summaryrefslogtreecommitdiff
path: root/src/gallium/drivers/softpipe/sp_tex_sample.c
blob: 274e56b40c71b8ad80de14368ebbe6321d208df2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
/**************************************************************************
 * 
 * Copyright 2007 VMware, Inc.
 * All Rights Reserved.
 * Copyright 2008-2010 VMware, Inc.  All rights reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sub license, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 * 
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 * 
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 * 
 **************************************************************************/

/**
 * Texture sampling
 *
 * Authors:
 *   Brian Paul
 *   Keith Whitwell
 */

#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
#include "util/u_math.h"
#include "util/u_format.h"
#include "util/u_memory.h"
#include "util/u_inlines.h"
#include "sp_quad.h"   /* only for #define QUAD_* tokens */
#include "sp_tex_sample.h"
#include "sp_texture.h"
#include "sp_tex_tile_cache.h"


/** Set to one to help debug texture sampling */
#define DEBUG_TEX 0


/*
 * Return fractional part of 'f'.  Used for computing interpolation weights.
 * Need to be careful with negative values.
 * Note, if this function isn't perfect you'll sometimes see 1-pixel bands
 * of improperly weighted linear-filtered textures.
 * The tests/texwrap.c demo is a good test.
 */
static INLINE float
frac(float f)
{
   return f - floorf(f);
}



/**
 * Linear interpolation macro
 */
static INLINE float
lerp(float a, float v0, float v1)
{
   return v0 + a * (v1 - v0);
}


/**
 * Do 2D/bilinear interpolation of float values.
 * v00, v10, v01 and v11 are typically four texture samples in a square/box.
 * a and b are the horizontal and vertical interpolants.
 * It's important that this function is inlined when compiled with
 * optimization!  If we find that's not true on some systems, convert
 * to a macro.
 */
static INLINE float
lerp_2d(float a, float b,
        float v00, float v10, float v01, float v11)
{
   const float temp0 = lerp(a, v00, v10);
   const float temp1 = lerp(a, v01, v11);
   return lerp(b, temp0, temp1);
}


/**
 * As above, but 3D interpolation of 8 values.
 */
static INLINE float
lerp_3d(float a, float b, float c,
        float v000, float v100, float v010, float v110,
        float v001, float v101, float v011, float v111)
{
   const float temp0 = lerp_2d(a, b, v000, v100, v010, v110);
   const float temp1 = lerp_2d(a, b, v001, v101, v011, v111);
   return lerp(c, temp0, temp1);
}



/**
 * Compute coord % size for repeat wrap modes.
 * Note that if coord is negative, coord % size doesn't give the right
 * value.  To avoid that problem we add a large multiple of the size
 * (rather than using a conditional).
 */
static INLINE int
repeat(int coord, unsigned size)
{
   return (coord + size * 1024) % size;
}


/**
 * Apply texture coord wrapping mode and return integer texture indexes
 * for a vector of four texcoords (S or T or P).
 * \param wrapMode  PIPE_TEX_WRAP_x
 * \param s  the incoming texcoords
 * \param size  the texture image size
 * \param icoord  returns the integer texcoords
 */
static void
wrap_nearest_repeat(float s, unsigned size, int *icoord)
{
   /* s limited to [0,1) */
   /* i limited to [0,size-1] */
   int i = util_ifloor(s * size);
   *icoord = repeat(i, size);
}


static void
wrap_nearest_clamp(float s, unsigned size, int *icoord)
{
   /* s limited to [0,1] */
   /* i limited to [0,size-1] */
   if (s <= 0.0F)
      *icoord = 0;
   else if (s >= 1.0F)
      *icoord = size - 1;
   else
      *icoord = util_ifloor(s * size);
}


static void
wrap_nearest_clamp_to_edge(float s, unsigned size, int *icoord)
{
   /* s limited to [min,max] */
   /* i limited to [0, size-1] */
   const float min = 1.0F / (2.0F * size);
   const float max = 1.0F - min;
   if (s < min)
      *icoord = 0;
   else if (s > max)
      *icoord = size - 1;
   else
      *icoord = util_ifloor(s * size);
}


static void
wrap_nearest_clamp_to_border(float s, unsigned size, int *icoord)
{
   /* s limited to [min,max] */
   /* i limited to [-1, size] */
   const float min = -1.0F / (2.0F * size);
   const float max = 1.0F - min;
   if (s <= min)
      *icoord = -1;
   else if (s >= max)
      *icoord = size;
   else
      *icoord = util_ifloor(s * size);
}


static void
wrap_nearest_mirror_repeat(float s, unsigned size, int *icoord)
{
   const float min = 1.0F / (2.0F * size);
   const float max = 1.0F - min;
   const int flr = util_ifloor(s);
   float u = frac(s);
   if (flr & 1)
      u = 1.0F - u;
   if (u < min)
      *icoord = 0;
   else if (u > max)
      *icoord = size - 1;
   else
      *icoord = util_ifloor(u * size);
}


static void
wrap_nearest_mirror_clamp(float s, unsigned size, int *icoord)
{
   /* s limited to [0,1] */
   /* i limited to [0,size-1] */
   const float u = fabsf(s);
   if (u <= 0.0F)
      *icoord = 0;
   else if (u >= 1.0F)
      *icoord = size - 1;
   else
      *icoord = util_ifloor(u * size);
}


static void
wrap_nearest_mirror_clamp_to_edge(float s, unsigned size, int *icoord)
{
   /* s limited to [min,max] */
   /* i limited to [0, size-1] */
   const float min = 1.0F / (2.0F * size);
   const float max = 1.0F - min;
   const float u = fabsf(s);
   if (u < min)
      *icoord = 0;
   else if (u > max)
      *icoord = size - 1;
   else
      *icoord = util_ifloor(u * size);
}


static void
wrap_nearest_mirror_clamp_to_border(float s, unsigned size, int *icoord)
{
   /* s limited to [min,max] */
   /* i limited to [0, size-1] */
   const float min = -1.0F / (2.0F * size);
   const float max = 1.0F - min;
   const float u = fabsf(s);
   if (u < min)
      *icoord = -1;
   else if (u > max)
      *icoord = size;
   else
      *icoord = util_ifloor(u * size);
}


/**
 * Used to compute texel locations for linear sampling
 * \param wrapMode  PIPE_TEX_WRAP_x
 * \param s  the texcoord
 * \param size  the texture image size
 * \param icoord0  returns first texture index
 * \param icoord1  returns second texture index (usually icoord0 + 1)
 * \param w  returns blend factor/weight between texture indices
 * \param icoord  returns the computed integer texture coord
 */
static void
wrap_linear_repeat(float s, unsigned size,
                   int *icoord0, int *icoord1, float *w)
{
   float u = s * size - 0.5F;
   *icoord0 = repeat(util_ifloor(u), size);
   *icoord1 = repeat(*icoord0 + 1, size);
   *w = frac(u);
}


static void
wrap_linear_clamp(float s, unsigned size,
                  int *icoord0, int *icoord1, float *w)
{
   float u = CLAMP(s, 0.0F, 1.0F);
   u = u * size - 0.5f;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   *w = frac(u);
}


static void
wrap_linear_clamp_to_edge(float s, unsigned size,
                          int *icoord0, int *icoord1, float *w)
{
   float u = CLAMP(s, 0.0F, 1.0F);
   u = u * size - 0.5f;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   if (*icoord0 < 0)
      *icoord0 = 0;
   if (*icoord1 >= (int) size)
      *icoord1 = size - 1;
   *w = frac(u);
}


static void
wrap_linear_clamp_to_border(float s, unsigned size,
                            int *icoord0, int *icoord1, float *w)
{
   const float min = -1.0F / (2.0F * size);
   const float max = 1.0F - min;
   float u = CLAMP(s, min, max);
   u = u * size - 0.5f;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   *w = frac(u);
}


static void
wrap_linear_mirror_repeat(float s, unsigned size,
                          int *icoord0, int *icoord1, float *w)
{
   const int flr = util_ifloor(s);
   float u = frac(s);
   if (flr & 1)
      u = 1.0F - u;
   u = u * size - 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   if (*icoord0 < 0)
      *icoord0 = 0;
   if (*icoord1 >= (int) size)
      *icoord1 = size - 1;
   *w = frac(u);
}


static void
wrap_linear_mirror_clamp(float s, unsigned size,
                         int *icoord0, int *icoord1, float *w)
{
   float u = fabsf(s);
   if (u >= 1.0F)
      u = (float) size;
   else
      u *= size;
   u -= 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   *w = frac(u);
}


static void
wrap_linear_mirror_clamp_to_edge(float s, unsigned size,
                                 int *icoord0, int *icoord1, float *w)
{
   float u = fabsf(s);
   if (u >= 1.0F)
      u = (float) size;
   else
      u *= size;
   u -= 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   if (*icoord0 < 0)
      *icoord0 = 0;
   if (*icoord1 >= (int) size)
      *icoord1 = size - 1;
   *w = frac(u);
}


static void
wrap_linear_mirror_clamp_to_border(float s, unsigned size,
                                   int *icoord0, int *icoord1, float *w)
{
   const float min = -1.0F / (2.0F * size);
   const float max = 1.0F - min;
   float u = fabsf(s);
   if (u <= min)
      u = min * size;
   else if (u >= max)
      u = max * size;
   else
      u *= size;
   u -= 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   *w = frac(u);
}


/**
 * PIPE_TEX_WRAP_CLAMP for nearest sampling, unnormalized coords.
 */
static void
wrap_nearest_unorm_clamp(float s, unsigned size, int *icoord)
{
   int i = util_ifloor(s);
   *icoord = CLAMP(i, 0, (int) size-1);
}


/**
 * PIPE_TEX_WRAP_CLAMP_TO_BORDER for nearest sampling, unnormalized coords.
 */
static void
wrap_nearest_unorm_clamp_to_border(float s, unsigned size, int *icoord)
{
   *icoord = util_ifloor( CLAMP(s, -0.5F, (float) size + 0.5F) );
}


/**
 * PIPE_TEX_WRAP_CLAMP_TO_EDGE for nearest sampling, unnormalized coords.
 */
static void
wrap_nearest_unorm_clamp_to_edge(float s, unsigned size, int *icoord)
{
   *icoord = util_ifloor( CLAMP(s, 0.5F, (float) size - 0.5F) );
}


/**
 * PIPE_TEX_WRAP_CLAMP for linear sampling, unnormalized coords.
 */
static void
wrap_linear_unorm_clamp(float s, unsigned size,
                        int *icoord0, int *icoord1, float *w)
{
   /* Not exactly what the spec says, but it matches NVIDIA output */
   float u = CLAMP(s - 0.5F, 0.0f, (float) size - 1.0f);
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   *w = frac(u);
}


/**
 * PIPE_TEX_WRAP_CLAMP_TO_BORDER for linear sampling, unnormalized coords.
 */
static void
wrap_linear_unorm_clamp_to_border(float s, unsigned size,
                                  int *icoord0, int *icoord1, float *w)
{
   float u = CLAMP(s, -0.5F, (float) size + 0.5F);
   u -= 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   if (*icoord1 > (int) size - 1)
      *icoord1 = size - 1;
   *w = frac(u);
}


/**
 * PIPE_TEX_WRAP_CLAMP_TO_EDGE for linear sampling, unnormalized coords.
 */
static void
wrap_linear_unorm_clamp_to_edge(float s, unsigned size,
                                int *icoord0, int *icoord1, float *w)
{
   float u = CLAMP(s, +0.5F, (float) size - 0.5F);
   u -= 0.5F;
   *icoord0 = util_ifloor(u);
   *icoord1 = *icoord0 + 1;
   if (*icoord1 > (int) size - 1)
      *icoord1 = size - 1;
   *w = frac(u);
}


/**
 * Do coordinate to array index conversion.  For array textures.
 */
static INLINE void
wrap_array_layer(float coord, unsigned size, int *layer)
{
   int c = util_ifloor(coord + 0.5F);
   *layer = CLAMP(c, 0, (int) size - 1);
}


