Loading...
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 | // SPDX-License-Identifier: GPL-2.0-or-later /* * Performance event support - powerpc architecture code * * Copyright 2008-2009 Paul Mackerras, IBM Corporation. */ #include <linux/kernel.h> #include <linux/sched.h> #include <linux/sched/clock.h> #include <linux/perf_event.h> #include <linux/percpu.h> #include <linux/hardirq.h> #include <linux/uaccess.h> #include <asm/reg.h> #include <asm/pmc.h> #include <asm/machdep.h> #include <asm/firmware.h> #include <asm/ptrace.h> #include <asm/code-patching.h> #include <asm/hw_irq.h> #include <asm/interrupt.h> #ifdef CONFIG_PPC64 #include "internal.h" #endif #define BHRB_MAX_ENTRIES 32 #define BHRB_TARGET 0x0000000000000002 #define BHRB_PREDICTION 0x0000000000000001 #define BHRB_EA 0xFFFFFFFFFFFFFFFCUL struct cpu_hw_events { int n_events; int n_percpu; int disabled; int n_added; int n_limited; u8 pmcs_enabled; struct perf_event *event[MAX_HWEVENTS]; u64 events[MAX_HWEVENTS]; unsigned int flags[MAX_HWEVENTS]; struct mmcr_regs mmcr; struct perf_event *limited_counter[MAX_LIMITED_HWCOUNTERS]; u8 limited_hwidx[MAX_LIMITED_HWCOUNTERS]; u64 alternatives[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES]; unsigned long amasks[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES]; unsigned long avalues[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES]; unsigned int txn_flags; int n_txn_start; /* BHRB bits */ u64 bhrb_filter; /* BHRB HW branch filter */ unsigned int bhrb_users; void *bhrb_context; struct perf_branch_stack bhrb_stack; struct perf_branch_entry bhrb_entries[BHRB_MAX_ENTRIES]; u64 ic_init; /* Store the PMC values */ unsigned long pmcs[MAX_HWEVENTS]; }; static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events); static struct power_pmu *ppmu; /* * Normally, to ignore kernel events we set the FCS (freeze counters * in supervisor mode) bit in MMCR0, but if the kernel runs with the * hypervisor bit set in the MSR, or if we are running on a processor * where the hypervisor bit is forced to 1 (as on Apple G5 processors), * then we need to use the FCHV bit to ignore kernel events. */ static unsigned int freeze_events_kernel = MMCR0_FCS; /* * 32-bit doesn't have MMCRA but does have an MMCR2, * and a few other names are different. * Also 32-bit doesn't have MMCR3, SIER2 and SIER3. * Define them as zero knowing that any code path accessing * these registers (via mtspr/mfspr) are done under ppmu flag * check for PPMU_ARCH_31 and we will not enter that code path * for 32-bit. */ #ifdef CONFIG_PPC32 #define MMCR0_FCHV 0 #define MMCR0_PMCjCE MMCR0_PMCnCE #define MMCR0_FC56 0 #define MMCR0_PMAO 0 #define MMCR0_EBE 0 #define MMCR0_BHRBA 0 #define MMCR0_PMCC 0 #define MMCR0_PMCC_U6 0 #define SPRN_MMCRA SPRN_MMCR2 #define SPRN_MMCR3 0 #define SPRN_SIER2 0 #define SPRN_SIER3 0 #define MMCRA_SAMPLE_ENABLE 0 #define MMCRA_BHRB_DISABLE 0 #define MMCR0_PMCCEXT 0 static inline unsigned long perf_ip_adjust(struct pt_regs *regs) { return 0; } static inline void perf_get_data_addr(struct perf_event *event, struct pt_regs *regs, u64 *addrp) { } static inline u32 perf_get_misc_flags(struct pt_regs *regs) { return 0; } static inline void perf_read_regs(struct pt_regs *regs) { regs->result = 0; } static inline int siar_valid(struct pt_regs *regs) { return 1; } static bool is_ebb_event(struct perf_event *event) { return false; } static int ebb_event_check(struct perf_event *event) { return 0; } static void ebb_event_add(struct perf_event *event) { } static void ebb_switch_out(unsigned long mmcr0) { } static unsigned long ebb_switch_in(bool ebb, struct cpu_hw_events *cpuhw) { return cpuhw->mmcr.mmcr0; } static inline void power_pmu_bhrb_enable(struct perf_event *event) {} static inline void power_pmu_bhrb_disable(struct perf_event *event) {} static void power_pmu_sched_task(struct perf_event_context *ctx, bool sched_in) {} static inline void power_pmu_bhrb_read(struct perf_event *event, struct cpu_hw_events *cpuhw) {} static void pmao_restore_workaround(bool ebb) { } #endif /* CONFIG_PPC32 */ bool is_sier_available(void) { if (!ppmu) return false; if (ppmu->flags & PPMU_HAS_SIER) return true; return false; } /* * Return PMC value corresponding to the * index passed. */ unsigned long get_pmcs_ext_regs(int idx) { struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); return cpuhw->pmcs[idx]; } static bool regs_use_siar(struct pt_regs *regs) { /* * When we take a performance monitor exception the regs are setup * using perf_read_regs() which overloads some fields, in particular * regs->result to tell us whether to use SIAR. * * However if the regs are from another exception, eg. a syscall, then * they have not been setup using perf_read_regs() and so regs->result * is something random. */ return ((TRAP(regs) == INTERRUPT_PERFMON) && regs->result); } /* * Things that are specific to 64-bit implementations. */ #ifdef CONFIG_PPC64 static inline unsigned long perf_ip_adjust(struct pt_regs *regs) { unsigned long mmcra = regs->dsisr; if ((ppmu->flags & PPMU_HAS_SSLOT) && (mmcra & MMCRA_SAMPLE_ENABLE)) { unsigned long slot = (mmcra & MMCRA_SLOT) >> MMCRA_SLOT_SHIFT; if (slot > 1) return 4 * (slot - 1); } return 0; } /* * The user wants a data address recorded. * If we're not doing instruction sampling, give them the SDAR * (sampled data address). If we are doing instruction sampling, then * only give them the SDAR if it corresponds to the instruction * pointed to by SIAR; this is indicated by the [POWER6_]MMCRA_SDSYNC, the * [POWER7P_]MMCRA_SDAR_VALID bit in MMCRA, or the SDAR_VALID bit in SIER. */ static inline void perf_get_data_addr(struct perf_event *event, struct pt_regs *regs, u64 *addrp) { unsigned long mmcra = regs->dsisr; bool sdar_valid; if (ppmu->flags & PPMU_HAS_SIER) sdar_valid = regs->dar & SIER_SDAR_VALID; else { unsigned long sdsync; if (ppmu->flags & PPMU_SIAR_VALID) sdsync = POWER7P_MMCRA_SDAR_VALID; else if (ppmu->flags & PPMU_ALT_SIPR) sdsync = POWER6_MMCRA_SDSYNC; else if (ppmu->flags & PPMU_NO_SIAR) sdsync = MMCRA_SAMPLE_ENABLE; else sdsync = MMCRA_SDSYNC; sdar_valid = mmcra & sdsync; } if (!(mmcra & MMCRA_SAMPLE_ENABLE) || sdar_valid) *addrp = mfspr(SPRN_SDAR); if (is_kernel_addr(mfspr(SPRN_SDAR)) && event->attr.exclude_kernel) *addrp = 0; } static bool regs_sihv(struct pt_regs *regs) { unsigned long sihv = MMCRA_SIHV; if (ppmu->flags & PPMU_HAS_SIER) return !!(regs->dar & SIER_SIHV); if (ppmu->flags & PPMU_ALT_SIPR) sihv = POWER6_MMCRA_SIHV; return !!(regs->dsisr & sihv); } static bool regs_sipr(struct pt_regs *regs) { unsigned long sipr = MMCRA_SIPR; if (ppmu->flags & PPMU_HAS_SIER) return !!(regs->dar & SIER_SIPR); if (ppmu->flags & PPMU_ALT_SIPR) sipr = POWER6_MMCRA_SIPR; return !!(regs->dsisr & sipr); } static inline u32 perf_flags_from_msr(struct pt_regs *regs) { if (regs->msr & MSR_PR) return PERF_RECORD_MISC_USER; if ((regs->msr & MSR_HV) && freeze_events_kernel != MMCR0_FCHV) return PERF_RECORD_MISC_HYPERVISOR; return PERF_RECORD_MISC_KERNEL; } static inline u32 perf_get_misc_flags(struct pt_regs *regs) { bool use_siar = regs_use_siar(regs); unsigned long mmcra = regs->dsisr; int marked = mmcra & MMCRA_SAMPLE_ENABLE; if (!use_siar) return perf_flags_from_msr(regs); /* * Check the address in SIAR to identify the * privilege levels since the SIER[MSR_HV, MSR_PR] * bits are not set for marked events in power10 * DD1. */ if (marked && (ppmu->flags & PPMU_P10_DD1)) { unsigned long siar = mfspr(SPRN_SIAR); if (siar) { if (is_kernel_addr(siar)) return PERF_RECORD_MISC_KERNEL; return PERF_RECORD_MISC_USER; } else { if (is_kernel_addr(regs->nip)) return PERF_RECORD_MISC_KERNEL; return PERF_RECORD_MISC_USER; } } /* * If we don't have flags in MMCRA, rather than using * the MSR, we intuit the flags from the address in * SIAR which should give slightly more reliable * results */ if (ppmu->flags & PPMU_NO_SIPR) { unsigned long siar = mfspr(SPRN_SIAR); if (is_kernel_addr(siar)) return PERF_RECORD_MISC_KERNEL; return PERF_RECORD_MISC_USER; } /* PR has priority over HV, so order below