summaryrefslogtreecommitdiffstats
path: root/firmware/thread.c
blob: a25a191b8018ffefce28719ffa83d3d65aa40914 (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
/***************************************************************************
 *             __________               __   ___.
 *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
 *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
 *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
 *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
 *                     \/            \/     \/    \/            \/
 * $Id$
 *
 * Copyright (C) 2002 by Ulf Ralberg
 *
 * All files in this archive are subject to the GNU General Public License.
 * See the file COPYING in the source tree root for full license agreement.
 *
 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
 * KIND, either express or implied.
 *
 ****************************************************************************/
#include "config.h"
#include <stdbool.h>
#include "thread.h"
#include "panic.h"
#include "sprintf.h"
#include "system.h"
#include "kernel.h"
#include "cpu.h"
#include "string.h"
#ifdef RB_PROFILE
#include <profile.h>
#endif
/****************************************************************************
 *                              ATTENTION!!                                 *
 *    See notes below on implementing processor-specific portions!          *
 ***************************************************************************/

/* Define THREAD_EXTRA_CHECKS as 1 to enable additional state checks */
#ifdef DEBUG
#define THREAD_EXTRA_CHECKS 1 /* Always 1 for DEBUG */
#else
#define THREAD_EXTRA_CHECKS 0
#endif

/**
 * General locking order to guarantee progress. Order must be observed but
 * all stages are not nescessarily obligatory. Going from 1) to 3) is
 * perfectly legal.
 *
 * 1) IRQ
 * This is first because of the likelyhood of having an interrupt occur that
 * also accesses one of the objects farther down the list. Any non-blocking
 * synchronization done may already have a lock on something during normal
 * execution and if an interrupt handler running on the same processor as
 * the one that has the resource locked were to attempt to access the
 * resource, the interrupt handler would wait forever waiting for an unlock
 * that will never happen. There is no danger if the interrupt occurs on
 * a different processor because the one that has the lock will eventually
 * unlock and the other processor's handler may proceed at that time. Not
 * nescessary when the resource in question is definitely not available to
 * interrupt handlers.
 *  
 * 2) Kernel Object
 * 1) May be needed beforehand if the kernel object allows dual-use such as
 * event queues. The kernel object must have a scheme to protect itself from
 * access by another processor and is responsible for serializing the calls
 * to block_thread(_w_tmo) and wakeup_thread both to themselves and to each
 * other. Objects' queues are also protected here.
 * 
 * 3) Thread Slot
 * This locks access to the thread's slot such that its state cannot be
 * altered by another processor when a state change is in progress such as
 * when it is in the process of going on a blocked list. An attempt to wake
 * a thread while it is still blocking will likely desync its state with
 * the other resources used for that state.
 *
 * 4) Core Lists
 * These lists are specific to a particular processor core and are accessible
 * by all processor cores and interrupt handlers. The running (rtr) list is
 * the prime example where a thread may be added by any means.
 */

/*---------------------------------------------------------------------------
 * Processor specific: core_sleep/core_wake/misc. notes
 *
 * ARM notes:
 * FIQ is not dealt with by the scheduler code and is simply restored if it
 * must by masked for some reason - because threading modifies a register
 * that FIQ may also modify and there's no way to accomplish it atomically.
 * s3c2440 is such a case.
 *
 * Audio interrupts are generally treated at a higher priority than others
 * usage of scheduler code with interrupts higher than HIGHEST_IRQ_LEVEL
 * are not in general safe. Special cases may be constructed on a per-
 * source basis and blocking operations are not available.
 *
 * core_sleep procedure to implement for any CPU to ensure an asychronous
 * wakup never results in requiring a wait until the next tick (up to
 * 10000uS!). May require assembly and careful instruction ordering.
 *
 * 1) On multicore, stay awake if directed to do so by another. If so, goto
 *    step 4.
 * 2) If processor requires, atomically reenable interrupts and perform step
 *    3.
 * 3) Sleep the CPU core. If wakeup itself enables interrupts (stop #0x2000
 *    on Coldfire) goto step 5.
 * 4) Enable interrupts.
 * 5) Exit procedure.
 *
 * core_wake and multprocessor notes for sleep/wake coordination:
 * If possible, to wake up another processor, the forcing of an interrupt on
 * the woken core by the waker core is the easiest way to ensure a non-
 * delayed wake and immediate execution of any woken threads. If that isn't
 * available then some careful non-blocking synchonization is needed (as on
 * PP targets at the moment).
 *---------------------------------------------------------------------------
 */

/* Cast to the the machine pointer size, whose size could be < 4 or > 32
 * (someday :). */
#define DEADBEEF ((uintptr_t)0xdeadbeefdeadbeefull)
struct core_entry cores[NUM_CORES] IBSS_ATTR;
struct thread_entry threads[MAXTHREADS] IBSS_ATTR;

static const char main_thread_name[] = "main";
extern uintptr_t stackbegin[];
extern uintptr_t stackend[];

static inline void core_sleep(IF_COP_VOID(unsigned int core))
        __attribute__((always_inline));

void check_tmo_threads(void)
        __attribute__((noinline));

static inline void block_thread_on_l(struct thread_entry *thread, unsigned state)
        __attribute__((always_inline));

static void add_to_list_tmo(struct thread_entry *thread)
        __attribute__((noinline));

static void core_schedule_wakeup(struct thread_entry *thread)
        __attribute__((noinline));

#if NUM_CORES > 1
static inline void run_blocking_ops(
    unsigned int core, struct thread_entry *thread)
        __attribute__((always_inline));
#endif

static void thread_stkov(struct thread_entry *thread)
        __attribute__((noinline));

static inline void store_context(void* addr)
        __attribute__((always_inline));

static inline void load_context(const void* addr)
        __attribute__((always_inline));

void switch_thread(void)
        __attribute__((noinline));

/****************************************************************************
 * Processor-specific section
 */

#if defined(MAX_PHYS_SECTOR_SIZE) && MEM == 64
/* Support a special workaround object for large-sector disks */
#define IF_NO_SKIP_YIELD(...) __VA_ARGS__
#else
#define IF_NO_SKIP_YIELD(...)
#endif

#if defined(CPU_ARM)
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
static void __attribute__((naked,used)) start_thread(void)
{
    /* r0 = context */
    asm volatile (
        "ldr    sp, [r0, #32]          \n" /* Load initial sp */
        "ldr    r4, [r0, #40]          \n" /* start in r4 since it's non-volatile */
        "mov    r1, #0                 \n" /* Mark thread as running */
        "str    r1, [r0, #40]          \n"
#if NUM_CORES > 1
        "ldr    r0, =invalidate_icache \n" /* Invalidate this core's cache. */
        "mov    lr, pc                 \n" /* This could be the first entry into */
        "bx     r0                     \n" /* plugin or codec code for this core. */
#endif
        "mov    lr, pc                 \n" /* Call thread function */
        "bx     r4                     \n"
    ); /* No clobber list - new thread doesn't care */
    thread_exit();
    //asm volatile (".ltorg"); /* Dump constant pool */
}

/* For startup, place context pointer in r4 slot, start_thread pointer in r5
 * slot, and thread function pointer in context.start. See load_context for
 * what happens when thread is initially going to run. */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.r[0] = (uint32_t)&(thread)->context,  \
       (thread)->context.r[1] = (uint32_t)start_thread, \
       (thread)->context.start = (uint32_t)function; })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile(
        "stmia  %0, { r4-r11, sp, lr } \n"
        : : "r" (addr)
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile(
        "ldr     r0, [%0, #40]          \n" /* Load start pointer */
        "cmp     r0, #0                 \n" /* Check for NULL */
        "ldmneia %0, { r0, pc }         \n" /* If not already running, jump to start */ 
        "ldmia   %0, { r4-r11, sp, lr } \n" /* Load regs r4 to r14 from context */
        : : "r" (addr) : "r0" /* only! */
    );
}

#if defined (CPU_PP)

#if NUM_CORES > 1
extern uintptr_t cpu_idlestackbegin[];
extern uintptr_t cpu_idlestackend[];
extern uintptr_t cop_idlestackbegin[];
extern uintptr_t cop_idlestackend[];
static uintptr_t * const idle_stacks[NUM_CORES] =
{
    [CPU] = cpu_idlestackbegin,
    [COP] = cop_idlestackbegin
};

#if CONFIG_CPU == PP5002
/* Bytes to emulate the PP502x mailbox bits */
struct core_semaphores
{
    volatile uint8_t intend_wake;  /* 00h */
    volatile uint8_t stay_awake;   /* 01h */
    volatile uint8_t intend_sleep; /* 02h */
    volatile uint8_t unused;       /* 03h */
};

static struct core_semaphores core_semaphores[NUM_CORES] IBSS_ATTR;
#endif /* CONFIG_CPU == PP5002 */

#endif /* NUM_CORES */

#if CONFIG_CORELOCK == SW_CORELOCK
/* Software core locks using Peterson's mutual exclusion algorithm */

/*---------------------------------------------------------------------------
 * Initialize the corelock structure.
 *---------------------------------------------------------------------------
 */
void corelock_init(struct corelock *cl)
{
    memset(cl, 0, sizeof (*cl));
}

#if 1 /* Assembly locks to minimize overhead */
/*---------------------------------------------------------------------------
 * Wait for the corelock to become free and acquire it when it does.
 *---------------------------------------------------------------------------
 */
void corelock_lock(struct corelock *cl) __attribute__((naked));
void corelock_lock(struct corelock *cl)
{
    /* Relies on the fact that core IDs are complementary bitmasks (0x55,0xaa) */
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "strb   r1, [r0, r1, lsr #7] \n" /* cl->myl[core] = core */
        "eor    r2, r1, #0xff        \n" /* r2 = othercore */
        "strb   r2, [r0, #2]         \n" /* cl->turn = othercore */
    "1:                              \n"
        "ldrb   r3, [r0, r2, lsr #7] \n" /* cl->myl[othercore] == 0 ? */
        "cmp    r3, #0               \n" /* yes? lock acquired */
        "bxeq   lr                   \n"
        "ldrb   r3, [r0, #2]         \n" /* || cl->turn == core ? */
        "cmp    r3, r1               \n"
        "bxeq   lr                   \n" /* yes? lock acquired */
        "b      1b                   \n" /* keep trying */
        : : "i"(&PROCESSOR_ID)
    );
    (void)cl;
}

/*---------------------------------------------------------------------------
 * Try to aquire the corelock. If free, caller gets it, otherwise return 0.
 *---------------------------------------------------------------------------
 */
int corelock_try_lock(struct corelock *cl) __attribute__((naked));
int corelock_try_lock(struct corelock *cl)
{
    /* Relies on the fact that core IDs are complementary bitmasks (0x55,0xaa) */
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "mov    r3, r0               \n"
        "strb   r1, [r0, r1, lsr #7] \n" /* cl->myl[core] = core */
        "eor    r2, r1, #0xff        \n" /* r2 = othercore */
        "strb   r2, [r0, #2]         \n" /* cl->turn = othercore */
        "ldrb   r0, [r3, r2, lsr #7] \n" /* cl->myl[othercore] == 0 ? */
        "eors   r0, r0, r2           \n" /* yes? lock acquired */
        "bxne   lr                   \n"
        "ldrb   r0, [r3, #2]         \n" /* || cl->turn == core? */
        "ands   r0, r0, r1           \n"
        "streqb r0, [r3, r1, lsr #7] \n" /* if not, cl->myl[core] = 0 */
        "bx     lr                   \n" /* return result */
        : : "i"(&PROCESSOR_ID)
    );

    return 0;
    (void)cl;
}

/*---------------------------------------------------------------------------
 * Release ownership of the corelock
 *---------------------------------------------------------------------------
 */
void corelock_unlock(struct corelock *cl) __attribute__((naked));
void corelock_unlock(struct corelock *cl)
{
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "mov    r2, #0               \n" /* cl->myl[core] = 0 */
        "strb   r2, [r0, r1, lsr #7] \n"
        "bx     lr                   \n"
        : : "i"(&PROCESSOR_ID)
    );
    (void)cl;
}
#else /* C versions for reference */
/*---------------------------------------------------------------------------
 * Wait for the corelock to become free and aquire it when it does.
 *---------------------------------------------------------------------------
 */
void corelock_lock(struct corelock *cl)
{
    const unsigned int core = CURRENT_CORE;
    const unsigned int othercore = 1 - core;

    cl->myl[core] = core;
    cl->turn = othercore;

    for (;;)
    {
        if (cl->myl[othercore] == 0 || cl->turn == core)
            break;
    }
}