/**
 * Examine the quad's texture coordinates to compute the partial
 * derivatives w.r.t X and Y, then compute lambda (level of detail).
 */
static float
compute_lambda_1d(const struct sp_sampler_view *sview,
                  const float s[TGSI_QUAD_SIZE],
                  const float t[TGSI_QUAD_SIZE],
                  const float p[TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sview->base.texture;
   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
   float rho = MAX2(dsdx, dsdy) * u_minify(texture->width0, sview->base.u.tex.first_level);

   return util_fast_log2(rho);
}


static float
compute_lambda_2d(const struct sp_sampler_view *sview,
                  const float s[TGSI_QUAD_SIZE],
                  const float t[TGSI_QUAD_SIZE],
                  const float p[TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sview->base.texture;
   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
   float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
   float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
   float maxx = MAX2(dsdx, dsdy) * u_minify(texture->width0, sview->base.u.tex.first_level);
   float maxy = MAX2(dtdx, dtdy) * u_minify(texture->height0, sview->base.u.tex.first_level);
   float rho  = MAX2(maxx, maxy);

   return util_fast_log2(rho);
}


static float
compute_lambda_3d(const struct sp_sampler_view *sview,
                  const float s[TGSI_QUAD_SIZE],
                  const float t[TGSI_QUAD_SIZE],
                  const float p[TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sview->base.texture;
   float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
   float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
   float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
   float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
   float dpdx = fabsf(p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]);
   float dpdy = fabsf(p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT]);
   float maxx = MAX2(dsdx, dsdy) * u_minify(texture->width0, sview->base.u.tex.first_level);
   float maxy = MAX2(dtdx, dtdy) * u_minify(texture->height0, sview->base.u.tex.first_level);
   float maxz = MAX2(dpdx, dpdy) * u_minify(texture->depth0, sview->base.u.tex.first_level);
   float rho;

   rho = MAX2(maxx, maxy);
   rho = MAX2(rho, maxz);

   return util_fast_log2(rho);
}


/**
 * Compute lambda for a vertex texture sampler.
 * Since there aren't derivatives to use, just return 0.
 */
static float
compute_lambda_vert(const struct sp_sampler_view *sview,
                    const float s[TGSI_QUAD_SIZE],
                    const float t[TGSI_QUAD_SIZE],
                    const float p[TGSI_QUAD_SIZE])
{
   return 0.0f;
}



/**
 * Get a texel from a texture, using the texture tile cache.
 *
 * \param addr  the template tex address containing cube, z, face info.
 * \param x  the x coord of texel within 2D image
 * \param y  the y coord of texel within 2D image
 * \param rgba  the quad to put the texel/color into
 *
 * XXX maybe move this into sp_tex_tile_cache.c and merge with the
 * sp_get_cached_tile_tex() function.
 */




static INLINE const float *
get_texel_2d_no_border(const struct sp_sampler_view *sp_sview,
                       union tex_tile_address addr, int x, int y)
{
   const struct softpipe_tex_cached_tile *tile;
   addr.bits.x = x / TEX_TILE_SIZE;
   addr.bits.y = y / TEX_TILE_SIZE;
   y %= TEX_TILE_SIZE;
   x %= TEX_TILE_SIZE;

   tile = sp_get_cached_tile_tex(sp_sview->cache, addr);

   return &tile->data.color[y][x][0];
}


static INLINE const float *
get_texel_2d(const struct sp_sampler_view *sp_sview,
             const struct sp_sampler *sp_samp,
             union tex_tile_address addr, int x, int y)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;

   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
       y < 0 || y >= (int) u_minify(texture->height0, level)) {
      return sp_samp->base.border_color.f;
   }
   else {
      return get_texel_2d_no_border( sp_sview, addr, x, y );
   }
}


/*
 * Here's the complete logic (HOLY CRAP) for finding next face and doing the
 * corresponding coord wrapping, implemented by get_next_face,
 * get_next_xcoord, get_next_ycoord.
 * Read like that (first line):
 * If face is +x and s coord is below zero, then
 * new face is +z, new s is max , new t is old t
 * (max is always cube size - 1).
 *
 * +x s- -> +z: s = max,   t = t
 * +x s+ -> -z: s = 0,     t = t
 * +x t- -> +y: s = max,   t = max-s
 * +x t+ -> -y: s = max,   t = s
 *
 * -x s- -> -z: s = max,   t = t
 * -x s+ -> +z: s = 0,     t = t
 * -x t- -> +y: s = 0,     t = s
 * -x t+ -> -y: s = 0,     t = max-s
 *
 * +y s- -> -x: s = t,     t = 0
 * +y s+ -> +x: s = max-t, t = 0
 * +y t- -> -z: s = max-s, t = 0
 * +y t+ -> +z: s = s,     t = 0
 *
 * -y s- -> -x: s = max-t, t = max
 * -y s+ -> +x: s = t,     t = max
 * -y t- -> +z: s = s,     t = max
 * -y t+ -> -z: s = max-s, t = max

 * +z s- -> -x: s = max,   t = t
 * +z s+ -> +x: s = 0,     t = t
 * +z t- -> +y: s = s,     t = max
 * +z t+ -> -y: s = s,     t = 0

 * -z s- -> +x: s = max,   t = t
 * -z s+ -> -x: s = 0,     t = t
 * -z t- -> +y: s = max-s, t = 0
 * -z t+ -> -y: s = max-s, t = max
 */


/*
 * seamless cubemap neighbour array.
 * this array is used to find the adjacent face in each of 4 directions,
 * left, right, up, down. (or -x, +x, -y, +y).
 */
static const unsigned face_array[PIPE_TEX_FACE_MAX][4] = {
   /* pos X first then neg X is Z different, Y the same */
   /* PIPE_TEX_FACE_POS_X,*/
   { PIPE_TEX_FACE_POS_Z, PIPE_TEX_FACE_NEG_Z,
     PIPE_TEX_FACE_POS_Y, PIPE_TEX_FACE_NEG_Y },
   /* PIPE_TEX_FACE_NEG_X */
   { PIPE_TEX_FACE_NEG_Z, PIPE_TEX_FACE_POS_Z,
     PIPE_TEX_FACE_POS_Y, PIPE_TEX_FACE_NEG_Y },

   /* pos Y first then neg Y is X different, X the same */
   /* PIPE_TEX_FACE_POS_Y */
   { PIPE_TEX_FACE_NEG_X, PIPE_TEX_FACE_POS_X,
     PIPE_TEX_FACE_NEG_Z, PIPE_TEX_FACE_POS_Z },

   /* PIPE_TEX_FACE_NEG_Y */
   { PIPE_TEX_FACE_NEG_X, PIPE_TEX_FACE_POS_X,
     PIPE_TEX_FACE_POS_Z, PIPE_TEX_FACE_NEG_Z },

   /* pos Z first then neg Y is X different, X the same */
   /* PIPE_TEX_FACE_POS_Z */
   { PIPE_TEX_FACE_NEG_X, PIPE_TEX_FACE_POS_X,
     PIPE_TEX_FACE_POS_Y, PIPE_TEX_FACE_NEG_Y },

   /* PIPE_TEX_FACE_NEG_Z */
   { PIPE_TEX_FACE_POS_X, PIPE_TEX_FACE_NEG_X,
     PIPE_TEX_FACE_POS_Y, PIPE_TEX_FACE_NEG_Y }
};

static INLINE unsigned
get_next_face(unsigned face, int idx)
{
   return face_array[face][idx];
}

/*
 * return a new xcoord based on old face, old coords, cube size
 * and fall_off_index (0 for x-, 1 for x+, 2 for y-, 3 for y+)
 */
static INLINE int
get_next_xcoord(unsigned face, unsigned fall_off_index, int max, int xc, int yc)
{
   if ((face == 0 && fall_off_index != 1) ||
       (face == 1 && fall_off_index == 0) ||
       (face == 4 && fall_off_index == 0) ||
       (face == 5 && fall_off_index == 0)) {
      return max;
   }
   if ((face == 1 && fall_off_index != 0) ||
       (face == 0 && fall_off_index == 1) ||
       (face == 4 && fall_off_index == 1) ||
       (face == 5 && fall_off_index == 1)) {
      return 0;
   }
   if ((face == 4 && fall_off_index >= 2) ||
       (face == 2 && fall_off_index == 3) ||
       (face == 3 && fall_off_index == 2)) {
      return xc;
   }
   if ((face == 5 && fall_off_index >= 2) ||
       (face == 2 && fall_off_index == 2) ||
       (face == 3 && fall_off_index == 3)) {
      return max - xc;
   }
   if ((face == 2 && fall_off_index == 0) ||
       (face == 3 && fall_off_index == 1)) {
      return yc;
   }
   /* (face == 2 && fall_off_index == 1) ||
      (face == 3 && fall_off_index == 0)) */
   return max - yc;
}

/*
 * return a new ycoord based on old face, old coords, cube size
 * and fall_off_index (0 for x-, 1 for x+, 2 for y-, 3 for y+)
 */
static INLINE int
get_next_ycoord(unsigned face, unsigned fall_off_index, int max, int xc, int yc)
{
   if ((fall_off_index <= 1) && (face <= 1 || face >= 4)) {
      return yc;
   }
   if (face == 2 ||
       (face == 4 && fall_off_index == 3) ||
       (face == 5 && fall_off_index == 2)) {
      return 0;
   }
   if (face == 3 ||
       (face == 4 && fall_off_index == 2) ||
       (face == 5 && fall_off_index == 3)) {
      return max;
   }
   if ((face == 0 && fall_off_index == 3) ||
       (face == 1 && fall_off_index == 2)) {
      return xc;
   }
   /* (face == 0 && fall_off_index == 2) ||
      (face == 1 && fall_off_index == 3) */
   return max - xc;
}


static INLINE const float *
get_texel_cube_seamless(const struct sp_sampler_view *sp_sview,
                        union tex_tile_address addr, int x, int y,
                        float *corner)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;
   unsigned face = addr.bits.face;
   int new_x, new_y, max_x;

   max_x = (int) u_minify(texture->width0, level);

   assert(texture->width0 == texture->height0);
   new_x = x;
   new_y = y;

   /* change the face */
   if (x < 0) {
      /*
       * Cheat with corners. They are difficult and I believe because we don't get
       * per-pixel faces we can actually have multiple corner texels per pixel,
       * which screws things up majorly in any case (as the per spec behavior is
       * to average the 3 remaining texels, which we might not have).
       * Hence just make sure that the 2nd coord is clamped, will simply pick the
       * sample which would have fallen off the x coord, but not y coord.
       * So the filter weight of the samples will be wrong, but at least this
       * ensures that only valid texels near the corner are used.
       */
      if (y < 0 || y >= max_x) {
         y = CLAMP(y, 0, max_x - 1);
      }
      new_x = get_next_xcoord(face, 0, max_x -1, x, y);
      new_y = get_next_ycoord(face, 0, max_x -1, x, y);
      face = get_next_face(face, 0);
   } else if (x >= max_x) {
      if (y < 0 || y >= max_x) {
         y = CLAMP(y, 0, max_x - 1);
      }
      new_x = get_next_xcoord(face, 1, max_x -1, x, y);
      new_y = get_next_ycoord(face, 1, max_x -1, x, y);
      face = get_next_face(face, 1);
   } else if (y < 0) {
      new_x = get_next_xcoord(face, 2, max_x -1, x, y);
      new_y = get_next_ycoord(face, 2, max_x -1, x, y);
      face = get_next_face(face, 2);
   } else if (y >= max_x) {
      new_x = get_next_xcoord(face, 3, max_x -1, x, y);
      new_y = get_next_ycoord(face, 3, max_x -1, x, y);
      face = get_next_face(face, 3);
   }

   addr.bits.face = face;
   return get_texel_2d_no_border( sp_sview, addr, new_x, new_y );
}

/* Gather a quad of adjacent texels within a tile:
 */
static INLINE void
get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_view *sp_sview,
                                        union tex_tile_address addr,
                                        unsigned x, unsigned y,
                                        const float *out[4])
{
    const struct softpipe_tex_cached_tile *tile;

   addr.bits.x = x / TEX_TILE_SIZE;
   addr.bits.y = y / TEX_TILE_SIZE;
   y %= TEX_TILE_SIZE;
   x %= TEX_TILE_SIZE;

   tile = sp_get_cached_tile_tex(sp_sview->cache, addr);
      
   out[0] = &tile->data.color[y  ][x  ][0];
   out[1] = &tile->data.color[y  ][x+1][0];
   out[2] = &tile->data.color[y+1][x  ][0];
   out[3] = &tile->data.color[y+1][x+1][0];
}


/* Gather a quad of potentially non-adjacent texels:
 */
static INLINE void
get_texel_quad_2d_no_border(const struct sp_sampler_view *sp_sview,
                            union tex_tile_address addr,
                            int x0, int y0,
                            int x1, int y1,
                            const float *out[4])
{
   out[0] = get_texel_2d_no_border( sp_sview, addr, x0, y0 );
   out[1] = get_texel_2d_no_border( sp_sview, addr, x1, y0 );
   out[2] = get_texel_2d_no_border( sp_sview, addr, x0, y1 );
   out[3] = get_texel_2d_no_border( sp_sview, addr, x1, y1 );
}

