diff --git a/client/X11/xf_client.c b/client/X11/xf_client.c index 5e4e50e38..531c8bc19 100644 --- a/client/X11/xf_client.c +++ b/client/X11/xf_client.c @@ -375,7 +375,7 @@ static BOOL xf_sw_desktop_resize(rdpContext* context) } if (!(xfc->image = XCreateImage(xfc->display, xfc->visual, xfc->depth, ZPixmap, - 0, gdi->primary_buffer, gdi->width, + 0, (char*)gdi->primary_buffer, gdi->width, gdi->height, xfc->scanline_pad, gdi->stride))) { goto out; diff --git a/client/X11/xf_gdi.c b/client/X11/xf_gdi.c index 5bd0b07ab..db51a08b8 100644 --- a/client/X11/xf_gdi.c +++ b/client/X11/xf_gdi.c @@ -437,7 +437,6 @@ static BOOL xf_gdi_opaque_rect(rdpContext* context, const OPAQUE_RECT_ORDER* opaque_rect) { XColor color; - rdpGdi* gdi = context->gdi; xfContext* xfc = (xfContext*) context; BOOL ret = TRUE; @@ -467,7 +466,6 @@ static BOOL xf_gdi_multi_opaque_rect(rdpContext* context, UINT32 i; xfContext* xfc = (xfContext*) context; BOOL ret = TRUE; - rdpGdi* gdi = context->gdi; XColor color; if (!xf_decode_color(xfc, multi_opaque_rect->color, &color)) diff --git a/client/common/file.c b/client/common/file.c index 3547875b0..40bd9dbcb 100644 --- a/client/common/file.c +++ b/client/common/file.c @@ -625,7 +625,6 @@ static BOOL freerdp_client_parse_rdp_file_buffer_unicode(rdpFile* file, const BY if (length > 1) { const WCHAR* beg = line; - const WCHAR* end = &line[length - 1]; if (!freerdp_client_parse_rdp_file_add_line_unicode(file, line, index)) goto fail; diff --git a/libfreerdp/codec/bitmap.c b/libfreerdp/codec/bitmap.c index b324cea56..aeb8e9381 100644 --- a/libfreerdp/codec/bitmap.c +++ b/libfreerdp/codec/bitmap.c @@ -459,8 +459,6 @@ int freerdp_bitmap_compress(const char* srcData, int width, int height, wStream* s, int bpp, int byte_limit, int start_line, wStream* temp_s, int e) { - char *line; - char *last_line; char fom_mask[8192]; /* good for up to 64K bitmap */ int lines_sent; int pixel; @@ -485,7 +483,6 @@ int freerdp_bitmap_compress(const char* srcData, int width, int height, Stream_SetPosition(temp_s, 0); fom_mask_len = 0; - last_line = 0; lines_sent = 0; end = width + e; count = 0; @@ -502,9 +499,10 @@ int freerdp_bitmap_compress(const char* srcData, int width, int height, if ((bpp == 15) || (bpp == 16)) { + const char* line = srcData + width * start_line * 2; + const char *last_line = NULL; mix = (bpp == 15) ? 0xBA1F : 0xFFFF; out_count = end * 2; - line = srcData + width * start_line * 2; while (start_line >= 0 && out_count < 32768) { @@ -794,9 +792,10 @@ int freerdp_bitmap_compress(const char* srcData, int width, int height, } else if (bpp == 24) { + const char* line = srcData + width * start_line * 4; + const char *last_line = NULL; mix = 0xFFFFFF; out_count = end * 3; - line = srcData + width * start_line * 4; while (start_line >= 0 && out_count < 32768) { diff --git a/libfreerdp/primitives/prim_colors_opt.c b/libfreerdp/primitives/prim_colors_opt.c index 24e19c53c..a1d40a58c 100644 --- a/libfreerdp/primitives/prim_colors_opt.c +++ b/libfreerdp/primitives/prim_colors_opt.c @@ -207,28 +207,22 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_BGRX( BYTE* pDst, UINT32 dstStep, const prim_size_t* roi) /* region of interest */ { - __m128i zero, max, r_cr, g_cb, g_cr, b_cb, c4096; - const __m128i* y_buf, *cb_buf, *cr_buf; - __m128i* d_buf; - int srcbump, dstbump, yp, imax; - size_t dstPad, yPad, cbPad, crPad; - zero = _mm_setzero_si128(); - max = _mm_set1_epi16(255); - y_buf = (const __m128i*)(pSrc[0]); - cb_buf = (const __m128i*)(pSrc[1]); - cr_buf = (const __m128i*)(pSrc[2]); - d_buf = (__m128i*)pDst; - r_cr = _mm_set1_epi16(22986); /* 1.403 << 14 */ - g_cb = _mm_set1_epi16(-5636); /* -0.344 << 14 */ - g_cr = _mm_set1_epi16(-11698); /* -0.714 << 14 */ - b_cb = _mm_set1_epi16(28999); /* 1.770 << 14 */ - c4096 = _mm_set1_epi16(4096); - srcbump = srcStep / sizeof(__m128i); - dstbump = dstStep / sizeof(__m128i); - dstPad = (dstStep - roi->width * 4) / sizeof(__m128i); - yPad = 0; - cbPad = 0; - crPad = 0; + const __m128i zero = _mm_setzero_si128(); + const __m128i max = _mm_set1_epi16(255); + const __m128i r_cr = _mm_set1_epi16(22986); /* 1.403 << 14 */ + const __m128i g_cb = _mm_set1_epi16(-5636); /* -0.344 << 14 */ + const __m128i g_cr = _mm_set1_epi16(-11698); /* -0.714 << 14 */ + const __m128i b_cb = _mm_set1_epi16(28999); /* 1.770 << 14 */ + const __m128i c4096 = _mm_set1_epi16(4096); + const INT16* y_buf = (INT16*)pSrc[0]; + const INT16* cb_buf = (INT16*)pSrc[1]; + const INT16* cr_buf = (INT16*)pSrc[2]; + const UINT32 pad = roi->width % 16; + const UINT32 step = sizeof(__m128i) / sizeof(INT16); + const UINT32 imax = (roi->width - pad) * sizeof(INT16) / sizeof(__m128i); + BYTE* d_buf = pDst; + int yp; + const size_t dstPad = (dstStep - roi->width * 4); #ifdef DO_PREFETCH /* Prefetch Y's, Cb's, and Cr's. */ @@ -236,28 +230,27 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_BGRX( { int i; - for (i = 0; i < roi->width * sizeof(INT16) / sizeof(__m128i); + for (i = 0; i < imax; i += (CACHE_LINE_BYTES / sizeof(__m128i))) { - _mm_prefetch((char*)(&y_buf[i]), _MM_HINT_NTA); - _mm_prefetch((char*)(&cb_buf[i]), _MM_HINT_NTA); - _mm_prefetch((char*)(&cr_buf[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)y_buf)[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)cb_buf)[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)cr_buf)[i]), _MM_HINT_NTA); } - y_buf += srcbump; - cb_buf += srcbump; - cr_buf += srcbump; + y_buf += srcStep / sizeof(INT16); + cb_buf += srcStep / sizeof(INT16); + cr_buf += srcStep / sizeof(INT16); } - y_buf = (__m128i*)(pSrc[0]); - cb_buf = (__m128i*)(pSrc[1]); - cr_buf = (__m128i*)(pSrc[2]); + y_buf = (INT16*)pSrc[0]; + cb_buf = (INT16*)pSrc[1]; + cr_buf = (INT16*)pSrc[2]; #endif /* DO_PREFETCH */ - imax = roi->width * sizeof(INT16) / sizeof(__m128i); for (yp = 0; yp < roi->height; ++yp) { - int i; + UINT32 i; for (i = 0; i < imax; i += 2) { @@ -282,13 +275,16 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_BGRX( */ /* y = (y_r_buf[i] + 4096) >> 2 */ __m128i y1, y2, cb1, cb2, cr1, cr2, r1, r2, g1, g2, b1, b2; - y1 = _mm_load_si128(y_buf++); + y1 = _mm_load_si128((__m128i*)y_buf); + y_buf += step; y1 = _mm_add_epi16(y1, c4096); y1 = _mm_srai_epi16(y1, 2); /* cb = cb_g_buf[i]; */ - cb1 = _mm_load_si128(cb_buf++); + cb1 = _mm_load_si128((__m128i*)cb_buf); + cb_buf += step; /* cr = cr_b_buf[i]; */ - cr1 = _mm_load_si128(cr_buf++); + cr1 = _mm_load_si128((__m128i*)cr_buf); + cr_buf += step; /* (y + HIWORD(cr*22986)) >> 3 */ r1 = _mm_add_epi16(y1, _mm_mulhi_epi16(cr1, r_cr)); r1 = _mm_srai_epi16(r1, 3); @@ -305,13 +301,16 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_BGRX( b1 = _mm_srai_epi16(b1, 3); /* b_buf[i] = CLIP(b); */ _mm_between_epi16(b1, zero, max); - y2 = _mm_load_si128(y_buf++); + y2 = _mm_load_si128((__m128i*)y_buf); + y_buf += step; y2 = _mm_add_epi16(y2, c4096); y2 = _mm_srai_epi16(y2, 2); /* cb = cb_g_buf[i]; */ - cb2 = _mm_load_si128(cb_buf++); + cb2 = _mm_load_si128((__m128i*)cb_buf); + cb_buf += step; /* cr = cr_b_buf[i]; */ - cr2 = _mm_load_si128(cr_buf++); + cr2 = _mm_load_si128((__m128i*)cr_buf); + cr_buf += step; /* (y + HIWORD(cr*22986)) >> 3 */ r2 = _mm_add_epi16(y2, _mm_mulhi_epi16(cr2, r_cr)); r2 = _mm_srai_epi16(r2, 3); @@ -355,16 +354,36 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_BGRX( R2 = R3; /* R2 = R3 */ R2 = _mm_unpacklo_epi16(R1, R2); /* R2 = B5G5R5FFB4G4R4FF */ R3 = _mm_unpackhi_epi16(R1, R3); /* R3 = B7G7R7FFB6G6R6FF */ - _mm_store_si128(d_buf++, R0); /* B1G1R1FFB0G0R0FF */ - _mm_store_si128(d_buf++, R4); /* B3G3R3FFB2G2R2FF */ - _mm_store_si128(d_buf++, R2); /* B5G5R5FFB4G4R4FF */ - _mm_store_si128(d_buf++, R3); /* B7G7R7FFB6G6R6FF */ + _mm_store_si128((__m128i*)d_buf, R0); /* B1G1R1FFB0G0R0FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R4); /* B3G3R3FFB2G2R2FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R2); /* B5G5R5FFB4G4R4FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R3); /* B7G7R7FFB6G6R6FF */ + d_buf += sizeof(__m128i); } } - y_buf += yPad; - cb_buf += cbPad; - cr_buf += crPad; + for (i = 0; i < pad; i++) + { + const INT32 divisor = 16; + const INT32 Y = ((*y_buf++) + 4096) << divisor; + const INT32 Cb = (*cb_buf++); + const INT32 Cr = (*cr_buf++); + const INT32 CrR = Cr * (INT32)(1.402525f * (1 << divisor)); + const INT32 CrG = Cr * (INT32)(0.714401f * (1 << divisor)); + const INT32 CbG = Cb * (INT32)(0.343730f * (1 << divisor)); + const INT32 CbB = Cb * (INT32)(1.769905f * (1 << divisor)); + const INT16 R = ((INT16)((CrR + Y) >> divisor) >> 5); + const INT16 G = ((INT16)((Y - CbG - CrG) >> divisor) >> 5); + const INT16 B = ((INT16)((CbB + Y) >> divisor) >> 5); + *d_buf++ = CLIP(B); + *d_buf++ = CLIP(G); + *d_buf++ = CLIP(R); + *d_buf++ = 0xFF; + } + d_buf += dstPad; } @@ -377,28 +396,22 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_RGBX( BYTE* pDst, UINT32 dstStep, const prim_size_t* roi) /* region of interest */ { - __m128i zero, max, r_cr, g_cb, g_cr, b_cb, c4096; - const __m128i* y_buf, *cb_buf, *cr_buf; - __m128i* d_buf; - int srcbump, dstbump, yp, imax; - size_t dstPad, yPad, cbPad, crPad; - zero = _mm_setzero_si128(); - max = _mm_set1_epi16(255); - y_buf = (const __m128i*)(pSrc[0]); - cb_buf = (const __m128i*)(pSrc[1]); - cr_buf = (const __m128i*)(pSrc[2]); - d_buf = (__m128i*)pDst; - r_cr = _mm_set1_epi16(22986); /* 1.403 << 14 */ - g_cb = _mm_set1_epi16(-5636); /* -0.344 << 14 */ - g_cr = _mm_set1_epi16(-11698); /* -0.714 << 14 */ - b_cb = _mm_set1_epi16(28999); /* 1.770 << 14 */ - c4096 = _mm_set1_epi16(4096); - srcbump = srcStep / sizeof(__m128i); - dstbump = dstStep / sizeof(__m128i); - dstPad = (dstStep - roi->width * 4) / sizeof(__m128i); - yPad = 0; - cbPad = 0; - crPad = 0; + const __m128i zero = _mm_setzero_si128(); + const __m128i max = _mm_set1_epi16(255); + const __m128i r_cr = _mm_set1_epi16(22986); /* 1.403 << 14 */ + const __m128i g_cb = _mm_set1_epi16(-5636); /* -0.344 << 14 */ + const __m128i g_cr = _mm_set1_epi16(-11698); /* -0.714 << 14 */ + const __m128i b_cb = _mm_set1_epi16(28999); /* 1.770 << 14 */ + const __m128i c4096 = _mm_set1_epi16(4096); + const INT16* y_buf = (INT16*)pSrc[0]; + const INT16* cb_buf = (INT16*)pSrc[1]; + const INT16* cr_buf = (INT16*)pSrc[2]; + const UINT32 pad = roi->width % 16; + const UINT32 step = sizeof(__m128i) / sizeof(INT16); + const UINT32 imax = (roi->width - pad) * sizeof(INT16) / sizeof(__m128i); + BYTE* d_buf = pDst; + int yp; + const size_t dstPad = (dstStep - roi->width * 4); #ifdef DO_PREFETCH /* Prefetch Y's, Cb's, and Cr's. */ @@ -406,28 +419,27 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_RGBX( { int i; - for (i = 0; i < roi->width * sizeof(INT16) / sizeof(__m128i); + for (i = 0; i < imax; i += (CACHE_LINE_BYTES / sizeof(__m128i))) { - _mm_prefetch((char*)(&y_buf[i]), _MM_HINT_NTA); - _mm_prefetch((char*)(&cb_buf[i]), _MM_HINT_NTA); - _mm_prefetch((char*)(&cr_buf[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)y_buf)[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)cb_buf)[i]), _MM_HINT_NTA); + _mm_prefetch((char*)(&((__m128i*)cr_buf)[i]), _MM_HINT_NTA); } - y_buf += srcbump; - cb_buf += srcbump; - cr_buf += srcbump; + y_buf += srcStep / sizeof(INT16); + cb_buf += srcStep / sizeof(INT16); + cr_buf += srcStep / sizeof(INT16); } - y_buf = (__m128i*)(pSrc[0]); - cb_buf = (__m128i*)(pSrc[1]); - cr_buf = (__m128i*)(pSrc[2]); + y_buf = (INT16*)(pSrc[0]); + cb_buf = (INT16*)(pSrc[1]); + cr_buf = (INT16*)(pSrc[2]); #endif /* DO_PREFETCH */ - imax = roi->width * sizeof(INT16) / sizeof(__m128i); for (yp = 0; yp < roi->height; ++yp) { - int i; + UINT32 i; for (i = 0; i < imax; i += 2) { @@ -452,13 +464,16 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_RGBX( */ /* y = (y_r_buf[i] + 4096) >> 2 */ __m128i y1, y2, cb1, cb2, cr1, cr2, r1, r2, g1, g2, b1, b2; - y1 = _mm_load_si128(y_buf++); + y1 = _mm_load_si128((__m128i*)y_buf); + y_buf += step; y1 = _mm_add_epi16(y1, c4096); y1 = _mm_srai_epi16(y1, 2); /* cb = cb_g_buf[i]; */ - cb1 = _mm_load_si128(cb_buf++); + cb1 = _mm_load_si128((__m128i*)cb_buf); + cb_buf += step; /* cr = cr_b_buf[i]; */ - cr1 = _mm_load_si128(cr_buf++); + cr1 = _mm_load_si128((__m128i*)cr_buf); + cr_buf += step; /* (y + HIWORD(cr*22986)) >> 3 */ r1 = _mm_add_epi16(y1, _mm_mulhi_epi16(cr1, r_cr)); r1 = _mm_srai_epi16(r1, 3); @@ -475,13 +490,16 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_RGBX( b1 = _mm_srai_epi16(b1, 3); /* b_buf[i] = CLIP(b); */ _mm_between_epi16(b1, zero, max); - y2 = _mm_load_si128(y_buf++); + y2 = _mm_load_si128((__m128i*)y_buf); + y_buf += step; y2 = _mm_add_epi16(y2, c4096); y2 = _mm_srai_epi16(y2, 2); /* cb = cb_g_buf[i]; */ - cb2 = _mm_load_si128(cb_buf++); + cb2 = _mm_load_si128((__m128i*)cb_buf); + cb_buf += step; /* cr = cr_b_buf[i]; */ - cr2 = _mm_load_si128(cr_buf++); + cr2 = _mm_load_si128((__m128i*)cr_buf); + cr_buf += step; /* (y + HIWORD(cr*22986)) >> 3 */ r2 = _mm_add_epi16(y2, _mm_mulhi_epi16(cr2, r_cr)); r2 = _mm_srai_epi16(r2, 3); @@ -525,16 +543,36 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R_RGBX( R2 = R3; /* R2 = R3 */ R2 = _mm_unpacklo_epi16(R1, R2); /* R2 = R5G5B5FFR4G4B4FF */ R3 = _mm_unpackhi_epi16(R1, R3); /* R3 = R7G7B7FFR6G6B6FF */ - _mm_store_si128(d_buf++, R0); /* R1G1B1FFR0G0B0FF */ - _mm_store_si128(d_buf++, R4); /* R3G3B3FFR2G2B2FF */ - _mm_store_si128(d_buf++, R2); /* R5G5B5FFR4G4B4FF */ - _mm_store_si128(d_buf++, R3); /* R7G7B7FFR6G6B6FF */ + _mm_store_si128((__m128i*)d_buf, R0); /* R1G1B1FFR0G0B0FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R4); /* R3G3B3FFR2G2B2FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R2); /* R5G5B5FFR4G4B4FF */ + d_buf += sizeof(__m128i); + _mm_store_si128((__m128i*)d_buf, R3); /* R7G7B7FFR6G6B6FF */ + d_buf += sizeof(__m128i); } } - y_buf += yPad; - cb_buf += cbPad; - cr_buf += crPad; + for (i = 0; i < pad; i++) + { + const INT32 divisor = 16; + const INT32 Y = ((*y_buf++) + 4096) << divisor; + const INT32 Cb = (*cb_buf++); + const INT32 Cr = (*cr_buf++); + const INT32 CrR = Cr * (INT32)(1.402525f * (1 << divisor)); + const INT32 CrG = Cr * (INT32)(0.714401f * (1 << divisor)); + const INT32 CbG = Cb * (INT32)(0.343730f * (1 << divisor)); + const INT32 CbB = Cb * (INT32)(1.769905f * (1 << divisor)); + const INT16 R = ((INT16)((CrR + Y) >> divisor) >> 5); + const INT16 G = ((INT16)((Y - CbG - CrG) >> divisor) >> 5); + const INT16 B = ((INT16)((CbB + Y) >> divisor) >> 5); + *d_buf++ = CLIP(R); + *d_buf++ = CLIP(G); + *d_buf++ = CLIP(B); + *d_buf++ = 0xFF; + } + d_buf += dstPad; } @@ -550,9 +588,8 @@ static pstatus_t sse2_yCbCrToRGB_16s8u_P3AC4R( || ((ULONG_PTR)(pSrc[1]) & 0x0f) || ((ULONG_PTR)(pSrc[2]) & 0x0f) || ((ULONG_PTR)(pDst) & 0x0f) - || (roi->width & 0x07) - || (srcStep & 127) - || (dstStep & 127)) + || (srcStep & 0x0f) + || (dstStep & 0x0f)) { /* We can't maintain 16-byte alignment. */ return generic->yCbCrToRGB_16s8u_P3AC4R(pSrc, srcStep, @@ -718,6 +755,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_BGRX( const UINT16* pr = (const UINT16*)(pSrc[0]); const UINT16* pg = (const UINT16*)(pSrc[1]); const UINT16* pb = (const UINT16*)(pSrc[2]); + const UINT32 pad = roi->width % 16; const __m128i a = _mm_set1_epi32(0xFFFFFFFFU); BYTE* out; UINT32 srcbump, dstbump, y; @@ -729,7 +767,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_BGRX( { UINT32 x; - for (x = 0; x < roi->width; x += 16) + for (x = 0; x < roi->width - pad; x += 16) { __m128i r, g, b; /* The comments below pretend these are 8-byte registers @@ -737,25 +775,25 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_BGRX( */ { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pb); + R0 = _mm_load_si128((__m128i*)pb); pb += 8; /* R0 = 00B300B200B100B0 */ - R1 = _mm_loadu_si128((__m128i*)pb); + R1 = _mm_load_si128((__m128i*)pb); pb += 8; /* R1 = 00B700B600B500B4 */ b = _mm_packus_epi16(R0, R1); /* b = B7B6B5B4B3B2B1B0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pg); + R0 = _mm_load_si128((__m128i*)pg); pg += 8; /* R1 = 00G300G200G100G0 */ - R1 = _mm_loadu_si128((__m128i*)pg); + R1 = _mm_load_si128((__m128i*)pg); pg += 8; /* R2 = 00G700G600G500G4 */ g = _mm_packus_epi16(R0, R1); /* g = G7G6G5G4G3G2G1G0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pr); + R0 = _mm_load_si128((__m128i*)pr); pr += 8; /* R0 = 00R300R200R100R0 */ - R1 = _mm_loadu_si128((__m128i*)pr); + R1 = _mm_load_si128((__m128i*)pr); pr += 8; /* R3 = 00R700R600R500R4 */ r = _mm_packus_epi16(R0, R1); /* r = R7R6R5R4R3R2R1R0 */ } @@ -790,6 +828,17 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_BGRX( } } + for (x = 0; x < pad; x++) + { + const BYTE R = CLIP(*pr++); + const BYTE G = CLIP(*pg++); + const BYTE B = CLIP(*pb++); + *out++ = B; + *out++ = G; + *out++ = R; + *out++ = 0xFF; + } + /* Jump to next row. */ pr += srcbump; pg += srcbump; @@ -810,6 +859,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_RGBX( const UINT16* pr = (const UINT16*)(pSrc[0]); const UINT16* pg = (const UINT16*)(pSrc[1]); const UINT16* pb = (const UINT16*)(pSrc[2]); + const UINT32 pad = roi->width % 16; const __m128i a = _mm_set1_epi32(0xFFFFFFFFU); BYTE* out; UINT32 srcbump, dstbump, y; @@ -821,7 +871,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_RGBX( { UINT32 x; - for (x = 0; x < roi->width; x += 16) + for (x = 0; x < roi->width - pad; x += 16) { __m128i r, g, b; /* The comments below pretend these are 8-byte registers @@ -829,25 +879,25 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_RGBX( */ { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pb); + R0 = _mm_load_si128((__m128i*)pb); pb += 8; /* R0 = 00B300B200B100B0 */ - R1 = _mm_loadu_si128((__m128i*)pb); + R1 = _mm_load_si128((__m128i*)pb); pb += 8; /* R1 = 00B700B600B500B4 */ b = _mm_packus_epi16(R0, R1); /* b = B7B6B5B4B3B2B1B0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pg); + R0 = _mm_load_si128((__m128i*)pg); pg += 8; /* R1 = 00G300G200G100G0 */ - R1 = _mm_loadu_si128((__m128i*)pg); + R1 = _mm_load_si128((__m128i*)pg); pg += 8; /* R2 = 00G700G600G500G4 */ g = _mm_packus_epi16(R0, R1); /* g = G7G6G5G4G3G2G1G0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pr); + R0 = _mm_load_si128((__m128i*)pr); pr += 8; /* R0 = 00R300R200R100R0 */ - R1 = _mm_loadu_si128((__m128i*)pr); + R1 = _mm_load_si128((__m128i*)pr); pr += 8; /* R3 = 00R700R600R500R4 */ r = _mm_packus_epi16(R0, R1); /* r = R7R6R5R4R3R2R1R0 */ } @@ -882,6 +932,17 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_RGBX( } } + for (x = 0; x < pad; x++) + { + const BYTE R = CLIP(*pr++); + const BYTE G = CLIP(*pg++); + const BYTE B = CLIP(*pb++); + *out++ = R; + *out++ = G; + *out++ = B; + *out++ = 0xFF; + } + /* Jump to next row. */ pr += srcbump; pg += srcbump; @@ -902,6 +963,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XBGR( const UINT16* pr = (const UINT16*)(pSrc[0]); const UINT16* pg = (const UINT16*)(pSrc[1]); const UINT16* pb = (const UINT16*)(pSrc[2]); + const UINT32 pad = roi->width % 16; const __m128i a = _mm_set1_epi32(0xFFFFFFFFU); BYTE* out; UINT32 srcbump, dstbump, y; @@ -913,7 +975,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XBGR( { UINT32 x; - for (x = 0; x < roi->width; x += 16) + for (x = 0; x < roi->width - pad; x += 16) { __m128i r, g, b; /* The comments below pretend these are 8-byte registers @@ -921,25 +983,25 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XBGR( */ { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pb); + R0 = _mm_load_si128((__m128i*)pb); pb += 8; /* R0 = 00B300B200B100B0 */ - R1 = _mm_loadu_si128((__m128i*)pb); + R1 = _mm_load_si128((__m128i*)pb); pb += 8; /* R1 = 00B700B600B500B4 */ b = _mm_packus_epi16(R0, R1); /* b = B7B6B5B4B3B2B1B0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pg); + R0 = _mm_load_si128((__m128i*)pg); pg += 8; /* R1 = 00G300G200G100G0 */ - R1 = _mm_loadu_si128((__m128i*)pg); + R1 = _mm_load_si128((__m128i*)pg); pg += 8; /* R2 = 00G700G600G500G4 */ g = _mm_packus_epi16(R0, R1); /* g = G7G6G5G4G3G2G1G0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pr); + R0 = _mm_load_si128((__m128i*)pr); pr += 8; /* R0 = 00R300R200R100R0 */ - R1 = _mm_loadu_si128((__m128i*)pr); + R1 = _mm_load_si128((__m128i*)pr); pr += 8; /* R3 = 00R700R600R500R4 */ r = _mm_packus_epi16(R0, R1); /* r = R7R6R5R4R3R2R1R0 */ } @@ -974,6 +1036,17 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XBGR( } } + for (x = 0; x < pad; x++) + { + const BYTE R = CLIP(*pr++); + const BYTE G = CLIP(*pg++); + const BYTE B = CLIP(*pb++); + *out++ = 0xFF; + *out++ = B; + *out++ = G; + *out++ = R; + } + /* Jump to next row. */ pr += srcbump; pg += srcbump; @@ -995,6 +1068,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XRGB( const UINT16* pg = (const UINT16*)(pSrc[1]); const UINT16* pb = (const UINT16*)(pSrc[2]); const __m128i a = _mm_set1_epi32(0xFFFFFFFFU); + const UINT32 pad = roi->width % 16; BYTE* out; UINT32 srcbump, dstbump, y; out = (BYTE*) pDst; @@ -1005,7 +1079,7 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XRGB( { UINT32 x; - for (x = 0; x < roi->width; x += 16) + for (x = 0; x < roi->width - pad; x += 16) { __m128i r, g, b; /* The comments below pretend these are 8-byte registers @@ -1013,25 +1087,25 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XRGB( */ { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pb); + R0 = _mm_load_si128((__m128i*)pb); pb += 8; /* R0 = 00B300B200B100B0 */ - R1 = _mm_loadu_si128((__m128i*)pb); + R1 = _mm_load_si128((__m128i*)pb); pb += 8; /* R1 = 00B700B600B500B4 */ b = _mm_packus_epi16(R0, R1); /* b = B7B6B5B4B3B2B1B0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pg); + R0 = _mm_load_si128((__m128i*)pg); pg += 8; /* R1 = 00G300G200G100G0 */ - R1 = _mm_loadu_si128((__m128i*)pg); + R1 = _mm_load_si128((__m128i*)pg); pg += 8; /* R2 = 00G700G600G500G4 */ g = _mm_packus_epi16(R0, R1); /* g = G7G6G5G4G3G2G1G0 */ } { __m128i R0, R1; - R0 = _mm_loadu_si128((__m128i*)pr); + R0 = _mm_load_si128((__m128i*)pr); pr += 8; /* R0 = 00R300R200R100R0 */ - R1 = _mm_loadu_si128((__m128i*)pr); + R1 = _mm_load_si128((__m128i*)pr); pr += 8; /* R3 = 00R700R600R500R4 */ r = _mm_packus_epi16(R0, R1); /* r = R7R6R5R4R3R2R1R0 */ } @@ -1066,6 +1140,17 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R_XRGB( } } + for (x = 0; x < pad; x++) + { + const BYTE R = CLIP(*pr++); + const BYTE G = CLIP(*pg++); + const BYTE B = CLIP(*pb++); + *out++ = 0xFF; + *out++ = R; + *out++ = G; + *out++ = B; + } + /* Jump to next row. */ pr += srcbump; pg += srcbump; @@ -1084,6 +1169,10 @@ static pstatus_t sse2_RGBToRGB_16s8u_P3AC4R( UINT32 DstFormat, const prim_size_t* roi) { + if (((UINT64)pSrc[0] & 0x0f) || ((UINT64)pSrc[0] & 0x0f) || ((UINT64)pSrc[0] & 0x0f) || + (srcStep & 0x0f) || ((UINT64)pDst & 0x0f) || (dstStep & 0x0f)) + return generic->RGBToRGB_16s8u_P3AC4R(pSrc, srcStep, pDst, dstStep, DstFormat, roi); + switch (DstFormat) { case PIXEL_FORMAT_BGRA32: @@ -1427,7 +1516,7 @@ static pstatus_t neon_RGBToRGB_16s8u_P3AC4R( return neon_RGBToRGB_16s8u_P3AC4R_X(pSrc, srcStep, pDst, dstStep, roi, 3, 2, 1, 0); default: - return generic->yCbCrToRGB_16s8u_P3AC4R(pSrc, srcStep, pDst, dstStep, DstFormat, roi); + return generic->RGBToRGB_16s8u_P3AC4R(pSrc, srcStep, pDst, dstStep, DstFormat, roi); } } #endif /* WITH_NEON */ diff --git a/libfreerdp/primitives/test/TestPrimitivesYCbCr.c b/libfreerdp/primitives/test/TestPrimitivesYCbCr.c index 4a68b6eac..5dc00031a 100644 --- a/libfreerdp/primitives/test/TestPrimitivesYCbCr.c +++ b/libfreerdp/primitives/test/TestPrimitivesYCbCr.c @@ -2155,34 +2155,55 @@ static int test_bmp_cmp_dump(const BYTE* actual, const BYTE* expected, int size, return count; } -static int test_PrimitivesYCbCr(const primitives_t* prims, UINT32 format, prim_size_t roi) +static int test_PrimitivesYCbCr(const primitives_t* prims, UINT32 format, prim_size_t roi, + BOOL compare) { pstatus_t status = -1; int cnt[3]; float err[3]; BYTE* actual; + BYTE* actual1; BYTE* expected; int margin = 1; - const INT16* pYCbCr[3]; + INT16* pYCbCr[3]; const UINT32 srcStride = roi.width * 2; const UINT32 dstStride = roi.width * GetBytesPerPixel(format); const UINT32 srcSize = srcStride * roi.height; const UINT32 dstSize = dstStride * roi.height; PROFILER_DEFINE(prof); + PROFILER_DEFINE(prof1); + PROFILER_DEFINE(prof2); //return test_YCbCr_pixels(); expected = (BYTE*) TEST_XRGB_IMAGE; actual = _aligned_malloc(dstSize, 16); - PROFILER_CREATE(prof, "YCbCr"); + actual1 = _aligned_malloc(dstSize, 16); + PROFILER_CREATE(prof, "yCbCrToRGB_16s8u"); + PROFILER_CREATE(prof1, "yCbCrToRGB16s16s"); + PROFILER_CREATE(prof2, "RGBToRGB_16s8u"); - if (!actual) + if (!actual || !actual1) goto fail; ZeroMemory(actual, dstSize); - pYCbCr[0] = TEST_Y_COMPONENT; - pYCbCr[1] = TEST_CB_COMPONENT; - pYCbCr[2] = TEST_CR_COMPONENT; + ZeroMemory(actual1, dstSize); + pYCbCr[0] = _aligned_malloc(srcSize, 16); + pYCbCr[1] = _aligned_malloc(srcSize, 16); + pYCbCr[2] = _aligned_malloc(srcSize, 16); + + if (!pYCbCr[0] || !pYCbCr[1] || !pYCbCr[2]) + goto fail; + + winpr_RAND((BYTE*)pYCbCr[0], srcSize); + winpr_RAND((BYTE*)pYCbCr[1], srcSize); + winpr_RAND((BYTE*)pYCbCr[2], srcSize); + + if (compare) + { + memcpy(pYCbCr[0], TEST_Y_COMPONENT, srcSize); + memcpy(pYCbCr[1], TEST_CB_COMPONENT, srcSize); + memcpy(pYCbCr[2], TEST_CR_COMPONENT, srcSize); + } - if (1) { PROFILER_ENTER(prof); status = prims->yCbCrToRGB_16s8u_P3AC4R((const INT16**) pYCbCr, srcStride, @@ -2190,7 +2211,7 @@ static int test_PrimitivesYCbCr(const primitives_t* prims, UINT32 format, prim_s &roi); PROFILER_EXIT(prof); } - else + { INT16* pSrcDst[3]; pSrcDst[0] = _aligned_malloc(srcSize, 16); @@ -2199,16 +2220,18 @@ static int test_PrimitivesYCbCr(const primitives_t* prims, UINT32 format, prim_s CopyMemory(pSrcDst[0], pYCbCr[0], srcSize); CopyMemory(pSrcDst[1], pYCbCr[1], srcSize); CopyMemory(pSrcDst[2], pYCbCr[2], srcSize); - PROFILER_ENTER(prof); + PROFILER_ENTER(prof1); status = prims->yCbCrToRGB_16s16s_P3P3((const INT16**) pSrcDst, srcStride, pSrcDst, srcStride, &roi); + PROFILER_EXIT(prof1); if (status != PRIMITIVES_SUCCESS) goto fail2; + PROFILER_ENTER(prof2); status = prims->RGBToRGB_16s8u_P3AC4R((const INT16**) pSrcDst, srcStride, - actual, dstStride, format, &roi); - PROFILER_EXIT(prof); + actual1, dstStride, format, &roi); + PROFILER_EXIT(prof2); fail2: _aligned_free(pSrcDst[0]); _aligned_free(pSrcDst[1]); @@ -2218,32 +2241,65 @@ static int test_PrimitivesYCbCr(const primitives_t* prims, UINT32 format, prim_s goto fail; } - cnt[2] = test_bmp_cmp_count(actual, expected, dstSize, 2, margin); /* red */ - err[2] = ((float) cnt[2]) / ((float) dstSize / 4) * 100.0f; - cnt[1] = test_bmp_cmp_count(actual, expected, dstSize, 1, margin); /* green */ - err[1] = ((float) cnt[1]) / ((float) dstSize / 4) * 100.0f; - cnt[0] = test_bmp_cmp_count(actual, expected, dstSize, 0, margin); /* blue */ - err[0] = ((float) cnt[0]) / ((float) dstSize / 4) * 100.0f; - - if (cnt[0] || cnt[1] || cnt[2]) + if (compare) { - printf("Red Error Dump:\n"); - test_bmp_cmp_dump(actual, expected, dstSize, 2, margin); /* red */ - printf("Green Error Dump:\n"); - test_bmp_cmp_dump(actual, expected, dstSize, 1, margin); /* green */ - printf("Blue Error Dump:\n"); - test_bmp_cmp_dump(actual, expected, dstSize, 0, margin); /* blue */ - printf("R: diff: %d (%f%%)\n", cnt[2], err[2]); - printf("G: diff: %d (%f%%)\n", cnt[1], err[1]); - printf("B: diff: %d (%f%%)\n", cnt[0], err[0]); + cnt[2] = test_bmp_cmp_count(actual, expected, dstSize, 2, margin); /* red */ + err[2] = ((float) cnt[2]) / ((float) dstSize / 4) * 100.0f; + cnt[1] = test_bmp_cmp_count(actual, expected, dstSize, 1, margin); /* green */ + err[1] = ((float) cnt[1]) / ((float) dstSize / 4) * 100.0f; + cnt[0] = test_bmp_cmp_count(actual, expected, dstSize, 0, margin); /* blue */ + err[0] = ((float) cnt[0]) / ((float) dstSize / 4) * 100.0f; + + if (cnt[0] || cnt[1] || cnt[2]) + { + printf("Summary information