Merge pull request #3026 from lioncash/constexpr

MathUtil: Make Clamp and IsPow2 constexpr functions.

Source/Core/AudioCommon/DPL2Decoder.cpp

@@ -134,7 +134,8 @@ static float* DesignFIR(unsigned int *n, float* fc, float opt)
 	// Sanity check
 	if (*n == 0)
 		return nullptr;
-	MathUtil::Clamp(&fc[0], float(0.001), float(1));
+
+	fc[0] = MathUtil::Clamp(fc[0], 0.001f, 1.0f);
 
 	float *w = (float*)calloc(sizeof(float), *n);
 

Source/Core/AudioCommon/Mixer.cpp

@@ -65,16 +65,14 @@ unsigned int CMixer::MixerFifo::Mix(short* samples, unsigned int numSamples, boo
 		int sampleL = ((l1 << 16) + (l2 - l1) * (u16)m_frac) >> 16;
 		sampleL = (sampleL * lvolume) >> 8;
 		sampleL += samples[currentSample + 1];
-		MathUtil::Clamp(&sampleL, -32767, 32767);
+		samples[currentSample + 1] = MathUtil::Clamp(sampleL, -32767, 32767);
-		samples[currentSample + 1] = sampleL;
 
 		s16 r1 = Common::swap16(m_buffer[(indexR + 1) & INDEX_MASK]); //current
 		s16 r2 = Common::swap16(m_buffer[(indexR2 + 1) & INDEX_MASK]); //next
 		int sampleR = ((r1 << 16) + (r2 - r1) * (u16)m_frac) >> 16;
 		sampleR = (sampleR * rvolume) >> 8;
 		sampleR += samples[currentSample];
-		MathUtil::Clamp(&sampleR, -32767, 32767);
+		samples[currentSample] = MathUtil::Clamp(sampleR, -32767, 32767);
-		samples[currentSample] = sampleR;
 
 		m_frac += ratio;
 		indexR += 2 * (u16)(m_frac >> 16);
@@ -89,11 +87,10 @@ unsigned int CMixer::MixerFifo::Mix(short* samples, unsigned int numSamples, boo
 	s[1] = (s[1] * lvolume) >> 8;
 	for (; currentSample < numSamples * 2; currentSample += 2)
 	{
-		int sampleR = s[0] + samples[currentSample];
+		int sampleR = MathUtil::Clamp(s[0] + samples[currentSample + 0], -32767, 32767);
-		MathUtil::Clamp(&sampleR, -32767, 32767);
+		int sampleL = MathUtil::Clamp(s[1] + samples[currentSample + 1], -32767, 32767);
-		samples[currentSample] = sampleR;
+
-		int sampleL = s[1] + samples[currentSample + 1];
+		samples[currentSample + 0] = sampleR;
-		MathUtil::Clamp(&sampleL, -32767, 32767);
 		samples[currentSample + 1] = sampleL;
 	}
 

Source/Core/Common/MathUtil.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <algorithm>
 #include <cstdlib>
 #include <vector>
 
@@ -12,20 +13,14 @@
 namespace MathUtil
 {
 template<class T>
-inline void Clamp(T* val, const T& min, const T& max)
+constexpr T Clamp(const T val, const T& min, const T& max)
 {
-	if (*val < min)
+	return std::max(min, std::min(max, val));
-		*val = min;
-	else if (*val > max)
-		*val = max;
 }
 
-template<class T>
+constexpr bool IsPow2(u32 imm)
-inline T Clamp(const T val, const T& min, const T& max)
 {
-	T ret = val;
+	return (imm & (imm - 1)) == 0;
-	Clamp(&ret, min, max);
-	return ret;
 }
 
 // The most significant bit of the fraction is an is-quiet bit on all architectures we care about.
@@ -143,20 +138,20 @@ struct Rectangle
 	// this Clamp.
 	void ClampLL(T x1, T y1, T x2, T y2)
 	{
-		Clamp(&left, x1, x2);
+		left   = Clamp(left, x1, x2);
-		Clamp(&right, x1, x2);
+		right  = Clamp(right, x1, x2);
-		Clamp(&top, y2, y1);
+		top    = Clamp(top, y2, y1);
-		Clamp(&bottom, y2, y1);
+		bottom = Clamp(bottom, y2, y1);
 	}
 
