/* * Copyright (c) 2017 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * * This file contains Original Code and/or Modifications of Original Code * as defined in and that are subject to the Apple Public Source License * Version 2.0 (the 'License'). You may not use this file except in * compliance with the License. The rights granted to you under the License * may not be used to create, or enable the creation or redistribution of, * unlawful or unlicensed copies of an Apple operating system, or to * circumvent, violate, or enable the circumvention or violation of, any * terms of an Apple operating system software license agreement. * * Please obtain a copy of the License at * http://www.opensource.apple.com/apsl/ and read it before using this file. * * The Original Code and all software distributed under the License are * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. * Please see the License for the specific language governing rights and * limitations under the License. * * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ */ #include #include #include #include #include #include /* NULL */ #include /* dev_t */ #include /* must come after sys/stat.h */ #include /* must come after sys/stat.h */ #include #include #include #include #include #include #include static int mt_cdev_open(dev_t dev, int flags, int devtype, proc_t p); static int mt_cdev_close(dev_t dev, int flags, int devtype, proc_t p); static int mt_cdev_ioctl(dev_t dev, unsigned long cmd, char *uptr, int fflag, proc_t p); #define MT_NODE "monotonic" static const struct cdevsw mt_cdevsw = { .d_open = mt_cdev_open, .d_close = mt_cdev_close, .d_ioctl = mt_cdev_ioctl, .d_read = eno_rdwrt, .d_write = eno_rdwrt, .d_stop = eno_stop, .d_reset = eno_reset, .d_ttys = NULL, .d_select = eno_select, .d_mmap = eno_mmap, .d_strategy = eno_strat, .d_type = 0 }; /* * Written at initialization, read-only thereafter. */ LCK_GRP_DECLARE(mt_lock_grp, MT_NODE); static int mt_dev_major; static mt_device_t mt_get_device(dev_t devnum) { const int minor_dev = minor(devnum); if (minor_dev < 0 || minor_dev >= MT_NDEVS) { return NULL; } return &mt_devices[minor_dev]; } static void mt_device_lock(mt_device_t dev) { lck_mtx_lock(&dev->mtd_lock); } static void mt_device_unlock(mt_device_t dev) { lck_mtx_unlock(&dev->mtd_lock); } static void mt_device_assert_lock_held(__assert_only mt_device_t dev) { LCK_MTX_ASSERT(&dev->mtd_lock, LCK_MTX_ASSERT_OWNED); } static void mt_device_assert_inuse(__assert_only mt_device_t dev) { assert(dev->mtd_inuse == true); } int mt_dev_init(void) { mt_dev_major = cdevsw_add(-1 /* allocate a major number */, &mt_cdevsw); if (mt_dev_major < 0) { panic("monotonic: cdevsw_add failed: %d", mt_dev_major); } for (int i = 0; i < MT_NDEVS; i++) { if (mt_devices[i].mtd_init(&mt_devices[i])) { continue; } assert(mt_devices[i].mtd_ncounters > 0); dev_t dev = makedev(mt_dev_major, i); void *node = devfs_make_node(dev, DEVFS_CHAR, UID_ROOT, GID_WINDOWSERVER, 0666, MT_NODE "/%s", mt_devices[i].mtd_name); if (!node) { panic("monotonic: devfs_make_node failed for '%s'", mt_devices[i].mtd_name); } lck_mtx_init(&mt_devices[i].mtd_lock, &mt_lock_grp, LCK_ATTR_NULL); } return 0; } static int mt_cdev_open(dev_t devnum, __unused int flags, __unused int devtype, __unused proc_t p) { int error = 0; mt_device_t dev = mt_get_device(devnum); if (!dev) { return ENODEV; } if (!cpc_hw_acquire(CPC_HW_UPMU, "monotonic")) { return EBUSY; } mt_device_lock(dev); if (dev->mtd_inuse) { error = EALREADY; } else if (!mt_acquire_counters()) { error = ECONNREFUSED; } else { dev->mtd_reset(); dev->mtd_inuse = true; } mt_device_unlock(dev); if (error != 0) { cpc_hw_release(CPC_HW_UPMU, "monotonic"); } return error; } static int mt_cdev_close(dev_t devnum, __unused int flags, __unused int devtype, __unused struct proc *p) { mt_device_t dev = mt_get_device(devnum); if (!dev) { return ENODEV; } cpc_hw_release(CPC_HW_UPMU, "monotonic"); mt_device_lock(dev); mt_device_assert_inuse(dev); dev->mtd_inuse = false; dev->mtd_reset(); mt_release_counters(); mt_device_unlock(dev); return 0; } static int mt_ctl_add(mt_device_t dev, user_addr_t uptr) { int error; uint32_t ctr; union monotonic_ctl_add ctl; mt_device_assert_lock_held(dev); error = copyin(uptr, &ctl, sizeof(ctl.in)); if (error) { return error; } error = dev->mtd_add(&ctl.in.config, &ctr); if (error) { return error; } ctl.out.ctr = ctr; error = copyout(&ctl, uptr, sizeof(ctl.out)); if (error) { return error; } return 0; } static int mt_ctl_counts(mt_device_t dev, user_addr_t uptr) { int error; union monotonic_ctl_counts ctl; mt_device_assert_lock_held(dev); error = copyin(uptr, &ctl, sizeof(ctl.in)); if (error) { return error; } if (ctl.in.ctr_mask == 0) { return EINVAL; } { uint64_t counts[dev->mtd_nmonitors][dev->mtd_ncounters]; memset(counts, 0, dev->mtd_ncounters * dev->mtd_nmonitors * sizeof(counts[0][0])); error = dev->mtd_read(ctl.in.ctr_mask, (uint64_t *)counts); if (error) { return error; } error = copyout(&counts, uptr, sizeof(counts)); if (error) { return error; } } return 0; } static int mt_ctl_enable(mt_device_t dev, user_addr_t uptr) { int error; union monotonic_ctl_enable ctl; mt_device_assert_lock_held(dev); error = copyin(uptr, &ctl, sizeof(ctl)); if (error) { return error; } dev->mtd_enable(ctl.in.enable); return 0; } static int mt_ctl_reset(mt_device_t dev) { mt_device_assert_lock_held(dev); dev->mtd_reset(); return 0; } static int mt_cdev_ioctl(dev_t devnum, unsigned long cmd, char *arg, __unused int flags, __unused proc_t p) { int error = ENODEV; user_addr_t uptr = *(user_addr_t *)(void *)arg; mt_device_t dev = mt_get_device(devnum); if (!dev) { return ENODEV; } mt_device_lock(dev); switch (cmd) { case MT_IOC_RESET: error = mt_ctl_reset(dev); break; case MT_IOC_ADD: error = mt_ctl_add(dev, uptr); break; case MT_IOC_ENABLE: error = mt_ctl_enable(dev, uptr); break; case MT_IOC_COUNTS: error = mt_ctl_counts(dev, uptr); break; case MT_IOC_GET_INFO: { union monotonic_ctl_info info = { .out = { .nmonitors = dev->mtd_nmonitors, .ncounters = dev->mtd_ncounters, }, }; error = copyout(&info, uptr, sizeof(info)); break; } default: error = ENODEV; break; } mt_device_unlock(dev); return error; } static void _convert_usage_to_counts(struct recount_usage *usage, uint64_t *counts) { #if CONFIG_PERVASIVE_CPI counts[MT_CORE_INSTRS] = usage->ru_metrics[RCT_LVL_KERNEL].rm_instructions; counts[MT_CORE_CYCLES] = usage->ru_metrics[RCT_LVL_KERNEL].rm_cycles; #else /* CONFIG_PERVASIVE_CPI */ #pragma unused(usage, counts) #endif /* !CONFIG_PERVASIVE_CPI */ } enum mt_sysctl { MT_SUPPORTED, MT_PMIS, MT_RETROGRADE, MT_TASK_THREAD, MT_DEBUG, MT_KDBG_TEST, MT_FIX_CPU_PERF, MT_FIX_THREAD_PERF, MT_FIX_TASK_PERF, }; static int mt_sysctl SYSCTL_HANDLER_ARGS { #pragma unused(oidp, arg2) uint64_t start[MT_CORE_NFIXED] = { 0 }, end[MT_CORE_NFIXED] = { 0 }; uint64_t counts[2] = { 0 }; switch ((enum mt_sysctl)arg1) { case