yCbCrToRGB_16s8u_P3AC4R\n"); + printf("Red Error Dump:\n"); + test_bmp_cmp_dump(actual, expected, dstSize, 2, margin); /* red */ + printf("Green Error Dump:\n"); + test_bmp_cmp_dump(actual, expected, dstSize, 1, margin); /* green */ + printf("Blue Error Dump:\n"); + test_bmp_cmp_dump(actual, expected, dstSize, 0, margin); /* blue */ + printf("R: diff: %d (%f%%)\n", cnt[2], err[2]); + printf("G: diff: %d (%f%%)\n", cnt[1], err[1]); + printf("B: diff: %d (%f%%)\n", cnt[0], err[0]); + } + + cnt[2] = test_bmp_cmp_count(actual1, expected, dstSize, 2, margin); /* red */ + err[2] = ((float) cnt[2]) / ((float) dstSize / 4) * 100.0f; + cnt[1] = test_bmp_cmp_count(actual1, expected, dstSize, 1, margin); /* green */ + err[1] = ((float) cnt[1]) / ((float) dstSize / 4) * 100.0f; + cnt[0] = test_bmp_cmp_count(actual1, expected, dstSize, 0, margin); /* blue */ + err[0] = ((float) cnt[0]) / ((float) dstSize / 4) * 100.0f; + + if (cnt[0] || cnt[1] || cnt[2]) + { + printf("Summary information yCbCrToRGB_16s16s_P3P3 & RGBToRGB_16s8u_P3AC4R\n"); + printf("Red Error Dump:\n"); + test_bmp_cmp_dump(actual1, expected, dstSize, 2, margin); /* red */ + printf("Green Error Dump:\n"); + test_bmp_cmp_dump(actual1, expected, dstSize, 1, margin); /* green */ + printf("Blue Error Dump:\n"); + test_bmp_cmp_dump(actual1, expected, dstSize, 0, margin); /* blue */ + printf("R: diff: %d (%f%%)\n", cnt[2], err[2]); + printf("G: diff: %d (%f%%)\n", cnt[1], err[1]); + printf("B: diff: %d (%f%%)\n", cnt[0], err[0]); + } } PROFILER_PRINT_HEADER; PROFILER_PRINT(prof); + PROFILER_PRINT(prof1); + PROFILER_PRINT(prof2); PROFILER_PRINT_FOOTER; fail: + _aligned_free((BYTE*)pYCbCr[0]); + _aligned_free((BYTE*)pYCbCr[1]); + _aligned_free((BYTE*)pYCbCr[2]); _aligned_free(actual); + _aligned_free(actual1); PROFILER_FREE(prof); + PROFILER_FREE(prof1); + PROFILER_FREE(prof2); return status; } @@ -2262,30 +2318,105 @@ int TestPrimitivesYCbCr(int argc, char* argv[]) }; const primitives_t* prims = primitives_get(); const primitives_t* generics = primitives_get_generic(); - prim_size_t roi = { 64, 64 }; UINT32 x; - for (x = 0; x < sizeof(formats) / sizeof(formats[0]); x++) + if (argc < 2) { - int rc; - printf("----------------------- GENERIC %s -------------------\n", - GetColorFormatName(formats[x])); - rc = test_PrimitivesYCbCr(generics, formats[x], roi); + { + /* Do content comparison. */ + for (x = 0; x < sizeof(formats) / sizeof(formats[0]); x++) + { + prim_size_t roi = { 64, 64 }; + int rc; + printf("----------------------- GENERIC %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(generics, formats[x], roi, TRUE); - if (rc != PRIMITIVES_SUCCESS) - return rc; + if (rc != PRIMITIVES_SUCCESS) + return rc; - printf("------------------------- END %s ----------------------\n", - GetColorFormatName(formats[x])); - printf("---------------------- OPTIMIZED %s -------------------\n", - GetColorFormatName(formats[x])); - rc = test_PrimitivesYCbCr(prims, formats[x], roi); + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + printf("---------------------- OPTIMIZED %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(prims, formats[x], roi, TRUE); - if (rc != PRIMITIVES_SUCCESS) - return rc; + if (rc != PRIMITIVES_SUCCESS) + return rc; - printf("------------------------- END %s ----------------------\n", - GetColorFormatName(formats[x])); + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + } + } + /* Do random data conversion with random sizes */ + { + prim_size_t roi; + + do + { + winpr_RAND((BYTE*)&roi.width, sizeof(roi.width)); + roi.width %= 2048; + } + while (roi.width < 16); + + do + { + winpr_RAND((BYTE*)&roi.height, sizeof(roi.height)); + roi.height %= 2048; + } + while (roi.height < 16); + + for (x = 0; x < sizeof(formats) / sizeof(formats[0]); x++) + { + int rc; + printf("----------------------- GENERIC %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(generics, formats[x], roi, FALSE); + + if (rc != PRIMITIVES_SUCCESS) + return rc; + + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + printf("---------------------- OPTIMIZED %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(prims, formats[x], roi, FALSE); + + if (rc != PRIMITIVES_SUCCESS) + return rc; + + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + } + } + } + /* Do a performance run with full HD */ + else + { + prim_size_t roi = { 1928, 1080 }; + + for (x = 0; x < sizeof(formats) / sizeof(formats[0]); x++) + { + int rc; + printf("----------------------- GENERIC %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(generics, formats[x], roi, FALSE); + + if (rc != PRIMITIVES_SUCCESS) + return rc; + + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + printf("---------------------- OPTIMIZED %s [%"PRIu32"x%"PRIu32"] COMPARE CONTENT ----\n", + GetColorFormatName(formats[x]), roi.width, roi.height); + rc = test_PrimitivesYCbCr(prims, formats[x], roi, FALSE); + + if (rc != PRIMITIVES_SUCCESS) + return rc; + + printf("------------------------- END %s ----------------------\n", + GetColorFormatName(formats[x])); + } } return 0; diff --git a/libfreerdp/primitives/test/TestPrimitivesYCoCg.c b/libfreerdp/primitives/test/TestPrimitivesYCoCg.c index 8239e95dd..de09ccaf2 100644 --- a/libfreerdp/primitives/test/TestPrimitivesYCoCg.c +++ b/libfreerdp/primitives/test/TestPrimitivesYCoCg.c @@ -27,7 +27,7 @@ /* ------------------------------------------------------------------------- */ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) { - pstatus_t status; + pstatus_t status = -1; BYTE* out_sse = NULL; BYTE* in = NULL; BYTE* out_c = NULL; @@ -45,8 +45,6 @@ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) }; PROFILER_DEFINE(genericProf); PROFILER_DEFINE(optProf); - PROFILER_CREATE(genericProf, "YCoCgRToRGB_8u_AC4R-GENERIC"); - PROFILER_CREATE(optProf, "YCoCgRToRGB_8u_AC4R-OPT"); in = _aligned_malloc(size, 16); out_c = _aligned_malloc(size, 16); out_sse = _aligned_malloc(size, 16); @@ -61,6 +59,8 @@ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) const UINT32 format = formats[x]; const UINT32 dstStride = width * GetBytesPerPixel(format); const char* formatName = GetColorFormatName(format); + PROFILER_CREATE(genericProf, "YCoCgRToRGB_8u_AC4R-GENERIC"); + PROFILER_CREATE(optProf, "YCoCgRToRGB_8u_AC4R-OPT"); PROFILER_ENTER(genericProf); status = generic->YCoCgToRGB_8u_AC4R( in, srcStride, @@ -68,7 +68,7 @@ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) PROFILER_EXIT(genericProf); if (status != PRIMITIVES_SUCCESS) - goto fail; + goto loop_fail; PROFILER_ENTER(optProf); status = optimized->YCoCgToRGB_8u_AC4R( @@ -77,7 +77,7 @@ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) PROFILER_EXIT(optProf); if (status != PRIMITIVES_SUCCESS) - goto fail; + goto loop_fail; if (memcmp(out_c, out_sse, dstStride * height) != 0) { @@ -95,15 +95,21 @@ static BOOL test_YCoCgRToRGB_8u_AC4R_func(UINT32 width, UINT32 height) } } + printf("--------------------------- [%s] [%"PRIu32"x%"PRIu32"] ---------------------------\n", + formatName, width, height); PROFILER_PRINT_HEADER; PROFILER_PRINT(genericProf); PROFILER_PRINT(optProf); PROFILER_PRINT_FOOTER; + loop_fail: + PROFILER_FREE(genericProf); + PROFILER_FREE(optProf); + + if (status != PRIMITIVES_SUCCESS) + goto fail; } fail: - PROFILER_FREE(genericProf); - PROFILER_FREE(optProf); _aligned_free(in); _aligned_free(out_c); _aligned_free(out_sse); @@ -114,6 +120,35 @@ int TestPrimitivesYCoCg(int argc, char* argv[]) { prim_test_setup(FALSE); + /* Random resolution tests */ + if (argc < 2) + { + UINT32 x; + + for (x = 0; x < 10; x++) + { + UINT32 w, h; + + do + { + winpr_RAND((BYTE*)&w, sizeof(w)); + w %= 2048; + } + while (w < 16); + + do + { + winpr_RAND((BYTE*)&h, sizeof(h)); + h %= 2048; + } + while (h < 16); + + if (!test_YCoCgRToRGB_8u_AC4R_func(w, h)) + return 1; + } + } + + /* Test once with full HD */ if (!test_YCoCgRToRGB_8u_AC4R_func(1920, 1080)) return 1; diff --git a/libfreerdp/primitives/test/TestPrimitivesYUV.c b/libfreerdp/primitives/test/TestPrimitivesYUV.c index 9c0309f02..07bde055e 100644 --- a/libfreerdp/primitives/test/TestPrimitivesYUV.c +++ b/libfreerdp/primitives/test/TestPrimitivesYUV.c @@ -254,9 +254,6 @@ static BOOL TestPrimitiveYUVCombine(primitives_t* prims, prim_size_t roi) } PROFILER_EXIT(yuvCombine); - PROFILER_PRINT_HEADER; - PROFILER_PRINT(yuvCombine); - PROFILER_PRINT_FOOTER; for (x = 0; x < 3; x++) { @@ -284,9 +281,6 @@ static BOOL TestPrimitiveYUVCombine(primitives_t* prims, prim_size_t roi) } PROFILER_EXIT(yuvSplit); - PROFILER_PRINT_HEADER; - PROFILER_PRINT(yuvSplit); - PROFILER_PRINT_FOOTER; for (x = 0; x < 3; x++) { @@ -346,6 +340,10 @@ static BOOL TestPrimitiveYUVCombine(primitives_t* prims, prim_size_t roi) } } + PROFILER_PRINT_HEADER; + PROFILER_PRINT(yuvSplit); + PROFILER_PRINT(yuvCombine); + PROFILER_PRINT_FOOTER; rc = TRUE; fail: PROFILER_FREE(yuvCombine); @@ -375,7 +373,7 @@ static BOOL TestPrimitiveYUV(primitives_t* prims, prim_size_t roi, BOOL use444) size_t size; size_t uvsize, uvwidth; size_t padding = 100 * 16; - size_t stride; + UINT32 stride; const UINT32 formats[] = { PIXEL_FORMAT_XRGB32, @@ -396,10 +394,6 @@ static BOOL TestPrimitiveYUV(primitives_t* prims, prim_size_t roi, BOOL use444) aheight = roi.height + 16 - roi.height % 16; stride = awidth * sizeof(UINT32); size = awidth * aheight; - PROFILER_CREATE(rgbToYUV420, "RGBToYUV420"); - PROFILER_CREATE(rgbToYUV444, "RGBToYUV444"); - PROFILER_CREATE(yuv420ToRGB, "YUV420ToRGB"); - PROFILER_CREATE(yuv444ToRGB, "YUV444ToRGB"); if (use444) { @@ -456,22 +450,25 @@ static BOOL TestPrimitiveYUV(primitives_t* prims, prim_size_t roi, BOOL use444) for (x = 0; x < sizeof(formats) / sizeof(formats[0]); x++) { + pstatus_t rc; const UINT32 DstFormat = formats[x]; printf("Testing destination color format %s\n", GetColorFormatName(DstFormat)); + PROFILER_CREATE(rgbToYUV420, "RGBToYUV420"); + PROFILER_CREATE(rgbToYUV444, "RGBToYUV444"); + PROFILER_CREATE(yuv420ToRGB, "YUV420ToRGB"); + PROFILER_CREATE(yuv444ToRGB, "YUV444ToRGB"); if (use444) { PROFILER_ENTER(rgbToYUV444); - - if (prims->RGBToYUV444_8u_P3AC4R(rgb, DstFormat, - stride, yuv, yuv_step, - &roi) != PRIMITIVES_SUCCESS) - { - PROFILER_EXIT(rgbToYUV444); - goto fail; - } - + rc = prims->RGBToYUV444_8u_P3AC4R(rgb, DstFormat, + stride, yuv, yuv_step, + &roi); PROFILER_EXIT(rgbToYUV444); + + if (rc != PRIMITIVES_SUCCESS) + goto loop_fail; + PROFILER_PRINT_HEADER; PROFILER_PRINT(rgbToYUV444); PROFILER_PRINT_FOOTER; @@ -479,45 +476,52 @@ static BOOL TestPrimitiveYUV(primitives_t* prims, prim_size_t roi, BOOL use444) else { PROFILER_ENTER(rgbToYUV420); - - if (prims->RGBToYUV420_8u_P3AC4R(rgb, DstFormat, - stride, yuv, yuv_step, - &roi) != PRIMITIVES_SUCCESS) - { - PROFILER_EXIT(rgbToYUV420); - goto fail; - } - + rc = prims->RGBToYUV420_8u_P3AC4R(rgb, DstFormat, + stride, yuv, yuv_step, + &roi); PROFILER_EXIT(rgbToYUV420); + + if (rc != PRIMITIVES_SUCCESS) + goto loop_fail; + PROFILER_PRINT_HEADER; PROFILER_PRINT(rgbToYUV420); PROFILER_PRINT_FOOTER; } if (!check_padding(rgb, size * sizeof(UINT32), padding, "rgb")) - goto fail; + { + rc = -1; + goto loop_fail; + } if ((!check_padding(yuv[0], size, padding, "Y")) || (!check_padding(yuv[1], uvsize, padding, "U")) || (!check_padding(yuv[2], uvsize, padding, "V"))) - goto fail; + { + rc = -1; + goto loop_fail; + } if (use444) { PROFILER_ENTER(yuv444ToRGB); + rc = prims->YUV444ToRGB_8u_P3AC4R((const BYTE**)yuv, yuv_step, rgb_dst, stride, + DstFormat, + &roi); + PROFILER_EXIT(yuv444ToRGB); - if (prims->YUV444ToRGB_8u_P3AC4R((const BYTE**)yuv, yuv_step, rgb_dst, stride, - DstFormat, - &roi) != PRIMITIVES_SUCCESS) - { - PROFILER_EXIT(yuv444ToRGB); - goto fail; - } + if (rc != PRIMITIVES_SUCCESS) + goto loop_fail; + loop_fail: PROFILER_EXIT(yuv444ToRGB); PROFILER_PRINT_HEADER; PROFILER_PRINT(yuv444ToRGB); PROFILER_PRINT_FOOTER; + + if (rc != PRIMITIVES_SUCCESS) + goto fail; } else { @@ -552,14 +556,15 @@ static BOOL TestPrimitiveYUV(primitives_t* prims, prim_size_t roi, BOOL use444) if (!similarRGB(srgb, drgb, roi.width, DstFormat)) goto fail; } + + PROFILER_FREE(rgbToYUV420); + PROFILER_FREE(rgbToYUV444); + PROFILER_FREE(yuv420ToRGB); + PROFILER_FREE(yuv444ToRGB); } rc = TRUE; fail: - PROFILER_FREE(rgbToYUV420); - PROFILER_FREE(rgbToYUV444); - PROFILER_FREE(yuv420ToRGB); - PROFILER_FREE(yuv444ToRGB); free_padding(rgb, padding); free_padding(rgb_dst, padding); free_padding(yuv[0], padding); diff --git a/winpr/libwinpr/sspi/sspi.c b/winpr/libwinpr/sspi/sspi.c index 7c8b0af83..8feb847ae 100644 --- a/winpr/libwinpr/sspi/sspi.c +++ b/winpr/libwinpr/sspi/sspi.c @@ -47,7 +47,6 @@ static SecurityFunctionTableA* g_SspiA = NULL; static BOOL ShouldUseNativeSspi(void); static BOOL InitializeSspiModule_Native(void); #endif -static void InitializeSspiModule(DWORD flags); #if defined(WITH_NATIVE_SSPI) BOOL ShouldUseNativeSspi(void) @@ -146,12 +145,6 @@ static BOOL CALLBACK InitializeSspiModuleInt(PINIT_ONCE once, PVOID param, PVOID return TRUE; } -void InitializeSspiModule(DWORD flags) -{ - BOOL status = FALSE; - InitOnceExecuteOnce(&g_Initialized, InitializeSspiModuleInt, &flags, NULL); -} - const char* GetSecurityStatusString(SECURITY_STATUS status) { switch (status) diff --git a/winpr/libwinpr/utils/test/TestLinkedList.c b/winpr/libwinpr/utils/test/TestLinkedList.c index 05cf4c645..fe2f72b62 100644 --- a/winpr/libwinpr/utils/test/TestLinkedList.c +++ b/winpr/libwinpr/utils/test/TestLinkedList.c @@ -6,15 +6,11 @@ int TestLinkedList(int argc, char* argv[]) { int count; - int number; wLinkedList* list; - list = LinkedList_New(); - - LinkedList_AddFirst(list, (void*) (size_t) 1); - LinkedList_AddLast(list, (void*) (size_t) 2); - LinkedList_AddLast(list, (void*) (size_t) 3); - + LinkedList_AddFirst(list, (void*)(size_t) 1); + LinkedList_AddLast(list, (void*)(size_t) 2); + LinkedList_AddLast(list, (void*)(size_t) 3); count = LinkedList_Count(list); if (count != 3) @@ -29,14 +25,12 @@ int TestLinkedList(int argc, char* argv[]) { printf("\t%"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); - printf("LinkedList First: %"PRIuz" Last: %"PRIuz"\n", - (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); - + (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); LinkedList_RemoveFirst(list); LinkedList_RemoveLast(list); - count = LinkedList_Count(list); if (count != 1) @@ -51,14 +45,12 @@ int TestLinkedList(int argc, char* argv[]) { printf("\t%"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); - printf("LinkedList First: %"PRIuz" Last: %"PRIuz"\n", - (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); - + (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); LinkedList_RemoveFirst(list); LinkedList_RemoveLast(list); - count = LinkedList_Count(list); if (count != 0) @@ -67,10 +59,9 @@ int TestLinkedList(int argc, char* argv[]) return -1; } - LinkedList_AddFirst(list, (void*) (size_t) 4); - LinkedList_AddLast(list, (void*) (size_t) 5); - LinkedList_AddLast(list, (void*) (size_t) 6); - + LinkedList_AddFirst(list, (void*)(size_t) 4); + LinkedList_AddLast(list, (void*)(size_t) 5); + LinkedList_AddLast(list, (void*)(size_t) 6); count = LinkedList_Count(list); if (count != 3) @@ -85,51 +76,47 @@ int TestLinkedList(int argc, char* argv[]) { printf("\t%"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); - printf("LinkedList First: %"PRIuz" Last: %"PRIuz"\n", - (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); - - LinkedList_Remove(list, (void*) (size_t) 5); - + (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); + LinkedList_Remove(list, (void*)(size_t) 5); LinkedList_Enumerator_Reset(list); while (LinkedList_Enumerator_MoveNext(list)) { printf("\t%"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); - printf("LinkedList First: %"PRIuz" Last: %"PRIuz"\n", - (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); - + (size_t) LinkedList_First(list), (size_t) LinkedList_Last(list)); LinkedList_Free(list); - /* Test enumerator robustness */ - /* enumerator on an empty list */ list = LinkedList_New(); LinkedList_Enumerator_Reset(list); + while (LinkedList_Enumerator_MoveNext(list)) { printf("\terror: %"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); LinkedList_Free(list); - /* Use an enumerator without reset */ list = LinkedList_New(); - LinkedList_AddFirst(list, (void*) (size_t) 4); - LinkedList_AddLast(list, (void*) (size_t) 5); - LinkedList_AddLast(list, (void*) (size_t) 6); + LinkedList_AddFirst(list, (void*)(size_t) 4); + LinkedList_AddLast(list, (void*)(size_t) 5); + LinkedList_AddLast(list, (void*)(size_t) 6); + while (LinkedList_Enumerator_MoveNext(list)) { printf("\t%"PRIuz"\n", (size_t) LinkedList_Enumerator_Current(list)); } + printf("\n"); LinkedList_Free(list); - - return 0; }