is important */ if (regs_sipr(regs)) return PERF_RECORD_MISC_USER; if (regs_sihv(regs) && (freeze_events_kernel != MMCR0_FCHV)) return PERF_RECORD_MISC_HYPERVISOR; return PERF_RECORD_MISC_KERNEL; } /* * Overload regs->dsisr to store MMCRA so we only need to read it once * on each interrupt. * Overload regs->dar to store SIER if we have it. * Overload regs->result to specify whether we should use the MSR (result * is zero) or the SIAR (result is non zero). */ static inline void perf_read_regs(struct pt_regs *regs) { unsigned long mmcra = mfspr(SPRN_MMCRA); int marked = mmcra & MMCRA_SAMPLE_ENABLE; int use_siar; regs->dsisr = mmcra; if (ppmu->flags & PPMU_HAS_SIER) regs->dar = mfspr(SPRN_SIER); /* * If this isn't a PMU exception (eg a software event) the SIAR is * not valid. Use pt_regs. * * If it is a marked event use the SIAR. * * If the PMU doesn't update the SIAR for non marked events use * pt_regs. * * If regs is a kernel interrupt, always use SIAR. Some PMUs have an * issue with regs_sipr not being in synch with SIAR in interrupt entry * and return sequences, which can result in regs_sipr being true for * kernel interrupts and SIAR, which has the effect of causing samples * to pile up at mtmsrd MSR[EE] 0->1 or pending irq replay around * interrupt entry/exit. * * If the PMU has HV/PR flags then check to see if they * place the exception in userspace. If so, use pt_regs. In * continuous sampling mode the SIAR and the PMU exception are * not synchronised, so they may be many instructions apart. * This can result in confusing backtraces. We still want * hypervisor samples as well as samples in the kernel with * interrupts off hence the userspace check. */ if (TRAP(regs) != INTERRUPT_PERFMON) use_siar = 0; else if ((ppmu->flags & PPMU_NO_SIAR)) use_siar = 0; else if (marked) use_siar = 1; else if ((ppmu->flags & PPMU_NO_CONT_SAMPLING)) use_siar = 0; else if (!user_mode(regs)) use_siar = 1; else if (!(ppmu->flags & PPMU_NO_SIPR) && regs_sipr(regs)) use_siar = 0; else use_siar = 1; regs->result = use_siar; } /* * On processors like P7+ that have the SIAR-Valid bit, marked instructions * must be sampled only if the SIAR-valid bit is set. * * For unmarked instructions and for processors that don't have the SIAR-Valid * bit, assume that SIAR is valid. */ static inline int siar_valid(struct pt_regs *regs) { unsigned long mmcra = regs->dsisr; int marked = mmcra & MMCRA_SAMPLE_ENABLE; if (marked) { /* * SIER[SIAR_VALID] is not set for some * marked events on power10 DD1, so drop * the check for SIER[SIAR_VALID] and return true. */ if (ppmu->flags & PPMU_P10_DD1) return 0x1; else if (ppmu->flags & PPMU_HAS_SIER) return regs->dar & SIER_SIAR_VALID; if (ppmu->flags & PPMU_SIAR_VALID) return mmcra & POWER7P_MMCRA_SIAR_VALID; } return 1; } /* Reset all possible BHRB entries */ static void power_pmu_bhrb_reset(void) { asm volatile(PPC_CLRBHRB); } static void power_pmu_bhrb_enable(struct perf_event *event) { struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); if (!ppmu->bhrb_nr) return; /* Clear BHRB if we changed task context to avoid data leaks */ if (event->ctx->task && cpuhw->bhrb_context != event->ctx) { power_pmu_bhrb_reset(); cpuhw->bhrb_context = event->ctx; } cpuhw->bhrb_users++; perf_sched_cb_inc(event->ctx->pmu); } static void power_pmu_bhrb_disable(struct perf_event *event) { struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); if (!ppmu->bhrb_nr) return; WARN_ON_ONCE(!cpuhw->bhrb_users); cpuhw->bhrb_users--; perf_sched_cb_dec(event->ctx->pmu); if (!cpuhw->disabled && !cpuhw->bhrb_users) { /* BHRB cannot be turned off when other * events are active on the PMU. */ /* avoid stale pointer */ cpuhw->bhrb_context = NULL; } } /* Called from ctxsw to prevent one process's branch entries to * mingle with the other process's entries during context switch. */ static void power_pmu_sched_task(struct perf_event_context *ctx, bool sched_in) { if (!ppmu->bhrb_nr) return; if (sched_in) power_pmu_bhrb_reset(); } /* Calculate the to address for a branch */ static __u64 power_pmu_bhrb_to(u64 addr) { unsigned int instr; __u64 target; if (is_kernel_addr(addr)) { if (copy_from_kernel_nofault(&instr, (void *)addr, sizeof(instr))) return 0; return branch_target(&instr); } /* Userspace: need copy instruction here then translate it */ if (copy_from_user_nofault(&instr, (unsigned int __user *)addr, sizeof(instr))) return 0; target = branch_target(&instr); if ((!target) || (instr & BRANCH_ABSOLUTE)) return target; /* Translate relative branch target from kernel to user address */ return target - (unsigned long)&instr + addr; } /* Processing BHRB entries */ static void power_pmu_bhrb_read(struct perf_event *event, struct cpu_hw_events *cpuhw) { u64 val; u64 addr; int r_index, u_index, pred; r_index = 0; u_index = 0; while (r_index < ppmu->bhrb_nr) { /* Assembly read function */ val = read_bhrb(r_index++); if (!val) /* Terminal marker: End of valid BHRB entries */ break; else { addr = val & BHRB_EA; pred = val & BHRB_PREDICTION; if (!addr) /* invalid entry */ continue; /* * BHRB rolling buffer could very much contain the kernel * addresses at this point. Check the privileges before * exporting it to userspace (avoid exposure of regions * where we could have speculative execution) * Incase of ISA v3.1, BHRB will capture only user-space * addresses, hence include a check before filtering code */ if (!(ppmu->flags & PPMU_ARCH_31) && is_kernel_addr(addr) && event->attr.exclude_kernel) continue; /* Branches are read most recent first (ie. mfbhrb 0 is * the most recent branch). * There are two types of valid entries: * 1) a target entry which is the to address of a * computed goto like a blr,bctr,btar. The next * entry read from the bhrb will be branch * corresponding to this target (ie. the actual * blr/bctr/btar instruction). * 2) a from address which is an actual branch. If a * target entry proceeds this, then this is the * matching branch for that target. If this is not * following a target entry, then this is a branch * where the target is given as an immediate field * in the instruction (ie. an i or b form branch). * In this case we need to read the instruction from * memory to determine the target/to address. */ if (val & BHRB_TARGET) { /* Target branches use two entries * (ie. computed gotos/XL form) */ cpuhw->bhrb_entries[u_index].to = addr; cpuhw->bhrb_entries[u_index].mispred = pred; cpuhw->bhrb_entries[u_index].predicted = ~pred; /* Get from address in next entry */ val = read_bhrb(r_index++); addr = val & BHRB_EA; if (val & BHRB_TARGET) { /* Shouldn't have two targets in a row.. Reset index and try again */ r_index--; addr = 0; } cpuhw->bhrb_entries[u_index].from = addr; } else { /* Branches to immediate field (ie I or B form) */ cpuhw->bhrb_entries[u_index].from = addr; cpuhw->bhrb_entries[u_index].to = power_pmu_bhrb_to(addr); cpuhw->bhrb_entries[u_index].mispred = pred; cpuhw->bhrb_entries[u_index].predicted = ~pred; } u_index++; } } cpuhw->bhrb_stack.nr = u_index; cpuhw->bhrb_stack.hw_idx = -1ULL; return; } static bool is_ebb_event(struct perf_event *event) { /* * This could be a per-PMU callback, but we'd rather avoid the cost. We * check that the PMU supports EBB, meaning those that don't can still * use bit 63 of the event code for something else if they wish. */ return (ppmu->flags & PPMU_ARCH_207S) && ((event->attr.config >> PERF_EVENT_CONFIG_EBB_SHIFT) & 1); } static int ebb_event_check(struct perf_event *event) { struct perf_event *leader = event->group_leader; /* Event and group leader must agree on EBB */ if (is_ebb_event(leader) != is_ebb_event(event)) return -EINVAL; if (is_ebb_event(event)) { if (!(event->attach_state & PERF_ATTACH_TASK)) return -EINVAL; if (!leader->attr.pinned || !leader->attr.exclusive) return -EINVAL; if (event->attr.freq || event->attr.inherit || event->attr.sample_type || event->attr.sample_period || event->attr.enable_on_exec) return -EINVAL; } return 0; } static void ebb_event_add(struct perf_event *event) { if (!is_ebb_event(event) || current->thread.used_ebb) return; /* * IFF this is the first time we've added an EBB event, set * PMXE in the user MMCR0 so we can detect when it's cleared by * userspace. We need this so that we can context switch while * userspace is in the EBB handler (where PMXE is 0). */ current->thread.used_ebb = 1; current->thread.mmcr0 |= MMCR0_PMXE; } static void ebb_switch_out(unsigned long mmcr0) { if (!