 	// If the rectangle is in a coordinate system with an upper-left origin,
 	// use this Clamp.
 	void ClampUL(T x1, T y1, T x2, T y2)
 	{
-		Clamp(&left, x1, x2);
+		left   = Clamp(left, x1, x2);
-		Clamp(&right, x1, x2);
+		right  = Clamp(right, x1, x2);
-		Clamp(&top, y1, y2);
+		top    = Clamp(top, y1, y2);
-		Clamp(&bottom, y1, y2);
+		bottom = Clamp(bottom, y1, y2);
 	}
 };
 
@@ -167,8 +162,6 @@ float MathFloatVectorSum(const std::vector<float>&);
 #define ROUND_UP(x, a)   (((x) + (a) - 1) & ~((a) - 1))
 #define ROUND_DOWN(x, a) ((x) & ~((a) - 1))
 
-inline bool IsPow2(u32 imm) {return (imm & (imm - 1)) == 0;}
-
 // Rounds down. 0 -> undefined
 inline int IntLog2(u64 val)
 {

Source/Core/Core/DSP/DSPAccelerator.cpp

@@ -36,8 +36,7 @@ static s16 ADPCM_Step(u32& _rSamplePos)
 
 	// 0x400 = 0.5  in 11-bit fixed point
 	int val = (scale * temp) + ((0x400 + coef1 * (s16)g_dsp.ifx_regs[DSP_YN1] + coef2 * (s16)g_dsp.ifx_regs[DSP_YN2]) >> 11);
-
+	val = MathUtil::Clamp(val, -0x7FFF, 0x7FFF);
-	MathUtil::Clamp(&val, -0x7FFF, 0x7FFF);
 
 	g_dsp.ifx_regs[DSP_YN2] = g_dsp.ifx_regs[DSP_YN1];
 	g_dsp.ifx_regs[DSP_YN1] = val;

Source/Core/Core/HW/DSPHLE/UCodes/AX.cpp

@@ -503,11 +503,8 @@ void AXUCode::OutputSamples(u32 lr_addr, u32 surround_addr)
 	// Output samples clamped to 16 bits and interlaced RLRLRLRLRL...
 	for (u32 i = 0; i < 5 * 32; ++i)
 	{
-		int left  = m_samples_left[i];
+		int left  = MathUtil::Clamp(m_samples_left[i], -32767, 32767);
-		int right = m_samples_right[i];
+		int right = MathUtil::Clamp(m_samples_right[i], -32767, 32767);
-
-		MathUtil::Clamp(&left, -32767, 32767);
-		MathUtil::Clamp(&right, -32767, 32767);
 
 		buffer[2 * i + 0] = Common::swap16(right);
 		buffer[2 * i + 1] = Common::swap16(left);

Source/Core/Core/HW/DSPHLE/UCodes/AXVoice.h

@@ -185,7 +185,7 @@ u16 AcceleratorGetSample()
 				temp -= 16;
 
 			int val = (scale * temp) + ((0x400 + coef1 * acc_pb->adpcm.yn1 + coef2 * acc_pb->adpcm.yn2) >> 11);
-			MathUtil::Clamp(&val, -0x7FFF, 0x7FFF);
+			val = MathUtil::Clamp(val, -0x7FFF, 0x7FFF);
 
 			acc_pb->adpcm.yn2 = acc_pb->adpcm.yn1;
 			acc_pb->adpcm.yn1 = val;

Source/Core/Core/HW/DSPHLE/UCodes/AXWii.cpp

@@ -618,11 +618,8 @@ void AXWiiUCode::OutputSamples(u32 lr_addr, u32 surround_addr, u16 volume,
 		left = ((s64)left * volume_ramp[i]) >> 15;
 		right = ((s64)right * volume_ramp[i]) >> 15;
 