/* Can involve a lot of unnecessary checks for border color:
 */
static INLINE void
get_texel_quad_2d(const struct sp_sampler_view *sp_sview,
                  const struct sp_sampler *sp_samp,
                  union tex_tile_address addr,
                  int x0, int y0,
                  int x1, int y1,
                  const float *out[4])
{
   out[0] = get_texel_2d( sp_sview, sp_samp, addr, x0, y0 );
   out[1] = get_texel_2d( sp_sview, sp_samp, addr, x1, y0 );
   out[3] = get_texel_2d( sp_sview, sp_samp, addr, x1, y1 );
   out[2] = get_texel_2d( sp_sview, sp_samp, addr, x0, y1 );
}



/* 3d variants:
 */
static INLINE const float *
get_texel_3d_no_border(const struct sp_sampler_view *sp_sview,
                       union tex_tile_address addr, int x, int y, int z)
{
   const struct softpipe_tex_cached_tile *tile;

   addr.bits.x = x / TEX_TILE_SIZE;
   addr.bits.y = y / TEX_TILE_SIZE;
   addr.bits.z = z;
   y %= TEX_TILE_SIZE;
   x %= TEX_TILE_SIZE;

   tile = sp_get_cached_tile_tex(sp_sview->cache, addr);

   return &tile->data.color[y][x][0];
}


static INLINE const float *
get_texel_3d(const struct sp_sampler_view *sp_sview,
             const struct sp_sampler *sp_samp,
             union tex_tile_address addr, int x, int y, int z)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;

   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
       y < 0 || y >= (int) u_minify(texture->height0, level) ||
       z < 0 || z >= (int) u_minify(texture->depth0, level)) {
      return sp_samp->base.border_color.f;
   }
   else {
      return get_texel_3d_no_border( sp_sview, addr, x, y, z );
   }
}


/* Get texel pointer for 1D array texture */
static INLINE const float *
get_texel_1d_array(const struct sp_sampler_view *sp_sview,
                   const struct sp_sampler *sp_samp,
                   union tex_tile_address addr, int x, int y)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;

   if (x < 0 || x >= (int) u_minify(texture->width0, level)) {
      return sp_samp->base.border_color.f;
   }
   else {
      return get_texel_2d_no_border(sp_sview, addr, x, y);
   }
}


/* Get texel pointer for 2D array texture */
static INLINE const float *
get_texel_2d_array(const struct sp_sampler_view *sp_sview,
                   const struct sp_sampler *sp_samp,
                   union tex_tile_address addr, int x, int y, int layer)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;

   assert(layer < (int) texture->array_size);
   assert(layer >= 0);

   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
       y < 0 || y >= (int) u_minify(texture->height0, level)) {
      return sp_samp->base.border_color.f;
   }
   else {
      return get_texel_3d_no_border(sp_sview, addr, x, y, layer);
   }
}


/* Get texel pointer for cube array texture */
static INLINE const float *
get_texel_cube_array(const struct sp_sampler_view *sp_sview,
                     const struct sp_sampler *sp_samp,
                     union tex_tile_address addr, int x, int y, int layer)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   unsigned level = addr.bits.level;

   assert(layer < (int) texture->array_size);
   assert(layer >= 0);

   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
       y < 0 || y >= (int) u_minify(texture->height0, level)) {
      return sp_samp->base.border_color.f;
   }
   else {
      return get_texel_3d_no_border(sp_sview, addr, x, y, layer);
   }
}
/**
 * Given the logbase2 of a mipmap's base level size and a mipmap level,
 * return the size (in texels) of that mipmap level.
 * For example, if level[0].width = 256 then base_pot will be 8.
 * If level = 2, then we'll return 64 (the width at level=2).
 * Return 1 if level > base_pot.
 */
static INLINE unsigned
pot_level_size(unsigned base_pot, unsigned level)
{
   return (base_pot >= level) ? (1 << (base_pot - level)) : 1;
}


static void
print_sample(const char *function, const float *rgba)
{
   debug_printf("%s %g %g %g %g\n",
                function,
                rgba[0], rgba[TGSI_NUM_CHANNELS], rgba[2*TGSI_NUM_CHANNELS], rgba[3*TGSI_NUM_CHANNELS]);
}


static void
print_sample_4(const char *function, float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   debug_printf("%s %g %g %g %g, %g %g %g %g, %g %g %g %g, %g %g %g %g\n",
                function,
                rgba[0][0], rgba[1][0], rgba[2][0], rgba[3][0],
                rgba[0][1], rgba[1][1], rgba[2][1], rgba[3][1],
                rgba[0][2], rgba[1][2], rgba[2][2], rgba[3][2],
                rgba[0][3], rgba[1][3], rgba[2][3], rgba[3][3]);
}


/* Some image-filter fastpaths:
 */
static INLINE void
img_filter_2d_linear_repeat_POT(struct sp_sampler_view *sp_sview,
                                struct sp_sampler *sp_samp,
                                float s,
                                float t,
                                float p,
                                unsigned level,
                                unsigned face_id,
                                float *rgba)
{
   unsigned xpot = pot_level_size(sp_sview->xpot, level);
   unsigned ypot = pot_level_size(sp_sview->ypot, level);
   int xmax = (xpot - 1) & (TEX_TILE_SIZE - 1); /* MIN2(TEX_TILE_SIZE, xpot) - 1; */
   int ymax = (ypot - 1) & (TEX_TILE_SIZE - 1); /* MIN2(TEX_TILE_SIZE, ypot) - 1; */
   union tex_tile_address addr;
   int c;

   float u = s * xpot - 0.5F;
   float v = t * ypot - 0.5F;

   int uflr = util_ifloor(u);
   int vflr = util_ifloor(v);

   float xw = u - (float)uflr;
   float yw = v - (float)vflr;

   int x0 = uflr & (xpot - 1);
   int y0 = vflr & (ypot - 1);

   const float *tx[4];
      
   addr.value = 0;
   addr.bits.level = level;

   /* Can we fetch all four at once:
    */
   if (x0 < xmax && y0 < ymax) {
      get_texel_quad_2d_no_border_single_tile(sp_sview, addr, x0, y0, tx);
   }
   else {
      unsigned x1 = (x0 + 1) & (xpot - 1);
      unsigned y1 = (y0 + 1) & (ypot - 1);
      get_texel_quad_2d_no_border(sp_sview, addr, x0, y0, x1, y1, tx);
   }

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++) {
      rgba[TGSI_NUM_CHANNELS*c] = lerp_2d(xw, yw, 
                                       tx[0][c], tx[1][c], 
                                       tx[2][c], tx[3][c]);
   }

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static INLINE void
img_filter_2d_nearest_repeat_POT(struct sp_sampler_view *sp_sview,
                                 struct sp_sampler *sp_samp,
                                 float s,
                                 float t,
                                 float p,
                                 unsigned level,
                                 unsigned face_id,
                                 float rgba[TGSI_QUAD_SIZE])
{
   unsigned xpot = pot_level_size(sp_sview->xpot, level);
   unsigned ypot = pot_level_size(sp_sview->ypot, level);
   const float *out;
   union tex_tile_address addr;
   int c;

   float u = s * xpot;
   float v = t * ypot;

   int uflr = util_ifloor(u);
   int vflr = util_ifloor(v);

   int x0 = uflr & (xpot - 1);
   int y0 = vflr & (ypot - 1);

   addr.value = 0;
   addr.bits.level = level;

   out = get_texel_2d_no_border(sp_sview, addr, x0, y0);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static INLINE void
img_filter_2d_nearest_clamp_POT(struct sp_sampler_view *sp_sview,
                                struct sp_sampler *sp_samp,
                                float s,
                                float t,
                                float p,
                                unsigned level,
                                unsigned face_id,
                                float rgba[TGSI_QUAD_SIZE])
{
   unsigned xpot = pot_level_size(sp_sview->xpot, level);
   unsigned ypot = pot_level_size(sp_sview->ypot, level);
   union tex_tile_address addr;
   int c;

   float u = s * xpot;
   float v = t * ypot;

   int x0, y0;
   const float *out;

   addr.value = 0;
   addr.bits.level = level;

   x0 = util_ifloor(u);
   if (x0 < 0) 
      x0 = 0;
   else if (x0 > (int) xpot - 1)
      x0 = xpot - 1;

   y0 = util_ifloor(v);
   if (y0 < 0) 
      y0 = 0;
   else if (y0 > (int) ypot - 1)
      y0 = ypot - 1;
   
   out = get_texel_2d_no_border(sp_sview, addr, x0, y0);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static void
img_filter_1d_nearest(struct sp_sampler_view *sp_sview,
                      struct sp_sampler *sp_samp,
                      float s,
                      float t,
                      float p,
                      unsigned level,
                      unsigned face_id,
                      float rgba[TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width;
   int x;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);

   assert(width > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->nearest_texcoord_s(s, width, &x);

   out = get_texel_2d(sp_sview, sp_samp, addr, x, 0);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static void
img_filter_1d_array_nearest(struct sp_sampler_view *sp_sview,
                            struct sp_sampler *sp_samp,
                            float s,
                            float t,
                            float p,
                            unsigned level,
                            unsigned face_id,
                            float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width;
   int x, layer;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);

   assert(width > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->nearest_texcoord_s(s, width, &x);
   wrap_array_layer(t, texture->array_size, &layer);

   out = get_texel_1d_array(sp_sview, sp_samp, addr, x, layer);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static void
img_filter_2d_nearest(struct sp_sampler_view *sp_sview,
                      struct sp_sampler *sp_samp,
                      float s,
                      float t,
                      float p,
                      unsigned level,
                      unsigned face_id,
                      float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x, y;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);
 
   addr.value = 0;
   addr.bits.level = level;

   sp_samp->nearest_texcoord_s(s, width, &x);
   sp_samp->nearest_texcoord_t(t, height, &y);

   out = get_texel_2d(sp_sview, sp_samp, addr, x, y);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static void
img_filter_2d_array_nearest(struct sp_sampler_view *sp_sview,
                            struct sp_sampler *sp_samp,
                            float s,
                            float t,
                            float p,
                            unsigned level,
                            unsigned face_id,
                            float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x, y, layer;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);
 
   addr.value = 0;
   addr.bits.level = level;

   sp_samp->nearest_texcoord_s(s, width, &x);
   sp_samp->nearest_texcoord_t(t, height, &y);
   wrap_array_layer(p, texture->array_size, &layer);

   out = get_texel_2d_array(sp_sview, sp_samp, addr, x, y, layer);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}


static INLINE union tex_tile_address
face(union tex_tile_address addr, unsigned face )
{
   addr.bits.face = face;
   return addr;
}


static void
img_filter_cube_nearest(struct sp_sampler_view *sp_sview,
                        struct sp_sampler *sp_samp,
                        float s,
                        float t,
                        float p,
                        unsigned level,
                        unsigned face_id,
                        float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x, y;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);
 
   addr.value = 0;
   addr.bits.level = level;

   /*
    * If NEAREST filtering is done within a miplevel, always apply wrap
    * mode CLAMP_TO_EDGE.
    */
   if (sp_samp->base.seamless_cube_map) {
      wrap_nearest_clamp_to_edge(s, width, &x);
      wrap_nearest_clamp_to_edge(t, height, &y);
   } else {
      /* Would probably make sense to ignore mode and just do edge clamp */
      sp_samp->nearest_texcoord_s(s, width, &x);
      sp_samp->nearest_texcoord_t(t, height, &y);
   }

   out = get_texel_2d(sp_sview, sp_samp, face(addr, face_id), x, y);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}

static void
img_filter_cube_array_nearest(struct sp_sampler_view *sp_sview,
                              struct sp_sampler *sp_samp,
                              float s,
                              float t,
                              float p,
                              unsigned level,
                              unsigned face_id,
                              float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x, y, layer;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);
 
   addr.value = 0;
   addr.bits.level = level;

   sp_samp->nearest_texcoord_s(s, width, &x);
   sp_samp->nearest_texcoord_t(t, height, &y);
   wrap_array_layer(p, texture->array_size, &layer);

   out = get_texel_cube_array(sp_sview, sp_samp, addr, x, y, layer * 6 + face_id);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];

   if (DEBUG_TEX) {
      print_sample(__FUNCTION__, rgba);
   }
}

static void
img_filter_3d_nearest(struct sp_sampler_view *sp_sview,
                      struct sp_sampler *sp_samp,
                      float s,
                      float t,
                      float p,
                      unsigned level,
                      unsigned face_id,
                      float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height, depth;
   int x, y, z;
   union tex_tile_address addr;
   const float *out;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);
   depth = u_minify(texture->depth0, level);

   assert(width > 0);
   assert(height > 0);
   assert(depth > 0);

   sp_samp->nearest_texcoord_s(s, width,  &x);
   sp_samp->nearest_texcoord_t(t, height, &y);
   sp_samp->nearest_texcoord_p(p, depth,  &z);

   addr.value = 0;
   addr.bits.level = level;

   out = get_texel_3d(sp_sview, sp_samp, addr, x, y, z);
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = out[c];
}


static void
img_filter_1d_linear(struct sp_sampler_view *sp_sview,
                     struct sp_sampler *sp_samp,
                     float s,
                     float t,
                     float p,
                     unsigned level,
                     unsigned face_id,
                     float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width;
   int x0, x1;
   float xw; /* weights */
   union tex_tile_address addr;
   const float *tx0, *tx1;
   int c;

   width = u_minify(texture->width0, level);

   assert(width > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->linear_texcoord_s(s, width, &x0, &x1, &xw);

   tx0 = get_texel_2d(sp_sview, sp_samp, addr, x0, 0);
   tx1 = get_texel_2d(sp_sview, sp_samp, addr, x1, 0);