(mmcr0 & MMCR0_EBE)) return; current->thread.siar = mfspr(SPRN_SIAR); current->thread.sier = mfspr(SPRN_SIER); current->thread.sdar = mfspr(SPRN_SDAR); current->thread.mmcr0 = mmcr0 & MMCR0_USER_MASK; current->thread.mmcr2 = mfspr(SPRN_MMCR2) & MMCR2_USER_MASK; if (ppmu->flags & PPMU_ARCH_31) { current->thread.mmcr3 = mfspr(SPRN_MMCR3); current->thread.sier2 = mfspr(SPRN_SIER2); current->thread.sier3 = mfspr(SPRN_SIER3); } } static unsigned long ebb_switch_in(bool ebb, struct cpu_hw_events *cpuhw) { unsigned long mmcr0 = cpuhw->mmcr.mmcr0; if (!ebb) goto out; /* Enable EBB and read/write to all 6 PMCs and BHRB for userspace */ mmcr0 |= MMCR0_EBE | MMCR0_BHRBA | MMCR0_PMCC_U6; /* * Add any bits from the user MMCR0, FC or PMAO. This is compatible * with pmao_restore_workaround() because we may add PMAO but we never * clear it here. */ mmcr0 |= current->thread.mmcr0; /* * Be careful not to set PMXE if userspace had it cleared. This is also * compatible with pmao_restore_workaround() because it has already * cleared PMXE and we leave PMAO alone. */ if (!(current->thread.mmcr0 & MMCR0_PMXE)) mmcr0 &= ~MMCR0_PMXE; mtspr(SPRN_SIAR, current->thread.siar); mtspr(SPRN_SIER, current->thread.sier); mtspr(SPRN_SDAR, current->thread.sdar); /* * Merge the kernel & user values of MMCR2. The semantics we implement * are that the user MMCR2 can set bits, ie. cause counters to freeze, * but not clear bits. If a task wants to be able to clear bits, ie. * unfreeze counters, it should not set exclude_xxx in its events and * instead manage the MMCR2 entirely by itself. */ mtspr(SPRN_MMCR2, cpuhw->mmcr.mmcr2 | current->thread.mmcr2); if (ppmu->flags & PPMU_ARCH_31) { mtspr(SPRN_MMCR3, current->thread.mmcr3); mtspr(SPRN_SIER2, current->thread.sier2); mtspr(SPRN_SIER3, current->thread.sier3); } out: return mmcr0; } static void pmao_restore_workaround(bool ebb) { unsigned pmcs[6]; if (!cpu_has_feature(CPU_FTR_PMAO_BUG)) return; /* * On POWER8E there is a hardware defect which affects the PMU context * switch logic, ie. power_pmu_disable/enable(). * * When a counter overflows PMXE is cleared and FC/PMAO is set in MMCR0 * by the hardware. Sometime later the actual PMU exception is * delivered. * * If we context switch, or simply disable/enable, the PMU prior to the * exception arriving, the exception will be lost when we clear PMAO. * * When we reenable the PMU, we will write the saved MMCR0 with PMAO * set, and this _should_ generate an exception. However because of the * defect no exception is generated when we write PMAO, and we get * stuck with no counters counting but no exception delivered. * * The workaround is to detect this case and tweak the hardware to * create another pending PMU exception. * * We do that by setting up PMC6 (cycles) for an imminent overflow and * enabling the PMU. That causes a new exception to be generated in the * chip, but we don't take it yet because we have interrupts hard * disabled. We then write back the PMU state as we want it to be seen * by the exception handler. When we reenable interrupts the exception * handler will be called and see the correct state. * * The logic is the same for EBB, except that the exception is gated by * us having interrupts hard disabled as well as the fact that we are * not in userspace. The exception is finally delivered when we return * to userspace. */ /* Only if PMAO is set and PMAO_SYNC is clear */ if ((current->thread.mmcr0 & (MMCR0_PMAO | MMCR0_PMAO_SYNC)) != MMCR0_PMAO) return; /* If we're doing EBB, only if BESCR[GE] is set */ if (ebb && !(current->thread.bescr & BESCR_GE)) return; /* * We are already soft-disabled in power_pmu_enable(). We need to hard * disable to actually prevent the PMU exception from firing. */ hard_irq_disable(); /* * This is a bit gross, but we know we're on POWER8E and have 6 PMCs. * Using read/write_pmc() in a for loop adds 12 function calls and * almost doubles our code size. */ pmcs[0] = mfspr(SPRN_PMC1); pmcs[1] = mfspr(SPRN_PMC2); pmcs[2] = mfspr(SPRN_PMC3); pmcs[3] = mfspr(SPRN_PMC4); pmcs[4] = mfspr(SPRN_PMC5); pmcs[5] = mfspr(SPRN_PMC6); /* Ensure all freeze bits are unset */ mtspr(SPRN_MMCR2, 0); /* Set up PMC6 to overflow in one cycle */ mtspr(SPRN_PMC6, 0x7FFFFFFE); /* Enable exceptions and unfreeze PMC6 */ mtspr(SPRN_MMCR0, MMCR0_PMXE | MMCR0_PMCjCE | MMCR0_PMAO); /* Now we need to refreeze and restore the PMCs */ mtspr(SPRN_MMCR0, MMCR0_FC | MMCR0_PMAO); mtspr(SPRN_PMC1, pmcs[0]); mtspr(SPRN_PMC2, pmcs[1]); mtspr(SPRN_PMC3, pmcs[2]); mtspr(SPRN_PMC4, pmcs[3]); mtspr(SPRN_PMC5, pmcs[4]); mtspr(SPRN_PMC6, pmcs[5]); } /* * If the perf subsystem wants performance monitor interrupts as soon as * possible (e.g., to sample the instruction address and stack chain), * this should return true. The IRQ masking code can then enable MSR[EE] * in some places (e.g., interrupt handlers) that allows PMI interrupts * through to improve accuracy of profiles, at the cost of some performance. * * The PMU counters can be enabled by other means (e.g., sysfs raw SPR * access), but in that case there is no need for prompt PMI handling. * * This currently returns true if any perf counter is being used. It * could possibly return false if only events are being counted rather than * samples being taken, but for now this is good enough. */ bool power_pmu_wants_prompt_pmi(void) { struct cpu_hw_events *cpuhw; /* * This could simply test local_paca->pmcregs_in_use if that were not * under ifdef KVM. */ if (!ppmu) return false; cpuhw = this_cpu_ptr(&cpu_hw_events); return cpuhw->n_events; } #endif /* CONFIG_PPC64 */ static void perf_event_interrupt(struct pt_regs *regs); /* * Read one performance monitor counter (PMC). */ static unsigned long read_pmc(int idx) { unsigned long val; switch (idx) { case 1: val = mfspr(SPRN_PMC1); break; case 2: val = mfspr(SPRN_PMC2); break; case 3: val = mfspr(SPRN_PMC3); break; case 4: val = mfspr(SPRN_PMC4); break; case 5: val = mfspr(SPRN_PMC5); break; case 6: val = mfspr(SPRN_PMC6); break; #ifdef CONFIG_PPC64 case 7: val = mfspr(SPRN_PMC7); break; case 8: val = mfspr(SPRN_PMC8); break; #endif /* CONFIG_PPC64 */ default: printk(KERN_ERR "oops trying to read PMC%d\n", idx); val = 0; } return val; } /* * Write one PMC. */ static void write_pmc(int idx, unsigned long val) { switch (idx) { case 1: mtspr(SPRN_PMC1, val); break; case 2: mtspr(SPRN_PMC2, val); break; case 3: mtspr(SPRN_PMC3, val); break; case 4: mtspr(SPRN_PMC4, val); break; case 5: mtspr(SPRN_PMC5, val); break; case 6: mtspr(SPRN_PMC6, val); break; #ifdef CONFIG_PPC64 case 7: mtspr(SPRN_PMC7, val); break; case 8: mtspr(SPRN_PMC8, val); break; #endif /* CONFIG_PPC64 */ default: printk(KERN_ERR "oops trying to write PMC%d\n", idx); } } static int any_pmc_overflown(struct cpu_hw_events *cpuhw) { int i, idx; for (i = 0; i < cpuhw->n_events; i++) { idx = cpuhw->event[i]->hw.idx; if ((idx) && ((int)read_pmc(idx) < 0)) return idx; } return 0; } /* Called from sysrq_handle_showregs() */ void perf_event_print_debug(void) { unsigned long sdar, sier, flags; u32 pmcs[MAX_HWEVENTS]; int i; if (!ppmu) { pr_info("Performance monitor hardware not registered.