gallium/util: move endian detect into a separate file
[mesa.git] / src / gallium / winsys / amdgpu / drm / addrlib / core / addrelemlib.h
1 /*
2 * Copyright © 2014 Advanced Micro Devices, Inc.
3 * All Rights Reserved.
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining
6 * a copy of this software and associated documentation files (the
7 * "Software"), to deal in the Software without restriction, including
8 * without limitation the rights to use, copy, modify, merge, publish,
9 * distribute, sub license, and/or sell copies of the Software, and to
10 * permit persons to whom the Software is furnished to do so, subject to
11 * the following conditions:
12 *
13 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
14 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
15 * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
16 * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
17 * AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
20 * USE OR OTHER DEALINGS IN THE SOFTWARE.
21 *
22 * The above copyright notice and this permission notice (including the
23 * next paragraph) shall be included in all copies or substantial portions
24 * of the Software.
25 */
26
27 /**
28 ***************************************************************************************************
29 * @file addrelemlib.h
30 * @brief Contains the class for element/pixel related functions
31 ***************************************************************************************************
32 */
33
34 #ifndef __ELEM_LIB_H__
35 #define __ELEM_LIB_H__
36
37 #include "addrinterface.h"
38 #include "addrobject.h"
39 #include "addrcommon.h"
40
41 class AddrLib;
42
43 // The masks for property bits within the Properties INT_32
44 union ADDR_COMPONENT_FLAGS
45 {
46 struct
47 {
48 UINT_32 byteAligned : 1; ///< all components are byte aligned
49 UINT_32 exportNorm : 1; ///< components support R6xx NORM compression
50 UINT_32 floatComp : 1; ///< there is at least one floating point component
51 };
52
53 UINT_32 value;
54 };
55
56 // Copy from legacy lib's AddrNumberType
57 enum AddrNumberType
58 {
59 // The following number types have the range [-1..1]
60 ADDR_NO_NUMBER, // This component doesn't exist and has no default value
61 ADDR_EPSILON, // Force component value to integer 0x00000001
62 ADDR_ZERO, // Force component value to integer 0x00000000
63 ADDR_ONE, // Force component value to floating point 1.0
64 // Above values don't have any bits per component (keep ADDR_ONE the last of these)