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp(xw, tx0[c], tx1[c]);
}


static void
img_filter_1d_array_linear(struct sp_sampler_view *sp_sview,
                           struct sp_sampler *sp_samp,
                           float s,
                           float t,
                           float p,
                           unsigned level,
                           unsigned face_id,
                           float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width;
   int x0, x1, layer;
   float xw; /* weights */
   union tex_tile_address addr;
   const float *tx0, *tx1;
   int c;

   width = u_minify(texture->width0, level);

   assert(width > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->linear_texcoord_s(s, width, &x0, &x1, &xw);
   wrap_array_layer(t, texture->array_size, &layer);

   tx0 = get_texel_1d_array(sp_sview, sp_samp, addr, x0, layer);
   tx1 = get_texel_1d_array(sp_sview, sp_samp, addr, x1, layer);

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp(xw, tx0[c], tx1[c]);
}


static void
img_filter_2d_linear(struct sp_sampler_view *sp_sview,
                     struct sp_sampler *sp_samp,
                     float s,
                     float t,
                     float p,
                     unsigned level,
                     unsigned face_id,
                     float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x0, y0, x1, y1;
   float xw, yw; /* weights */
   union tex_tile_address addr;
   const float *tx0, *tx1, *tx2, *tx3;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->linear_texcoord_s(s, width,  &x0, &x1, &xw);
   sp_samp->linear_texcoord_t(t, height, &y0, &y1, &yw);

   tx0 = get_texel_2d(sp_sview, sp_samp, addr, x0, y0);
   tx1 = get_texel_2d(sp_sview, sp_samp, addr, x1, y0);
   tx2 = get_texel_2d(sp_sview, sp_samp, addr, x0, y1);
   tx3 = get_texel_2d(sp_sview, sp_samp, addr, x1, y1);

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp_2d(xw, yw,
                                          tx0[c], tx1[c],
                                          tx2[c], tx3[c]);
}


static void
img_filter_2d_array_linear(struct sp_sampler_view *sp_sview,
                           struct sp_sampler *sp_samp,
                           float s,
                           float t,
                           float p,
                           unsigned level,
                           unsigned face_id,
                           float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x0, y0, x1, y1, layer;
   float xw, yw; /* weights */
   union tex_tile_address addr;
   const float *tx0, *tx1, *tx2, *tx3;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->linear_texcoord_s(s, width,  &x0, &x1, &xw);
   sp_samp->linear_texcoord_t(t, height, &y0, &y1, &yw);
   wrap_array_layer(p, texture->array_size, &layer);

   tx0 = get_texel_2d_array(sp_sview, sp_samp, addr, x0, y0, layer);
   tx1 = get_texel_2d_array(sp_sview, sp_samp, addr, x1, y0, layer);
   tx2 = get_texel_2d_array(sp_sview, sp_samp, addr, x0, y1, layer);
   tx3 = get_texel_2d_array(sp_sview, sp_samp, addr, x1, y1, layer);

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp_2d(xw, yw,
                                          tx0[c], tx1[c],
                                          tx2[c], tx3[c]);
}


static void
img_filter_cube_linear(struct sp_sampler_view *sp_sview,
                       struct sp_sampler *sp_samp,
                       float s,
                       float t,
                       float p,
                       unsigned level,
                       unsigned face_id,
                       float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x0, y0, x1, y1;
   float xw, yw; /* weights */
   union tex_tile_address addr, addrj;
   const float *tx0, *tx1, *tx2, *tx3;
   float corner0[TGSI_QUAD_SIZE], corner1[TGSI_QUAD_SIZE],
         corner2[TGSI_QUAD_SIZE], corner3[TGSI_QUAD_SIZE];
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);

   addr.value = 0;
   addr.bits.level = level;

   /*
    * For seamless if LINEAR filtering is done within a miplevel,
    * always apply wrap mode CLAMP_TO_BORDER.
    */
   if (sp_samp->base.seamless_cube_map) {
      /* Note this is a bit overkill, actual clamping is not required */
      wrap_linear_clamp_to_border(s, width, &x0, &x1, &xw);
      wrap_linear_clamp_to_border(t, height, &y0, &y1, &yw);
   } else {
      /* Would probably make sense to ignore mode and just do edge clamp */
      sp_samp->linear_texcoord_s(s, width,  &x0, &x1, &xw);
      sp_samp->linear_texcoord_t(t, height, &y0, &y1, &yw);
   }

   addrj = face(addr, face_id);

   if (sp_samp->base.seamless_cube_map) {
      tx0 = get_texel_cube_seamless(sp_sview, addrj, x0, y0, corner0);
      tx1 = get_texel_cube_seamless(sp_sview, addrj, x1, y0, corner1);
      tx2 = get_texel_cube_seamless(sp_sview, addrj, x0, y1, corner2);
      tx3 = get_texel_cube_seamless(sp_sview, addrj, x1, y1, corner3);
   } else {
      tx0 = get_texel_2d(sp_sview, sp_samp, addrj, x0, y0);
      tx1 = get_texel_2d(sp_sview, sp_samp, addrj, x1, y0);
      tx2 = get_texel_2d(sp_sview, sp_samp, addrj, x0, y1);
      tx3 = get_texel_2d(sp_sview, sp_samp, addrj, x1, y1);
   }
   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp_2d(xw, yw,
                                          tx0[c], tx1[c],
                                          tx2[c], tx3[c]);
}


static void
img_filter_cube_array_linear(struct sp_sampler_view *sp_sview,
                             struct sp_sampler *sp_samp,
                             float s,
                             float t,
                             float p,
                             unsigned level,
                             unsigned face_id,
                             float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height;
   int x0, y0, x1, y1, layer;
   float xw, yw; /* weights */
   union tex_tile_address addr;
   const float *tx0, *tx1, *tx2, *tx3;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);

   assert(width > 0);
   assert(height > 0);

   addr.value = 0;
   addr.bits.level = level;

   sp_samp->linear_texcoord_s(s, width,  &x0, &x1, &xw);
   sp_samp->linear_texcoord_t(t, height, &y0, &y1, &yw);
   wrap_array_layer(p, texture->array_size, &layer);

   tx0 = get_texel_cube_array(sp_sview, sp_samp, addr, x0, y0, layer * 6 + face_id);
   tx1 = get_texel_cube_array(sp_sview, sp_samp, addr, x1, y0, layer * 6 + face_id);
   tx2 = get_texel_cube_array(sp_sview, sp_samp, addr, x0, y1, layer * 6 + face_id);
   tx3 = get_texel_cube_array(sp_sview, sp_samp, addr, x1, y1, layer * 6 + face_id);

   /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] = lerp_2d(xw, yw,
                                          tx0[c], tx1[c],
                                          tx2[c], tx3[c]);
}

static void
img_filter_3d_linear(struct sp_sampler_view *sp_sview,
                     struct sp_sampler *sp_samp,
                     float s,
                     float t,
                     float p,
                     unsigned level,
                     unsigned face_id,
                     float *rgba)
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int width, height, depth;
   int x0, x1, y0, y1, z0, z1;
   float xw, yw, zw; /* interpolation weights */
   union tex_tile_address addr;
   const float *tx00, *tx01, *tx02, *tx03, *tx10, *tx11, *tx12, *tx13;
   int c;

   width = u_minify(texture->width0, level);
   height = u_minify(texture->height0, level);
   depth = u_minify(texture->depth0, level);

   addr.value = 0;
   addr.bits.level = level;

   assert(width > 0);
   assert(height > 0);
   assert(depth > 0);

   sp_samp->linear_texcoord_s(s, width,  &x0, &x1, &xw);
   sp_samp->linear_texcoord_t(t, height, &y0, &y1, &yw);
   sp_samp->linear_texcoord_p(p, depth,  &z0, &z1, &zw);


   tx00 = get_texel_3d(sp_sview, sp_samp, addr, x0, y0, z0);
   tx01 = get_texel_3d(sp_sview, sp_samp, addr, x1, y0, z0);
   tx02 = get_texel_3d(sp_sview, sp_samp, addr, x0, y1, z0);
   tx03 = get_texel_3d(sp_sview, sp_samp, addr, x1, y1, z0);
      
   tx10 = get_texel_3d(sp_sview, sp_samp, addr, x0, y0, z1);
   tx11 = get_texel_3d(sp_sview, sp_samp, addr, x1, y0, z1);
   tx12 = get_texel_3d(sp_sview, sp_samp, addr, x0, y1, z1);
   tx13 = get_texel_3d(sp_sview, sp_samp, addr, x1, y1, z1);
      
      /* interpolate R, G, B, A */
   for (c = 0; c < TGSI_QUAD_SIZE; c++)
      rgba[TGSI_NUM_CHANNELS*c] =  lerp_3d(xw, yw, zw,
                                           tx00[c], tx01[c],
                                           tx02[c], tx03[c],
                                           tx10[c], tx11[c],
                                           tx12[c], tx13[c]);
}


/* Calculate level of detail for every fragment,
 * with lambda already computed.
 * Note that lambda has already been biased by global LOD bias.
 * \param biased_lambda per-quad lambda.
 * \param lod_in per-fragment lod_bias or explicit_lod.
 * \param lod returns the per-fragment lod.
 */
static INLINE void
compute_lod(const struct pipe_sampler_state *sampler,
            enum tgsi_sampler_control control,
            const float biased_lambda,
            const float lod_in[TGSI_QUAD_SIZE],
            float lod[TGSI_QUAD_SIZE])
{
   float min_lod = sampler->min_lod;
   float max_lod = sampler->max_lod;
   uint i;

   switch (control) {
   case tgsi_sampler_lod_none:
   case tgsi_sampler_lod_zero:
   /* XXX FIXME */
   case tgsi_sampler_derivs_explicit:
      lod[0] = lod[1] = lod[2] = lod[3] = CLAMP(biased_lambda, min_lod, max_lod);
      break;
   case tgsi_sampler_lod_bias:
      for (i = 0; i < TGSI_QUAD_SIZE; i++) {
         lod[i] = biased_lambda + lod_in[i];
         lod[i] = CLAMP(lod[i], min_lod, max_lod);
      }
      break;
   case tgsi_sampler_lod_explicit:
      for (i = 0; i < TGSI_QUAD_SIZE; i++) {
         lod[i] = CLAMP(lod_in[i], min_lod, max_lod);
      }
      break;
   default:
      assert(0);
      lod[0] = lod[1] = lod[2] = lod[3] = 0.0f;
   }
}


/* Calculate level of detail for every fragment.
 * \param lod_in per-fragment lod_bias or explicit_lod.
 * \param lod results per-fragment lod.
 */
static INLINE void
compute_lambda_lod(struct sp_sampler_view *sp_sview,
                   struct sp_sampler *sp_samp,
                   const float s[TGSI_QUAD_SIZE],
                   const float t[TGSI_QUAD_SIZE],
                   const float p[TGSI_QUAD_SIZE],
                   const float lod_in[TGSI_QUAD_SIZE],
                   enum tgsi_sampler_control control,
                   float lod[TGSI_QUAD_SIZE])
{
   const struct pipe_sampler_state *sampler = &sp_samp->base;
   float lod_bias = sampler->lod_bias;
   float min_lod = sampler->min_lod;
   float max_lod = sampler->max_lod;
   float lambda;
   uint i;

   switch (control) {
   case tgsi_sampler_lod_none:
      /* XXX FIXME */
   case tgsi_sampler_derivs_explicit:
      lambda = sp_sview->compute_lambda(sp_sview, s, t, p) + lod_bias;
      lod[0] = lod[1] = lod[2] = lod[3] = CLAMP(lambda, min_lod, max_lod);
      break;
   case tgsi_sampler_lod_bias:
      lambda = sp_sview->compute_lambda(sp_sview, s, t, p) + lod_bias;
      for (i = 0; i < TGSI_QUAD_SIZE; i++) {
         lod[i] = lambda + lod_in[i];
         lod[i] = CLAMP(lod[i], min_lod, max_lod);
      }
      break;
   case tgsi_sampler_lod_explicit:
      for (i = 0; i < TGSI_QUAD_SIZE; i++) {
         lod[i] = CLAMP(lod_in[i], min_lod, max_lod);
      }
      break;
   case tgsi_sampler_lod_zero:
      /* this is all static state in the sampler really need clamp here? */
      lod[0] = lod[1] = lod[2] = lod[3] = CLAMP(lod_bias, min_lod, max_lod);
      break;
   default:
      assert(0);
      lod[0] = lod[1] = lod[2] = lod[3] = 0.0f;
   }
}


static void
mip_filter_linear(struct sp_sampler_view *sp_sview,
                  struct sp_sampler *sp_samp,
                  img_filter_func min_filter,
                  img_filter_func mag_filter,
                  const float s[TGSI_QUAD_SIZE],
                  const float t[TGSI_QUAD_SIZE],
                  const float p[TGSI_QUAD_SIZE],
                  const float c0[TGSI_QUAD_SIZE],
                  const float lod_in[TGSI_QUAD_SIZE],
                  enum tgsi_sampler_control control,
                  float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int j;
   float lod[TGSI_QUAD_SIZE];

   compute_lambda_lod(sp_sview, sp_samp, s, t, p, lod_in, control, lod);