/*---------------------------------------------------------------------------
 * Try to aquire the corelock. If free, caller gets it, otherwise return 0.
 *---------------------------------------------------------------------------
 */
int corelock_try_lock(struct corelock *cl)
{
    const unsigned int core = CURRENT_CORE;
    const unsigned int othercore = 1 - core;

    cl->myl[core] = core;
    cl->turn = othercore;

    if (cl->myl[othercore] == 0 || cl->turn == core)
    {
        return 1;
    }

    cl->myl[core] = 0;
    return 0;
}

/*---------------------------------------------------------------------------
 * Release ownership of the corelock
 *---------------------------------------------------------------------------
 */
void corelock_unlock(struct corelock *cl)
{
    cl->myl[CURRENT_CORE] = 0;
}
#endif /* ASM / C selection */

#endif /* CONFIG_CORELOCK == SW_CORELOCK */

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated and if
 * no other core requested a wakeup for it to perform a task.
 *---------------------------------------------------------------------------
 */
#ifdef CPU_PP502x
#if NUM_CORES == 1
static inline void core_sleep(void)
{
    PROC_CTL(CURRENT_CORE) = PROC_SLEEP;
    nop; nop; nop;
    enable_irq();
}
#else
static inline void core_sleep(unsigned int core)
{
#if 1
    asm volatile (
        "mov    r0, #4                     \n" /* r0 = 0x4 << core */
        "mov    r0, r0, lsl %[c]           \n"
        "str    r0, [%[mbx], #4]           \n" /* signal intent to sleep */
        "ldr    r1, [%[mbx], #0]           \n" /* && !(MBX_MSG_STAT & (0x10<<core)) ? */
        "tst    r1, r0, lsl #2             \n"  
        "moveq  r1, #0x80000000            \n" /* Then sleep */
        "streq  r1, [%[ctl], %[c], lsl #2] \n"
        "moveq  r1, #0                     \n" /* Clear control reg */
        "streq  r1, [%[ctl], %[c], lsl #2] \n"
        "orr    r1, r0, r0, lsl #2         \n" /* Signal intent to wake - clear wake flag */
        "str    r1, [%[mbx], #8]           \n"
    "1:                                    \n" /* Wait for wake procedure to finish */
        "ldr    r1, [%[mbx], #0]           \n"
        "tst    r1, r0, lsr #2             \n"
        "bne    1b                         \n"
        :
        :  [ctl]"r"(&PROC_CTL(CPU)), [mbx]"r"(MBX_BASE), [c]"r"(core)
        : "r0", "r1");
#else /* C version for reference */
    /* Signal intent to sleep */
    MBX_MSG_SET = 0x4 << core;

    /* Something waking or other processor intends to wake us? */
    if ((MBX_MSG_STAT & (0x10 << core)) == 0)
    {
        PROC_CTL(core) = PROC_SLEEP; nop; /* Snooze */
        PROC_CTL(core) = 0;               /* Clear control reg */
    }

    /* Signal wake - clear wake flag */
    MBX_MSG_CLR = 0x14 << core;

    /* Wait for other processor to finish wake procedure */
    while (MBX_MSG_STAT & (0x1 << core));
#endif /* ASM/C selection */
    enable_irq();
}
#endif /* NUM_CORES */
#elif CONFIG_CPU == PP5002
#if NUM_CORES == 1
static inline void core_sleep(void)
{
    asm volatile (
        /* Sleep: PP5002 crashes if the instruction that puts it to sleep is
         * located at 0xNNNNNNN0. 4/8/C works. This sequence makes sure
         * that the correct alternative is executed. Don't change the order
         * of the next 4 instructions! */
        "tst    pc, #0x0c     \n"
        "mov    r0, #0xca     \n"
        "strne  r0, [%[ctl]]  \n"
        "streq  r0, [%[ctl]]  \n"
        "nop                  \n" /* nop's needed because of pipeline */
        "nop                  \n"
        "nop                  \n"
        :
        : [ctl]"r"(&PROC_CTL(CURRENT_CORE))
        : "r0"
    );
    enable_irq();
}
#else
/* PP5002 has no mailboxes - emulate using bytes */
static inline void core_sleep(unsigned int core)
{
#if 1
    asm volatile (
        "mov    r0, #1                     \n" /* Signal intent to sleep */
        "strb   r0, [%[sem], #2]           \n"
        "ldrb   r0, [%[sem], #1]           \n" /* && stay_awake == 0? */
        "cmp    r0, #0                     \n"
        "bne    2f                         \n"
        /* Sleep: PP5002 crashes if the instruction that puts it to sleep is
         * located at 0xNNNNNNN0. 4/8/C works. This sequence makes sure
         * that the correct alternative is executed. Don't change the order
         * of the next 4 instructions! */
        "tst    pc, #0x0c                  \n"
        "mov    r0, #0xca                  \n"
        "strne  r0, [%[ctl], %[c], lsl #2] \n"
        "streq  r0, [%[ctl], %[c], lsl #2] \n"
        "nop                               \n" /* nop's needed because of pipeline */
        "nop                               \n"
        "nop                               \n"
    "2:                                    \n"
        "mov    r0, #0                     \n" /* Clear stay_awake and sleep intent */
        "strb   r0, [%[sem], #1]           \n"
        "strb   r0, [%[sem], #2]           \n"
    "1:                                    \n" /* Wait for wake procedure to finish */
        "ldrb   r0, [%[sem], #0]           \n"
        "cmp    r0, #0                     \n"
        "bne    1b                         \n"
        :
        : [sem]"r"(&core_semaphores[core]), [c]"r"(core),
          [ctl]"r"(&PROC_CTL(CPU))
        : "r0"
        );
#else /* C version for reference */
    /* Signal intent to sleep */
    core_semaphores[core].intend_sleep = 1;

    /* Something waking or other processor intends to wake us? */
    if (core_semaphores[core].stay_awake == 0)
    {
        PROC_CTL(core) = PROC_SLEEP; /* Snooze */
        nop; nop; nop;
    }

    /* Signal wake - clear wake flag */
    core_semaphores[core].stay_awake = 0;
    core_semaphores[core].intend_sleep = 0;

    /* Wait for other processor to finish wake procedure */
    while (core_semaphores[core].intend_wake != 0);

    /* Enable IRQ */
#endif /* ASM/C selection */
    enable_irq();
}
#endif /* NUM_CORES */
#endif /* PP CPU type */

/*---------------------------------------------------------------------------
 * Wake another processor core that is sleeping or prevent it from doing so
 * if it was already destined. FIQ, IRQ should be disabled before calling.
 *---------------------------------------------------------------------------
 */
#if NUM_CORES == 1
/* Shared single-core build debugging version */
void core_wake(void)
{
    /* No wakey - core already wakey */
}
#elif defined (CPU_PP502x)
void core_wake(unsigned int othercore)
{
#if 1
    /* avoid r0 since that contains othercore */
    asm volatile (
        "mrs    r3, cpsr                    \n" /* Disable IRQ */
        "orr    r1, r3, #0x80               \n"
        "msr    cpsr_c, r1                  \n"
        "mov    r2, #0x11                   \n" /* r2 = (0x11 << othercore) */
        "mov    r2, r2, lsl %[oc]           \n" /* Signal intent to wake othercore */
        "str    r2, [%[mbx], #4]            \n"
    "1:                                     \n" /* If it intends to sleep, let it first */
        "ldr    r1, [%[mbx], #0]            \n" /* (MSG_MSG_STAT & (0x4 << othercore)) != 0 ? */
        "eor    r1, r1, #0xc                \n"
        "tst    r1, r2, lsr #2              \n"
        "ldr    r1, [%[ctl], %[oc], lsl #2] \n" /* && (PROC_CTL(othercore) & PROC_SLEEP) == 0 ? */
        "tsteq  r1, #0x80000000             \n"
        "beq    1b                          \n" /* Wait for sleep or wake */
        "tst    r1, #0x80000000             \n" /* If sleeping, wake it */
        "movne  r1, #0x0                    \n"
        "strne  r1, [%[ctl], %[oc], lsl #2] \n"
        "mov    r1, r2, lsr #4              \n"
        "str    r1, [%[mbx], #8]            \n" /* Done with wake procedure */
        "msr    cpsr_c, r3                  \n" /* Restore IRQ */
        :
        : [ctl]"r"(&PROC_CTL(CPU)), [mbx]"r"(MBX_BASE),
          [oc]"r"(othercore)
        : "r1", "r2", "r3");
#else /* C version for reference */
    /* Disable interrupts - avoid reentrancy from the tick */
    int oldlevel = disable_irq_save();

    /* Signal intent to wake other processor - set stay awake */
    MBX_MSG_SET = 0x11 << othercore;

    /* If it intends to sleep, wait until it does or aborts */
    while ((MBX_MSG_STAT & (0x4 << othercore)) != 0 &&
           (PROC_CTL(othercore) & PROC_SLEEP) == 0);

    /* If sleeping, wake it up */
    if (PROC_CTL(othercore) & PROC_SLEEP)
        PROC_CTL(othercore) = 0;

    /* Done with wake procedure */
    MBX_MSG_CLR = 0x1 << othercore;
    restore_irq(oldlevel);
#endif /* ASM/C selection */
}
#elif CONFIG_CPU == PP5002
/* PP5002 has no mailboxes - emulate using bytes */
void core_wake(unsigned int othercore)
{
#if 1
    /* avoid r0 since that contains othercore */
    asm volatile (
        "mrs    r3, cpsr                \n" /* Disable IRQ */
        "orr    r1, r3, #0x80           \n"
        "msr    cpsr_c, r1              \n"
        "mov    r1, #1                  \n" /* Signal intent to wake other core */
        "orr    r1, r1, r1, lsl #8      \n" /* and set stay_awake */
        "strh   r1, [%[sem], #0]        \n"
        "mov    r2, #0x8000             \n"
    "1:                                 \n" /* If it intends to sleep, let it first */
        "ldrb   r1, [%[sem], #2]        \n" /* intend_sleep != 0 ? */
        "cmp    r1, #1                  \n"
        "ldr    r1, [%[st]]             \n" /* && not sleeping ? */
        "tsteq  r1, r2, lsr %[oc]       \n"
        "beq    1b                      \n" /* Wait for sleep or wake */
        "tst    r1, r2, lsr %[oc]       \n"
        "ldrne  r2, =0xcf004054         \n" /* If sleeping, wake it */
        "movne  r1, #0xce               \n"
        "strne  r1, [r2, %[oc], lsl #2] \n"
        "mov    r1, #0                  \n" /* Done with wake procedure */
        "strb   r1, [%[sem], #0]        \n"
        "msr    cpsr_c, r3              \n" /* Restore IRQ */
        :
        : [sem]"r"(&core_semaphores[othercore]),
          [st]"r"(&PROC_STAT),
          [oc]"r"(othercore)
        : "r1", "r2", "r3"
    );
#else /* C version for reference */
    /* Disable interrupts - avoid reentrancy from the tick */
    int oldlevel = disable_irq_save();

