_data != NULL) {
+#ifdef DEBUG_VERBOSE
+ printf("Reusing codeblock %d,%d at resno=%d, bandno=%d\n",
+ cblk->x0, cblk->y0, resno, bandno);
+#endif
continue;
}
- /* Zero-init required */
- cblk->decoded_data = opj_calloc(1, cblk_w * cblk_h * sizeof(OPJ_INT32));
- if (cblk->decoded_data == NULL) {
- if (p_manager_mutex) {
- opj_mutex_lock(p_manager_mutex);
- }
- opj_event_msg(p_manager, EVT_ERROR,
- "Cannot allocate cblk->decoded_data\n");
- if (p_manager_mutex) {
- opj_mutex_unlock(p_manager_mutex);
- }
- *pret = OPJ_FALSE;
- return;
+ if (cblk_w == 0 || cblk_h == 0) {
+ continue;
}
+#ifdef DEBUG_VERBOSE
+ printf("Decoding codeblock %d,%d at resno=%d, bandno=%d\n",
+ cblk->x0, cblk->y0, resno, bandno);
+#endif
}
job = (opj_t1_cblk_decode_processing_job_t*) opj_calloc(1,
*pret = OPJ_FALSE;
return;
}
+ job->whole_tile_decoding = tcd->whole_tile_decoding;
job->resno = resno;
job->cblk = cblk;
job->band = band;
job->check_pterm = check_pterm;
job->mustuse_cblkdatabuffer = opj_thread_pool_get_thread_count(tp) > 1;
opj_thread_pool_submit_job(tp, opj_t1_clbl_decode_processor, job);
+#ifdef DEBUG_VERBOSE
+ codeblocks_decoded ++;
+#endif
if (!(*pret)) {
return;
}
} /* bandno */
} /* resno */
+#ifdef DEBUG_VERBOSE
+ printf("Leave opj_t1_decode_cblks(). Number decoded: %d\n", codeblocks_decoded);
+#endif
return;
}
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;
+ OPJ_SIZE_T tileIndex = 0;
OPJ_INT32 x = cblk->x0 - band->x0;
OPJ_INT32 y = cblk->y0 - band->y0;
cblk_h = t1->h;
tileLineAdvance = tile_w - cblk_w;
- tiledp = &tilec->data[(OPJ_UINT32)y * tile_w + (OPJ_UINT32)x];
+ tiledp = &tilec->data[(OPJ_SIZE_T)y * tile_w + (OPJ_SIZE_T)x];
t1->data = tiledp;
t1->data_stride = tile_w;
if (tccp->qmfbid == 1) {