   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      int level0 = sp_sview->base.u.tex.first_level + (int)lod[j];

      if (lod[j] < 0.0)
         mag_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                    sp_sview->base.u.tex.first_level,
                    sp_sview->faces[j], &rgba[0][j]);

      else if (level0 >= (int) texture->last_level)
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j], texture->last_level,
                    sp_sview->faces[j], &rgba[0][j]);

      else {
         float levelBlend = frac(lod[j]);
         float rgbax[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
         int c;

         min_filter(sp_sview, sp_samp, s[j], t[j], p[j], level0,
                    sp_sview->faces[j], &rgbax[0][0]);
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j], level0+1,
                    sp_sview->faces[j], &rgbax[0][1]);

         for (c = 0; c < 4; c++) {
            rgba[c][j] = lerp(levelBlend, rgbax[c][0], rgbax[c][1]);
         }
      }
   }

   if (DEBUG_TEX) {
      print_sample_4(__FUNCTION__, rgba);
   }
}


/**
 * Compute nearest mipmap level from texcoords.
 * Then sample the texture level for four elements of a quad.
 * \param c0  the LOD bias factors, or absolute LODs (depending on control)
 */
static void
mip_filter_nearest(struct sp_sampler_view *sp_sview,
                   struct sp_sampler *sp_samp,
                   img_filter_func min_filter,
                   img_filter_func mag_filter,
                   const float s[TGSI_QUAD_SIZE],
                   const float t[TGSI_QUAD_SIZE],
                   const float p[TGSI_QUAD_SIZE],
                   const float c0[TGSI_QUAD_SIZE],
                   const float lod_in[TGSI_QUAD_SIZE],
                   enum tgsi_sampler_control control,
                   float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   float lod[TGSI_QUAD_SIZE];
   int j;

   compute_lambda_lod(sp_sview, sp_samp, s, t, p, lod_in, control, lod);

   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      if (lod[j] < 0.0)
         mag_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                    sp_sview->base.u.tex.first_level,
                    sp_sview->faces[j], &rgba[0][j]);
      else {
         int level = sp_sview->base.u.tex.first_level + (int)(lod[j] + 0.5F);
         level = MIN2(level, (int)texture->last_level);
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                    level, sp_sview->faces[j], &rgba[0][j]);
      }
   }

   if (DEBUG_TEX) {
      print_sample_4(__FUNCTION__, rgba);
   }
}


static void
mip_filter_none(struct sp_sampler_view *sp_sview,
                struct sp_sampler *sp_samp,
                img_filter_func min_filter,
                img_filter_func mag_filter,
                const float s[TGSI_QUAD_SIZE],
                const float t[TGSI_QUAD_SIZE],
                const float p[TGSI_QUAD_SIZE],
                const float c0[TGSI_QUAD_SIZE],
                const float lod_in[TGSI_QUAD_SIZE],
                enum tgsi_sampler_control control,
                float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   float lod[TGSI_QUAD_SIZE];
   int j;

   compute_lambda_lod(sp_sview, sp_samp, s, t, p, lod_in, control, lod);

   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      if (lod[j] < 0.0) { 
         mag_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                    sp_sview->base.u.tex.first_level,
                    sp_sview->faces[j], &rgba[0][j]);
      }
      else {
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                    sp_sview->base.u.tex.first_level,
                    sp_sview->faces[j], &rgba[0][j]);
      }
   }
}


static void
mip_filter_none_no_filter_select(struct sp_sampler_view *sp_sview,
                                 struct sp_sampler *sp_samp,
                                 img_filter_func min_filter,
                                 img_filter_func mag_filter,
                                 const float s[TGSI_QUAD_SIZE],
                                 const float t[TGSI_QUAD_SIZE],
                                 const float p[TGSI_QUAD_SIZE],
                                 const float c0[TGSI_QUAD_SIZE],
                                 const float lod_in[TGSI_QUAD_SIZE],
                                 enum tgsi_sampler_control control,
                                 float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   int j;

   for (j = 0; j < TGSI_QUAD_SIZE; j++)
      mag_filter(sp_sview, sp_samp, s[j], t[j], p[j],
                 sp_sview->base.u.tex.first_level,
                 sp_sview->faces[j], &rgba[0][j]);
}


/* For anisotropic filtering */
#define WEIGHT_LUT_SIZE 1024

static float *weightLut = NULL;

/**
 * Creates the look-up table used to speed-up EWA sampling
 */
static void
create_filter_table(void)
{
   unsigned i;
   if (!weightLut) {
      weightLut = (float *) MALLOC(WEIGHT_LUT_SIZE * sizeof(float));

      for (i = 0; i < WEIGHT_LUT_SIZE; ++i) {
         float alpha = 2;
         float r2 = (float) i / (float) (WEIGHT_LUT_SIZE - 1);
         float weight = (float) exp(-alpha * r2);
         weightLut[i] = weight;
      }
   }
}


/**
 * Elliptical weighted average (EWA) filter for producing high quality
 * anisotropic filtered results.
 * Based on the Higher Quality Elliptical Weighted Average Filter
 * published by Paul S. Heckbert in his Master's Thesis
 * "Fundamentals of Texture Mapping and Image Warping" (1989)
 */
static void
img_filter_2d_ewa(struct sp_sampler_view *sp_sview,
                  struct sp_sampler *sp_samp,
                  img_filter_func min_filter,
                  img_filter_func mag_filter,
                  const float s[TGSI_QUAD_SIZE],
                  const float t[TGSI_QUAD_SIZE],
                  const float p[TGSI_QUAD_SIZE],
                  unsigned level,
                  const float dudx, const float dvdx,
                  const float dudy, const float dvdy,
                  float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;

   // ??? Won't the image filters blow up if level is negative?
   unsigned level0 = level > 0 ? level : 0;
   float scaling = 1.0f / (1 << level0);
   int width = u_minify(texture->width0, level0);
   int height = u_minify(texture->height0, level0);

   float ux = dudx * scaling;
   float vx = dvdx * scaling;
   float uy = dudy * scaling;
   float vy = dvdy * scaling;

   /* compute ellipse coefficients to bound the region: 
    * A*x*x + B*x*y + C*y*y = F.
    */
   float A = vx*vx+vy*vy+1;
   float B = -2*(ux*vx+uy*vy);
   float C = ux*ux+uy*uy+1;
   float F = A*C-B*B/4.0f;

   /* check if it is an ellipse */
   /* ASSERT(F > 0.0); */

   /* Compute the ellipse's (u,v) bounding box in texture space */
   float d = -B*B+4.0f*C*A;
   float box_u = 2.0f / d * sqrtf(d*C*F); /* box_u -> half of bbox with   */
   float box_v = 2.0f / d * sqrtf(A*d*F); /* box_v -> half of bbox height */

   float rgba_temp[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
   float s_buffer[TGSI_QUAD_SIZE];
   float t_buffer[TGSI_QUAD_SIZE];
   float weight_buffer[TGSI_QUAD_SIZE];
   unsigned buffer_next;
   int j;
   float den; /* = 0.0F; */
   float ddq;
   float U; /* = u0 - tex_u; */
   int v;

   /* Scale ellipse formula to directly index the Filter Lookup Table.
    * i.e. scale so that F = WEIGHT_LUT_SIZE-1
    */
   double formScale = (double) (WEIGHT_LUT_SIZE - 1) / F;
   A *= formScale;
   B *= formScale;
   C *= formScale;
   /* F *= formScale; */ /* no need to scale F as we don't use it below here */

   /* For each quad, the du and dx values are the same and so the ellipse is
    * also the same. Note that texel/image access can only be performed using
    * a quad, i.e. it is not possible to get the pixel value for a single
    * tex coord. In order to have a better performance, the access is buffered
    * using the s_buffer/t_buffer and weight_buffer. Only when the buffer is
    * full, then the pixel values are read from the image.
    */
   ddq = 2 * A;
   
   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      /* Heckbert MS thesis, p. 59; scan over the bounding box of the ellipse
       * and incrementally update the value of Ax^2+Bxy*Cy^2; when this
       * value, q, is less than F, we're inside the ellipse
       */
      float tex_u = -0.5F + s[j] * texture->width0 * scaling;
      float tex_v = -0.5F + t[j] * texture->height0 * scaling;

      int u0 = (int) floorf(tex_u - box_u);
      int u1 = (int) ceilf(tex_u + box_u);
      int v0 = (int) floorf(tex_v - box_v);
      int v1 = (int) ceilf(tex_v + box_v);

      float num[4] = {0.0F, 0.0F, 0.0F, 0.0F};
      buffer_next = 0;
      den = 0;
      U = u0 - tex_u;
      for (v = v0; v <= v1; ++v) {
         float V = v - tex_v;
         float dq = A * (2 * U + 1) + B * V;
         float q = (C * V + B * U) * V + A * U * U;

         int u;
         for (u = u0; u <= u1; ++u) {
            /* Note that the ellipse has been pre-scaled so F =
             * WEIGHT_LUT_SIZE - 1
             */
            if (q < WEIGHT_LUT_SIZE) {
               /* as a LUT is used, q must never be negative;
                * should not happen, though
                */
               const int qClamped = q >= 0.0F ? q : 0;
               float weight = weightLut[qClamped];

               weight_buffer[buffer_next] = weight;
               s_buffer[buffer_next] = u / ((float) width);
               t_buffer[buffer_next] = v / ((float) height);
            
               buffer_next++;
               if (buffer_next == TGSI_QUAD_SIZE) {
                  /* 4 texel coords are in the buffer -> read it now */
                  unsigned jj;
                  /* it is assumed that samp->min_img_filter is set to
                   * img_filter_2d_nearest or one of the
                   * accelerated img_filter_2d_nearest_XXX functions.
                   */
                  for (jj = 0; jj < buffer_next; jj++) {
                     min_filter(sp_sview, sp_samp, s_buffer[jj], t_buffer[jj], p[jj],
                                level, sp_sview->faces[j], &rgba_temp[0][jj]);
                     num[0] += weight_buffer[jj] * rgba_temp[0][jj];
                     num[1] += weight_buffer[jj] * rgba_temp[1][jj];
                     num[2] += weight_buffer[jj] * rgba_temp[2][jj];
                     num[3] += weight_buffer[jj] * rgba_temp[3][jj];
                  }

                  buffer_next = 0;
               }

               den += weight;
            }
            q += dq;
            dq += ddq;
         }
      }

      /* if the tex coord buffer contains unread values, we will read
       * them now.
       */
      if (buffer_next > 0) {
         unsigned jj;
         /* it is assumed that samp->min_img_filter is set to
          * img_filter_2d_nearest or one of the
          * accelerated img_filter_2d_nearest_XXX functions.
          */
         for (jj = 0; jj < buffer_next; jj++) {
            min_filter(sp_sview, sp_samp, s_buffer[jj], t_buffer[jj], p[jj],
                       level, sp_sview->faces[j], &rgba_temp[0][jj]);
            num[0] += weight_buffer[jj] * rgba_temp[0][jj];
            num[1] += weight_buffer[jj] * rgba_temp[1][jj];
            num[2] += weight_buffer[jj] * rgba_temp[2][jj];
            num[3] += weight_buffer[jj] * rgba_temp[3][jj];
         }
      }

      if (den <= 0.0F) {
         /* Reaching this place would mean that no pixels intersected
          * the ellipse.  This should never happen because the filter
          * we use always intersects at least one pixel.
          */