    /* Signal intent to wake other processor - set stay awake */
    core_semaphores[othercore].intend_wake = 1;
    core_semaphores[othercore].stay_awake = 1;

    /* If it intends to sleep, wait until it does or aborts */
    while (core_semaphores[othercore].intend_sleep != 0 &&
           (PROC_STAT & PROC_SLEEPING(othercore)) == 0);

    /* If sleeping, wake it up */
    if (PROC_STAT & PROC_SLEEPING(othercore))
        PROC_CTL(othercore) = PROC_WAKE;

    /* Done with wake procedure */
    core_semaphores[othercore].intend_wake = 0;
    restore_irq(oldlevel);
#endif  /* ASM/C selection */
}
#endif /* CPU type */

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Switches to a stack that always resides in the Rockbox core.
 *
 * Needed when a thread suicides on a core other than the main CPU since the
 * stack used when idling is the stack of the last thread to run. This stack
 * may not reside in the core firmware in which case the core will continue
 * to use a stack from an unloaded module until another thread runs on it.
 *---------------------------------------------------------------------------
 */
static inline void switch_to_idle_stack(const unsigned int core)
{
    asm volatile (
        "str  sp, [%0] \n" /* save original stack pointer on idle stack */
        "mov  sp, %0   \n" /* switch stacks */
        : : "r"(&idle_stacks[core][IDLE_STACK_WORDS-1]));
    (void)core;
}

/*---------------------------------------------------------------------------
 * Perform core switch steps that need to take place inside switch_thread.
 *
 * These steps must take place while before changing the processor and after
 * having entered switch_thread since switch_thread may not do a normal return
 * because the stack being used for anything the compiler saved will not belong
 * to the thread's destination core and it may have been recycled for other
 * purposes by the time a normal context load has taken place. switch_thread
 * will also clobber anything stashed in the thread's context or stored in the
 * nonvolatile registers if it is saved there before the call since the
 * compiler's order of operations cannot be known for certain.
 */
static void core_switch_blk_op(unsigned int core, struct thread_entry *thread)
{
    /* Flush our data to ram */
    flush_icache();
    /* Stash thread in r4 slot */
    thread->context.r[0] = (uint32_t)thread;
    /* Stash restart address in r5 slot */
    thread->context.r[1] = thread->context.start;
    /* Save sp in context.sp while still running on old core */
    thread->context.sp = idle_stacks[core][IDLE_STACK_WORDS-1];
}

/*---------------------------------------------------------------------------
 * Machine-specific helper function for switching the processor a thread is
 * running on. Basically, the thread suicides on the departing core and is
 * reborn on the destination. Were it not for gcc's ill-behavior regarding
 * naked functions written in C where it actually clobbers non-volatile
 * registers before the intended prologue code, this would all be much
 * simpler.  Generic setup is done in switch_core itself.
 */

/*---------------------------------------------------------------------------
 * This actually performs the core switch.
 */
static void __attribute__((naked))
    switch_thread_core(unsigned int core, struct thread_entry *thread)
{
    /* Pure asm for this because compiler behavior isn't sufficiently predictable.
     * Stack access also isn't permitted until restoring the original stack and
     * context. */
    asm volatile (
        "stmfd  sp!, { r4-r12, lr }    \n" /* Stack all non-volatile context on current core */
        "ldr    r2, =idle_stacks       \n" /* r2 = &idle_stacks[core][IDLE_STACK_WORDS] */
        "ldr    r2, [r2, r0, lsl #2]   \n"
        "add    r2, r2, %0*4           \n"
        "stmfd  r2!, { sp }            \n" /* save original stack pointer on idle stack */
        "mov    sp, r2                 \n" /* switch stacks */
        "adr    r2, 1f                 \n" /* r2 = new core restart address */
        "str    r2, [r1, #40]          \n" /* thread->context.start = r2 */
        "ldr    pc, =switch_thread     \n" /* r0 = thread after call - see load_context */
    "1:                                \n"
        "ldr    sp, [r0, #32]          \n" /* Reload original sp from context structure */
        "mov    r1, #0                 \n" /* Clear start address */
        "str    r1, [r0, #40]          \n"
        "ldr    r0, =invalidate_icache \n" /* Invalidate new core's cache */
        "mov    lr, pc                 \n"
        "bx     r0                     \n"
        "ldmfd  sp!, { r4-r12, pc }    \n" /* Restore non-volatile context to new core and return */
        ".ltorg                        \n" /* Dump constant pool */
        : : "i"(IDLE_STACK_WORDS)
    );
    (void)core; (void)thread;
}

/*---------------------------------------------------------------------------
 * Do any device-specific inits for the threads and synchronize the kernel
 * initializations.
 *---------------------------------------------------------------------------
 */
static void core_thread_init(unsigned int core)
{
    if (core == CPU)
    {
        /* Wake up coprocessor and let it initialize kernel and threads */
#ifdef CPU_PP502x
        MBX_MSG_CLR = 0x3f;
#endif
        COP_CTL = PROC_WAKE;
        /* Sleep until COP has finished */
        CPU_CTL = PROC_SLEEP;
        nop; nop; nop;
    }
    else
    {
        /* Wake the CPU and return */
        CPU_CTL = PROC_WAKE;
    }
}
#endif /* NUM_CORES */

#elif CONFIG_CPU == S3C2440

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    /* FIQ also changes the CLKCON register so FIQ must be disabled
       when changing it here */
    asm volatile (
        "mrs    r0, cpsr        \n"
        "orr    r2, r0, #0x40   \n" /* Disable FIQ */
        "bic    r0, r0, #0x80   \n" /* Prepare IRQ enable */
        "msr    cpsr_c, r2      \n"
        "mov    r1, #0x4c000000 \n" /* CLKCON = 0x4c00000c */
        "ldr    r2, [r1, #0xc]  \n" /* Set IDLE bit */
        "orr    r2, r2, #4      \n"
        "str    r2, [r1, #0xc]  \n"
        "msr    cpsr_c, r0      \n" /* Enable IRQ, restore FIQ */
        "mov    r2, #0          \n" /* wait for IDLE */
    "1:                         \n"
        "add    r2, r2, #1      \n"
        "cmp    r2, #10         \n"
        "bne    1b              \n"
        "orr    r2, r0, #0xc0   \n" /* Disable IRQ, FIQ */
        "msr    cpsr_c, r2      \n"
        "ldr    r2, [r1, #0xc]  \n" /* Reset IDLE bit */
        "bic    r2, r2, #4      \n"
        "str    r2, [r1, #0xc]  \n"
        "msr    cpsr_c, r0      \n" /* Enable IRQ, restore FIQ */
        :  :  : "r0", "r1", "r2");
}
#elif defined(CPU_TCC77X)
static inline void core_sleep(void)
{
    #warning TODO: Implement core_sleep
    enable_irq();
}
#elif defined(CPU_TCC780X)
static inline void core_sleep(void)
{
    /* Single core only for now. Use the generic ARMv5 wait for IRQ */
    asm volatile (
        "mov r0, #0                \n"
        "mcr p15, 0, r0, c7, c0, 4 \n" /* Wait for interrupt */
        : : : "r0"
    );
    enable_irq();
}
#elif CONFIG_CPU == IMX31L
static inline void core_sleep(void)
{
    asm volatile (
        "mov r0, #0                \n"
        "mcr p15, 0, r0, c7, c0, 4 \n" /* Wait for interrupt */
        : : : "r0"
    );
    enable_irq();
}
#else
static inline void core_sleep(void)
{
    #warning core_sleep not implemented, battery life will be decreased
    enable_irq();
}
#endif /* CONFIG_CPU == */

#elif defined(CPU_COLDFIRE)
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
void start_thread(void); /* Provide C access to ASM label */
static void __attribute__((used)) __start_thread(void)
{
    /* a0=macsr, a1=context */
    asm volatile (
    "start_thread:             \n" /* Start here - no naked attribute */
        "move.l  %a0, %macsr   \n" /* Set initial mac status reg */
        "lea.l   48(%a1), %a1  \n"
        "move.l  (%a1)+, %sp   \n" /* Set initial stack */
        "move.l  (%a1), %a2    \n" /* Fetch thread function pointer */
        "clr.l   (%a1)         \n" /* Mark thread running */
        "jsr     (%a2)         \n" /* Call thread function */
    );
    thread_exit();
}

/* Set EMAC unit to fractional mode with saturation for each new thread,
 * since that's what'll be the most useful for most things which the dsp
 * will do. Codecs should still initialize their preferred modes
 * explicitly. Context pointer is placed in d2 slot and start_thread
 * pointer in d3 slot. thread function pointer is placed in context.start.
 * See load_context for what happens when thread is initially going to
 * run.
 */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.macsr = EMAC_FRACTIONAL | EMAC_SATURATE, \
       (thread)->context.d[0] = (uint32_t)&(thread)->context, \
       (thread)->context.d[1] = (uint32_t)start_thread,       \
       (thread)->context.start = (uint32_t)(function); })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile (
        "move.l  %%macsr,%%d0                  \n"
        "movem.l %%d0/%%d2-%%d7/%%a2-%%a7,(%0) \n"
        : : "a" (addr) : "d0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile (
        "move.l  52(%0), %%d0                   \n"  /* Get start address */
        "beq.b   1f                             \n"  /* NULL -> already running */
        "movem.l (%0), %%a0-%%a2                \n"  /* a0=macsr, a1=context, a2=start_thread */
        "jmp     (%%a2)                         \n"  /* Start the thread */
    "1:                                         \n"
        "movem.l (%0), %%d0/%%d2-%%d7/%%a2-%%a7 \n"  /* Load context */
        "move.l  %%d0, %%macsr                  \n"
        : : "a" (addr) : "d0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    /* Supervisor mode, interrupts enabled upon wakeup */
    asm volatile ("stop #0x2000");
};

#elif CONFIG_CPU == SH7034
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
void start_thread(void); /* Provide C access to ASM label */
static void __attribute__((used)) __start_thread(void)
{
    /* r8 = context */
    asm volatile (
    "_start_thread:            \n" /* Start here - no naked attribute */
        "mov.l  @(4, r8), r0   \n" /* Fetch thread function pointer */
        "mov.l  @(28, r8), r15 \n" /* Set initial sp */
        "mov    #0, r1         \n" /* Start the thread */
        "jsr    @r0            \n"
        "mov.l  r1, @(36, r8)  \n" /* Clear start address */
    );
    thread_exit();
}

/* Place context pointer in r8 slot, function pointer in r9 slot, and
 * start_thread pointer in context_start */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.r[0] = (uint32_t)&(thread)->context, \
       (thread)->context.r[1] = (uint32_t)(function),         \
       (thread)->context.start = (uint32_t)start_thread; })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile (
        "add     #36, %0   \n" /* Start at last reg. By the time routine */
        "sts.l   pr, @-%0  \n" /* is done, %0 will have the original value */
        "mov.l   r15,@-%0  \n"
        "mov.l   r14,@-%0  \n"
        "mov.l   r13,@-%0  \n"
        "mov.l   r12,@-%0  \n"
        "mov.l   r11,@-%0  \n"
        "mov.l   r10,@-%0  \n"
        "mov.l   r9, @-%0  \n"
        "mov.l   r8, @-%0  \n"
        : : "r" (addr)
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile (
        "mov.l  @(36, %0), r0 \n" /* Get start address */
        "tst    r0, r0        \n"
        "bt     .running      \n" /* NULL -> already running */
        "jmp    @r0           \n" /* r8 = context */
    ".running:                \n"
        "mov.l  @%0+, r8      \n" /* Executes in delay slot and outside it */
        "mov.l  @%0+, r9      \n"
        "mov.l  @%0+, r10     \n"
        "mov.l  @%0+, r11     \n"
        "mov.l  @%0+, r12     \n"
        "mov.l  @%0+, r13     \n"
        "mov.l  @%0+, r14     \n"
        "mov.l  @%0+, r15     \n"
        "lds.l  @%0+, pr      \n"
        : : "r" (addr) : "r0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Put core in a power-saving state.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    asm volatile (
        "and.b  #0x7f, @(r0, gbr) \n" /* Clear SBY (bit 7) in SBYCR */
        "mov    #0, r1            \n" /* Enable interrupts */
        "ldc    r1, sr            \n" /* Following instruction cannot be interrupted */
        "sleep                    \n" /* Execute standby */
        : : "z"(&SBYCR-GBR) : "r1");
}

