Index: head/sys/sys/vmmeter.h =================================================================== --- head/sys/sys/vmmeter.h (revision 171632) +++ head/sys/sys/vmmeter.h (revision 171633) @@ -1,216 +1,217 @@ /*- * Copyright (c) 1982, 1986, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)vmmeter.h 8.2 (Berkeley) 7/10/94 * $FreeBSD$ */ #ifndef _SYS_VMMETER_H_ #define _SYS_VMMETER_H_ /* * System wide statistics counters. * Locking: * a - locked by atomic operations * c - constant after initialization * f - locked by vm_page_queue_free_mtx * p - locked by being in the PCPU and atomicity respect to interrupts * q - locked by vm_page_queue_mtx */ struct vmmeter { /* * General system activity. */ u_int v_swtch; /* (p) context switches */ u_int v_trap; /* (p) calls to trap */ u_int v_syscall; /* (p) calls to syscall() */ u_int v_intr; /* (p) device interrupts */ u_int v_soft; /* (p) software interrupts */ /* * Virtual memory activity. */ u_int v_vm_faults; /* (p) address memory faults */ u_int v_cow_faults; /* (p) copy-on-writes faults */ u_int v_cow_optim; /* (p) optimized copy-on-writes faults */ u_int v_zfod; /* (p) pages zero filled on demand */ u_int v_ozfod; /* (p) optimized zero fill pages */ u_int v_swapin; /* (p) swap pager pageins */ u_int v_swapout; /* (p) swap pager pageouts */ u_int v_swappgsin; /* (p) swap pager pages paged in */ u_int v_swappgsout; /* (p) swap pager pages paged out */ u_int v_vnodein; /* (p) vnode pager pageins */ u_int v_vnodeout; /* (p) vnode pager pageouts */ u_int v_vnodepgsin; /* (p) vnode_pager pages paged in */ u_int v_vnodepgsout; /* (p) vnode pager pages paged out */ u_int v_intrans; /* (p) intransit blocking page faults */ u_int v_reactivated; /* (q) pages reactivated from free list */ u_int v_pdwakeups; /* (f) times daemon has awaken from sleep */ u_int v_pdpages; /* (q) pages analyzed by daemon */ + u_int v_tcached; /* (q) total pages cached */ u_int v_dfree; /* (q) pages freed by daemon */ u_int v_pfree; /* (q) pages freed by exiting processes */ u_int v_tfree; /* (p) total pages freed */ /* * Distribution of page usages. */ u_int v_page_size; /* (c) page size in bytes */ u_int v_page_count; /* (c) total number of pages in system */ u_int v_free_reserved; /* (c) pages reserved for deadlock */ u_int v_free_target; /* (c) pages desired free */ u_int v_free_min; /* (c) pages desired free */ u_int v_free_count; /* (f) pages free */ u_int v_wire_count; /* (a) pages wired down */ u_int v_active_count; /* (q) pages active */ u_int v_inactive_target; /* (c) pages desired inactive */ u_int v_inactive_count; /* (q) pages inactive */ u_int v_cache_count; /* (q) pages on buffer cache queue */ u_int v_cache_min; /* (c) min pages desired on cache queue */ u_int v_cache_max; /* (c) max pages in cached obj */ u_int v_pageout_free_min; /* (c) min pages reserved for kernel */ u_int v_interrupt_free_min; /* (c) reserved pages for int code */ u_int v_free_severe; /* (c) severe page depletion point */ /* * Fork/vfork/rfork activity. */ u_int v_forks; /* (p) fork() calls */ u_int v_vforks; /* (p) vfork() calls */ u_int v_rforks; /* (p) rfork() calls */ u_int v_kthreads; /* (p) fork() calls by kernel */ u_int v_forkpages; /* (p) VM pages affected by fork() */ u_int v_vforkpages; /* (p) VM pages affected by vfork() */ u_int v_rforkpages; /* (p) VM pages affected by rfork() */ u_int v_kthreadpages; /* (p) VM pages affected by fork() by kernel */ }; #ifdef _KERNEL extern struct vmmeter cnt; /* * Return TRUE if we are under our reserved low-free-pages threshold */ static __inline int vm_page_count_reserved(void) { return (cnt.v_free_reserved > (cnt.v_free_count + cnt.v_cache_count)); } /* * Return TRUE if we are under our severe low-free-pages threshold * * This routine is typically used at the user<->system interface to determine * whether we need to block in order to avoid a low memory deadlock. */ static __inline int vm_page_count_severe(void) { return (cnt.v_free_severe > (cnt.v_free_count + cnt.v_cache_count)); } /* * Return TRUE if we are under our minimum low-free-pages threshold. * * This routine is typically used within the system to determine whether * we can execute potentially very expensive code in terms of memory. It * is also used by the pageout daemon to calculate when to sleep, when * to wake waiters up, and when (after making a pass) to become more * desparate. */ static __inline int vm_page_count_min(void) { return (cnt.v_free_min > (cnt.v_free_count + cnt.v_cache_count)); } /* * Return TRUE if we have not reached our free page target during * free page recovery operations. */ static __inline int vm_page_count_target(void) { return (cnt.v_free_target > (cnt.v_free_count + cnt.v_cache_count)); } /* * Return the number of pages we need to free-up or cache * A positive number indicates that we do not have enough free pages. */ static __inline int vm_paging_target(void) { return ( (cnt.v_free_target + cnt.v_cache_min) - (cnt.v_free_count + cnt.v_cache_count) ); } /* * Returns TRUE if the pagedaemon needs to be woken up. */ static __inline int vm_paging_needed(void) { return ( (cnt.v_free_reserved + cnt.v_cache_min) > (cnt.v_free_count + cnt.v_cache_count) ); } #endif /* systemwide totals computed every five seconds */ struct vmtotal { int16_t t_rq; /* length of the run queue */ int16_t t_dw; /* jobs in ``disk wait'' (neg priority) */ int16_t t_pw; /* jobs in page wait */ int16_t t_sl; /* jobs sleeping in core */ int16_t t_sw; /* swapped out runnable/short block jobs */ int32_t t_vm; /* total virtual memory */ int32_t t_avm; /* active virtual memory */ int32_t t_rm; /* total real memory in use */ int32_t t_arm; /* active real memory */ int32_t t_vmshr; /* shared virtual memory */ int32_t t_avmshr; /* active shared virtual memory */ int32_t t_rmshr; /* shared real memory */ int32_t t_armshr; /* active shared real memory */ int32_t t_free; /* free memory pages */ }; #endif Index: head/sys/vm/vm_meter.c =================================================================== --- head/sys/vm/vm_meter.c (revision 171632) +++ head/sys/vm/vm_meter.c (revision 171633) @@ -1,386 +1,388 @@ /*- * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)vm_meter.c 8.4 (Berkeley) 1/4/94 */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct vmmeter cnt; int maxslp = MAXSLP; SYSCTL_UINT(_vm, VM_V_FREE_MIN, v_free_min, CTLFLAG_RW, &cnt.v_free_min, 0, ""); SYSCTL_UINT(_vm, VM_V_FREE_TARGET, v_free_target, CTLFLAG_RW, &cnt.v_free_target, 0, ""); SYSCTL_UINT(_vm, VM_V_FREE_RESERVED, v_free_reserved, CTLFLAG_RW, &cnt.v_free_reserved, 0, ""); SYSCTL_UINT(_vm, VM_V_INACTIVE_TARGET, v_inactive_target, CTLFLAG_RW, &cnt.v_inactive_target, 0, ""); SYSCTL_UINT(_vm, VM_V_CACHE_MIN, v_cache_min, CTLFLAG_RW, &cnt.v_cache_min, 0, ""); SYSCTL_UINT(_vm, VM_V_CACHE_MAX, v_cache_max, CTLFLAG_RW, &cnt.v_cache_max, 0, ""); SYSCTL_UINT(_vm, VM_V_PAGEOUT_FREE_MIN, v_pageout_free_min, CTLFLAG_RW, &cnt.v_pageout_free_min, 0, ""); SYSCTL_UINT(_vm, OID_AUTO, v_free_severe, CTLFLAG_RW, &cnt.v_free_severe, 0, ""); static int sysctl_vm_loadavg(SYSCTL_HANDLER_ARGS) { #ifdef SCTL_MASK32 u_int32_t la[4]; if (req->flags & SCTL_MASK32) { la[0] = averunnable.ldavg[0]; la[1] = averunnable.ldavg[1]; la[2] = averunnable.ldavg[2]; la[3] = averunnable.fscale; return SYSCTL_OUT(req, la, sizeof(la)); } else #endif return SYSCTL_OUT(req, &averunnable, sizeof(averunnable)); } SYSCTL_PROC(_vm, VM_LOADAVG, loadavg, CTLTYPE_STRUCT|CTLFLAG_RD, NULL, 0, sysctl_vm_loadavg, "S,loadavg", "Machine loadaverage history"); static int vmtotal(SYSCTL_HANDLER_ARGS) { /* XXXKSE almost completely broken */ struct proc *p; struct vmtotal total; vm_map_entry_t entry; vm_object_t object; vm_map_t map; int paging; struct thread *td; struct vmspace *vm; bzero(&total, sizeof(total)); /* * Mark all objects as inactive. */ GIANT_REQUIRED; mtx_lock(&vm_object_list_mtx); TAILQ_FOREACH(object, &vm_object_list, object_list) { if (!VM_OBJECT_TRYLOCK(object)) { /* * Avoid a lock-order