MT_SUPPORTED: return sysctl_io_number(req, (int)mt_core_supported, sizeof(int), NULL, NULL); case MT_PMIS: return sysctl_io_number(req, mt_count_pmis(), sizeof(uint64_t), NULL, NULL); case MT_RETROGRADE: { uint64_t value = os_atomic_load_wide(&mt_retrograde, relaxed); return sysctl_io_number(req, value, sizeof(mt_retrograde), NULL, NULL); } case MT_TASK_THREAD: return sysctl_io_number(req, (int)mt_core_supported, sizeof(int), NULL, NULL); case MT_DEBUG: { int value = mt_debug; int r = sysctl_io_number(req, value, sizeof(value), &value, NULL); if (r) { return r; } mt_debug = value; return 0; } case MT_KDBG_TEST: { if (req->newptr == USER_ADDR_NULL) { return EINVAL; } int intrs_en = ml_set_interrupts_enabled(FALSE); MT_KDBG_TMPCPU_START(0x3fff); MT_KDBG_TMPCPU_END(0x3fff); ml_set_interrupts_enabled(intrs_en); return 0; } case MT_FIX_CPU_PERF: { int intrs_en = ml_set_interrupts_enabled(FALSE); mt_fixed_counts(start); mt_fixed_counts(end); ml_set_interrupts_enabled(intrs_en); goto copyout_counts; } case MT_FIX_THREAD_PERF: { int intrs_en = ml_set_interrupts_enabled(FALSE); struct recount_usage start_usage = { 0 }; struct recount_usage end_usage = { 0 }; recount_current_thread_usage(&start_usage); recount_current_thread_usage(&end_usage); ml_set_interrupts_enabled(intrs_en); _convert_usage_to_counts(&start_usage, start); _convert_usage_to_counts(&end_usage, end); goto copyout_counts; } case MT_FIX_TASK_PERF: { int intrs_en = ml_set_interrupts_enabled(FALSE); struct recount_usage start_usage = { 0 }; struct recount_usage end_usage = { 0 }; recount_current_task_usage(&start_usage); recount_current_task_usage(&end_usage); ml_set_interrupts_enabled(intrs_en); _convert_usage_to_counts(&start_usage, start); _convert_usage_to_counts(&end_usage, end); goto copyout_counts; } default: return ENOENT; } copyout_counts: counts[0] = end[MT_CORE_INSTRS] - start[MT_CORE_INSTRS]; counts[1] = end[MT_CORE_CYCLES] - start[MT_CORE_CYCLES]; return copyout(counts, req->oldptr, MIN(req->oldlen, sizeof(counts))); } SYSCTL_DECL(_kern_monotonic); SYSCTL_NODE(_kern, OID_AUTO, monotonic, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "monotonic"); #define MT_SYSCTL(NAME, ARG, FLAGS, SIZE, SIZESTR, DESC) \ SYSCTL_PROC(_kern_monotonic, OID_AUTO, NAME, \ CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | (FLAGS), \ (void *)(ARG), SIZE, mt_sysctl, SIZESTR, DESC) MT_SYSCTL(supported, MT_SUPPORTED, 0, sizeof(int), "I", "whether monotonic is supported"); MT_SYSCTL(debug, MT_DEBUG, CTLFLAG_MASKED, sizeof(int), "I", "whether monotonic is printing debug messages"); MT_SYSCTL(pmis, MT_PMIS, 0, sizeof(uint64_t), "Q", "number of PMIs seen"); MT_SYSCTL(retrograde_updates, MT_RETROGRADE, 0, sizeof(uint64_t), "Q", "number of times a counter appeared to go backwards"); MT_SYSCTL(task_thread_counting, MT_TASK_THREAD, 0, sizeof(int), "I", "whether task and thread counting is enabled"); MT_SYSCTL(kdebug_test, MT_KDBG_TEST, CTLFLAG_MASKED, sizeof(int), "O", "whether task and thread counting is enabled"); MT_SYSCTL(fixed_cpu_perf, MT_FIX_CPU_PERF, CTLFLAG_MASKED, sizeof(uint64_t) * 2, "O", "overhead of accessing the current CPU's counters"); MT_SYSCTL(fixed_thread_perf, MT_FIX_THREAD_PERF, CTLFLAG_MASKED, sizeof(uint64_t) * 2, "O", "overhead of accessing the current thread's counters"); MT_SYSCTL(fixed_task_perf, MT_FIX_TASK_PERF, CTLFLAG_MASKED, sizeof(uint64_t) * 2, "O", "overhead of accessing the current task's counters");