#endif /* CONFIG_CPU == */

/*
 * End Processor-specific section
 ***************************************************************************/

#if THREAD_EXTRA_CHECKS
static void thread_panicf(const char *msg, struct thread_entry *thread)
{
    IF_COP( const unsigned int core = thread->core; )
    static char name[32];
    thread_get_name(name, 32, thread);
    panicf ("%s %s" IF_COP(" (%d)"), msg, name IF_COP(, core));
}
static void thread_stkov(struct thread_entry *thread)
{
    thread_panicf("Stkov", thread);
}
#define THREAD_PANICF(msg, thread) \
    thread_panicf(msg, thread)
#define THREAD_ASSERT(exp, msg, thread) \
    ({ if (!({ exp; })) thread_panicf((msg), (thread)); })
#else
static void thread_stkov(struct thread_entry *thread)
{
    IF_COP( const unsigned int core = thread->core; )
    static char name[32];
    thread_get_name(name, 32, thread);
    panicf("Stkov %s" IF_COP(" (%d)"), name IF_COP(, core));
}
#define THREAD_PANICF(msg, thread)
#define THREAD_ASSERT(exp, msg, thread)
#endif /* THREAD_EXTRA_CHECKS */

/* Thread locking */
#if NUM_CORES > 1
#define LOCK_THREAD(thread) \
    ({ corelock_lock(&(thread)->slot_cl); })
#define TRY_LOCK_THREAD(thread) \
    ({ corelock_try_lock(&thread->slot_cl); })
#define UNLOCK_THREAD(thread) \
    ({ corelock_unlock(&(thread)->slot_cl); })
#define UNLOCK_THREAD_AT_TASK_SWITCH(thread) \
    ({ unsigned int _core = (thread)->core; \
       cores[_core].blk_ops.flags |= TBOP_UNLOCK_CORELOCK; \
       cores[_core].blk_ops.cl_p = &(thread)->slot_cl; })
#else
#define LOCK_THREAD(thread) \
    ({ })
#define TRY_LOCK_THREAD(thread) \
    ({ })
#define UNLOCK_THREAD(thread) \
    ({ })
#define UNLOCK_THREAD_AT_TASK_SWITCH(thread) \
    ({ })
#endif

/* RTR list */
#define RTR_LOCK(core) \
    ({ corelock_lock(&cores[core].rtr_cl); })
#define RTR_UNLOCK(core) \
    ({ corelock_unlock(&cores[core].rtr_cl); })

#ifdef HAVE_PRIORITY_SCHEDULING
#define rtr_add_entry(core, priority) \
    prio_add_entry(&cores[core].rtr, (priority))

#define rtr_subtract_entry(core, priority) \
    prio_subtract_entry(&cores[core].rtr, (priority))

#define rtr_move_entry(core, from, to) \
    prio_move_entry(&cores[core].rtr, (from), (to))
#else
#define rtr_add_entry(core, priority)
#define rtr_add_entry_inl(core, priority)
#define rtr_subtract_entry(core, priority)
#define rtr_subtract_entry_inl(core, priotity)
#define rtr_move_entry(core, from, to)
#define rtr_move_entry_inl(core, from, to)
#endif

/*---------------------------------------------------------------------------
 * Thread list structure - circular:
 *    +------------------------------+
 *    |                              |
 *    +--+---+<-+---+<-+---+<-+---+<-+
 * Head->| T |  | T |  | T |  | T |
 *    +->+---+->+---+->+---+->+---+--+
 *    |                              |
 *    +------------------------------+
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Adds a thread to a list of threads using "insert last". Uses the "l"
 * links.
 *---------------------------------------------------------------------------
 */
static void add_to_list_l(struct thread_entry **list,
                          struct thread_entry *thread)
{
    struct thread_entry *l = *list;

    if (l == NULL)
    {
        /* Insert into unoccupied list */
        thread->l.prev = thread;
        thread->l.next = thread;
        *list = thread;
        return;
    }

    /* Insert last */
    thread->l.prev = l->l.prev;
    thread->l.next = l;
    l->l.prev->l.next = thread;
    l->l.prev = thread;
}

/*---------------------------------------------------------------------------
 * Removes a thread from a list of threads. Uses the "l" links.
 *---------------------------------------------------------------------------
 */
static void remove_from_list_l(struct thread_entry **list,
                               struct thread_entry *thread)
{
    struct thread_entry *prev, *next;

    next = thread->l.next;

    if (thread == next)
    {
        /* The only item */
        *list = NULL;
        return;
    }

    if (thread == *list)
    {
        /* List becomes next item */
        *list = next;
    }

    prev = thread->l.prev;
    
    /* Fix links to jump over the removed entry. */
    next->l.prev = prev;
    prev->l.next = next;
}

/*---------------------------------------------------------------------------
 * Timeout list structure - circular reverse (to make "remove item" O(1)),
 * NULL-terminated forward (to ease the far more common forward traversal):
 *    +------------------------------+
 *    |                              |
 *    +--+---+<-+---+<-+---+<-+---+<-+
 * Head->| T |  | T |  | T |  | T |
 *       +---+->+---+->+---+->+---+-X
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Add a thread from the core's timout list by linking the pointers in its
 * tmo structure.
 *---------------------------------------------------------------------------
 */
static void add_to_list_tmo(struct thread_entry *thread)
{
    struct thread_entry *tmo = cores[IF_COP_CORE(thread->core)].timeout;
    THREAD_ASSERT(thread->tmo.prev == NULL,
                  "add_to_list_tmo->already listed", thread);

    thread->tmo.next = NULL;

    if (tmo == NULL)
    {
        /* Insert into unoccupied list */
        thread->tmo.prev = thread;
        cores[IF_COP_CORE(thread->core)].timeout = thread;
        return;
    }

    /* Insert Last */
    thread->tmo.prev = tmo->tmo.prev;
    tmo->tmo.prev->tmo.next = thread;
    tmo->tmo.prev = thread;
}

/*---------------------------------------------------------------------------
 * Remove a thread from the core's timout list by unlinking the pointers in
 * its tmo structure. Sets thread->tmo.prev to NULL to indicate the timeout
 * is cancelled.
 *---------------------------------------------------------------------------
 */
static void remove_from_list_tmo(struct thread_entry *thread)
{
    struct thread_entry **list = &cores[IF_COP_CORE(thread->core)].timeout;
    struct thread_entry *prev = thread->tmo.prev;
    struct thread_entry *next = thread->tmo.next;

    THREAD_ASSERT(prev != NULL, "remove_from_list_tmo->not listed", thread);

    if (next != NULL)
        next->tmo.prev = prev;

    if (thread == *list)
    {
        /* List becomes next item and empty if next == NULL */
        *list = next;
        /* Mark as unlisted */
        thread->tmo.prev = NULL;
    }
    else
    {
        if (next == NULL)
            (*list)->tmo.prev = prev;
        prev->tmo.next = next;
        /* Mark as unlisted */
        thread->tmo.prev = NULL;
    }
}


#ifdef HAVE_PRIORITY_SCHEDULING
/*---------------------------------------------------------------------------
 * Priority distribution structure (one category for each possible priority):
 *
 *       +----+----+----+ ... +-----+
 * hist: | F0 | F1 | F2 |     | F31 |
 *       +----+----+----+ ... +-----+
 * mask: | b0 | b1 | b2 |     | b31 |
 *       +----+----+----+ ... +-----+
 *
 * F = count of threads at priority category n (frequency)
 * b = bitmask of non-zero priority categories (occupancy)
 *
 *        / if H[n] != 0 : 1
 * b[n] = |
 *        \ else         : 0 
 *
 *---------------------------------------------------------------------------
 * Basic priority inheritance priotocol (PIP):
 *
 * Mn = mutex n, Tn = thread n
 *
 * A lower priority thread inherits the priority of the highest priority
 * thread blocked waiting for it to complete an action (such as release a
 * mutex or respond to a message via queue_send):
 *
 * 1) T2->M1->T1
 *
 * T1 owns M1, T2 is waiting for M1 to realease M1. If T2 has a higher
 * priority than T1 then T1 inherits the priority of T2.
 *
 * 2) T3
 *    \/
 *    T2->M1->T1
 *
 * Situation is like 1) but T2 and T3 are both queued waiting for M1 and so
 * T1 inherits the higher of T2 and T3.
 *
 * 3) T3->M2->T2->M1->T1
 *
 * T1 owns M1, T2 owns M2. If T3 has a higher priority than both T1 and T2,
 * then T1 inherits the priority of T3 through T2.
 *
 * Blocking chains can grow arbitrarily complex (though it's best that they
 * not form at all very often :) and build-up from these units.
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Increment frequency at category "priority"
 *---------------------------------------------------------------------------
 */
static inline unsigned int prio_add_entry(
    struct priority_distribution *pd, int priority)
{
    unsigned int count;
    /* Enough size/instruction count difference for ARM makes it worth it to
     * use different code (192 bytes for ARM). Only thing better is ASM. */
#ifdef CPU_ARM
    count = pd->hist[priority];
    if (++count == 1)
        pd->mask |= 1 << priority;
    pd->hist[priority] = count;
#else /* This one's better for Coldfire */
    if ((count = ++pd->hist[priority]) == 1)
        pd->mask |= 1 << priority;
#endif

    return count;
}

/*---------------------------------------------------------------------------
 * Decrement frequency at category "priority"
 *---------------------------------------------------------------------------
 */
static inline unsigned int prio_subtract_entry(
    struct priority_distribution *pd, int priority)
{
    unsigned int count;

#ifdef CPU_ARM
    count = pd->hist[priority];
    if (--count == 0)
        pd->mask &= ~(1 << priority);
    pd->hist[priority] = count;
#else
    if ((count = --pd->hist[priority]) == 0)
        pd->mask &= ~(1 << priority);
#endif

    return count;
}

/*---------------------------------------------------------------------------
 * Remove from one category and add to another
 *---------------------------------------------------------------------------
 */
static inline void prio_move_entry(
    struct priority_distribution *pd, int from, int to)
{
    uint32_t mask = pd->mask;

#ifdef CPU_ARM
    unsigned int count;

    count = pd->hist[from];
    if (--count == 0)
        mask &= ~(1 << from);
    pd->hist[from] = count;

    count = pd->hist[to];
    if (++count == 1)
        mask |= 1 << to;
    pd->hist[to] = count;
#else
    if (--pd->hist[from] == 0)
        mask &= ~(1 << from);

    if (++pd->hist[to] == 1)
        mask |= 1 << to;
#endif

    pd->mask = mask;
}

/*---------------------------------------------------------------------------
 * Change the priority and rtr entry for a running thread
 *---------------------------------------------------------------------------
 */
static inline void set_running_thread_priority(
    struct thread_entry *thread, int priority)
{
    const unsigned int core = IF_COP_CORE(thread->core);
    RTR_LOCK(core);
    rtr_move_entry(core, thread->priority, priority);
    thread->priority = priority;
    RTR_UNLOCK(core);
}