reversal. Consequently, * the reported number of active pages may be * greater than the actual number. */ continue; } vm_object_clear_flag(object, OBJ_ACTIVE); VM_OBJECT_UNLOCK(object); } mtx_unlock(&vm_object_list_mtx); /* * Calculate process statistics. */ sx_slock(&allproc_lock); FOREACH_PROC_IN_SYSTEM(p) { if (p->p_flag & P_SYSTEM) continue; PROC_SLOCK(p); switch (p->p_state) { case PRS_NEW: PROC_SUNLOCK(p); continue; break; default: FOREACH_THREAD_IN_PROC(p, td) { /* Need new statistics XXX */ thread_lock(td); switch (td->td_state) { case TDS_INHIBITED: /* * XXX stats no longer synchronized. */ if (TD_ON_LOCK(td) || (td->td_inhibitors == TDI_SWAPPED)) { total.t_sw++; } else if (TD_IS_SLEEPING(td) || TD_AWAITING_INTR(td) || TD_IS_SUSPENDED(td)) { if (td->td_priority <= PZERO) total.t_dw++; else total.t_sl++; } break; case TDS_CAN_RUN: total.t_sw++; break; case TDS_RUNQ: case TDS_RUNNING: total.t_rq++; thread_unlock(td); continue; default: break; } thread_unlock(td); } } PROC_SUNLOCK(p); /* * Note active objects. */ paging = 0; vm = vmspace_acquire_ref(p); if (vm == NULL) continue; map = &vm->vm_map; vm_map_lock_read(map); for (entry = map->header.next; entry != &map->header; entry = entry->next) { if ((entry->eflags & MAP_ENTRY_IS_SUB_MAP) || (object = entry->object.vm_object) == NULL) continue; VM_OBJECT_LOCK(object); vm_object_set_flag(object, OBJ_ACTIVE); paging |= object->paging_in_progress; VM_OBJECT_UNLOCK(object); } vm_map_unlock_read(map); vmspace_free(vm); if (paging) total.t_pw++; } sx_sunlock(&allproc_lock); /* * Calculate object memory usage statistics. */ mtx_lock(&vm_object_list_mtx); TAILQ_FOREACH(object, &vm_object_list, object_list) { /* * Perform unsynchronized reads on the object to avoid * a lock-order reversal. In this case, the lack of * synchronization should not impair the accuracy of * the reported statistics. */ if (object->type == OBJT_DEVICE) { /* * Devices, like /dev/mem, will badly skew our totals. */ continue; } if (object->ref_count == 0) { /* * Also skip unreferenced objects, including * vnodes representing mounted file systems. */ continue; } total.t_vm += object->size; total.t_rm += object->resident_page_count; if (object->flags & OBJ_ACTIVE) { total.t_avm += object->size; total.t_arm += object->resident_page_count; } if (object->shadow_count > 1) { /* shared object */ total.t_vmshr += object->size; total.t_rmshr += object->resident_page_count; if (object->flags & OBJ_ACTIVE) { total.t_avmshr += object->size; total.t_armshr += object->resident_page_count; } } } mtx_unlock(&vm_object_list_mtx); total.t_free = cnt.v_free_count + cnt.v_cache_count; return (sysctl_handle_opaque(oidp, &total, sizeof(total), req)); } /* * vcnt() - accumulate statistics from all cpus and the global cnt * structure. * * The vmmeter structure is now per-cpu as well as global. Those * statistics which can be kept on a per-cpu basis (to avoid cache * stalls between cpus) can be moved to the per-cpu vmmeter. Remaining * statistics, such as v_free_reserved, are left in the global * structure. * * (sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req) */ static int vcnt(SYSCTL_HANDLER_ARGS) { int count = *(int *)arg1; int offset = (char *)arg1 - (char *)&cnt; #ifdef SMP int i; for (i = 0; i < mp_ncpus; ++i) { struct pcpu *pcpu = pcpu_find(i); count += *(int *)((char *)&pcpu->pc_cnt + offset); } #else count += *(int *)((char *)PCPU_PTR(cnt) + offset); #endif return (SYSCTL_OUT(req, &count, sizeof(int))); } SYSCTL_PROC(_vm, VM_TOTAL, vmtotal, CTLTYPE_OPAQUE|CTLFLAG_RD, 0, sizeof(struct vmtotal), vmtotal, "S,vmtotal", "System virtual memory statistics"); SYSCTL_NODE(_vm, OID_AUTO, stats, CTLFLAG_RW, 0, "VM meter stats"); static SYSCTL_NODE(_vm_stats, OID_AUTO, sys, CTLFLAG_RW, 0, "VM meter sys stats"); static SYSCTL_NODE(_vm_stats, OID_AUTO, vm, CTLFLAG_RW, 0, "VM meter vm stats"); SYSCTL_NODE(_vm_stats, OID_AUTO, misc, CTLFLAG_RW, 0, "VM meter misc