         /*rgba[0]=0;
         rgba[1]=0;
         rgba[2]=0;
         rgba[3]=0;*/
         /* not enough pixels in resampling, resort to direct interpolation */
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j], level,
                    sp_sview->faces[j], &rgba_temp[0][j]);
         den = 1;
         num[0] = rgba_temp[0][j];
         num[1] = rgba_temp[1][j];
         num[2] = rgba_temp[2][j];
         num[3] = rgba_temp[3][j];
      }

      rgba[0][j] = num[0] / den;
      rgba[1][j] = num[1] / den;
      rgba[2][j] = num[2] / den;
      rgba[3][j] = num[3] / den;
   }
}


/**
 * Sample 2D texture using an anisotropic filter.
 */
static void
mip_filter_linear_aniso(struct sp_sampler_view *sp_sview,
                        struct sp_sampler *sp_samp,
                        img_filter_func min_filter,
                        img_filter_func mag_filter,
                        const float s[TGSI_QUAD_SIZE],
                        const float t[TGSI_QUAD_SIZE],
                        const float p[TGSI_QUAD_SIZE],
                        const float c0[TGSI_QUAD_SIZE],
                        const float lod_in[TGSI_QUAD_SIZE],
                        enum tgsi_sampler_control control,
                        float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int level0;
   float lambda;
   float lod[TGSI_QUAD_SIZE];

   float s_to_u = u_minify(texture->width0, sp_sview->base.u.tex.first_level);
   float t_to_v = u_minify(texture->height0, sp_sview->base.u.tex.first_level);
   float dudx = (s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]) * s_to_u;
   float dudy = (s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]) * s_to_u;
   float dvdx = (t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]) * t_to_v;
   float dvdy = (t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]) * t_to_v;
   
   if (control == tgsi_sampler_lod_bias ||
       control == tgsi_sampler_lod_none ||
       /* XXX FIXME */
       control == tgsi_sampler_derivs_explicit) {
      /* note: instead of working with Px and Py, we will use the 
       * squared length instead, to avoid sqrt.
       */
      float Px2 = dudx * dudx + dvdx * dvdx;
      float Py2 = dudy * dudy + dvdy * dvdy;

      float Pmax2;
      float Pmin2;
      float e;
      const float maxEccentricity = sp_samp->base.max_anisotropy * sp_samp->base.max_anisotropy;
      
      if (Px2 < Py2) {
         Pmax2 = Py2;
         Pmin2 = Px2;
      }
      else {
         Pmax2 = Px2;
         Pmin2 = Py2;
      }
      
      /* if the eccentricity of the ellipse is too big, scale up the shorter
       * of the two vectors to limit the maximum amount of work per pixel
       */
      e = Pmax2 / Pmin2;
      if (e > maxEccentricity) {
         /* float s=e / maxEccentricity;
            minor[0] *= s;
            minor[1] *= s;
            Pmin2 *= s; */
         Pmin2 = Pmax2 / maxEccentricity;
      }
      
      /* note: we need to have Pmin=sqrt(Pmin2) here, but we can avoid
       * this since 0.5*log(x) = log(sqrt(x))
       */
      lambda = 0.5F * util_fast_log2(Pmin2) + sp_samp->base.lod_bias;
      compute_lod(&sp_samp->base, control, lambda, lod_in, lod);
   }
   else {
      assert(control == tgsi_sampler_lod_explicit ||
             control == tgsi_sampler_lod_zero);
      compute_lod(&sp_samp->base, control, sp_samp->base.lod_bias, lod_in, lod);
   }
   
   /* XXX: Take into account all lod values.
    */
   lambda = lod[0];
   level0 = sp_sview->base.u.tex.first_level + (int)lambda;

   /* If the ellipse covers the whole image, we can
    * simply return the average of the whole image.
    */
   if (level0 >= (int) texture->last_level) {
      int j;
      for (j = 0; j < TGSI_QUAD_SIZE; j++)
         min_filter(sp_sview, sp_samp, s[j], t[j], p[j], texture->last_level,
                    sp_sview->faces[j], &rgba[0][j]);
   }
   else {
      /* don't bother interpolating between multiple LODs; it doesn't
       * seem to be worth the extra running time.
       */
      img_filter_2d_ewa(sp_sview, sp_samp, min_filter, mag_filter,
                        s, t, p, level0,
                        dudx, dvdx, dudy, dvdy, rgba);
   }

   if (DEBUG_TEX) {
      print_sample_4(__FUNCTION__, rgba);
   }
}


/**
 * Specialized version of mip_filter_linear with hard-wired calls to
 * 2d lambda calculation and 2d_linear_repeat_POT img filters.
 */
static void
mip_filter_linear_2d_linear_repeat_POT(
   struct sp_sampler_view *sp_sview,
   struct sp_sampler *sp_samp,
   img_filter_func min_filter,
   img_filter_func mag_filter,
   const float s[TGSI_QUAD_SIZE],
   const float t[TGSI_QUAD_SIZE],
   const float p[TGSI_QUAD_SIZE],
   const float c0[TGSI_QUAD_SIZE],
   const float lod_in[TGSI_QUAD_SIZE],
   enum tgsi_sampler_control control,
   float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_resource *texture = sp_sview->base.texture;
   int j;
   float lod[TGSI_QUAD_SIZE];

   compute_lambda_lod(sp_sview, sp_samp, s, t, p, lod_in, control, lod);

   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      int level0 = sp_sview->base.u.tex.first_level + (int)lod[j];

      /* Catches both negative and large values of level0:
       */
      if ((unsigned)level0 >= texture->last_level) { 
         if (level0 < 0)
            img_filter_2d_linear_repeat_POT(sp_sview, sp_samp, s[j], t[j], p[j],
                                            sp_sview->base.u.tex.first_level,
                                            sp_sview->faces[j], &rgba[0][j]);
         else
            img_filter_2d_linear_repeat_POT(sp_sview, sp_samp, s[j], t[j], p[j],
                                            sp_sview->base.texture->last_level,
                                            sp_sview->faces[j], &rgba[0][j]);

      }
      else {
         float levelBlend = frac(lod[j]);
         float rgbax[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
         int c;

         img_filter_2d_linear_repeat_POT(sp_sview, sp_samp, s[j], t[j], p[j], level0,
                                         sp_sview->faces[j], &rgbax[0][0]);
         img_filter_2d_linear_repeat_POT(sp_sview, sp_samp, s[j], t[j], p[j], level0+1,
                                         sp_sview->faces[j], &rgbax[0][1]);

         for (c = 0; c < TGSI_NUM_CHANNELS; c++)
            rgba[c][j] = lerp(levelBlend, rgbax[c][0], rgbax[c][1]);
      }
   }

   if (DEBUG_TEX) {
      print_sample_4(__FUNCTION__, rgba);
   }
}


/**
 * Do shadow/depth comparisons.
 */
static void
sample_compare(struct sp_sampler_view *sp_sview,
               struct sp_sampler *sp_samp,
               const float s[TGSI_QUAD_SIZE],
               const float t[TGSI_QUAD_SIZE],
               const float p[TGSI_QUAD_SIZE],
               const float c0[TGSI_QUAD_SIZE],
               const float c1[TGSI_QUAD_SIZE],
               enum tgsi_sampler_control control,
               float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   const struct pipe_sampler_state *sampler = &sp_samp->base;
   int j;
   int k[4];
   float pc[4];
   const struct util_format_description *format_desc;
   unsigned chan_type;

   /**
    * Compare texcoord 'p' (aka R) against texture value 'rgba[0]'
    * for 2D Array texture we need to use the 'c0' (aka Q).
    * When we sampled the depth texture, the depth value was put into all
    * RGBA channels.  We look at the red channel here.
    */

   if (sp_sview->base.texture->target == PIPE_TEXTURE_2D_ARRAY ||
       sp_sview->base.texture->target == PIPE_TEXTURE_CUBE) {
      pc[0] = c0[0];
      pc[1] = c0[1];
      pc[2] = c0[2];
      pc[3] = c0[3];
   } else if (sp_sview->base.texture->target == PIPE_TEXTURE_CUBE_ARRAY) {
      pc[0] = c1[0];
      pc[1] = c1[1];
      pc[2] = c1[2];
      pc[3] = c1[3];
   } else {
      pc[0] = p[0];
      pc[1] = p[1];
      pc[2] = p[2];
      pc[3] = p[3];
   }

   format_desc = util_format_description(sp_sview->base.format);
   /* not entirely sure we couldn't end up with non-valid swizzle here */
   chan_type = format_desc->swizzle[0] <= UTIL_FORMAT_SWIZZLE_W ?
                  format_desc->channel[format_desc->swizzle[0]].type :
                  UTIL_FORMAT_TYPE_FLOAT;
   if (chan_type != UTIL_FORMAT_TYPE_FLOAT) {
      /*
       * clamping is a result of conversion to texture format, hence
       * doesn't happen with floats. Technically also should do comparison
       * in texture format (quantization!).
       */
      pc[0] = CLAMP(pc[0], 0.0F, 1.0F);
      pc[1] = CLAMP(pc[1], 0.0F, 1.0F);
      pc[2] = CLAMP(pc[2], 0.0F, 1.0F);
      pc[3] = CLAMP(pc[3], 0.0F, 1.0F);
   }

   /* compare four texcoords vs. four texture samples */
   switch (sampler->compare_func) {
   case PIPE_FUNC_LESS:
      k[0] = pc[0] < rgba[0][0];
      k[1] = pc[1] < rgba[0][1];
      k[2] = pc[2] < rgba[0][2];
      k[3] = pc[3] < rgba[0][3];
      break;
   case PIPE_FUNC_LEQUAL:
      k[0] = pc[0] <= rgba[0][0];
      k[1] = pc[1] <= rgba[0][1];
      k[2] = pc[2] <= rgba[0][2];
      k[3] = pc[3] <= rgba[0][3];
      break;
   case PIPE_FUNC_GREATER:
      k[0] = pc[0] > rgba[0][0];
      k[1] = pc[1] > rgba[0][1];
      k[2] = pc[2] > rgba[0][2];
      k[3] = pc[3] > rgba[0][3];
      break;
   case PIPE_FUNC_GEQUAL:
      k[0] = pc[0] >= rgba[0][0];
      k[1] = pc[1] >= rgba[0][1];
      k[2] = pc[2] >= rgba[0][2];
      k[3] = pc[3] >= rgba[0][3];
      break;
   case PIPE_FUNC_EQUAL:
      k[0] = pc[0] == rgba[0][0];
      k[1] = pc[1] == rgba[0][1];
      k[2] = pc[2] == rgba[0][2];
      k[3] = pc[3] == rgba[0][3];
      break;
   case PIPE_FUNC_NOTEQUAL:
      k[0] = pc[0] != rgba[0][0];
      k[1] = pc[1] != rgba[0][1];
      k[2] = pc[2] != rgba[0][2];
      k[3] = pc[3] != rgba[0][3];
      break;
   case PIPE_FUNC_ALWAYS:
      k[0] = k[1] = k[2] = k[3] = 1;
      break;
   case PIPE_FUNC_NEVER:
      k[0] = k[1] = k[2] = k[3] = 0;
      break;
   default:
      k[0] = k[1] = k[2] = k[3] = 0;
      assert(0);
      break;
   }