-		MathUtil::Clamp(&left, -32767, 32767);
+		m_samples_left[i]  = MathUtil::Clamp(left, -32767, 32767);
-		MathUtil::Clamp(&right, -32767, 32767);
+		m_samples_right[i] = MathUtil::Clamp(right, -32767, 32767);
-
-		m_samples_left[i] = left;
-		m_samples_right[i] = right;
 	}
 
 	for (u32 i = 0; i < 3 * 32; ++i)
@@ -653,8 +650,7 @@ void AXWiiUCode::OutputWMSamples(u32* addresses)
 		u16* out = (u16*)HLEMemory_Get_Pointer(addresses[i]);
 		for (u32 j = 0; j < 3 * 6; ++j)
 		{
-			int sample = in[j];
+			int sample = MathUtil::Clamp(in[j], -32767, 32767);
-			MathUtil::Clamp(&sample, -32767, 32767);
 			out[j] = Common::swap16((u16)sample);
 		}
 	}

Source/Core/Core/HW/DSPHLE/UCodes/Zelda.cpp

@@ -1015,8 +1015,7 @@ void ZeldaAudioRenderer::ApplyReverb(bool post_rendering)
 					for (u16 j = 0; j < 8; ++j)
 						sample += (s32)buffer[i + j] * rpb.filter_coeffs[j];
 					sample >>= 15;
-					MathUtil::Clamp(&sample, -0x8000, 0x7fff);
+					buffer[i] = MathUtil::Clamp(sample, -0x8000, 0x7FFF);
-					buffer[i] = sample;
 				}
 			};
 
@@ -1445,8 +1444,8 @@ void ZeldaAudioRenderer::Resample(VPB* vpb, const s16* src, MixingBuffer* dst)
 			for (size_t i = 0; i < 4; ++i)
 				dst_sample_unclamped += (s64)2 * coeffs[i] * input[i];
 			dst_sample_unclamped >>= 16;
-			MathUtil::Clamp(&dst_sample_unclamped, (s64)-0x8000, (s64)0x7fff);
+
-			dst_sample = (s16)dst_sample_unclamped;
+			dst_sample = (s16)MathUtil::Clamp<s64>(dst_sample_unclamped, -0x8000, 0x7FFF);
 
 			pos += ratio;
 		}
@@ -1696,7 +1695,7 @@ void ZeldaAudioRenderer::DecodeAFC(VPB* vpb, s16* dst, size_t block_count)
 				yn1 * m_afc_coeffs[idx * 2] +
 				yn2 * m_afc_coeffs[idx * 2 + 1];
 			sample >>= 11;
-			MathUtil::Clamp(&sample, -0x8000, 0x7fff);
+			sample = MathUtil::Clamp(sample, -0x8000, 0x7fff);
 			*dst++ = (s16)sample;
 			yn2 = yn1;
 			yn1 = sample;

Source/Core/Core/HW/DSPHLE/UCodes/Zelda.h

@@ -46,8 +46,8 @@ private:
 		{
 			s32 tmp = (u32)(*buf)[i] * (u32)vol;
 			tmp >>= 16 - B;
-			MathUtil::Clamp(&tmp, -0x8000, 0x7fff);
+
-			(*buf)[i] = (s16)tmp;
+			(*buf)[i] = (s16)MathUtil::Clamp(tmp, -0x8000, 0x7FFF);
 		}
 	}
 	template <size_t N>
@@ -90,8 +90,7 @@ private:
 		while (count--)
 		{
 			s32 vol_src = ((s32)*src++ * (s32)vol) >> 15;
-			MathUtil::Clamp(&vol_src, -0x8000, 0x7fff);
+			*dst++ += MathUtil::Clamp(vol_src, -0x8000, 0x7FFF);
-			*dst++ += vol_src;
 		}
 	}
 

Source/Core/Core/HW/StreamADPCM.cpp

@@ -31,8 +31,7 @@ static s16 ADPDecodeSample(s32 bits, s32 q, s32& hist1, s32& hist2)
 		hist = (hist1 * 0x62) - (hist2 * 0x37);
 		break;
 	}
-	hist = (hist + 0x20) >> 6;
+	hist = MathUtil::Clamp((hist + 0x20) >> 6, -0x200000, 0x1fffff);
-	MathUtil::Clamp(&hist, -0x200000, 0x1fffff);
 
 	s32 cur = (((s16)(bits << 12) >> (q & 0xf)) << 6) + hist;
 