stats"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_swtch, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swtch, 0, vcnt, "IU", "Context switches"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_trap, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_trap, 0, vcnt, "IU", "Traps"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_syscall, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_syscall, 0, vcnt, "IU", "Syscalls"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_intr, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_intr, 0, vcnt, "IU", "Hardware interrupts"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_soft, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_soft, 0, vcnt, "IU", "Software interrupts"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vm_faults, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vm_faults, 0, vcnt, "IU", "VM faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cow_faults, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cow_faults, 0, vcnt, "IU", "COW faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cow_optim, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cow_optim, 0, vcnt, "IU", "Optimized COW faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_zfod, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_zfod, 0, vcnt, "IU", "Zero fill"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_ozfod, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_ozfod, 0, vcnt, "IU", "Optimized zero fill"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swapin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swapin, 0, vcnt, "IU", "Swapin operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swapout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swapout, 0, vcnt, "IU", "Swapout operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swappgsin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swappgsin, 0, vcnt, "IU", "Swapin pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swappgsout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swappgsout, 0, vcnt, "IU", "Swapout pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodein, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodein, 0, vcnt, "IU", "Vnodein operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodeout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodeout, 0, vcnt, "IU", "Vnodeout operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodepgsin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodepgsin, 0, vcnt, "IU", "Vnodein pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodepgsout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodepgsout, 0, vcnt, "IU", "Vnodeout pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_intrans, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_intrans, 0, vcnt, "IU", "In transit page blocking"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_reactivated, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_reactivated, 0, vcnt, "IU", "Reactivated pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pdwakeups, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pdwakeups, 0, vcnt, "IU", "Pagedaemon wakeups"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pdpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pdpages, 0, vcnt, "IU", "Pagedaemon page scans"); +SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_tcached, CTLTYPE_UINT|CTLFLAG_RD, + &cnt.v_tcached, 0, vcnt, "IU", "Total pages