/*---------------------------------------------------------------------------
 * Finds the highest priority thread in a list of threads. If the list is
 * empty, the PRIORITY_IDLE is returned.
 *
 * It is possible to use the struct priority_distribution within an object
 * instead of scanning the remaining threads in the list but as a compromise,
 * the resulting per-object memory overhead is saved at a slight speed
 * penalty under high contention.
 *---------------------------------------------------------------------------
 */
static int find_highest_priority_in_list_l(
    struct thread_entry * const thread)
{
    if (thread != NULL)
    {
        /* Go though list until the ending up at the initial thread */
        int highest_priority = thread->priority;
        struct thread_entry *curr = thread;

        do
        {
            int priority = curr->priority;

            if (priority < highest_priority)
                highest_priority = priority;

            curr = curr->l.next;
        }
        while (curr != thread);

        return highest_priority;
    }

    return PRIORITY_IDLE;
}

/*---------------------------------------------------------------------------
 * Register priority with blocking system and bubble it down the chain if
 * any until we reach the end or something is already equal or higher.
 *
 * NOTE: A simultaneous circular wait could spin deadlock on multiprocessor
 * targets but that same action also guarantees a circular block anyway and
 * those are prevented, right? :-)
 *---------------------------------------------------------------------------
 */
static struct thread_entry *
    blocker_inherit_priority(struct thread_entry *current)
{
    const int priority = current->priority;
    struct blocker *bl = current->blocker;
    struct thread_entry * const tstart = current;
    struct thread_entry *bl_t = bl->thread;

    /* Blocker cannot change since the object protection is held */
    LOCK_THREAD(bl_t);

    for (;;)
    {
        struct thread_entry *next;
        int bl_pr = bl->priority;

        if (priority >= bl_pr)
            break; /* Object priority already high enough */

        bl->priority = priority;

        /* Add this one */
        prio_add_entry(&bl_t->pdist, priority);

        if (bl_pr < PRIORITY_IDLE)
        {
            /* Not first waiter - subtract old one */
            prio_subtract_entry(&bl_t->pdist, bl_pr);
        }

        if (priority >= bl_t->priority)
            break; /* Thread priority high enough */

        if (bl_t->state == STATE_RUNNING)
        {
            /* Blocking thread is a running thread therefore there are no
             * further blockers. Change the "run queue" on which it
             * resides. */
            set_running_thread_priority(bl_t, priority);
            break;
        }

        bl_t->priority = priority;

        /* If blocking thread has a blocker, apply transitive inheritance */
        bl = bl_t->blocker;

        if (bl == NULL)
            break; /* End of chain or object doesn't support inheritance */

        next = bl->thread;

        if (next == tstart)
            break; /* Full-circle - deadlock! */

        UNLOCK_THREAD(current);

#if NUM_CORES > 1
        for (;;)
        {
            LOCK_THREAD(next);

            /* Blocker could change - retest condition */
            if (bl->thread == next)
                break;

            UNLOCK_THREAD(next);
            next = bl->thread;
        }
#endif
        current = bl_t;
        bl_t = next;
    }

    UNLOCK_THREAD(bl_t);

    return current;
}

/*---------------------------------------------------------------------------
 * Readjust priorities when waking a thread blocked waiting for another
 * in essence "releasing" the thread's effect on the object owner. Can be
 * performed from any context.
 *---------------------------------------------------------------------------
 */
struct thread_entry *
    wakeup_priority_protocol_release(struct thread_entry *thread)
{
    const int priority = thread->priority;
    struct blocker *bl = thread->blocker;
    struct thread_entry * const tstart = thread;
    struct thread_entry *bl_t = bl->thread;

    /* Blocker cannot change since object will be locked */
    LOCK_THREAD(bl_t);

    thread->blocker = NULL; /* Thread not blocked */

    for (;;)
    {
        struct thread_entry *next;
        int bl_pr = bl->priority;

        if (priority > bl_pr)
            break; /* Object priority higher */

        next = *thread->bqp;

        if (next == NULL)
        {
            /* No more threads in queue */
            prio_subtract_entry(&bl_t->pdist, bl_pr);
            bl->priority = PRIORITY_IDLE;
        }
        else
        {
            /* Check list for highest remaining priority */
            int queue_pr = find_highest_priority_in_list_l(next);

            if (queue_pr == bl_pr)
                break; /* Object priority not changing */

            /* Change queue priority */
            prio_move_entry(&bl_t->pdist, bl_pr, queue_pr);
            bl->priority = queue_pr;
        }

        if (bl_pr > bl_t->priority)
            break; /* thread priority is higher */

        bl_pr = find_first_set_bit(bl_t->pdist.mask);

        if (bl_pr == bl_t->priority)
            break; /* Thread priority not changing */

        if (bl_t->state == STATE_RUNNING)
        {
            /* No further blockers */
            set_running_thread_priority(bl_t, bl_pr);
            break;
        }

        bl_t->priority = bl_pr;

        /* If blocking thread has a blocker, apply transitive inheritance */
        bl = bl_t->blocker;

        if (bl == NULL)
            break; /* End of chain or object doesn't support inheritance */

        next = bl->thread;

        if (next == tstart)
            break; /* Full-circle - deadlock! */

        UNLOCK_THREAD(thread);

#if NUM_CORES > 1
        for (;;)
        {
            LOCK_THREAD(next);

            /* Blocker could change - retest condition */
            if (bl->thread == next)
                break;

            UNLOCK_THREAD(next);
            next = bl->thread;
        }
#endif
        thread = bl_t;
        bl_t = next;
    }

    UNLOCK_THREAD(bl_t);

#if NUM_CORES > 1
    if (thread != tstart)
    {
        /* Relock original if it changed */
        LOCK_THREAD(tstart);
    }
#endif

    return cores[CURRENT_CORE].running;
}

/*---------------------------------------------------------------------------
 * Transfer ownership to a thread waiting for an objects and transfer
 * inherited priority boost from other waiters. This algorithm knows that
 * blocking chains may only unblock from the very end.
 *
 * Only the owning thread itself may call this and so the assumption that
 * it is the running thread is made.
 *---------------------------------------------------------------------------
 */
struct thread_entry *
    wakeup_priority_protocol_transfer(struct thread_entry *thread)
{
    /* Waking thread inherits priority boost from object owner */
    struct blocker *bl = thread->blocker;
    struct thread_entry *bl_t = bl->thread;
    struct thread_entry *next;
    int bl_pr;

    THREAD_ASSERT(thread_get_current() == bl_t,
                  "UPPT->wrong thread", thread_get_current());

    LOCK_THREAD(bl_t);

    bl_pr = bl->priority;

    /* Remove the object's boost from the owning thread */
    if (prio_subtract_entry(&bl_t->pdist, bl_pr) == 0 &&
        bl_pr <= bl_t->priority)
    {
        /* No more threads at this priority are waiting and the old level is
         * at least the thread level */
        int priority = find_first_set_bit(bl_t->pdist.mask);

        if (priority != bl_t->priority)
        {
            /* Adjust this thread's priority */
            set_running_thread_priority(bl_t, priority);
        }
    }

    next = *thread->bqp;

    if (next == NULL)
    {
        /* Expected shortcut - no more waiters */
        bl_pr = PRIORITY_IDLE;
    }
    else
    {
        if (thread->priority <= bl_pr)
        {
            /* Need to scan threads remaining in queue */
            bl_pr = find_highest_priority_in_list_l(next);
        }

        if (prio_add_entry(&thread->pdist, bl_pr) == 1 &&
            bl_pr < thread->priority)
        {
            /* Thread priority must be raised */
            thread->priority = bl_pr;
        }
    }

    bl->thread = thread;    /* This thread pwns */
    bl->priority = bl_pr;   /* Save highest blocked priority */
    thread->blocker = NULL; /* Thread not blocked */

    UNLOCK_THREAD(bl_t);

    return bl_t;
}

/*---------------------------------------------------------------------------
 * No threads must be blocked waiting for this thread except for it to exit.
 * The alternative is more elaborate cleanup and object registration code.
 * Check this for risk of silent data corruption when objects with
 * inheritable blocking are abandoned by the owner - not precise but may
 * catch something.
 *---------------------------------------------------------------------------
 */
void check_for_obj_waiters(const char *function, struct thread_entry *thread)
{
    /* Only one bit in the mask should be set with a frequency on 1 which
     * represents the thread's own base priority */
    uint32_t mask = thread->pdist.mask;
    if ((mask & (mask - 1)) != 0 ||
        thread->pdist.hist[find_first_set_bit(mask)] > 1)
    {
        unsigned char name[32];
        thread_get_name(name, 32, thread);
        panicf("%s->%s with obj. waiters", function, name);
    }
}
#endif /* HAVE_PRIORITY_SCHEDULING */

/*---------------------------------------------------------------------------
 * Move a thread back to a running state on its core.
 *---------------------------------------------------------------------------
 */
static void core_schedule_wakeup(struct thread_entry *thread)
{
    const unsigned int core = IF_COP_CORE(thread->core);

    RTR_LOCK(core);

    thread->state = STATE_RUNNING;

    add_to_list_l(&cores[core].running, thread);
    rtr_add_entry(core, thread->priority);

    RTR_UNLOCK(core);

#if NUM_CORES > 1
    if (core != CURRENT_CORE)
        core_wake(core);
#endif
}

/*---------------------------------------------------------------------------
 * Check the core's timeout list when at least one thread is due to wake.
 * Filtering for the condition is done before making the call. Resets the
 * tick when the next check will occur.
 *---------------------------------------------------------------------------
 */
void check_tmo_threads(void)
{
    const unsigned int core = CURRENT_CORE;
    const long tick = current_tick; /* snapshot the current tick */
    long next_tmo_check = tick + 60*HZ; /* minimum duration: once/minute */
    struct thread_entry *next = cores[core].timeout;

    /* If there are no processes waiting for a timeout, just keep the check
       tick from falling into the past. */

    /* Break the loop once we have walked through the list of all
     * sleeping processes or have removed them all. */
    while (next != NULL)
    {
        /* Check sleeping threads. Allow interrupts between checks. */
        enable_irq();

        struct thread_entry *curr = next;

        next = curr->tmo.next;

        /* Lock thread slot against explicit wakeup */
        disable_irq();
        LOCK_THREAD(curr);

        unsigned state = curr->state;

        if (state < TIMEOUT_STATE_FIRST)
        {
            /* Cleanup threads no longer on a timeout but still on the
             * list. */
            remove_from_list_tmo(curr);
        }
        else if (TIME_BEFORE(tick, curr->tmo_tick))
        {
            /* Timeout still pending - this will be the usual case */
            if (TIME_BEFORE(curr->tmo_tick, next_tmo_check))
            {
                /* Earliest timeout found so far - move the next check up
                   to its time */
                next_tmo_check = curr->tmo_tick;
            }
        }
        else
        {
            /* Sleep timeout has been reached so bring the thread back to
             * life again. */
            if (state == STATE_BLOCKED_W_TMO)
            {
#if NUM_CORES > 1
                /* Lock the waiting thread's kernel object */
                struct corelock *ocl = curr->obj_cl;

                if (corelock_try_lock(ocl) == 0)
                {
                    /* Need to retry in the correct order though the need is
                     * unlikely */
                    UNLOCK_THREAD(curr);
                    corelock_lock(ocl);
                    LOCK_THREAD(curr);

                    if (curr->state != STATE_BLOCKED_W_TMO)
                    {
                        /* Thread was woken or removed explicitely while slot
                         * was unlocked */
                        corelock_unlock(ocl);
                        remove_from_list_tmo(curr);
                        UNLOCK_THREAD(curr);
                        continue;
                    }
                }
#endif /* NUM_CORES */

                remove_from_list_l(curr->bqp, curr);