\n"); return; } if (!ppmu->n_counter) return; local_irq_save(flags); pr_info("CPU: %d PMU registers, ppmu = %s n_counters = %d", smp_processor_id(), ppmu->name, ppmu->n_counter); for (i = 0; i < ppmu->n_counter; i++) pmcs[i] = read_pmc(i + 1); for (; i < MAX_HWEVENTS; i++) pmcs[i] = 0xdeadbeef; pr_info("PMC1: %08x PMC2: %08x PMC3: %08x PMC4: %08x\n", pmcs[0], pmcs[1], pmcs[2], pmcs[3]); if (ppmu->n_counter > 4) pr_info("PMC5: %08x PMC6: %08x PMC7: %08x PMC8: %08x\n", pmcs[4], pmcs[5], pmcs[6], pmcs[7]); pr_info("MMCR0: %016lx MMCR1: %016lx MMCRA: %016lx\n", mfspr(SPRN_MMCR0), mfspr(SPRN_MMCR1), mfspr(SPRN_MMCRA)); sdar = sier = 0; #ifdef CONFIG_PPC64 sdar = mfspr(SPRN_SDAR); if (ppmu->flags & PPMU_HAS_SIER) sier = mfspr(SPRN_SIER); if (ppmu->flags & PPMU_ARCH_207S) { pr_info("MMCR2: %016lx EBBHR: %016lx\n", mfspr(SPRN_MMCR2), mfspr(SPRN_EBBHR)); pr_info("EBBRR: %016lx BESCR: %016lx\n", mfspr(SPRN_EBBRR), mfspr(SPRN_BESCR)); } if (ppmu->flags & PPMU_ARCH_31) { pr_info("MMCR3: %016lx SIER2: %016lx SIER3: %016lx\n", mfspr(SPRN_MMCR3), mfspr(SPRN_SIER2), mfspr(SPRN_SIER3)); } #endif pr_info("SIAR: %016lx SDAR: %016lx SIER: %016lx\n", mfspr(SPRN_SIAR), sdar, sier); local_irq_restore(flags); } /* * Check if a set of events can all go on the PMU at once. * If they can't, this will look at alternative codes for the events * and see if any combination of alternative codes is feasible. * The feasible set is returned in event_id[]. */ static int power_check_constraints(struct cpu_hw_events *cpuhw, u64 event_id[], unsigned int cflags[], int n_ev, struct perf_event **event) { unsigned long mask, value, nv; unsigned long smasks[MAX_HWEVENTS], svalues[MAX_HWEVENTS]; int n_alt[MAX_HWEVENTS], choice[MAX_HWEVENTS]; int i, j; unsigned long addf = ppmu->add_fields; unsigned long tadd = ppmu->test_adder; unsigned long grp_mask = ppmu->group_constraint_mask; unsigned long grp_val = ppmu->group_constraint_val; if (n_ev > ppmu->n_counter) return -1; /* First see if the events will go on as-is */ for (i = 0; i < n_ev; ++i) { if ((cflags[i] & PPMU_LIMITED_PMC_REQD) && !ppmu->limited_pmc_event(event_id[i])) { ppmu->get_alternatives(event_id[i], cflags[i], cpuhw->alternatives[i]); event_id[i] = cpuhw->alternatives[i][0]; } if (ppmu->get_constraint(event_id[i], &cpuhw->amasks[i][0], &cpuhw->avalues[i][0], event[i]->attr.config1)) return -1; } value = mask = 0; for (i = 0; i < n_ev; ++i) { nv = (value | cpuhw->avalues[i][0]) + (value & cpuhw->avalues[i][0] & addf); if (((((nv + tadd) ^ value) & mask) & (~grp_mask)) != 0) break; if (((((nv + tadd) ^ cpuhw->avalues[i][0]) & cpuhw->amasks[i][0]) & (~grp_mask)) != 0) break; value = nv; mask |= cpuhw->amasks[i][0]; } if (i == n_ev) { if ((value & mask & grp_mask) != (mask & grp_val)) return -1; else return 0; /* all OK */ } /* doesn't work, gather alternatives... */ if (!ppmu->get_alternatives) return -1; for (i = 0; i < n_ev; ++i) { choice[i] = 0; n_alt[i] = ppmu->get_alternatives(event_id[i], cflags[i], cpuhw->alternatives[i]); for (j = 1; j < n_alt[i]; ++j) ppmu->get_constraint(cpuhw->alternatives[i][j], &cpuhw->amasks[i][j], &cpuhw->avalues[i][j], event[i]->attr.config1); } /* enumerate all possibilities and see if any will work */ i = 0; j = -1; value = mask = nv = 0; while (i < n_ev) { if (j >= 0) { /* we're backtracking, restore context */ value = svalues[i]; mask = smasks[i]; j = choice[i]; } /* * See if any alternative k for event_id i, * where k > j, will satisfy the constraints. */ while (++j < n_alt[i]) { nv = (value | cpuhw->avalues[i][j]) + (value & cpuhw->avalues[i][j] & addf); if ((((nv + tadd) ^ value) & mask) == 0 && (((nv + tadd) ^ cpuhw->avalues[i][j]) & cpuhw->amasks[i][j]) == 0) break; } if (j >= n_alt[i]) { /* * No feasible alternative, backtrack * to event_id i-1 and continue enumerating its * alternatives from where we got up to. */ if (--i < 0) return -1; } else { /* * Found a feasible alternative for event_id i, * remember where we got up to with this event_id, * go on to the next event_id, and start with * the first alternative for it. */ choice[i] = j; svalues[i] = value; smasks[i] = mask; value = nv; mask |= cpuhw->amasks[i][j]; ++i; j = -1; } } /* OK, we have a feasible combination, tell the caller the solution */ for (i = 0; i < n_ev; ++i) event_id[i] = cpuhw->alternatives[i][choice[i]]; return 0; } /* * Check if newly-added events have consistent settings for * exclude_{user,kernel,hv} with each other and any previously * added events. */ static int check_excludes(struct perf_event **ctrs, unsigned int cflags[], int n_prev, int n_new) { int eu = 0, ek = 0, eh = 0; int i, n, first; struct perf_event *event; /* * If the PMU we're on supports per event exclude settings then we * don't need to do any of this logic. NB. This assumes no PMU has both * per event exclude and limited PMCs. */ if (ppmu->flags & PPMU_ARCH_207S) return 0; n = n_prev + n_new; if (n <= 1) return 0; first = 1; for (i = 0; i < n; ++i) { if (cflags[i] & PPMU_LIMITED_PMC_OK) { cflags[i] &= ~PPMU_LIMITED_PMC_REQD; continue; } event = ctrs[i]; if (first) { eu = event->attr.exclude_user; ek = event->attr.exclude_kernel; eh = event->attr.exclude_hv; first = 0; } else if (event->attr.exclude_user != eu || event->attr.exclude_kernel != ek || event->attr.exclude_hv != eh) { return -EAGAIN; } } if (eu || ek || eh) for (i = 0; i < n; ++i) if (cflags[i] & PPMU_LIMITED_PMC_OK) cflags[i] |= PPMU_LIMITED_PMC_REQD; return 0; } static u64 check_and_compute_delta(u64 prev, u64 val) { u64 delta = (val - prev) & 0xfffffffful; /* * POWER7 can roll back counter values, if the new value is smaller * than the previous value it will cause the delta and the counter to * have bogus values unless we rolled a counter over. If a counter is * rolled back, it will be smaller, but within 256, which is the maximum * number of events to rollback at once. If we detect a rollback * return 0. This can lead to a small lack of precision in the * counters. */ if (prev > val && (prev - val) < 256) delta = 0; return delta; } static void power_pmu_read(struct perf_event *event) { s64 val, delta, prev; if (event->hw.state & PERF_HES_STOPPED) return; if (!event->hw.idx) return; if (is_ebb_event(event)) { val = read_pmc(event->hw.idx); local64_set(&event->hw.prev_count, val); return; } /* * Performance monitor interrupts come even when interrupts * are soft-disabled, as long as interrupts are hard-enabled. * Therefore we treat them like NMIs. */ do { prev = local64_read(&event->hw.prev_count); barrier(); val = read_pmc(event->hw.idx); delta = check_and_compute_delta(prev, val); if (!delta) return; } while (local64_cmpxchg(&event->hw.prev_count, prev, val) != prev); local64_add(delta, &event->count); /* * A number of places program the PMC with (0x80000000 - period_left). * We never want period_left to be less than 1 because we will program * the PMC with a value >= 0x800000000 and an edge detected PMC will * roll around to 0 before taking an exception. We have seen this * on POWER8. * * To fix this, clamp the minimum value of period_left to 1. */ do { prev = local64_read(&event->hw.period_left); val = prev - delta; if (val < 1) val = 1; } while (local64_cmpxchg(&event->hw.period_left, prev, val) != prev); } /* * On some machines, PMC5 and PMC6 can't be written, don't respect * the freeze conditions, and don't generate interrupts. This tells * us if `event' is using such a PMC. */ static int is_limited_pmc(int pmcnum) { return (ppmu->flags & PPMU_LIMITED_PMC5_6) && (pmcnum == 5 || pmcnum == 6); } static void freeze_limited_counters(struct cpu_hw_events *cpuhw, unsigned long pmc5, unsigned long pmc6) { struct perf_event *event; u64 val, prev, delta; int i; for (i = 0; i < cpuhw->n_limited; ++i) { event = cpuhw->limited_counter[i]; if (!event->hw.idx) continue; val = (event->hw.idx == 5) ? pmc5 : pmc6; prev = local64_read(&event->hw.prev_count); event->hw.idx = 0; delta = check_and_compute_delta(prev, val); if (delta) local64_add(delta, &event->count); } } static void thaw_limited_counters(struct cpu_hw_events *cpuhw, unsigned long pmc5, unsigned long pmc6) { struct perf_event *event; u64 val, prev; int i; for (i = 0; i < cpuhw->n_limited; ++i) { event = cpuhw->limited_counter[i]; event->hw.idx = cpuhw->limited_hwidx[i]; val = (event->hw.idx == 5) ? pmc5 : pmc6; prev = local64_read(&event->hw.prev_count); if (check_and_compute_delta(prev, val)) local64_set(&event->hw.prev_count, val); perf_event_update_userpage(event); } } /* * Since limited events don't respect the freeze conditions, we * have to read them immediately after freezing or unfreezing the * other events. We