65
66 ADDR_UNORM, // Unsigned normalized (repeating fraction) full precision
67 ADDR_SNORM, // Signed normalized (repeating fraction) full precision
68 ADDR_GAMMA, // Gamma-corrected, full precision
69
70 ADDR_UNORM_R5XXRB, // Unsigned normalized (repeating fraction) for r5xx RB
71 ADDR_SNORM_R5XXRB, // Signed normalized (repeating fraction) for r5xx RB
72 ADDR_GAMMA_R5XXRB, // Gamma-corrected for r5xx RB (note: unnormalized value)
73 ADDR_UNORM_R5XXBC, // Unsigned normalized (repeating fraction) for r5xx BC
74 ADDR_SNORM_R5XXBC, // Signed normalized (repeating fraction) for r5xx BC
75 ADDR_GAMMA_R5XXBC, // Gamma-corrected for r5xx BC (note: unnormalized value)
76
77 ADDR_UNORM_R6XX, // Unsigned normalized (repeating fraction) for R6xx
78 ADDR_UNORM_R6XXDB, // Unorms for 24-bit depth: one value differs from ADDR_UNORM_R6XX
79 ADDR_SNORM_R6XX, // Signed normalized (repeating fraction) for R6xx
80 ADDR_GAMMA8_R6XX, // Gamma-corrected for r6xx
81 ADDR_GAMMA8_R7XX_TP, // Gamma-corrected for r7xx TP 12bit unorm 8.4.
82
83 ADDR_U4FLOATC, // Unsigned float: 4-bit exponent, bias=15, no NaN, clamp [0..1]
84 ADDR_GAMMA_4SEG, // Gamma-corrected, four segment approximation
85 ADDR_U0FIXED, // Unsigned 0.N-bit fixed point
86
87 // The following number types have large ranges (LEAVE ADDR_USCALED first or fix Finish routine)
88 ADDR_USCALED, // Unsigned integer converted to/from floating point
89 ADDR_SSCALED, // Signed integer converted to/from floating point
90 ADDR_USCALED_R5XXRB, // Unsigned integer to/from floating point for r5xx RB
91 ADDR_SSCALED_R5XXRB, // Signed integer to/from floating point for r5xx RB
92 ADDR_UINT_BITS, // Keep in unsigned integer form, clamped to specified range
93 ADDR_SINT_BITS, // Keep in signed integer form, clamped to specified range
94 ADDR_UINTBITS, // @@ remove Keep in unsigned integer form, use modulus to reduce bits
95 ADDR_SINTBITS, // @@ remove Keep in signed integer form, use modulus to reduce bits
96
97 // The following number types and ADDR_U4FLOATC have exponents
98 // (LEAVE ADDR_S8FLOAT first or fix Finish routine)
99 ADDR_S8FLOAT, // Signed floating point with 8-bit exponent, bias=127
100 ADDR_S8FLOAT32, // 32-bit IEEE float, passes through NaN values
101 ADDR_S5FLOAT, // Signed floating point with 5-bit exponent, bias=15
102 ADDR_S5FLOATM, // Signed floating point with 5-bit exponent, bias=15, no NaN/Inf
103 ADDR_U5FLOAT, // Signed floating point with 5-bit exponent, bias=15
104 ADDR_U3FLOATM, // Unsigned floating point with 3-bit exponent, bias=3
105
106 ADDR_S5FIXED, // Signed 5.N-bit fixed point, with rounding
107
108 ADDR_END_NUMBER // Used for range comparisons
109 };
110
111 // Copy from legacy lib's AddrElement
112 enum AddrElemMode
113 {
114 // These formats allow both packing an unpacking
115 ADDR_ROUND_BY_HALF, // add 1/2 and truncate when packing this element
116 ADDR_ROUND_TRUNCATE, // truncate toward 0 for sign/mag, else toward neg
117 ADDR_ROUND_DITHER, // Pack by dithering -- requires (x,y) position
118
119 // These formats only allow unpacking, no packing
120 ADDR_UNCOMPRESSED, // Elements are not compressed: one data element per pixel/texel
121 ADDR_EXPANDED, // Elements are split up and stored in multiple data elements
122 ADDR_PACKED_STD, // Elements are compressed into ExpandX by ExpandY data elements
123 ADDR_PACKED_REV, // Like ADDR_PACKED, but X order of pixels is reverved
124 ADDR_PACKED_GBGR, // Elements are compressed 4:2:2 in G1B_G0R order (high to low)
125 ADDR_PACKED_BGRG, // Elements are compressed 4:2:2 in BG1_RG0 order (high to low)
126 ADDR_PACKED_BC1, // Each data element is uncompressed to a 4x4 pixel/texel array
127 ADDR_PACKED_BC2, // Each data element is uncompressed to a 4x4 pixel/texel array
128 ADDR_PACKED_BC3, // Each data element is uncompressed to a 4x4 pixel/texel array
129 ADDR_PACKED_BC4, // Each data element is uncompressed to a 4x4 pixel/texel array
130 ADDR_PACKED_BC5, // Each data element is uncompressed to a 4x4 pixel/texel array
131
132 // These formats provide various kinds of compression
133 ADDR_ZPLANE_R5XX, // Compressed Zplane using r5xx architecture format
134 ADDR_ZPLANE_R6XX, // Compressed Zplane using r6xx architecture format
135 //@@ Fill in the compression modes
136
137 ADDR_END_ELEMENT // Used for range comparisons
138 };
139
140 enum AddrDepthPlanarType
141 {
142 ADDR_DEPTH_PLANAR_NONE = 0, // No plane z/stencl
143 ADDR_DEPTH_PLANAR_R600 = 1, // R600 z and stencil planes are store within a tile
144 ADDR_DEPTH_PLANAR_R800 = 2, // R800 has separate z and stencil planes
145 };
146
147 /**
148 ***************************************************************************************************
149 * ADDR_PIXEL_FORMATINFO
150 *
151 * @brief
152 * Per component info
153 *
154 ***************************************************************************************************
155 */
156 struct ADDR_PIXEL_FORMATINFO
157 {
158 UINT_32 compBit[4];
159 AddrNumberType numType[4];
160 UINT_32 compStart[4];
161 AddrElemMode elemMode;
162 UINT_32 comps; ///< Number of components
163 };
164
165 /**
166 ***************************************************************************************************
167 * @brief This class contains asic indepentent element related attributes and operations
168 ***************************************************************************************************
169 */
170 class AddrElemLib : public AddrObject
171 {
172 protected:
173 AddrElemLib(AddrLib* const pAddrLib);
174
175 public:
176
177 /// Makes this class virtual
178 virtual ~AddrElemLib();
179
180 static AddrElemLib *Create(
181 const AddrLib* const pAddrLib);
182
183 /// The implementation is only for R6xx/R7xx, so make it virtual in case we need for R8xx
184 BOOL_32 PixGetExportNorm(
185 AddrColorFormat colorFmt,
186 AddrSurfaceNumber numberFmt, AddrSurfaceSwap swap) const;
187
188 /// Below method are asic independent, so make them just static.