   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
      rgba[0][j] = k[j];
      rgba[1][j] = k[j];
      rgba[2][j] = k[j];
      rgba[3][j] = 1.0F;
   }
}


static void
do_swizzling(const struct pipe_sampler_view *sview,
             float in[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE],
             float out[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   int j;
   const unsigned swizzle_r = sview->swizzle_r;
   const unsigned swizzle_g = sview->swizzle_g;
   const unsigned swizzle_b = sview->swizzle_b;
   const unsigned swizzle_a = sview->swizzle_a;

   switch (swizzle_r) {
   case PIPE_SWIZZLE_ZERO:
      for (j = 0; j < 4; j++)
         out[0][j] = 0.0f;
      break;
   case PIPE_SWIZZLE_ONE:
      for (j = 0; j < 4; j++)
         out[0][j] = 1.0f;
      break;
   default:
      assert(swizzle_r < 4);
      for (j = 0; j < 4; j++)
         out[0][j] = in[swizzle_r][j];
   }

   switch (swizzle_g) {
   case PIPE_SWIZZLE_ZERO:
      for (j = 0; j < 4; j++)
         out[1][j] = 0.0f;
      break;
   case PIPE_SWIZZLE_ONE:
      for (j = 0; j < 4; j++)
         out[1][j] = 1.0f;
      break;
   default:
      assert(swizzle_g < 4);
      for (j = 0; j < 4; j++)
         out[1][j] = in[swizzle_g][j];
   }

   switch (swizzle_b) {
   case PIPE_SWIZZLE_ZERO:
      for (j = 0; j < 4; j++)
         out[2][j] = 0.0f;
      break;
   case PIPE_SWIZZLE_ONE:
      for (j = 0; j < 4; j++)
         out[2][j] = 1.0f;
      break;
   default:
      assert(swizzle_b < 4);
      for (j = 0; j < 4; j++)
         out[2][j] = in[swizzle_b][j];
   }

   switch (swizzle_a) {
   case PIPE_SWIZZLE_ZERO:
      for (j = 0; j < 4; j++)
         out[3][j] = 0.0f;
      break;
   case PIPE_SWIZZLE_ONE:
      for (j = 0; j < 4; j++)
         out[3][j] = 1.0f;
      break;
   default:
      assert(swizzle_a < 4);
      for (j = 0; j < 4; j++)
         out[3][j] = in[swizzle_a][j];
   }
}


static wrap_nearest_func
get_nearest_unorm_wrap(unsigned mode)
{
   switch (mode) {
   case PIPE_TEX_WRAP_CLAMP:
      return wrap_nearest_unorm_clamp;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      return wrap_nearest_unorm_clamp_to_edge;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      return wrap_nearest_unorm_clamp_to_border;
   default:
      assert(0);
      return wrap_nearest_unorm_clamp;
   }
}


static wrap_nearest_func
get_nearest_wrap(unsigned mode)
{
   switch (mode) {
   case PIPE_TEX_WRAP_REPEAT:
      return wrap_nearest_repeat;
   case PIPE_TEX_WRAP_CLAMP:
      return wrap_nearest_clamp;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      return wrap_nearest_clamp_to_edge;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      return wrap_nearest_clamp_to_border;
   case PIPE_TEX_WRAP_MIRROR_REPEAT:
      return wrap_nearest_mirror_repeat;
   case PIPE_TEX_WRAP_MIRROR_CLAMP:
      return wrap_nearest_mirror_clamp;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
      return wrap_nearest_mirror_clamp_to_edge;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
      return wrap_nearest_mirror_clamp_to_border;
   default:
      assert(0);
      return wrap_nearest_repeat;
   }
}


static wrap_linear_func
get_linear_unorm_wrap(unsigned mode)
{
   switch (mode) {
   case PIPE_TEX_WRAP_CLAMP:
      return wrap_linear_unorm_clamp;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      return wrap_linear_unorm_clamp_to_edge;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      return wrap_linear_unorm_clamp_to_border;
   default:
      assert(0);
      return wrap_linear_unorm_clamp;
   }
}


static wrap_linear_func
get_linear_wrap(unsigned mode)
{
   switch (mode) {
   case PIPE_TEX_WRAP_REPEAT:
      return wrap_linear_repeat;
   case PIPE_TEX_WRAP_CLAMP:
      return wrap_linear_clamp;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      return wrap_linear_clamp_to_edge;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      return wrap_linear_clamp_to_border;
   case PIPE_TEX_WRAP_MIRROR_REPEAT:
      return wrap_linear_mirror_repeat;
   case PIPE_TEX_WRAP_MIRROR_CLAMP:
      return wrap_linear_mirror_clamp;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
      return wrap_linear_mirror_clamp_to_edge;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
      return wrap_linear_mirror_clamp_to_border;
   default:
      assert(0);
      return wrap_linear_repeat;
   }
}


/**
 * Is swizzling needed for the given state key?
 */
static INLINE bool
any_swizzle(const struct pipe_sampler_view *view)
{
   return (view->swizzle_r != PIPE_SWIZZLE_RED ||
           view->swizzle_g != PIPE_SWIZZLE_GREEN ||
           view->swizzle_b != PIPE_SWIZZLE_BLUE ||
           view->swizzle_a != PIPE_SWIZZLE_ALPHA);
}


static img_filter_func
get_img_filter(const struct sp_sampler_view *sp_sview,
               const struct pipe_sampler_state *sampler,
               unsigned filter)
{
   switch (sp_sview->base.texture->target) {
   case PIPE_BUFFER:
   case PIPE_TEXTURE_1D:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_1d_nearest;
      else
         return img_filter_1d_linear;
      break;
   case PIPE_TEXTURE_1D_ARRAY:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_1d_array_nearest;
      else
         return img_filter_1d_array_linear;
      break;
   case PIPE_TEXTURE_2D:
   case PIPE_TEXTURE_RECT:
      /* Try for fast path:
       */
      if (sp_sview->pot2d &&
          sampler->wrap_s == sampler->wrap_t &&
          sampler->normalized_coords) 
      {
         switch (sampler->wrap_s) {
         case PIPE_TEX_WRAP_REPEAT:
            switch (filter) {
            case PIPE_TEX_FILTER_NEAREST:
               return img_filter_2d_nearest_repeat_POT;
            case PIPE_TEX_FILTER_LINEAR:
               return img_filter_2d_linear_repeat_POT;
            default:
               break;
            }
            break;
         case PIPE_TEX_WRAP_CLAMP:
            switch (filter) {
            case PIPE_TEX_FILTER_NEAREST:
               return img_filter_2d_nearest_clamp_POT;
            default:
               break;
            }
         }
      }
      /* Otherwise use default versions:
       */
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_2d_nearest;
      else
         return img_filter_2d_linear;
      break;
   case PIPE_TEXTURE_2D_ARRAY:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_2d_array_nearest;
      else
         return img_filter_2d_array_linear;
      break;
   case PIPE_TEXTURE_CUBE:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_cube_nearest;
      else
         return img_filter_cube_linear;
      break;
   case PIPE_TEXTURE_CUBE_ARRAY:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_cube_array_nearest;
      else
         return img_filter_cube_array_linear;
      break;
   case PIPE_TEXTURE_3D:
      if (filter == PIPE_TEX_FILTER_NEAREST) 
         return img_filter_3d_nearest;
      else
         return img_filter_3d_linear;
      break;
   default:
      assert(0);
      return img_filter_1d_nearest;
   }
}


static void
sample_mip(struct sp_sampler_view *sp_sview,
           struct sp_sampler *sp_samp,
           const float s[TGSI_QUAD_SIZE],
           const float t[TGSI_QUAD_SIZE],
           const float p[TGSI_QUAD_SIZE],
           const float c0[TGSI_QUAD_SIZE],
           const float lod[TGSI_QUAD_SIZE],
           enum tgsi_sampler_control control,
           float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   mip_filter_func mip_filter;
   img_filter_func min_img_filter = NULL;
   img_filter_func mag_img_filter = NULL;

   if (sp_sview->pot2d & sp_samp->min_mag_equal_repeat_linear) {
      mip_filter = mip_filter_linear_2d_linear_repeat_POT;
   }
   else {
      mip_filter = sp_samp->mip_filter;
      min_img_filter = get_img_filter(sp_sview, &sp_samp->base, sp_samp->min_img_filter);
      if (sp_samp->min_mag_equal) {
         mag_img_filter = min_img_filter;
      }
      else {
         mag_img_filter = get_img_filter(sp_sview, &sp_samp->base, sp_samp->base.mag_img_filter);
      }
   }

   mip_filter(sp_sview, sp_samp, min_img_filter, mag_img_filter,
              s, t, p, c0, lod, control, rgba);

   if (sp_samp->base.compare_mode != PIPE_TEX_COMPARE_NONE) {
      sample_compare(sp_sview, sp_samp, s, t, p, c0, lod, control, rgba);
   }

   if (sp_sview->need_swizzle) {
      float rgba_temp[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
      memcpy(rgba_temp, rgba, sizeof(rgba_temp));
      do_swizzling(&sp_sview->base, rgba_temp, rgba);
   }

}


/**
 * Use 3D texcoords to choose a cube face, then sample the 2D cube faces.
 * Put face info into the sampler faces[] array.
 */
static void
sample_cube(struct sp_sampler_view *sp_sview,
            struct sp_sampler *sp_samp,
            const float s[TGSI_QUAD_SIZE],
            const float t[TGSI_QUAD_SIZE],
            const float p[TGSI_QUAD_SIZE],
            const float c0[TGSI_QUAD_SIZE],
            const float c1[TGSI_QUAD_SIZE],
            enum tgsi_sampler_control control,
            float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   unsigned j;
   float ssss[4], tttt[4];

   /* Not actually used, but the intermediate steps that do the
    * dereferencing don't know it.
    */
   static float pppp[4] = { 0, 0, 0, 0 };

   pppp[0] = c0[0];
   pppp[1] = c0[1];
   pppp[2] = c0[2];
   pppp[3] = c0[3];
   /*
     major axis
     direction    target                             sc     tc    ma
     ----------   -------------------------------    ---    ---   ---
     +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
     -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
     +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
     -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
     +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
     -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
   */

   /* Choose the cube face and compute new s/t coords for the 2D face.
    *
    * Use the same cube face for all four pixels in the quad.
    *
    * This isn't ideal, but if we want to use a different cube face
    * per pixel in the quad, we'd have to also compute the per-face
    * LOD here too.  That's because the four post-face-selection
    * texcoords are no longer related to each other (they're
    * per-face!)  so we can't use subtraction to compute the partial
    * deriviates to compute the LOD.  Doing so (near cube edges
    * anyway) gives us pretty much random values.
    */
   {
      /* use the average of the four pixel's texcoords to choose the face */
      const float rx = 0.25F * (s[0] + s[1] + s[2] + s[3]);
      const float ry = 0.25F * (t[0] + t[1] + t[2] + t[3]);
      const float rz = 0.25F * (p[0] + p[1] + p[2] + p[3]);
      const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz);

      if (arx >= ary && arx >= arz) {
         float sign = (rx >= 0.0F) ? 1.0F : -1.0F;
         uint face = (rx >= 0.0F) ? PIPE_TEX_FACE_POS_X : PIPE_TEX_FACE_NEG_X;
         for (j = 0; j < TGSI_QUAD_SIZE; j++) {
            const float ima = -0.5F / fabsf(s[j]);
            ssss[j] = sign *  p[j] * ima + 0.5F;
            tttt[j] =         t[j] * ima + 0.5F;
            sp_sview->faces[j] = face;
         }
      }
      else if (ary >= arx && ary >= arz) {
         float sign = (ry >= 0.0F) ? 1.0F : -1.0F;
         uint face = (ry >= 0.0F) ? PIPE_TEX_FACE_POS_Y : PIPE_TEX_FACE_NEG_Y;
         for (j = 0; j < TGSI_QUAD_SIZE; j++) {
            const float ima = -0.5F / fabsf(t[j]);
            ssss[j] =        -s[j] * ima + 0.5F;
            tttt[j] = sign * -p[j] * ima + 0.5F;
            sp_sview->faces[j] = face;
         }
      }
      else {
         float sign = (rz >= 0.0F) ? 1.0F : -1.0F;
         uint face = (rz >= 0.0F) ? PIPE_TEX_FACE_POS_Z : PIPE_TEX_FACE_NEG_Z;
         for (j = 0; j < TGSI_QUAD_SIZE; j++) {
            const float ima = -0.5F / fabsf(p[j]);
            ssss[j] = sign * -s[j] * ima + 0.5F;
            tttt[j] =         t[j] * ima + 0.5F;
            sp_sview->faces[j] = face;
         }
      }
   }

   sample_mip(sp_sview, sp_samp, ssss, tttt, pppp, c0, c1, control, rgba);
}


static void
sp_get_dims(struct sp_sampler_view *sp_sview, int level,
            int dims[4])
{
   const struct pipe_sampler_view *view = &sp_sview->base;
   const struct pipe_resource *texture = view->texture;

   /* undefined according to EXT_gpu_program */
   level += view->u.tex.first_level;
   if (level > view->u.tex.last_level)
      return;

   dims[0] = u_minify(texture->width0, level);

   switch(texture->target) {
   case PIPE_TEXTURE_1D_ARRAY:
      dims[1] = view->u.tex.last_layer - view->u.tex.first_layer + 1;
      /* fallthrough */
   case PIPE_TEXTURE_1D:
      return;
   case PIPE_TEXTURE_2D_ARRAY:
      dims[2] = view->u.tex.last_layer - view->u.tex.first_layer + 1;
      /* fallthrough */
   case PIPE_TEXTURE_2D:
   case PIPE_TEXTURE_CUBE:
   case PIPE_TEXTURE_RECT:
      dims[1] = u_minify(texture->height0, level);
      return;
   case PIPE_TEXTURE_3D:
      dims[1] = u_minify(texture->height0, level);
      dims[2] = u_minify(texture->depth0, level);
      return;
   case PIPE_TEXTURE_CUBE_ARRAY:
      dims[1] = u_minify(texture->height0, level);
      dims[2] = (view->u.tex.last_layer - view->u.tex.first_layer + 1) / 6;
      break;
   case PIPE_BUFFER:
      dims[0] /= util_format_get_blocksize(view->format);
      return;
   default:
      assert(!"unexpected texture target in sp_get_dims()");
      return;
   }
}