try to keep the values from the limited * events as consistent as possible by keeping the delay (in * cycles and instructions) between freezing/unfreezing and reading * the limited events as small and consistent as possible. * Therefore, if any limited events are in use, we read them * both, and always in the same order, to minimize variability, * and do it inside the same asm that writes MMCR0. */ static void write_mmcr0(struct cpu_hw_events *cpuhw, unsigned long mmcr0) { unsigned long pmc5, pmc6; if (!cpuhw->n_limited) { mtspr(SPRN_MMCR0, mmcr0); return; } /* * Write MMCR0, then read PMC5 and PMC6 immediately. * To ensure we don't get a performance monitor interrupt * between writing MMCR0 and freezing/thawing the limited * events, we first write MMCR0 with the event overflow * interrupt enable bits turned off. */ asm volatile("mtspr %3,%2; mfspr %0,%4; mfspr %1,%5" : "=&r" (pmc5), "=&r" (pmc6) : "r" (mmcr0 & ~(MMCR0_PMC1CE | MMCR0_PMCjCE)), "i" (SPRN_MMCR0), "i" (SPRN_PMC5), "i" (SPRN_PMC6)); if (mmcr0 & MMCR0_FC) freeze_limited_counters(cpuhw, pmc5, pmc6); else thaw_limited_counters(cpuhw, pmc5, pmc6); /* * Write the full MMCR0 including the event overflow interrupt * enable bits, if necessary. */ if (mmcr0 & (MMCR0_PMC1CE | MMCR0_PMCjCE)) mtspr(SPRN_MMCR0, mmcr0); } /* * Disable all events to prevent PMU interrupts and to allow * events to be added or removed. */ static void power_pmu_disable(struct pmu *pmu) { struct cpu_hw_events *cpuhw; unsigned long flags, mmcr0, val, mmcra; if (!ppmu) return; local_irq_save(flags); cpuhw = this_cpu_ptr(&cpu_hw_events); if (!cpuhw->disabled) { /* * Check if we ever enabled the PMU on this cpu. */ if (!cpuhw->pmcs_enabled) { ppc_enable_pmcs(); cpuhw->pmcs_enabled = 1; } /* * Set the 'freeze counters' bit, clear EBE/BHRBA/PMCC/PMAO/FC56 * Also clear PMXE to disable PMI's getting triggered in some * corner cases during PMU disable. */ val = mmcr0 = mfspr(SPRN_MMCR0); val |= MMCR0_FC; val &= ~(MMCR0_EBE | MMCR0_BHRBA | MMCR0_PMCC | MMCR0_PMAO | MMCR0_PMXE | MMCR0_FC56); /* Set mmcr0 PMCCEXT for p10 */ if (ppmu->flags & PPMU_ARCH_31) val |= MMCR0_PMCCEXT; /* * The barrier is to make sure the mtspr has been * executed and the PMU has frozen the events etc. * before we return. */ write_mmcr0(cpuhw, val); mb(); isync(); /* * Some corner cases could clear the PMU counter overflow * while a masked PMI is pending. One such case is when * a PMI happens during interrupt replay and perf counter * values are cleared by PMU callbacks before replay. * * Disable the interrupt by clearing the paca bit for PMI * since we are disabling the PMU now. Otherwise provide a * warning if there is PMI pending, but no counter is found * overflown. * * Since power_pmu_disable runs under local_irq_save, it * could happen that code hits a PMC overflow without PMI * pending in paca. Hence only clear PMI pending if it was * set. * * If a PMI is pending, then MSR[EE] must be disabled (because * the masked PMI handler disabling EE). So it is safe to * call clear_pmi_irq_pending(). */ if (pmi_irq_pending()) clear_pmi_irq_pending(); val = mmcra = cpuhw->mmcr.mmcra; /* * Disable instruction sampling if it was enabled */ if (cpuhw->mmcr.mmcra & MMCRA_SAMPLE_ENABLE) val &= ~MMCRA_SAMPLE_ENABLE; /* Disable BHRB via mmcra (BHRBRD) for p10 */ if (ppmu->flags & PPMU_ARCH_31) val |= MMCRA_BHRB_DISABLE; /* * Write SPRN_MMCRA if mmcra has either disabled * instruction sampling or BHRB. */ if (val != mmcra) { mtspr(SPRN_MMCRA, mmcra); mb(); isync(); } cpuhw->disabled = 1; cpuhw->n_added = 0; ebb_switch_out(mmcr0); #ifdef CONFIG_PPC64 /* * These are readable by userspace, may contain kernel * addresses and are not switched by context switch, so clear * them now to avoid leaking anything to userspace in general * including to another process. */ if (ppmu->flags & PPMU_ARCH_207S) { mtspr(SPRN_SDAR, 0); mtspr(SPRN_SIAR, 0); } #endif } local_irq_restore(flags); } /* * Re-enable all events if disable == 0. * If we were previously disabled and events were added, then * put the new config on the PMU. */ static void power_pmu_enable(struct pmu *pmu) { struct perf_event *event; struct cpu_hw_events *cpuhw; unsigned long flags; long i; unsigned long val, mmcr0; s64 left; unsigned int hwc_index[MAX_HWEVENTS]; int n_lim; int idx; bool ebb; if (!ppmu) return; local_irq_save(flags); cpuhw = this_cpu_ptr(&cpu_hw_events); if (!cpuhw->disabled) goto out; if (cpuhw->n_events == 0) { ppc_set_pmu_inuse(0); goto out; } cpuhw->disabled = 0; /* * EBB requires an exclusive group and all events must have the EBB * flag set, or not set, so we can just check a single event. Also we * know we have at least one event. */ ebb = is_ebb_event(cpuhw->event[0]); /* * If we didn't change anything, or only removed events, * no need to recalculate MMCR* settings and reset the PMCs. * Just reenable the PMU with the current MMCR* settings * (possibly updated for removal of events). */ if (!cpuhw->n_added) { /* * If there is any active event with an overflown PMC * value, set back PACA_IRQ_PMI which would have been * cleared in power_pmu_disable(). */ hard_irq_disable(); if (any_pmc_overflown(cpuhw)) set_pmi_irq_pending(); mtspr(SPRN_MMCRA, cpuhw->mmcr.mmcra & ~MMCRA_SAMPLE_ENABLE); mtspr(SPRN_MMCR1, cpuhw->mmcr.mmcr1); if (ppmu->flags & PPMU_ARCH_31) mtspr(SPRN_MMCR3, cpuhw->mmcr.mmcr3); goto out_enable; } /* * Clear all MMCR settings and recompute them for the new set of events. */ memset(&cpuhw->mmcr, 0, sizeof(cpuhw->mmcr)); if (ppmu->compute_mmcr(cpuhw->events, cpuhw->n_events, hwc_index, &cpuhw->mmcr, cpuhw->event, ppmu->flags)) { /* shouldn't ever get here */ printk(KERN_ERR "oops compute_mmcr failed\n"); goto out; } if (!(ppmu->flags & PPMU_ARCH_207S)) { /* * Add in MMCR0 freeze bits corresponding to the attr.exclude_* * bits for the first event. We have already checked that all * events have the same value for these bits as the first event. */ event = cpuhw->event[0]; if (event->attr.exclude_user) cpuhw->mmcr.mmcr0 |= MMCR0_FCP; if (event->attr.exclude_kernel) cpuhw->mmcr.mmcr0 |= freeze_events_kernel; if (event->attr.exclude_hv) cpuhw->mmcr.mmcr0 |= MMCR0_FCHV; } /* * Write the new configuration to MMCR* with the freeze * bit set and set the hardware events to their initial values. * Then unfreeze the events. */ ppc_set_pmu_inuse(1); mtspr(SPRN_MMCRA, cpuhw->mmcr.mmcra & ~MMCRA_SAMPLE_ENABLE); mtspr(SPRN_MMCR1, cpuhw->mmcr.mmcr1); mtspr(SPRN_MMCR0, (cpuhw->mmcr.mmcr0 & ~(MMCR0_PMC1CE | MMCR0_PMCjCE)) | MMCR0_FC); if (ppmu->flags & PPMU_ARCH_207S) mtspr(SPRN_MMCR2, cpuhw->mmcr.mmcr2); if (ppmu->flags & PPMU_ARCH_31) mtspr(SPRN_MMCR3, cpuhw->mmcr.mmcr3); /* * Read off any pre-existing events that need to move * to another PMC. */ for (i = 0; i < cpuhw->n_events; ++i) { event = cpuhw->event[i]; if (event->hw.idx && event->hw.idx != hwc_index[i] + 1) { power_pmu_read(event); write_pmc(event->hw.idx, 0); event->hw.idx = 0; } } /* * Initialize the PMCs for all the new and moved events. */ cpuhw->n_limited = n_lim = 0; for (i = 0; i < cpuhw->n_events; ++i) { event = cpuhw->event[i]; if (event->hw.idx) continue; idx = hwc_index[i] + 1; if (is_limited_pmc(idx)) { cpuhw->limited_counter[n_lim] = event; cpuhw->limited_hwidx[n_lim] = idx; ++n_lim; continue; } if (ebb) val = local64_read(&event->hw.prev_count); else { val = 0; if (event->hw.sample_period) { left = local64_read(&event->hw.period_left); if (left < 0x80000000L) val = 0x80000000L - left; } local64_set(&event->hw.prev_count, val); } event->hw.idx = idx; if (event->hw.state & PERF_HES_STOPPED) val = 0; write_pmc(idx, val); perf_event_update_userpage(event); } cpuhw->n_limited = n_lim; cpuhw->mmcr.mmcr0 |= MMCR0_PMXE | MMCR0_FCECE; out_enable: pmao_restore_workaround(ebb); mmcr0 = ebb_switch_in(ebb, cpuhw); mb(); if (cpuhw->bhrb_users) ppmu->config_bhrb(cpuhw->bhrb_filter); write_mmcr0(cpuhw, mmcr0); /* * Enable instruction sampling if necessary */ if (cpuhw->mmcr.mmcra & MMCRA_SAMPLE_ENABLE) { mb(); mtspr(SPRN_MMCRA, cpuhw->mmcr.mmcra); } out: local_irq_restore(flags); } static int