#ifdef HAVE_WAKEUP_EXT_CB
                if (curr->wakeup_ext_cb != NULL)
                    curr->wakeup_ext_cb(curr);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
                if (curr->blocker != NULL)
                    wakeup_priority_protocol_release(curr);
#endif
                corelock_unlock(ocl);
            }
            /* else state == STATE_SLEEPING */

            remove_from_list_tmo(curr);

            RTR_LOCK(core);

            curr->state = STATE_RUNNING;

            add_to_list_l(&cores[core].running, curr);
            rtr_add_entry(core, curr->priority);

            RTR_UNLOCK(core);
        }

        UNLOCK_THREAD(curr);
    }

    cores[core].next_tmo_check = next_tmo_check;
}

/*---------------------------------------------------------------------------
 * Performs operations that must be done before blocking a thread but after
 * the state is saved.
 *---------------------------------------------------------------------------
 */
#if NUM_CORES > 1
static inline void run_blocking_ops(
    unsigned int core, struct thread_entry *thread)
{
    struct thread_blk_ops *ops = &cores[core].blk_ops;
    const unsigned flags = ops->flags;

    if (flags == TBOP_CLEAR)
        return;

    switch (flags)
    {
    case TBOP_SWITCH_CORE:
        core_switch_blk_op(core, thread);
        /* Fall-through */
    case TBOP_UNLOCK_CORELOCK:
        corelock_unlock(ops->cl_p);
        break;
    }

    ops->flags = TBOP_CLEAR;
}
#endif /* NUM_CORES > 1 */

#ifdef RB_PROFILE
void profile_thread(void)
{
    profstart(cores[CURRENT_CORE].running - threads);
}
#endif

/*---------------------------------------------------------------------------
 * Prepares a thread to block on an object's list and/or for a specified
 * duration - expects object and slot to be appropriately locked if needed
 * and interrupts to be masked.
 *---------------------------------------------------------------------------
 */
static inline void block_thread_on_l(struct thread_entry *thread,
                                     unsigned state)
{
    /* If inlined, unreachable branches will be pruned with no size penalty
       because state is passed as a constant parameter. */
    const unsigned int core = IF_COP_CORE(thread->core);

    /* Remove the thread from the list of running threads. */
    RTR_LOCK(core);
    remove_from_list_l(&cores[core].running, thread);
    rtr_subtract_entry(core, thread->priority);
    RTR_UNLOCK(core);

    /* Add a timeout to the block if not infinite */
    switch (state)
    {
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        /* Put the thread into a new list of inactive threads. */
        add_to_list_l(thread->bqp, thread);

        if (state == STATE_BLOCKED)
            break;

        /* Fall-through */
    case STATE_SLEEPING:
        /* If this thread times out sooner than any other thread, update
           next_tmo_check to its timeout */
        if (TIME_BEFORE(thread->tmo_tick, cores[core].next_tmo_check))
        {
            cores[core].next_tmo_check = thread->tmo_tick;
        }

        if (thread->tmo.prev == NULL)
        {
            add_to_list_tmo(thread);
        }
        /* else thread was never removed from list - just keep it there */
        break;
    }

    /* Remember the the next thread about to block. */
    cores[core].block_task = thread;

    /* Report new state. */
    thread->state = state;
}

/*---------------------------------------------------------------------------
 * Switch thread in round robin fashion for any given priority. Any thread
 * that removed itself from the running list first must specify itself in
 * the paramter.
 *
 * INTERNAL: Intended for use by kernel and not for programs.
 *---------------------------------------------------------------------------
 */
void switch_thread(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *block = cores[core].block_task;
    struct thread_entry *thread = cores[core].running;

    /* Get context to save - next thread to run is unknown until all wakeups
     * are evaluated */
    if (block != NULL)
    {
        cores[core].block_task = NULL;

#if NUM_CORES > 1
        if (thread == block)
        {
            /* This was the last thread running and another core woke us before
             * reaching here. Force next thread selection to give tmo threads or
             * other threads woken before this block a first chance. */
            block = NULL;
        }
        else
#endif
        {
            /* Blocking task is the old one */
            thread = block;
        }
    }

#ifdef RB_PROFILE
    profile_thread_stopped(thread - threads);
#endif

    /* Begin task switching by saving our current context so that we can
     * restore the state of the current thread later to the point prior
     * to this call. */
    store_context(&thread->context);

    /* Check if the current thread stack is overflown */
    if (thread->stack[0] != DEADBEEF)
        thread_stkov(thread);

#if NUM_CORES > 1
    /* Run any blocking operations requested before switching/sleeping */
    run_blocking_ops(core, thread);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
    IF_NO_SKIP_YIELD( if (thread->skip_count != -1) )
    /* Reset the value of thread's skip count */
        thread->skip_count = 0;
#endif

    for (;;)
    {
        /* If there are threads on a timeout and the earliest wakeup is due,
         * check the list and wake any threads that need to start running
         * again. */
        if (!TIME_BEFORE(current_tick, cores[core].next_tmo_check))
        {
            check_tmo_threads();
        }

        disable_irq();
        RTR_LOCK(core);

        thread = cores[core].running;

        if (thread == NULL)
        {
            /* Enter sleep mode to reduce power usage - woken up on interrupt
             * or wakeup request from another core - expected to enable
             * interrupts. */
            RTR_UNLOCK(core);
            core_sleep(IF_COP(core));
        }
        else
        {
#ifdef HAVE_PRIORITY_SCHEDULING
            /* Select the new task based on priorities and the last time a
             * process got CPU time relative to the highest priority runnable
             * task. */
            struct priority_distribution *pd = &cores[core].rtr;
            int max = find_first_set_bit(pd->mask);

            if (block == NULL)
            {
                /* Not switching on a block, tentatively select next thread */
                thread = thread->l.next;
            }

            for (;;)
            {
                int priority = thread->priority;
                int diff;

                /* This ridiculously simple method of aging seems to work
                 * suspiciously well. It does tend to reward CPU hogs (under
                 * yielding) but that's generally not desirable at all. On the
                 * plus side, it, relatively to other threads, penalizes excess
                 * yielding which is good if some high priority thread is
                 * performing no useful work such as polling for a device to be
                 * ready. Of course, aging is only employed when higher and lower
                 * priority threads are runnable. The highest priority runnable
                 * thread(s) are never skipped. */
                if (priority <= max ||
                    IF_NO_SKIP_YIELD( thread->skip_count == -1 || )
                    (diff = priority - max, ++thread->skip_count > diff*diff))
                {
                    cores[core].running = thread;
                    break;
                }

                thread = thread->l.next;
            }
#else
            /* Without priority use a simple FCFS algorithm */
            if (block == NULL)
            {
                /* Not switching on a block, select next thread */
                thread = thread->l.next;
                cores[core].running = thread;
            }
#endif /* HAVE_PRIORITY_SCHEDULING */

            RTR_UNLOCK(core);
            enable_irq();
            break;
        }
    }

    /* And finally give control to the next thread. */
    load_context(&thread->context);

#ifdef RB_PROFILE
    profile_thread_started(thread - threads);
#endif
}

/*---------------------------------------------------------------------------
 * Sleeps a thread for at least a specified number of ticks with zero being
 * a wait until the next tick.
 *
 * INTERNAL: Intended for use by kernel and not for programs.
 *---------------------------------------------------------------------------
 */
void sleep_thread(int ticks)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;

    LOCK_THREAD(current);

    /* Set our timeout, remove from run list and join timeout list. */
    current->tmo_tick = current_tick + ticks + 1;
    block_thread_on_l(current, STATE_SLEEPING);

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Indefinitely block a thread on a blocking queue for explicit wakeup.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
void block_thread(struct thread_entry *current)
{
    /* Set the state to blocked and take us off of the run queue until we
     * are explicitly woken */
    LOCK_THREAD(current);

    /* Set the list for explicit wakeup */
    block_thread_on_l(current, STATE_BLOCKED);

#ifdef HAVE_PRIORITY_SCHEDULING
    if (current->blocker != NULL)
    {
        /* Object supports PIP */
        current = blocker_inherit_priority(current);
    }
#endif

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Block a thread on a blocking queue for a specified time interval or until
 * explicitly woken - whichever happens first.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
void block_thread_w_tmo(struct thread_entry *current, int timeout)
{
    /* Get the entry for the current running thread. */
    LOCK_THREAD(current);

    /* Set the state to blocked with the specified timeout */
    current->tmo_tick = current_tick + timeout;

    /* Set the list for explicit wakeup */
    block_thread_on_l(current, STATE_BLOCKED_W_TMO);

#ifdef HAVE_PRIORITY_SCHEDULING
    if (current->blocker != NULL)
    {
        /* Object supports PIP */
        current = blocker_inherit_priority(current);
    }
#endif

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Explicitly wakeup a thread on a blocking queue. Only effects threads of
 * STATE_BLOCKED and STATE_BLOCKED_W_TMO.
 *
 * This code should be considered a critical section by the caller meaning
 * that the object's corelock should be held.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
unsigned int wakeup_thread(struct thread_entry **list)
{
    struct thread_entry *thread = *list;
    unsigned int result = THREAD_NONE;

    /* Check if there is a blocked thread at all. */
    if (thread == NULL)
        return result;

    LOCK_THREAD(thread);

    /* Determine thread's current state. */
    switch (thread->state)
    {
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        remove_from_list_l(list, thread);

        result = THREAD_OK;

#ifdef HAVE_PRIORITY_SCHEDULING
        struct thread_entry *current;
        struct blocker *bl = thread->blocker;

        if (bl == NULL)
        {
            /* No inheritance - just boost the thread by aging */
            IF_NO_SKIP_YIELD( if (thread->skip_count != -1) )
                thread->skip_count = thread->priority;
            current = cores[CURRENT_CORE].running;
        }
        else
        {
            /* Call the specified unblocking PIP */
            current = bl->wakeup_protocol(thread);
        }

        if (current != NULL && thread->priority < current->priority
            IF_COP( && thread->core == current->core ))
        {
            /* Woken thread is higher priority and exists on the same CPU core;
             * recommend a task switch. Knowing if this is an interrupt call
             * would be helpful here. */
            result |= THREAD_SWITCH;
        }
#endif /* HAVE_PRIORITY_SCHEDULING */

        core_schedule_wakeup(thread);
        break;

    /* Nothing to do. State is not blocked. */
#if THREAD_EXTRA_CHECKS
    default:
        THREAD_PANICF("wakeup_thread->block invalid", thread);
    case STATE_RUNNING:
    case STATE_KILLED:
        break;
#endif
    }

    UNLOCK_THREAD(thread);
    return result;
}

/*---------------------------------------------------------------------------
 * Wakeup an entire queue of threads - returns bitwise-or of return bitmask
 * from each operation or THREAD_NONE of nothing was awakened. Object owning
 * the queue must be locked first.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
unsigned int thread_queue_wake(struct thread_entry **list)
{
    unsigned result = THREAD_NONE;

    for (;;)
    {
        unsigned int rc = wakeup_thread(list);

        if (rc == THREAD_NONE)
            break; /* No more threads */

        result |= rc;
    }

    return result;
}

/*---------------------------------------------------------------------------
 * Find an empty thread slot or MAXTHREADS if none found. The slot returned
 * will be locked on multicore.
 *---------------------------------------------------------------------------
 */
static struct thread_entry * find_empty_thread_slot(void)
{
    /* Any slot could be on an interrupt-accessible list */
    IF_COP( int oldlevel = disable_irq_save(); )
    struct thread_entry *thread = NULL;
    int n;

    for (n = 0; n < MAXTHREADS; n++)
    {
        /* Obtain current slot state - lock it on multicore */
        struct thread_entry *t = &threads[n];
        LOCK_THREAD(t);

        if (t->state == STATE_KILLED IF_COP( && t->name != THREAD_DESTRUCT ))
        {
            /* Slot is empty - leave it locked and caller will unlock */
            thread = t;
            break;
        }