cached"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_dfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_dfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_tfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_tfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_page_size, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_page_size, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_page_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_page_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_reserved, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_reserved, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_target, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_target, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_wire_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_wire_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_active_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_active_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_inactive_target, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_inactive_target, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_inactive_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_inactive_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_max, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_max, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pageout_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pageout_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_interrupt_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_interrupt_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_forks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_forks, 0, vcnt, "IU", "Number of fork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vforks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vforks, 0, vcnt, "IU", "Number of vfork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_rforks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_rforks, 0, vcnt, "IU", "Number of rfork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_kthreads, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_kthreads, 0, vcnt, "IU", "Number of fork() calls by kernel"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_forkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_forkpages, 0, vcnt, "IU", "VM pages affected by fork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vforkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vforkpages, 0, vcnt, "IU", "VM pages affected by vfork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_rforkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_rforkpages, 0, vcnt, "IU", "VM pages affected by rfork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_kthreadpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_kthreadpages, 0, vcnt, "IU", "VM pages affected by fork() by kernel"); SYSCTL_INT(_vm_stats_misc, OID_AUTO, zero_page_count, CTLFLAG_RD, &vm_page_zero_count, 0, ""); Index: head/sys/vm/vm_page.c =================================================================== --- head/sys/vm/vm_page.c (revision 171632) +++ head/sys/vm/vm_page.c (revision 171633) @@ -1,1822 +1,1823 @@ /*- * Copyright (c) 1991 Regents of the University of California. * All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_page.c 7.4 (Berkeley) 5/7/91 */ /*- * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * GENERAL RULES ON VM_PAGE MANIPULATION * * - a pageq mutex is required when adding or removing a page from a * page queue (vm_page_queue[]), regardless of other mutexes or the * busy state of a page. * * - a hash chain mutex is required when associating or disassociating * a page from the VM PAGE CACHE hash table (vm_page_buckets), * regardless of other mutexes or the busy state of a page. * * - either a hash chain mutex OR a busied page is required in order * to modify the page flags. A hash chain mutex must be obtained in * order to busy a page. A page's flags cannot be modified by a * hash chain mutex if the page is marked busy. * * - The object memq mutex is held when inserting or removing * pages from an object (vm_page_insert() or vm_page_remove()). This * is different from the object's main mutex. * * Generally speaking, you have to be aware of side effects when running * vm_page ops. A vm_page_lookup() will return with the hash chain * locked, whether it was able to lookup the page or not. vm_page_free(), * vm_page_cache(), vm_page_activate(), and a number of other routines * will release the hash chain mutex for you. Intermediate manipulation * routines such as vm_page_flag_set() expect the hash chain to be held * on entry and the hash chain will remain held on return. * * pageq scanning can only occur with the pageq in question locked. * We have a known bottleneck with the active queue, but the cache * and free queues are actually arrays already. */ /* * Resident memory management module. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Associated with page of user-allocatable memory is a * page structure. */ struct mtx vm_page_queue_mtx; struct mtx vm_page_queue_free_mtx; vm_page_t vm_page_array = 0; int vm_page_array_size = 0; long first_page = 0; int vm_page_zero_count = 0; static int boot_pages = UMA_BOOT_PAGES; TUNABLE_INT("vm.boot_pages", &boot_pages); SYSCTL_INT(_vm, OID_AUTO, boot_pages, CTLFLAG_RD, &boot_pages, 0, "number of pages allocated for bootstrapping the VM system"); /* * vm_set_page_size: * * Sets the page size, perhaps based upon the memory * size. Must be called before any use of page-size * dependent functions. */ void vm_set_page_size(void) { if (cnt.v_page_size == 0) cnt.v_page_size = PAGE_SIZE; if (((cnt.v_page_size - 1) & cnt.v_page_size) != 0) panic("vm_set_page_size: page size not a power of two"); } /* * vm_page_blacklist_lookup: * * See if a physical address in this page has been listed * in the blacklist tunable. Entries in the tunable are * separated by spaces or commas. If an invalid integer is * encountered then the rest of the string is skipped. */ static int vm_page_blacklist_lookup(char *list, vm_paddr_t pa) { vm_paddr_t bad; char *cp, *pos; for (pos = list; *pos != '\0'; pos = cp) { bad = strtoq(pos, &cp, 0); if (*cp != '\0') { if (*cp == ' ' || *cp == ',') { cp++; if (cp == pos) continue; } else break; } if (pa == trunc_page(bad)) return (1); } return (0); } /* * vm_page_startup: * * Initializes the resident memory module. * * Allocates memory for the page cells, and * for the object/offset-to-page hash table headers. * Each page cell is initialized and placed on the free list. */ vm_offset_t vm_page_startup(vm_offset_t vaddr) { vm_offset_t mapped; vm_size_t npages; vm_paddr_t page_range; vm_paddr_t new_end; int i; vm_paddr_t pa; int nblocks; vm_paddr_t last_pa; char *list; /* the biggest memory array is the second group of pages */ vm_paddr_t end; vm_paddr_t biggestsize; vm_paddr_t low_water, high_water; int biggestone; vm_paddr_t total; total = 0; biggestsize = 0; biggestone = 0; nblocks = 0; vaddr = round_page(vaddr); for (i = 0; phys_avail[i + 1]; i += 2) { phys_avail[i] = round_page(phys_avail[i]); phys_avail[i + 1] = trunc_page(phys_avail[i + 1]); } low_water = phys_avail[0]; high_water = phys_avail[1]; for (i = 0; phys_avail[i + 1]; i += 2) { vm_paddr_t size = phys_avail[i + 1] - phys_avail[i]; if (size > biggestsize) { biggestone = i; biggestsize = size; } if (phys_avail[i] < low_water) low_water = phys_avail[i]; if (phys_avail[i + 1] > high_water) high_water = phys_avail[i + 1]; ++nblocks; total += size; } end = phys_avail[biggestone+1]; /* * Initialize the locks. */ mtx_init(&vm_page_queue_mtx, "vm page queue mutex", NULL, MTX_DEF | MTX_RECURSE); mtx_init(&vm_page_queue_free_mtx, "vm page queue free mutex", NULL, MTX_DEF); /* * Initialize the queue headers for the free queue, the active queue * and the inactive queue. */ vm_pageq_init(); /* * Allocate memory for use when boot strapping the kernel memory * allocator. */ new_end = end - (boot_pages * UMA_SLAB_SIZE); new_end = trunc_page(new_end); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)mapped, end - new_end); uma_startup((void *)mapped, boot_pages); #if defined(__amd64__) || defined(__i386__) /* * Allocate a bitmap to indicate that a random physical page * needs to be included in a minidump. * * The amd64 port needs this to indicate which direct map pages * need to be dumped, via calls to dump_add_page()/dump_drop_page(). * * However, i386 still needs this