189 /// Remove static if we need different operation in hwl.
190
191 VOID Flt32ToDepthPixel(
192 AddrDepthFormat format, const ADDR_FLT_32 comps[2], UINT_8 *pPixel) const;
193
194 VOID Flt32ToColorPixel(
195 AddrColorFormat format, AddrSurfaceNumber surfNum, AddrSurfaceSwap surfSwap,
196 const ADDR_FLT_32 comps[4], UINT_8 *pPixel) const;
197
198 static VOID Flt32sToInt32s(
199 ADDR_FLT_32 value, UINT_32 bits, AddrNumberType numberType, UINT_32* pResult);
200
201 static VOID Int32sToPixel(
202 UINT_32 numComps, UINT_32* pComps, UINT_32* pCompBits, UINT_32* pCompStart,
203 ADDR_COMPONENT_FLAGS properties, UINT_32 resultBits, UINT_8* pPixel);
204
205 VOID PixGetColorCompInfo(
206 AddrColorFormat format, AddrSurfaceNumber number, AddrSurfaceSwap swap,
207 ADDR_PIXEL_FORMATINFO* pInfo) const;
208
209 VOID PixGetDepthCompInfo(
210 AddrDepthFormat format, ADDR_PIXEL_FORMATINFO* pInfo) const;
211
212 UINT_32 GetBitsPerPixel(
213 AddrFormat format, AddrElemMode* pElemMode,
214 UINT_32* pExpandX = NULL, UINT_32* pExpandY = NULL, UINT_32* pBitsUnused = NULL);
215
216 static VOID SetClearComps(
217 ADDR_FLT_32 comps[4], BOOL_32 clearColor, BOOL_32 float32);
218
219 VOID AdjustSurfaceInfo(
220 AddrElemMode elemMode, UINT_32 expandX, UINT_32 expandY,
221 UINT_32* pBpp, UINT_32* pBasePitch, UINT_32* pWidth, UINT_32* pHeight);
222
223 VOID RestoreSurfaceInfo(
224 AddrElemMode elemMode, UINT_32 expandX, UINT_32 expandY,
225 UINT_32* pBpp, UINT_32* pWidth, UINT_32* pHeight);
226
227 /// Checks if depth and stencil are planar inside a tile
228 BOOL_32 IsDepthStencilTilePlanar()
229 {
230 return (m_depthPlanarType == ADDR_DEPTH_PLANAR_R600) ? TRUE : FALSE;
231 }
232
233 /// Sets m_configFlags, copied from AddrLib
234 VOID SetConfigFlags(ADDR_CONFIG_FLAGS flags)
235 {
236 m_configFlags = flags;
237 }
238
239 static BOOL_32 IsCompressed(AddrFormat format);
240 static BOOL_32 IsBlockCompressed(AddrFormat format);
241 static BOOL_32 IsExpand3x(AddrFormat format);
242
243 protected:
244
245 static VOID GetCompBits(
246 UINT_32 c0, UINT_32 c1, UINT_32 c2, UINT_32 c3,
247 ADDR_PIXEL_FORMATINFO* pInfo,
248 AddrElemMode elemMode = ADDR_ROUND_BY_HALF);
249
250 static VOID GetCompType(
251 AddrColorFormat format, AddrSurfaceNumber numType,
252 ADDR_PIXEL_FORMATINFO* pInfo);
253
254 static VOID GetCompSwap(
255 AddrSurfaceSwap swap, ADDR_PIXEL_FORMATINFO* pInfo);
256
257 static VOID SwapComps(
258 UINT_32 c0, UINT_32 c1, ADDR_PIXEL_FORMATINFO* pInfo);
259
260 private:
261
262 UINT_32 m_fp16ExportNorm; ///< If allow FP16 to be reported as EXPORT_NORM
263 AddrDepthPlanarType m_depthPlanarType;
264
265 ADDR_CONFIG_FLAGS m_configFlags; ///< Copy of AddrLib's configFlags
266 AddrLib* const m_pAddrLib; ///< Pointer to parent addrlib instance
267 };
268
269 #endif
270