        /* Finished examining slot - no longer busy - unlock on multicore */
        UNLOCK_THREAD(t);
    }

    IF_COP( restore_irq(oldlevel); ) /* Reenable interrups - this slot is
                                          not accesible to them yet */
    return thread;
}


/*---------------------------------------------------------------------------
 * Place the current core in idle mode - woken up on interrupt or wake
 * request from another core.
 *---------------------------------------------------------------------------
 */
void core_idle(void)
{
    IF_COP( const unsigned int core = CURRENT_CORE; )
    disable_irq();
    core_sleep(IF_COP(core));
}

/*---------------------------------------------------------------------------
 * Create a thread. If using a dual core architecture, specify which core to
 * start the thread on.
 *
 * Return ID if context area could be allocated, else NULL.
 *---------------------------------------------------------------------------
 */
struct thread_entry* 
    create_thread(void (*function)(void), void* stack, size_t stack_size,
                  unsigned flags, const char *name
                  IF_PRIO(, int priority)
                  IF_COP(, unsigned int core))
{
    unsigned int i;
    unsigned int stack_words;
    uintptr_t stackptr, stackend;
    struct thread_entry *thread;
    unsigned state;
    int oldlevel;

    thread = find_empty_thread_slot();
    if (thread == NULL)
    {
        return NULL;
    }

    oldlevel = disable_irq_save();

    /* Munge the stack to make it easy to spot stack overflows */
    stackptr = ALIGN_UP((uintptr_t)stack, sizeof (uintptr_t));
    stackend = ALIGN_DOWN((uintptr_t)stack + stack_size, sizeof (uintptr_t));
    stack_size = stackend - stackptr;
    stack_words = stack_size / sizeof (uintptr_t);

    for (i = 0; i < stack_words; i++)
    {
        ((uintptr_t *)stackptr)[i] = DEADBEEF;
    }

    /* Store interesting information */
    thread->name = name;
    thread->stack = (uintptr_t *)stackptr;
    thread->stack_size = stack_size;
    thread->queue = NULL;
#ifdef HAVE_WAKEUP_EXT_CB
    thread->wakeup_ext_cb = NULL;
#endif
#ifdef HAVE_SCHEDULER_BOOSTCTRL
    thread->cpu_boost = 0;
#endif
#ifdef HAVE_PRIORITY_SCHEDULING
    memset(&thread->pdist, 0, sizeof(thread->pdist));
    thread->blocker = NULL;
    thread->base_priority = priority;
    thread->priority = priority;
    thread->skip_count = priority;
    prio_add_entry(&thread->pdist, priority);
#endif

#if NUM_CORES > 1
    thread->core = core;

    /* Writeback stack munging or anything else before starting */
    if (core != CURRENT_CORE)
    {
        flush_icache();
    }
#endif

    /* Thread is not on any timeout list but be a bit paranoid */
    thread->tmo.prev = NULL;

    state = (flags & CREATE_THREAD_FROZEN) ?
        STATE_FROZEN : STATE_RUNNING;
    
    thread->context.sp = (typeof (thread->context.sp))stackend;

    /* Load the thread's context structure with needed startup information */
    THREAD_STARTUP_INIT(core, thread, function);

    thread->state = state;

    if (state == STATE_RUNNING)
        core_schedule_wakeup(thread);

    UNLOCK_THREAD(thread);

    restore_irq(oldlevel);

    return thread;
}

#ifdef HAVE_SCHEDULER_BOOSTCTRL
/*---------------------------------------------------------------------------
 * Change the boost state of a thread boosting or unboosting the CPU
 * as required.
 *---------------------------------------------------------------------------
 */
static inline void boost_thread(struct thread_entry *thread, bool boost)
{
    if ((thread->cpu_boost != 0) != boost)
    {
        thread->cpu_boost = boost;
        cpu_boost(boost);
    }
}

void trigger_cpu_boost(void)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;
    boost_thread(current, true);
}

void cancel_cpu_boost(void)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;
    boost_thread(current, false);
}
#endif /* HAVE_SCHEDULER_BOOSTCTRL */

/*---------------------------------------------------------------------------
 * Block the current thread until another thread terminates. A thread may
 * wait on itself to terminate which prevents it from running again and it
 * will need to be killed externally.
 * Parameter is the ID as returned from create_thread().
 *---------------------------------------------------------------------------
 */
void thread_wait(struct thread_entry *thread)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;

    if (thread == NULL)
        thread = current;

    /* Lock thread-as-waitable-object lock */
    corelock_lock(&thread->waiter_cl);

    /* Be sure it hasn't been killed yet */
    if (thread->state != STATE_KILLED)
    {
        IF_COP( current->obj_cl = &thread->waiter_cl; )
        current->bqp = &thread->queue;

        disable_irq();
        block_thread(current);

        corelock_unlock(&thread->waiter_cl);

        switch_thread();
        return;
    }

    corelock_unlock(&thread->waiter_cl);
}

/*---------------------------------------------------------------------------
 * Exit the current thread. The Right Way to Do Things (TM).
 *---------------------------------------------------------------------------
 */
void thread_exit(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *current = cores[core].running;

    /* Cancel CPU boost if any */
    cancel_cpu_boost();

    disable_irq();

    corelock_lock(&current->waiter_cl);
    LOCK_THREAD(current);

#if defined (ALLOW_REMOVE_THREAD) && NUM_CORES > 1
    if (current->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - become a waiter */
        UNLOCK_THREAD(current);
        corelock_unlock(&current->waiter_cl);
        thread_wait(current);
        THREAD_PANICF("thread_exit->WK:*R", current);
    }
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
    check_for_obj_waiters("thread_exit", current);
#endif

    if (current->tmo.prev != NULL)
    {
        /* Cancel pending timeout list removal */
        remove_from_list_tmo(current);
    }

    /* Switch tasks and never return */
    block_thread_on_l(current, STATE_KILLED);

#if NUM_CORES > 1
    /* Switch to the idle stack if not on the main core (where "main"
     * runs) - we can hope gcc doesn't need the old stack beyond this
     * point. */
    if (core != CPU)
    {
        switch_to_idle_stack(core);
    }

    flush_icache();
#endif
    current->name = NULL;

    /* Signal this thread */
    thread_queue_wake(&current->queue);
    corelock_unlock(&current->waiter_cl);
    /* Slot must be unusable until thread is really gone */
    UNLOCK_THREAD_AT_TASK_SWITCH(current);
    switch_thread();
    /* This should never and must never be reached - if it is, the
     * state is corrupted */
    THREAD_PANICF("thread_exit->K:*R", current);
}

#ifdef ALLOW_REMOVE_THREAD
/*---------------------------------------------------------------------------
 * Remove a thread from the scheduler. Not The Right Way to Do Things in
 * normal programs.
 *
 * Parameter is the ID as returned from create_thread().
 *
 * Use with care on threads that are not under careful control as this may
 * leave various objects in an undefined state.
 *---------------------------------------------------------------------------
 */
void remove_thread(struct thread_entry *thread)
{
#if NUM_CORES > 1
    /* core is not constant here because of core switching */
    unsigned int core = CURRENT_CORE;
    unsigned int old_core = NUM_CORES;
    struct corelock *ocl = NULL;
#else
    const unsigned int core = CURRENT_CORE;
#endif
    struct thread_entry *current = cores[core].running;

    unsigned state;
    int oldlevel;

    if (thread == NULL)
        thread = current;

    if (thread == current)
        thread_exit(); /* Current thread - do normal exit */

    oldlevel = disable_irq_save();

    corelock_lock(&thread->waiter_cl);
    LOCK_THREAD(thread);

    state = thread->state;

    if (state == STATE_KILLED)
    {
        goto thread_killed;
    }

#if NUM_CORES > 1
    if (thread->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - become a waiter */
        UNLOCK_THREAD(thread);
        corelock_unlock(&thread->waiter_cl);
        restore_irq(oldlevel);
        thread_wait(thread);
        return;
    }

    thread->name = THREAD_DESTRUCT; /* Slot can't be used for now */

#ifdef HAVE_PRIORITY_SCHEDULING
    check_for_obj_waiters("remove_thread", thread);
#endif

    if (thread->core != core)
    {
        /* Switch cores and safely extract the thread there */
        /* Slot HAS to be unlocked or a deadlock could occur which means other
         * threads have to be guided into becoming thread waiters if they
         * attempt to remove it. */
        unsigned int new_core = thread->core;

        corelock_unlock(&thread->waiter_cl);

        UNLOCK_THREAD(thread);
        restore_irq(oldlevel);

        old_core = switch_core(new_core);

        oldlevel = disable_irq_save();

        corelock_lock(&thread->waiter_cl);
        LOCK_THREAD(thread);

        state = thread->state;
        core = new_core;
        /* Perform the extraction and switch ourselves back to the original
           processor */
    }
#endif /* NUM_CORES > 1 */

    if (thread->tmo.prev != NULL)
    {
        /* Clean thread off the timeout list if a timeout check hasn't
         * run yet */
        remove_from_list_tmo(thread);
    }

#ifdef HAVE_SCHEDULER_BOOSTCTRL
    /* Cancel CPU boost if any */
    boost_thread(thread, false);
#endif

IF_COP( retry_state: )

    switch (state)
    {
    case STATE_RUNNING:
        RTR_LOCK(core);
        /* Remove thread from ready to run tasks */
        remove_from_list_l(&cores[core].running, thread);
        rtr_subtract_entry(core, thread->priority);
        RTR_UNLOCK(core);
        break;
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        /* Remove thread from the queue it's blocked on - including its
         * own if waiting there */
#if NUM_CORES > 1
        if (&thread->waiter_cl != thread->obj_cl)
        {
            ocl = thread->obj_cl;

            if (corelock_try_lock(ocl) == 0)
            {
                UNLOCK_THREAD(thread);
                corelock_lock(ocl);
                LOCK_THREAD(thread);

                if (thread->state != state)
                {
                    /* Something woke the thread */
                    state = thread->state;
                    corelock_unlock(ocl);
                    goto retry_state;
                }
            }
        }
#endif
        remove_from_list_l(thread->bqp, thread);

#ifdef HAVE_WAKEUP_EXT_CB
        if (thread->wakeup_ext_cb != NULL)
            thread->wakeup_ext_cb(thread);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
        if (thread->blocker != NULL)
        {
            /* Remove thread's priority influence from its chain */
            wakeup_priority_protocol_release(thread);
        }
#endif

#if NUM_CORES > 1
        if (ocl != NULL)
            corelock_unlock(ocl);
#endif
        break;
    /* Otherwise thread is frozen and hasn't run yet */
    }

    thread->state = STATE_KILLED;

    /* If thread was waiting on itself, it will have been removed above.
     * The wrong order would result in waking the thread first and deadlocking
     * since the slot is already locked. */
    thread_queue_wake(&thread->queue);

    thread->name = NULL;

thread_killed: /* Thread was already killed */
    /* Removal complete - safe to unlock and reenable interrupts */
    corelock_unlock(&thread->waiter_cl);
    UNLOCK_THREAD(thread);
    restore_irq(oldlevel);

#if NUM_CORES > 1
    if (old_core < NUM_CORES)
    {
        /* Did a removal on another processor's thread - switch back to
           native core */
        switch_core(old_core);
    }
#endif
}
#endif /* ALLOW_REMOVE_THREAD */