/**
 * This function is only used for getting unfiltered texels via the
 * TXF opcode.  The GL spec says that out-of-bounds texel fetches
 * produce undefined results.  Instead of crashing, lets just clamp
 * coords to the texture image size.
 */
static void
sp_get_texels(struct sp_sampler_view *sp_sview,
              const int v_i[TGSI_QUAD_SIZE],
              const int v_j[TGSI_QUAD_SIZE],
              const int v_k[TGSI_QUAD_SIZE],
              const int lod[TGSI_QUAD_SIZE],
              const int8_t offset[3],
              float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   union tex_tile_address addr;
   const struct pipe_resource *texture = sp_sview->base.texture;
   int j, c;
   const float *tx;
   int width, height, depth;

   addr.value = 0;
   /* TODO write a better test for LOD */
   addr.bits.level = lod[0];

   width = u_minify(texture->width0, addr.bits.level);
   height = u_minify(texture->height0, addr.bits.level);
   depth = u_minify(texture->depth0, addr.bits.level);

   switch(texture->target) {
   case PIPE_BUFFER:
   case PIPE_TEXTURE_1D:
      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
         int x = CLAMP(v_i[j] + offset[0], 0, width - 1);
         tx = get_texel_2d_no_border(sp_sview, addr, x, 0);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = tx[c];
         }
      }
      break;
   case PIPE_TEXTURE_1D_ARRAY:
      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
         int x = CLAMP(v_i[j] + offset[0], 0, width - 1);
         int y = CLAMP(v_j[j], sp_sview->base.u.tex.first_layer, sp_sview->base.u.tex.last_layer);
         tx = get_texel_2d_no_border(sp_sview, addr, x, y);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = tx[c];
         }
      }
      break;
   case PIPE_TEXTURE_2D:
   case PIPE_TEXTURE_RECT:
      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
         int x = CLAMP(v_i[j] + offset[0], 0, width - 1);
         int y = CLAMP(v_j[j] + offset[1], 0, height - 1);
         tx = get_texel_2d_no_border(sp_sview, addr, x, y);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = tx[c];
         }
      }
      break;
   case PIPE_TEXTURE_2D_ARRAY:
      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
         int x = CLAMP(v_i[j] + offset[0], 0, width - 1);
         int y = CLAMP(v_j[j] + offset[1], 0, height - 1);
         int layer = CLAMP(v_k[j], sp_sview->base.u.tex.first_layer, sp_sview->base.u.tex.last_layer);
         tx = get_texel_3d_no_border(sp_sview, addr, x, y, layer);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = tx[c];
         }
      }
      break;
   case PIPE_TEXTURE_3D:
      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
         int x = CLAMP(v_i[j] + offset[0], 0, width - 1);
         int y = CLAMP(v_j[j] + offset[1], 0, height - 1);
         int z = CLAMP(v_k[j] + offset[2], 0, depth - 1);
         tx = get_texel_3d_no_border(sp_sview, addr, x, y, z);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = tx[c];
         }
      }
      break;
   case PIPE_TEXTURE_CUBE: /* TXF can't work on CUBE according to spec */
   default:
      assert(!"Unknown or CUBE texture type in TXF processing\n");
      break;
   }

   if (sp_sview->need_swizzle) {
      float rgba_temp[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
      memcpy(rgba_temp, rgba, sizeof(rgba_temp));
      do_swizzling(&sp_sview->base, rgba_temp, rgba);
   }
}


void *
softpipe_create_sampler_state(struct pipe_context *pipe,
                              const struct pipe_sampler_state *sampler)
{
   struct sp_sampler *samp = CALLOC_STRUCT(sp_sampler);

   samp->base = *sampler;

   /* Note that (for instance) linear_texcoord_s and
    * nearest_texcoord_s may be active at the same time, if the
    * sampler min_img_filter differs from its mag_img_filter.
    */
   if (sampler->normalized_coords) {
      samp->linear_texcoord_s = get_linear_wrap( sampler->wrap_s );
      samp->linear_texcoord_t = get_linear_wrap( sampler->wrap_t );
      samp->linear_texcoord_p = get_linear_wrap( sampler->wrap_r );

      samp->nearest_texcoord_s = get_nearest_wrap( sampler->wrap_s );
      samp->nearest_texcoord_t = get_nearest_wrap( sampler->wrap_t );
      samp->nearest_texcoord_p = get_nearest_wrap( sampler->wrap_r );
   }
   else {
      samp->linear_texcoord_s = get_linear_unorm_wrap( sampler->wrap_s );
      samp->linear_texcoord_t = get_linear_unorm_wrap( sampler->wrap_t );
      samp->linear_texcoord_p = get_linear_unorm_wrap( sampler->wrap_r );

      samp->nearest_texcoord_s = get_nearest_unorm_wrap( sampler->wrap_s );
      samp->nearest_texcoord_t = get_nearest_unorm_wrap( sampler->wrap_t );
      samp->nearest_texcoord_p = get_nearest_unorm_wrap( sampler->wrap_r );
   }

   samp->min_img_filter = sampler->min_img_filter;

   switch (sampler->min_mip_filter) {
   case PIPE_TEX_MIPFILTER_NONE:
      if (sampler->min_img_filter == sampler->mag_img_filter)
         samp->mip_filter = mip_filter_none_no_filter_select;
      else
         samp->mip_filter = mip_filter_none;
      break;

   case PIPE_TEX_MIPFILTER_NEAREST:
      samp->mip_filter = mip_filter_nearest;
      break;

   case PIPE_TEX_MIPFILTER_LINEAR:
      if (sampler->min_img_filter == sampler->mag_img_filter &&
          sampler->normalized_coords &&
          sampler->wrap_s == PIPE_TEX_WRAP_REPEAT &&
          sampler->wrap_t == PIPE_TEX_WRAP_REPEAT &&
          sampler->min_img_filter == PIPE_TEX_FILTER_LINEAR &&
          sampler->max_anisotropy <= 1) {
         samp->min_mag_equal_repeat_linear = TRUE;
      }
      samp->mip_filter = mip_filter_linear;

      /* Anisotropic filtering extension. */
      if (sampler->max_anisotropy > 1) {
         samp->mip_filter = mip_filter_linear_aniso;

         /* Override min_img_filter:
          * min_img_filter needs to be set to NEAREST since we need to access
          * each texture pixel as it is and weight it later; using linear
          * filters will have incorrect results.
          * By setting the filter to NEAREST here, we can avoid calling the
          * generic img_filter_2d_nearest in the anisotropic filter function,
          * making it possible to use one of the accelerated implementations
          */
         samp->min_img_filter = PIPE_TEX_FILTER_NEAREST;

         /* on first access create the lookup table containing the filter weights. */
        if (!weightLut) {
           create_filter_table();
        }
      }
      break;
   }
   if (samp->min_img_filter == sampler->mag_img_filter) {
      samp->min_mag_equal = TRUE;
   }

   return (void *)samp;
}


compute_lambda_func
softpipe_get_lambda_func(const struct pipe_sampler_view *view, unsigned shader)
{
   if (shader != PIPE_SHADER_FRAGMENT)
      return compute_lambda_vert;

   switch (view->texture->target) {
   case PIPE_BUFFER:
   case PIPE_TEXTURE_1D:
   case PIPE_TEXTURE_1D_ARRAY:
      return compute_lambda_1d;
   case PIPE_TEXTURE_2D:
   case PIPE_TEXTURE_2D_ARRAY:
   case PIPE_TEXTURE_RECT:
   case PIPE_TEXTURE_CUBE:
   case PIPE_TEXTURE_CUBE_ARRAY:
      return compute_lambda_2d;
   case PIPE_TEXTURE_3D:
      return compute_lambda_3d;
   default:
      assert(0);
      return compute_lambda_1d;
   }
}


struct pipe_sampler_view *
softpipe_create_sampler_view(struct pipe_context *pipe,
                             struct pipe_resource *resource,
                             const struct pipe_sampler_view *templ)
{
   struct sp_sampler_view *sview = CALLOC_STRUCT(sp_sampler_view);
   struct softpipe_resource *spr = (struct softpipe_resource *)resource;

   if (sview) {
      struct pipe_sampler_view *view = &sview->base;
      *view = *templ;
      view->reference.count = 1;
      view->texture = NULL;
      pipe_resource_reference(&view->texture, resource);
      view->context = pipe;

      if (any_swizzle(view)) {
         sview->need_swizzle = TRUE;
      }

      if (resource->target == PIPE_TEXTURE_CUBE ||
          resource->target == PIPE_TEXTURE_CUBE_ARRAY)
         sview->get_samples = sample_cube;
      else {
         sview->get_samples = sample_mip;
      }
      sview->pot2d = spr->pot &&
                     (resource->target == PIPE_TEXTURE_2D ||
                      resource->target == PIPE_TEXTURE_RECT);

      sview->xpot = util_logbase2( resource->width0 );
      sview->ypot = util_logbase2( resource->height0 );
   }

   return (struct pipe_sampler_view *) sview;
}


static void
sp_tgsi_get_dims(struct tgsi_sampler *tgsi_sampler,
                 const unsigned sview_index,
                 int level, int dims[4])
{
   struct sp_tgsi_sampler *sp_samp = (struct sp_tgsi_sampler *)tgsi_sampler;

   assert(sview_index < PIPE_MAX_SHADER_SAMPLER_VIEWS);
   /* always have a view here but texture is NULL if no sampler view was set. */
   if (!sp_samp->sp_sview[sview_index].base.texture) {
      dims[0] = dims[1] = dims[2] = dims[3] = 0;
      return;
   }
   sp_get_dims(&sp_samp->sp_sview[sview_index], level, dims);
}


static void
sp_tgsi_get_samples(struct tgsi_sampler *tgsi_sampler,
                    const unsigned sview_index,
                    const unsigned sampler_index,
                    const float s[TGSI_QUAD_SIZE],
                    const float t[TGSI_QUAD_SIZE],
                    const float p[TGSI_QUAD_SIZE],
                    const float c0[TGSI_QUAD_SIZE],
                    const float lod[TGSI_QUAD_SIZE],
                    float derivs[3][2][TGSI_QUAD_SIZE],
                    const int8_t offset[3],
                    enum tgsi_sampler_control control,
                    float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   struct sp_tgsi_sampler *sp_samp = (struct sp_tgsi_sampler *)tgsi_sampler;

   assert(sview_index < PIPE_MAX_SHADER_SAMPLER_VIEWS);
   assert(sampler_index < PIPE_MAX_SAMPLERS);
   assert(sp_samp->sp_sampler[sampler_index]);
   /* always have a view here but texture is NULL if no sampler view was set. */
   if (!sp_samp->sp_sview[sview_index].base.texture) {
      int i, j;
      for (j = 0; j < TGSI_NUM_CHANNELS; j++) {
         for (i = 0; i < TGSI_QUAD_SIZE; i++) {
            rgba[j][i] = 0.0f;
         }
      }
      return;
   }
   sp_samp->sp_sview[sview_index].get_samples(&sp_samp->sp_sview[sview_index],
                                              sp_samp->sp_sampler[sampler_index],
                                              s, t, p, c0, lod, control, rgba);
}


static void
sp_tgsi_get_texel(struct tgsi_sampler *tgsi_sampler,
                  const unsigned sview_index,
                  const int i[TGSI_QUAD_SIZE],
                  const int j[TGSI_QUAD_SIZE], const int k[TGSI_QUAD_SIZE],
                  const int lod[TGSI_QUAD_SIZE], const int8_t offset[3],
                  float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE])
{
   struct sp_tgsi_sampler *sp_samp = (struct sp_tgsi_sampler *)tgsi_sampler;

   assert(sview_index < PIPE_MAX_SHADER_SAMPLER_VIEWS);
   /* always have a view here but texture is NULL if no sampler view was set. */
   if (!sp_samp->sp_sview[sview_index].base.texture) {
      int i, j;
      for (j = 0; j < TGSI_NUM_CHANNELS; j++) {
         for (i = 0; i < TGSI_QUAD_SIZE; i++) {
            rgba[j][i] = 0.0f;
         }
      }
      return;
   }
   sp_get_texels(&sp_samp->sp_sview[sview_index], i, j, k, lod, offset, rgba);
}


struct sp_tgsi_sampler *
sp_create_tgsi_sampler(void)
{
   struct sp_tgsi_sampler *samp = CALLOC_STRUCT(sp_tgsi_sampler);
   if (!samp)
      return NULL;

   samp->base.get_dims = sp_tgsi_get_dims;
   samp->base.get_samples = sp_tgsi_get_samples;
   samp->base.get_texel = sp_tgsi_get_texel;

   return samp;
}