@@ -40,7 +39,7 @@ static s16 ADPDecodeSample(s32 bits, s32 q, s32& hist1, s32& hist2)
 	hist1 = cur;
 
 	cur >>= 6;
-	MathUtil::Clamp(&cur, -0x8000, 0x7fff);
+	cur = MathUtil::Clamp(cur, -0x8000, 0x7fff);
 
 	return (s16)cur;
 }

Source/Core/Core/PowerPC/Interpreter/Interpreter_LoadStorePaired.cpp

@@ -53,8 +53,8 @@ template<typename SType> SType ScaleAndClamp(double ps, u32 stScale)
 	float convPS = (float)ps * m_quantizeTable[stScale];
 	float min = (float)std::numeric_limits<SType>::min();
 	float max = (float)std::numeric_limits<SType>::max();
-	MathUtil::Clamp(&convPS, min, max);
+
-	return (SType)convPS;
+	return (SType)MathUtil::Clamp(convPS, min, max);
 }
 
 template<typename T> static T ReadUnpaired(u32 addr);

Source/Core/Core/PowerPC/Jit64/Jit_Integer.cpp

@@ -924,7 +924,7 @@ void Jit64::MultiplyImmediate(u32 imm, int a, int d, bool overflow)
 	if (!overflow)
 	{
 		// power of 2; just a shift
-		if (IsPow2(imm))
+		if (MathUtil::IsPow2(imm))
 		{
 			u32 shift = IntLog2(imm);
 			// use LEA if it saves an op

Source/Core/DolphinWX/FrameAui.cpp

@@ -677,8 +677,8 @@ void CFrame::SetPaneSize()
 				H = Perspectives[ActivePerspective].Height[j];
 
 			// Check limits
-			MathUtil::Clamp<u32>(&W, 5, 95);
+			W = MathUtil::Clamp<u32>(W, 5, 95);
-			MathUtil::Clamp<u32>(&H, 5, 95);
+			H = MathUtil::Clamp<u32>(H, 5, 95);
 
 			// Convert percentages to pixel lengths
 			W = (W * iClientX) / 100;

Source/Core/VideoBackends/Software/TransformUnit.cpp

@@ -336,12 +336,9 @@ void TransformColor(const InputVertexData *src, OutputVertexData *dst)
 					LightColor(dst->mvPosition, dst->normal[0], i, colorchan, lightCol);
 			}
 
-			int light_x = int(lightCol.x);
+			int light_x = MathUtil::Clamp(static_cast<int>(lightCol.x), 0, 255);
-			int light_y = int(lightCol.y);
+			int light_y = MathUtil::Clamp(static_cast<int>(lightCol.y), 0, 255);
-			int light_z = int(lightCol.z);
+			int light_z = MathUtil::Clamp(static_cast<int>(lightCol.z), 0, 255);
-			MathUtil::Clamp(&light_x, 0, 255);
-			MathUtil::Clamp(&light_y, 0, 255);
-			MathUtil::Clamp(&light_z, 0, 255);
 			chancolor[1] = (matcolor[1] * (light_x + (light_x >> 7))) >> 8;
 			chancolor[2] = (matcolor[2] * (light_y + (light_y >> 7))) >> 8;
 			chancolor[3] = (matcolor[3] * (light_z + (light_z >> 7))) >> 8;
@@ -373,8 +370,7 @@ void TransformColor(const InputVertexData *src, OutputVertexData *dst)
 					LightAlpha(dst->mvPosition, dst->normal[0], i, alphachan, lightCol);
 			}
 
-			int light_a = int(lightCol);
+			int light_a = MathUtil::Clamp(static_cast<int>(lightCol), 0, 255);
-			MathUtil::Clamp(&light_a, 0, 255);
 			chancolor[0] = (matcolor[0] * (light_a + (light_a >> 7))) >> 8;
 		}
 		else