#ifdef HAVE_PRIORITY_SCHEDULING
/*---------------------------------------------------------------------------
 * Sets the thread's relative base priority for the core it runs on. Any
 * needed inheritance changes also may happen.
 *---------------------------------------------------------------------------
 */
int thread_set_priority(struct thread_entry *thread, int priority)
{
    int old_base_priority = -1;

    /* A little safety measure */
    if (priority < HIGHEST_PRIORITY || priority > LOWEST_PRIORITY)
        return -1;

    if (thread == NULL)
        thread = cores[CURRENT_CORE].running;

    /* Thread could be on any list and therefore on an interrupt accessible
       one - disable interrupts */
    int oldlevel = disable_irq_save();

    LOCK_THREAD(thread);

    /* Make sure it's not killed */
    if (thread->state != STATE_KILLED)
    {
        int old_priority = thread->priority;

        old_base_priority = thread->base_priority;
        thread->base_priority = priority;

        prio_move_entry(&thread->pdist, old_base_priority, priority);
        priority = find_first_set_bit(thread->pdist.mask);

        if (old_priority == priority)
        {
            /* No priority change - do nothing */
        }
        else if (thread->state == STATE_RUNNING)
        {
            /* This thread is running - change location on the run
             * queue. No transitive inheritance needed. */
            set_running_thread_priority(thread, priority);
        }
        else
        {
            thread->priority = priority;

            if (thread->blocker != NULL)
            {
                /* Bubble new priority down the chain */
                struct blocker *bl = thread->blocker;   /* Blocker struct */
                struct thread_entry *bl_t = bl->thread; /* Blocking thread */
                struct thread_entry * const tstart = thread;   /* Initial thread */
                const int highest = MIN(priority, old_priority); /* Higher of new or old */

                for (;;)
                {
                    struct thread_entry *next; /* Next thread to check */
                    int bl_pr;    /* Highest blocked thread */
                    int queue_pr; /* New highest blocked thread */
#if NUM_CORES > 1
                    /* Owner can change but thread cannot be dislodged - thread
                     * may not be the first in the queue which allows other
                     * threads ahead in the list to be given ownership during the
                     * operation. If thread is next then the waker will have to
                     * wait for us and the owner of the object will remain fixed.
                     * If we successfully grab the owner -- which at some point
                     * is guaranteed -- then the queue remains fixed until we
                     * pass by. */
                    for (;;)
                    {
                        LOCK_THREAD(bl_t);

                        /* Double-check the owner - retry if it changed */
                        if (bl->thread == bl_t)
                            break;

                        UNLOCK_THREAD(bl_t);
                        bl_t = bl->thread;
                    }
#endif
                    bl_pr = bl->priority;

                    if (highest > bl_pr)
                        break; /* Object priority won't change */

                    /* This will include the thread being set */
                    queue_pr = find_highest_priority_in_list_l(*thread->bqp);

                    if (queue_pr == bl_pr)
                        break; /* Object priority not changing */

                    /* Update thread boost for this object */
                    bl->priority = queue_pr;
                    prio_move_entry(&bl_t->pdist, bl_pr, queue_pr);
                    bl_pr = find_first_set_bit(bl_t->pdist.mask);

                    if (bl_t->priority == bl_pr)
                        break; /* Blocking thread priority not changing */

                    if (bl_t->state == STATE_RUNNING)
                    {
                        /* Thread not blocked - we're done */
                        set_running_thread_priority(bl_t, bl_pr);
                        break;
                    }

                    bl_t->priority = bl_pr;
                    bl = bl_t->blocker; /* Blocking thread has a blocker? */

                    if (bl == NULL)
                        break; /* End of chain */

                    next = bl->thread;

                    if (next == tstart)
                        break; /* Full-circle */

                    UNLOCK_THREAD(thread);

                    thread = bl_t;
                    bl_t = next;
                } /* for (;;) */

                UNLOCK_THREAD(bl_t);
            }
        }
    }

    UNLOCK_THREAD(thread);

    restore_irq(oldlevel);

    return old_base_priority;
}

/*---------------------------------------------------------------------------
 * Returns the current base priority for a thread.
 *---------------------------------------------------------------------------
 */
int thread_get_priority(struct thread_entry *thread)
{
    /* Simple, quick probe. */
    if (thread == NULL)
        thread = cores[CURRENT_CORE].running;

    return thread->base_priority;
}
#endif /* HAVE_PRIORITY_SCHEDULING */

/*---------------------------------------------------------------------------
 * Starts a frozen thread - similar semantics to wakeup_thread except that
 * the thread is on no scheduler or wakeup queue at all. It exists simply by
 * virtue of the slot having a state of STATE_FROZEN.
 *---------------------------------------------------------------------------
 */
void thread_thaw(struct thread_entry *thread)
{
    int oldlevel = disable_irq_save();
    LOCK_THREAD(thread);

    if (thread->state == STATE_FROZEN)
        core_schedule_wakeup(thread);

    UNLOCK_THREAD(thread);
    restore_irq(oldlevel);
}

/*---------------------------------------------------------------------------
 * Return the ID of the currently executing thread.
 *---------------------------------------------------------------------------
 */
struct thread_entry * thread_get_current(void)
{
    return cores[CURRENT_CORE].running;
}

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Switch the processor that the currently executing thread runs on.
 *---------------------------------------------------------------------------
 */
unsigned int switch_core(unsigned int new_core)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *current = cores[core].running;

    if (core == new_core)
    {
        /* No change - just return same core */
        return core;
    }

    int oldlevel = disable_irq_save();
    LOCK_THREAD(current);

    if (current->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - deactivate and let process complete */
        UNLOCK_THREAD(current);
        restore_irq(oldlevel);
        thread_wait(current);
        /* Should never be reached */
        THREAD_PANICF("switch_core->D:*R", current);
    }

    /* Get us off the running list for the current core */
    RTR_LOCK(core);
    remove_from_list_l(&cores[core].running, current);
    rtr_subtract_entry(core, current->priority);
    RTR_UNLOCK(core);

    /* Stash return value (old core) in a safe place */
    current->retval = core;

    /* If a timeout hadn't yet been cleaned-up it must be removed now or
     * the other core will likely attempt a removal from the wrong list! */
    if (current->tmo.prev != NULL)
    {
        remove_from_list_tmo(current);
    }

    /* Change the core number for this thread slot */
    current->core = new_core;

    /* Do not use core_schedule_wakeup here since this will result in
     * the thread starting to run on the other core before being finished on
     * this one. Delay the  list unlock to keep the other core stuck
     * until this thread is ready. */
    RTR_LOCK(new_core);

    rtr_add_entry(new_core, current->priority);
    add_to_list_l(&cores[new_core].running, current);

    /* Make a callback into device-specific code, unlock the wakeup list so
     * that execution may resume on the new core, unlock our slot and finally
     * restore the interrupt level */
    cores[core].blk_ops.flags = TBOP_SWITCH_CORE;
    cores[core].blk_ops.cl_p  = &cores[new_core].rtr_cl;
    cores[core].block_task    = current;

    UNLOCK_THREAD(current);

    /* Alert other core to activity */
    core_wake(new_core);

    /* Do the stack switching, cache_maintenence and switch_thread call -
       requires native code */
    switch_thread_core(core, current);

    /* Finally return the old core to caller */
    return current->retval;
}
#endif /* NUM_CORES > 1 */

/*---------------------------------------------------------------------------
 * Initialize threading API. This assumes interrupts are not yet enabled. On
 * multicore setups, no core is allowed to proceed until create_thread calls
 * are safe to perform.
 *---------------------------------------------------------------------------
 */
void init_threads(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *thread;

    /* CPU will initialize first and then sleep */
    thread = find_empty_thread_slot();

    if (thread == NULL)
    {
        /* WTF? There really must be a slot available at this stage.
         * This can fail if, for example, .bss isn't zero'ed out by the loader
         * or threads is in the wrong section. */
        THREAD_PANICF("init_threads->no slot", NULL);
    }

    /* Initialize initially non-zero members of core */
    cores[core].next_tmo_check = current_tick; /* Something not in the past */

    /* Initialize initially non-zero members of slot */
    UNLOCK_THREAD(thread); /* No sync worries yet */
    thread->name = main_thread_name;
    thread->state = STATE_RUNNING;
    IF_COP( thread->core = core; )
#ifdef HAVE_PRIORITY_SCHEDULING
    corelock_init(&cores[core].rtr_cl);
    thread->base_priority = PRIORITY_USER_INTERFACE;
    prio_add_entry(&thread->pdist, PRIORITY_USER_INTERFACE);
    thread->priority = PRIORITY_USER_INTERFACE;
    rtr_add_entry(core, PRIORITY_USER_INTERFACE);
#endif
    corelock_init(&thread->waiter_cl);
    corelock_init(&thread->slot_cl);

    add_to_list_l(&cores[core].running, thread);

    if (core == CPU)
    {
        thread->stack = stackbegin;
        thread->stack_size = (uintptr_t)stackend - (uintptr_t)stackbegin;
#if NUM_CORES > 1  /* This code path will not be run on single core targets */
        /* Wait for other processors to finish their inits since create_thread
         * isn't safe to call until the kernel inits are done. The first
         * threads created in the system must of course be created by CPU. */
        core_thread_init(CPU);
    } 
    else
    {
        /* Initial stack is the idle stack */
        thread->stack = idle_stacks[core];
        thread->stack_size = IDLE_STACK_SIZE;
        /* After last processor completes, it should signal all others to
         * proceed or may signal the next and call thread_exit(). The last one
         * to finish will signal CPU. */
        core_thread_init(core);
        /* Other cores do not have a main thread - go idle inside switch_thread
         * until a thread can run on the core. */
        thread_exit();
#endif /* NUM_CORES */
    }
}

/* Shared stack scan helper for thread_stack_usage and idle_stack_usage */
#if NUM_CORES == 1
static inline int stack_usage(uintptr_t *stackptr, size_t stack_size)
#else
static int stack_usage(uintptr_t *stackptr, size_t stack_size)
#endif
{
    unsigned int stack_words = stack_size / sizeof (uintptr_t);
    unsigned int i;
    int usage = 0;

    for (i = 0; i < stack_words; i++)
    {
        if (stackptr[i] != DEADBEEF)
        {
            usage = ((stack_words - i) * 100) / stack_words;
            break;
        }
    }

    return usage;
}

/*---------------------------------------------------------------------------
 * Returns the maximum percentage of stack a thread ever used while running.
 * NOTE: Some large buffer allocations that don't use enough the buffer to
 * overwrite stackptr[0] will not be seen.
 *---------------------------------------------------------------------------
 */
int thread_stack_usage(const struct thread_entry *thread)
{
    return stack_usage(thread->stack, thread->stack_size);
}

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Returns the maximum percentage of the core's idle stack ever used during
 * runtime.
 *---------------------------------------------------------------------------
 */
int idle_stack_usage(unsigned int core)
{
    return stack_usage(idle_stacks[core], IDLE_STACK_SIZE);
}
#endif

/*---------------------------------------------------------------------------
 * Fills in the buffer with the specified thread's name. If the name is NULL,
 * empty, or the thread is in destruct state a formatted ID is written
 * instead.
 *---------------------------------------------------------------------------
 */
void thread_get_name(char *buffer, int size,
                     struct thread_entry *thread)
{
    if (size <= 0)
        return;

    *buffer = '\0';

    if (thread)
    {
        /* Display thread name if one or ID if none */
        const char *name = thread->name;
        const char *fmt = "%s";
        if (name == NULL IF_COP(|| name == THREAD_DESTRUCT) || *name == '\0')
        {
            name = (const char *)thread;
            fmt = "%08lX";
        }
        snprintf(buffer, size, fmt, name);
    }
}