collect_events(struct perf_event *group, int max_count, struct perf_event *ctrs[], u64 *events, unsigned int *flags) { int n = 0; struct perf_event *event; if (group->pmu->task_ctx_nr == perf_hw_context) { if (n >= max_count) return -1; ctrs[n] = group; flags[n] = group->hw.event_base; events[n++] = group->hw.config; } for_each_sibling_event(event, group) { if (event->pmu->task_ctx_nr == perf_hw_context && event->state != PERF_EVENT_STATE_OFF) { if (n >= max_count) return -1; ctrs[n] = event; flags[n] = event->hw.event_base; events[n++] = event->hw.config; } } return n; } /* * Add an event to the PMU. * If all events are not already frozen, then we disable and * re-enable the PMU in order to get hw_perf_enable to do the * actual work of reconfiguring the PMU. */ static int power_pmu_add(struct perf_event *event, int ef_flags) { struct cpu_hw_events *cpuhw; unsigned long flags; int n0; int ret = -EAGAIN; local_irq_save(flags); perf_pmu_disable(event->pmu); /* * Add the event to the list (if there is room) * and check whether the total set is still feasible. */ cpuhw = this_cpu_ptr(&cpu_hw_events); n0 = cpuhw->n_events; if (n0 >= ppmu->n_counter) goto out; cpuhw->event[n0] = event; cpuhw->events[n0] = event->hw.config; cpuhw->flags[n0] = event->hw.event_base; /* * This event may have been disabled/stopped in record_and_restart() * because we exceeded the ->event_limit. If re-starting the event, * clear the ->hw.state (STOPPED and UPTODATE flags), so the user * notification is re-enabled. */ if (!(ef_flags & PERF_EF_START)) event->hw.state = PERF_HES_STOPPED | PERF_HES_UPTODATE; else event->hw.state = 0; /* * If group events scheduling transaction was started, * skip the schedulability test here, it will be performed * at commit time(->commit_txn) as a whole */ if (cpuhw->txn_flags & PERF_PMU_TXN_ADD) goto nocheck; if (check_excludes(cpuhw->event, cpuhw->flags, n0, 1)) goto out; if (power_check_constraints(cpuhw, cpuhw->events, cpuhw->flags, n0 + 1, cpuhw->event)) goto out; event->hw.config = cpuhw->events[n0]; nocheck: ebb_event_add(event); ++cpuhw->n_events; ++cpuhw->n_added; ret = 0; out: if (has_branch_stack(event)) { u64 bhrb_filter = -1; if (ppmu->bhrb_filter_map) bhrb_filter = ppmu->bhrb_filter_map( event->attr.branch_sample_type); if (bhrb_filter != -1) { cpuhw->bhrb_filter = bhrb_filter; power_pmu_bhrb_enable(event); } } perf_pmu_enable(event->pmu); local_irq_restore(flags); return ret; } /* * Remove an event from the PMU. */ static void power_pmu_del(struct perf_event *event, int ef_flags) { struct cpu_hw_events *cpuhw; long i; unsigned long flags; local_irq_save(flags); perf_pmu_disable(event->pmu); power_pmu_read(event); cpuhw = this_cpu_ptr(&cpu_hw_events); for (i = 0; i < cpuhw->n_events; ++i) { if (event == cpuhw->event[i]) { while (++i < cpuhw->n_events) { cpuhw->event[i-1] = cpuhw->event[i]; cpuhw->events[i-1] = cpuhw->events[i]; cpuhw->flags[i-1] = cpuhw->flags[i]; } --cpuhw->n_events; ppmu->disable_pmc(event->hw.idx - 1, &cpuhw->mmcr); if (event->hw.idx) { write_pmc(event->hw.idx, 0); event->hw.idx = 0; } perf_event_update_userpage(event); break; } } for (i = 0; i < cpuhw->n_limited; ++i) if (event == cpuhw->limited_counter[i]) break; if (i < cpuhw->n_limited) { while (++i < cpuhw->n_limited) { cpuhw->limited_counter[i-1] = cpuhw->limited_counter[i]; cpuhw->limited_hwidx[i-1] = cpuhw->limited_hwidx[i]; } --cpuhw->n_limited; } if (cpuhw->n_events == 0) { /* disable exceptions if no events are running */ cpuhw->mmcr.mmcr0 &= ~(MMCR0_PMXE | MMCR0_FCECE); } if (has_branch_stack(event)) power_pmu_bhrb_disable(event); perf_pmu_enable(event->pmu); local_irq_restore(flags); } /* * POWER-PMU does not support disabling individual counters, hence * program their cycle counter to their max value and ignore the interrupts. */ static void power_pmu_start(struct perf_event *event, int ef_flags) { unsigned long flags; s64 left; unsigned long val; if (!event->hw.idx || !event->hw.sample_period) return; if (!(event->hw.state & PERF_HES_STOPPED)) return; if (ef_flags & PERF_EF_RELOAD) WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE)); local_irq_save(flags); perf_pmu_disable(event->pmu); event->hw.state = 0; left = local64_read(&event->hw.period_left); val = 0; if (left < 0x80000000L) val = 0x80000000L - left; write_pmc(event->hw.idx, val); perf_event_update_userpage(event); perf_pmu_enable(event->pmu); local_irq_restore(flags); } static void power_pmu_stop(struct perf_event *event, int ef_flags) { unsigned long flags; if (!event->hw.idx || !event->hw.sample_period) return; if (event->hw.state & PERF_HES_STOPPED) return; local_irq_save(flags); perf_pmu_disable(event->pmu); power_pmu_read(event); event->hw.state |= PERF_HES_STOPPED | PERF_HES_UPTODATE; write_pmc(event->hw.idx, 0); perf_event_update_userpage(event); perf_pmu_enable(event->pmu); local_irq_restore(flags); } /* * Start group events scheduling transaction * Set the flag to make pmu::enable() not perform the * schedulability test, it will be performed at commit time * * We only support PERF_PMU_TXN_ADD transactions. Save the * transaction flags but otherwise ignore non-PERF_PMU_TXN_ADD * transactions. */ static void power_pmu_start_txn(struct pmu *pmu, unsigned int txn_flags) { struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); WARN_ON_ONCE(cpuhw->txn_flags); /* txn already in flight */ cpuhw->txn_flags = txn_flags; if (txn_flags & ~PERF_PMU_TXN_ADD) return; perf_pmu_disable(pmu); cpuhw->n_txn_start = cpuhw->n_events; } /* * Stop group events scheduling transaction * Clear the flag and pmu::enable() will perform the * schedulability test. */ static void power_pmu_cancel_txn(struct pmu *pmu) { struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); unsigned int txn_flags; WARN_ON_ONCE(!cpuhw->txn_flags); /* no txn in flight */ txn_flags = cpuhw->txn_flags; cpuhw->txn_flags = 0; if (txn_flags & ~PERF_PMU_TXN_ADD) return; perf_pmu_enable(pmu); } /* * Commit group events scheduling transaction * Perform the group schedulability test as a whole * Return 0 if success */ static int power_pmu_commit_txn(struct pmu *pmu) { struct cpu_hw_events *cpuhw; long i, n; if (!ppmu) return -EAGAIN; cpuhw = this_cpu_ptr(&cpu_hw_events); WARN_ON_ONCE(!cpuhw->txn_flags); /* no txn in flight */ if (cpuhw->txn_flags & ~PERF_PMU_TXN_ADD) { cpuhw->txn_flags = 0; return 0; } n = cpuhw->n_events; if (check_excludes(cpuhw->event, cpuhw->flags, 0, n)) return -EAGAIN; i = power_check_constraints(cpuhw, cpuhw->events, cpuhw->flags, n, cpuhw->event); if (i < 0) return -EAGAIN; for (i = cpuhw->n_txn_start; i < n; ++i) cpuhw->event[i]->hw.config = cpuhw->events[i]; cpuhw->txn_flags = 0; perf_pmu_enable(pmu); return 0; } /* * Return 1 if we might be able to put event on a limited PMC, * or 0 if not. * An event can only go on a limited PMC if it counts something * that a limited PMC can count, doesn't require interrupts, and * doesn't exclude any processor mode. */ static int can_go_on_limited_pmc(struct perf_event *event, u64 ev, unsigned int flags) { int n; u64 alt[MAX_EVENT_ALTERNATIVES]; if (event->attr.exclude_user || event->attr.exclude_kernel || event->attr.exclude_hv || event->attr.sample_period) return 0; if (ppmu->limited_pmc_event(ev)) return 1; /* * The requested event_id isn't on a limited PMC already; * see if any alternative code goes on a limited PMC. */ if (!ppmu->get_alternatives) return 0; flags |= PPMU_LIMITED_PMC_OK | PPMU_LIMITED_PMC_REQD; n = ppmu->get_alternatives(ev, flags, alt); return n > 0; } /* * Find an alternative event_id that goes on a normal PMC, if possible, * and return the event_id code, or 0 if there is no such alternative. * (Note: event_id code 0 is "don't count" on all machines.) */ static u64 normal_pmc_alternative(u64 ev, unsigned long flags) { u64 alt[MAX_EVENT_ALTERNATIVES]; int n; flags &= ~(PPMU_LIMITED_PMC_OK | PPMU_LIMITED_PMC_REQD); n = ppmu->get_alternatives(ev, flags, alt); if (!n) return 0; return alt[0]; } /* Number of perf_events counting hardware events */ static atomic_t num_events; /* Used to avoid races in calling reserve/release_pmc_hardware */ static DEFINE_MUTEX(pmc_reserve_mutex); /* * Release the PMU if this is the last perf_event. */ static void hw_perf_event_destroy(struct perf_event *event) { if (!atomic_add_unless(&num_events, -1, 1)) { mutex_lock(&pmc_reserve_mutex); if (atomic_dec_return(&num_events) == 0) release_pmc_hardware(); mutex_unlock(&pmc_reserve_mutex); } } /* * Translate a generic cache event_id config to a raw event_id code. */ static int hw_perf_cache_event(u64 config, u64 *eventp) { unsigned long type, op, result; u64 ev; if (!ppmu->cache_events) return -EINVAL; /* unpack config */ type = config & 0xff; op = (config >> 8) & 0xff; result = (config >> 16) & 0xff; if (type >= PERF_COUNT_HW_CACHE_MAX || op >= PERF_COUNT_HW_CACHE_OP_MAX || result >= PERF_COUNT_HW_CACHE_RESULT_MAX) return -EINVAL; ev = (*ppmu->cache_events)[type][op][result]; if (ev == 0) return -EOPNOTSUPP; if (ev == -1) return -EINVAL; *eventp = ev; return 0; } static bool is_event_blacklisted(u64 ev) { int i; for (i=0; i < ppmu->n_blacklist_ev; i++) { if (ppmu->blacklist_ev[i] == ev) return true; } return false; } static int power_pmu_event_init(struct perf_event *event) { u64 ev; unsigned long flags, irq_flags; struct perf_event *ctrs[MAX_HWEVENTS]; u64 events[MAX_HWEVENTS]; unsigned int cflags[MAX_HWEVENTS]; int n; int err; struct cpu_hw_events *cpuhw; if (!ppmu) return -ENOENT; if (has_branch_stack(event)) { /* PMU has BHRB enabled */ if (!(ppmu->flags & PPMU_ARCH_207S)) return -EOPNOTSUPP; } switch (event->attr.type) { case PERF_TYPE_HARDWARE: ev = event->attr.config; if (ev >= ppmu->n_generic || ppmu->generic_events[ev] == 0) return -EOPNOTSUPP; if (ppmu->blacklist_ev && is_event_blacklisted(ev)) return -EINVAL; ev = ppmu->generic_events[ev]; break; case PERF_TYPE_HW_CACHE: err = hw_perf_cache_event(event->attr.config, &ev); if (err) return err; if (ppmu->blacklist_ev && is_event_blacklisted(ev)) return -EINVAL; break; case PERF_TYPE_RAW: ev = event->attr.config; if (ppmu->blacklist_ev && is_event_blacklisted(ev)) return -EINVAL; break; default: return -ENOENT; } /* * PMU config registers have fields that are * reserved and some specific values for bit fields are reserved. * For ex., MMCRA[61:62] is Random Sampling Mode (SM) * and value of 0b11 to this field is reserved. * Check for invalid values in attr.config. */ if (ppmu->check_attr_config && ppmu->check_attr_config(event)) return -EINVAL; event->hw.config_base = ev; event->hw.idx = 0; /* * If we are not running on a hypervisor, force the * exclude_hv bit to 0 so that we don't care what * the user set it to. */ if (!firmware_has_feature(FW_FEATURE_LPAR)) event->attr.exclude_hv = 0; /* * If this is a per-task event, then we can use * PM_RUN_* events interchangeably with their non RUN_* * equivalents, e.g. PM_RUN_CYC instead of PM_CYC. * XXX we should check if the task is an idle task. */ flags = 0; if (event->attach_state & PERF_ATTACH_TASK) flags |= PPMU_ONLY_COUNT_RUN; /* * If this machine has limited events, check whether this * event_id could go on a limited event. */ if (ppmu->flags & PPMU_LIMITED_PMC5_6) { if (can_go_on_limited_pmc(event, ev, flags)) { flags |= PPMU_LIMITED_PMC_OK; } else if (ppmu->limited_pmc_event(ev)) { /* * The requested event_id is on a limited PMC, * but we can't use a limited PMC; see if any * alternative goes on a normal PMC. */ ev = normal_pmc_alternative(ev, flags); if (!ev) return -EINVAL; } } /* Extra checks for EBB */ err = ebb_event_check(event); if (err) return err; /* * If this is in a group, check if it can go on with all the * other hardware events in the group. We assume the event * hasn't been linked into its leader's sibling list at this point. */ n = 0; if (event->group_leader != event) { n = collect_events(event->group_leader, ppmu->n_counter - 1, ctrs, events, cflags); if (n < 0) return -EINVAL; } events[n] = ev; ctrs[n] = event; cflags[n] = flags; if (check_excludes(ctrs, cflags, n, 1)) return -EINVAL; local_irq_save(irq_flags); cpuhw = this_cpu_ptr(&cpu_hw_events); err = power_check_constraints(cpuhw, events, cflags, n + 1, ctrs); if (has_branch_stack(event)) { u64 bhrb_filter = -1; /* * Currently no PMU supports having multiple branch filters * at the same time. Branch filters are set via MMCRA IFM[32:33] * bits for Power8 and above. Return EOPNOTSUPP when multiple * branch filters are requested in the event attr. * * When opening event via perf_event_open(), branch_sample_type * gets adjusted in perf_copy_attr(). Kernel will automatically * adjust the branch_sample_type based on the event modifier * settings to include PERF_SAMPLE_BRANCH_PLM_ALL. Hence drop * the check for PERF_SAMPLE_BRANCH_PLM_ALL. */ if (hweight64(event->attr.branch_sample_type & ~PERF_SAMPLE_BRANCH_PLM_ALL) > 1) { local_irq_restore(irq_flags); return -EOPNOTSUPP; } if (ppmu->bhrb_filter_map) bhrb_filter = ppmu->bhrb_filter_map( event->attr.branch_sample_type); if (bhrb_filter == -1) { local_irq_restore(irq_flags); return -EOPNOTSUPP; } cpuhw->bhrb_filter = bhrb_filter; } local_irq_restore(irq_flags); if (err) return -EINVAL; event->hw.config = events[n]; event->hw.event_base = cflags[n]; event->hw.last_period = event->hw.sample_period; local64_set(&event->hw.period_left, event->hw.last_period); /* * For EBB events we just context switch the PMC value, we don't do any * of the sample_period logic. We use hw.prev_count for this. */ if (is_ebb_event(event)) local64_set(&event->hw.prev_count, 0); /* * See if we need to reserve the PMU. * If no events are currently in use, then we have to take a * mutex to ensure that we don't race with another task doing * reserve_pmc_hardware or release_pmc_hardware. */ err = 0; if (!atomic_inc_not_zero(&num_events)) { mutex_lock(&pmc_reserve_mutex); if (atomic_read(&num_events) == 0 && reserve_pmc_hardware(perf_event_interrupt)) err = -EBUSY; else atomic_inc(&num_events); mutex_unlock(&pmc_reserve_mutex); } event->destroy = hw_perf_event_destroy; return err; } static int power_pmu_event_idx(struct perf_event *event) { return event->hw.idx; } ssize_t power_events_sysfs_show(struct device *dev, struct device_attribute *attr, char *page) { struct perf_pmu_events_attr *pmu_attr; pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr); return sprintf(page, "event=0x%02llx\n", pmu_attr->id); } static struct pmu power_pmu = { .pmu_enable = power_pmu_enable, .pmu_disable = power_pmu_disable, .event_init = power_pmu_event_init, .add = power_pmu_add, .del = power_pmu_del, .start = power_pmu_start, .stop = power_pmu_stop, .read = power_pmu_read, .start_txn = power_pmu_start_txn, .cancel_txn = power_pmu_cancel_txn, .commit_txn = power_pmu_commit_txn, .event_idx = power_pmu_event_idx, .sched_task = power_pmu_sched_task, }; #define PERF_SAMPLE_ADDR_TYPE (PERF_SAMPLE_ADDR | \ PERF_SAMPLE_PHYS_ADDR | \ PERF_SAMPLE_DATA_PAGE_SIZE) /* * A counter has overflowed; update its count and record * things if requested. Note that interrupts are hard-disabled * here so there is no possibility of being interrupted. */ static void record_and_restart(struct perf_event *event, unsigned long val, struct pt_regs *regs) { u64 period = event->hw.sample_period; s64 prev, delta, left; int record = 0; if (event->hw.state & PERF_HES_STOPPED) { write_pmc(event->hw.idx, 0); return; } /* we don't have to worry about interrupts here */ prev = local64_read(&event->hw.prev_count); delta = check_and_compute_delta(prev, val); local64_add(delta, &event->count); /* * See if the total period for this event has expired, * and update for the next period. */ val = 0; left = local64_read(&event->hw.period_left) - delta; if (delta == 0) left++; if (period) { if (left <= 0) { left += period; if (left <= 0) left = period; /* * If address is not requested in the sample via * PERF_SAMPLE_IP, just record that sample irrespective * of SIAR valid check. */ if (event->attr.sample_type & PERF_SAMPLE_IP) record = siar_valid(regs); else record = 1; event->hw.last_period = event->hw.sample_period; } if (left < 0x80000000LL) val = 0x80000000LL - left; } write_pmc(event->hw.idx, val); local64_set(&event->hw.prev_count, val); local64_set(&event->hw.period_left, left); perf_event_update_userpage(event); /* * Due to hardware limitation, sometimes SIAR could sample a kernel * address even when freeze on supervisor state (kernel) is set in * MMCR2. Check attr.exclude_kernel and address to drop the sample in * these cases. */ if (event->attr.exclude_kernel && (event->attr.sample_type & PERF_SAMPLE_IP) && is_kernel_addr(mfspr(SPRN_SIAR))) record = 0; /* * Finally record data if requested. */ if (record) { struct perf_sample_data data; perf_sample_data_init(&data, ~0ULL, event->hw.last_period); if (event->attr.sample_type & PERF_SAMPLE_ADDR_TYPE) perf_get_data_addr(event, regs, &data.addr); if (event->attr.sample_type & PERF_SAMPLE_BRANCH_STACK) { struct cpu_hw_events *cpuhw; cpuhw = this_cpu_ptr(&cpu_hw_events); power_pmu_bhrb_read(event, cpuhw); data.br_stack = &cpuhw->bhrb_stack; data.sample_flags |= PERF_SAMPLE_BRANCH_STACK; } if (event->attr.sample_type & PERF_SAMPLE_DATA_SRC && ppmu->get_mem_data_src) { ppmu->get_mem_data_src(&data.data_src, ppmu->flags, regs); data.sample_flags |= PERF_SAMPLE_DATA_SRC; } if (event->attr.sample_type & PERF_SAMPLE_WEIGHT_TYPE && ppmu->get_mem_weight) { ppmu->get_mem_weight(&data.weight.full, event->attr.sample_type); data.sample_flags |= PERF_SAMPLE_WEIGHT_TYPE; } if (perf_event_overflow(event, &data, regs)) power_pmu_stop(event, 0); } else if (period) { /* Account for interrupt in case of invalid SIAR */ if (perf_event_account_interrupt(event)) power_pmu_stop(event, 0); } } /* * Called from generic code to get the misc flags (i.e. processor mode) * for an event_id. */ unsigned long perf_misc_flags(struct pt_regs *regs) { u32 flags = perf_get_misc_flags(regs); if (flags) return flags; return user_mode(regs) ? PERF_RECORD_MISC_USER : PERF_RECORD_MISC_KERNEL; } /* * Called from generic code to get the instruction pointer * for an event_id. */ unsigned long perf_instruction_pointer(struct pt_regs *regs) { unsigned long siar = mfspr(SPRN_SIAR); if (regs_use_siar(regs) && siar_valid(regs) && siar) return siar + perf_ip_adjust(regs); else return regs->nip; } static bool pmc_overflow_power7(unsigned long val) { /* * Events on POWER7 can roll back if a speculative event doesn't * eventually complete. Unfortunately in some rare cases they will * raise a performance monitor exception. We need to catch this to * ensure we reset the PMC. In all cases the PMC will be 256 or less * cycles from overflow. * * We only do this if the first pass fails to find any overflowing * PMCs because a user might set a period of less than 256 and we * don't want to mistakenly reset them. */ if ((0x80000000 - val) <= 256) return true; return false; } static bool pmc_overflow(unsigned long val) { if ((int)val < 0) return true; return false; } /* * Performance monitor interrupt stuff */ static void __perf_event_interrupt(struct pt_regs *regs) { int i, j; struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); struct perf_event *event; int found, active; if (cpuhw->n_limited) freeze_limited_counters(cpuhw, mfspr(SPRN_PMC5), mfspr(SPRN_PMC6)); perf_read_regs(regs); /* Read all the PMCs since we'll need them a bunch of times */ for (i = 0; i < ppmu->n_counter; ++i) cpuhw->pmcs[i] = read_pmc(i + 1); /* Try to find what caused the IRQ */ found = 0; for (i = 0; i < ppmu->n_counter; ++i) { if (!pmc_overflow(cpuhw->pmcs[i])) continue; if (is_limited_pmc(i + 1)) continue; /* these won't generate IRQs */ /* * We've found one that's overflowed. For active * counters we need to log this. For inactive * counters, we need to reset it anyway */ found = 1; active = 0; for (j = 0; j < cpuhw->n_events; ++j) { event = cpuhw->event[j]; if (event->hw.idx == (i + 1)) { active = 1; record_and_restart(event, cpuhw->pmcs[i], regs); break; } } /* * Clear PACA_IRQ_PMI in case it was set by * set_pmi_irq_pending() when PMU was enabled * after accounting for interrupts. */ clear_pmi_irq_pending(); if (!active) /* reset non active counters that have overflowed */ write_pmc(i + 1, 0); } if (!found && pvr_version_is(PVR_POWER7)) { /* check active counters for special buggy p7 overflow */ for (i = 0; i < cpuhw->n_events; ++i) { event = cpuhw->event[i]; if (!event->hw.idx || is_limited_pmc(event->hw.idx)) continue; if (pmc_overflow_power7(cpuhw->pmcs[event->hw.idx - 1])) { /* event has overflowed in a buggy way*/ found = 1; record_and_restart(event, cpuhw->pmcs[event->hw.idx - 1], regs); } } } /* * During system wide profiling or while specific CPU is monitored for an * event, some corner cases could cause PMC to overflow in idle path. This * will trigger a PMI after waking up from idle. Since counter values are _not_ * saved/restored in idle path, can lead to below "Can't find PMC" message. */ if (unlikely(!found) && !arch_irq_disabled_regs(regs)) printk_ratelimited(KERN_WARNING "Can't find PMC that caused IRQ\n"); /* * Reset MMCR0 to its normal value. This will set PMXE and * clear FC (freeze counters) and PMAO (perf mon alert occurred) * and thus allow interrupts to occur again. * XXX might want to use MSR.PM to keep the events frozen until * we get back out of this interrupt. */ write_mmcr0(cpuhw, cpuhw->mmcr.mmcr0); /* Clear the cpuhw->pmcs */ memset(&cpuhw->pmcs, 0, sizeof(cpuhw->pmcs)); } static void perf_event_interrupt(struct pt_regs *regs) { u64 start_clock = sched_clock(); __perf_event_interrupt(regs); perf_sample_event_took(sched_clock() - start_clock); } static int power_pmu_prepare_cpu(unsigned int cpu) { struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu); if (ppmu) { memset(cpuhw, 0, sizeof(*cpuhw)); cpuhw->mmcr.mmcr0 = MMCR0_FC; } return 0; } static ssize_t pmu_name_show(struct device *cdev, struct device_attribute *attr, char *buf) { if (ppmu) return sysfs_emit(buf, "%s", ppmu->name); return 0; } static DEVICE_ATTR_RO(pmu_name); static struct attribute *pmu_caps_attrs[] = { &dev_attr_pmu_name.attr, NULL }; static const struct attribute_group pmu_caps_group = { .name = "caps", .attrs = pmu_caps_attrs, }; static const struct attribute_group *pmu_caps_groups[] = { &pmu_caps_group, NULL, }; int __init register_power_pmu(struct power_pmu *pmu) { if (ppmu) return -EBUSY; /* something's already registered */ ppmu = pmu; pr_info("%s performance monitor hardware support registered\n", pmu->name); power_pmu.attr_groups = ppmu->attr_groups; if (ppmu->flags & PPMU_ARCH_207S) power_pmu.attr_update = pmu_caps_groups; power_pmu.capabilities |= (ppmu->capabilities & PERF_PMU_CAP_EXTENDED_REGS); #ifdef MSR_HV /* * Use FCHV to ignore kernel events if MSR.HV is set. */ if (mfmsr() & MSR_HV) freeze_events_kernel = MMCR0_FCHV; #endif /* CONFIG_PPC64 */ perf_pmu_register(&power_pmu, "cpu", PERF_TYPE_RAW); cpuhp_setup_state(CPUHP_PERF_POWER, "perf/powerpc:prepare", power_pmu_prepare_cpu, NULL); return 0; } #ifdef CONFIG_PPC64 static bool pmu_override = false; static unsigned long pmu_override_val; static void do_pmu_override(void *data) { ppc_set_pmu_inuse(1); if (pmu_override_val) mtspr(SPRN_MMCR1, pmu_override_val); mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) & ~MMCR0_FC); } static int __init init_ppc64_pmu(void) { if (cpu_has_feature(CPU_FTR_HVMODE) && pmu_override) { pr_warn("disabling perf due to pmu_override= command line option.\n"); on_each_cpu(do_pmu_override, NULL, 1); return 0; } /* run through all the pmu drivers one at a time */ if (!init_power5_pmu()) return 0; else if (!init_power5p_pmu()) return 0; else if (!init_power6_pmu()) return 0; else if (!init_power7_pmu()) return 0; else if (!init_power8_pmu()) return 0; else if (!init_power9_pmu()) return 0; else if (!init_power10_pmu()) return 0; else if (!init_ppc970_pmu()) return 0; else return init_generic_compat_pmu(); } early_initcall(init_ppc64_pmu); static int __init pmu_setup(char *str) { unsigned long val; if (!early_cpu_has_feature(CPU_FTR_HVMODE)) return 0; pmu_override = true; if (kstrtoul(str, 0, &val)) val = 0; pmu_override_val = val; return 1; } __setup("pmu_override=", pmu_setup); #endif |