workspace internally within the * minidump code. In theory, they are not needed on i386, but are * included should the sf_buf code decide to use them. */ page_range = phys_avail[(nblocks - 1) * 2 + 1] / PAGE_SIZE; vm_page_dump_size = round_page(roundup2(page_range, NBBY) / NBBY); new_end -= vm_page_dump_size; vm_page_dump = (void *)(uintptr_t)pmap_map(&vaddr, new_end, new_end + vm_page_dump_size, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)vm_page_dump, vm_page_dump_size); #endif /* * Compute the number of pages of memory that will be available for * use (taking into account the overhead of a page structure per * page). */ first_page = low_water / PAGE_SIZE; #ifdef VM_PHYSSEG_SPARSE page_range = 0; for (i = 0; phys_avail[i + 1] != 0; i += 2) page_range += atop(phys_avail[i + 1] - phys_avail[i]); #elif defined(VM_PHYSSEG_DENSE) page_range = high_water / PAGE_SIZE - first_page; #else #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." #endif npages = (total - (page_range * sizeof(struct vm_page)) - (end - new_end)) / PAGE_SIZE; end = new_end; /* * Reserve an unmapped guard page to trap access to vm_page_array[-1]. */ vaddr += PAGE_SIZE; /* * Initialize the mem entry structures now, and put them in the free * queue. */ new_end = trunc_page(end - page_range * sizeof(struct vm_page)); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); vm_page_array = (vm_page_t) mapped; #ifdef __amd64__ /* * pmap_map on amd64 comes out of the direct-map, not kvm like i386, * so the pages must be tracked for a crashdump to include this data. * This includes the vm_page_array and the early UMA bootstrap pages. */ for (pa = new_end; pa < phys_avail[biggestone + 1]; pa += PAGE_SIZE) dump_add_page(pa); #endif phys_avail[biggestone + 1] = new_end; /* * Clear all of the page structures */ bzero((caddr_t) vm_page_array, page_range * sizeof(struct vm_page)); for (i = 0; i < page_range; i++) vm_page_array[i].order = VM_NFREEORDER; vm_page_array_size = page_range; /* * This assertion tests the hypothesis that npages and total are * redundant. XXX */ page_range = 0; for (i = 0; phys_avail[i + 1] != 0; i += 2) page_range += atop(phys_avail[i + 1] - phys_avail[i]); KASSERT(page_range == npages, ("vm_page_startup: inconsistent page counts")); /* * Initialize the physical memory allocator. */ vm_phys_init(); /* * Add every available physical page that is not blacklisted to * the free lists. */ cnt.v_page_count = 0; cnt.v_free_count = 0; list = getenv("vm.blacklist"); for (i = 0; phys_avail[i + 1] != 0; i += 2) { pa = phys_avail[i]; last_pa = phys_avail[i + 1]; while (pa < last_pa) { if (list != NULL && vm_page_blacklist_lookup(list, pa)) printf("Skipping page with pa 0x%jx\n", (uintmax_t)pa); else vm_phys_add_page(pa); pa += PAGE_SIZE; } } freeenv(list); return (vaddr); } void vm_page_flag_set(vm_page_t m, unsigned short bits) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); m->flags |= bits; } void vm_page_flag_clear(vm_page_t m, unsigned short bits) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); m->flags &= ~bits; } void vm_page_busy(vm_page_t m) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); KASSERT((m->oflags & VPO_BUSY) == 0, ("vm_page_busy: page already busy!!!")); m->oflags |= VPO_BUSY; } /* * vm_page_flash: * * wakeup anyone waiting for the page. */ void vm_page_flash(vm_page_t m) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (m->oflags & VPO_WANTED) { m->oflags &= ~VPO_WANTED; wakeup(m); } } /* * vm_page_wakeup: * * clear the VPO_BUSY flag and wakeup anyone waiting for the * page. * */ void vm_page_wakeup(vm_page_t m) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); KASSERT(m->oflags & VPO_BUSY, ("vm_page_wakeup: page not busy!!!")); m->oflags &= ~VPO_BUSY; vm_page_flash(m); } void vm_page_io_start(vm_page_t m) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); m->busy++; } void vm_page_io_finish(vm_page_t m) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); m->busy--; if (m->busy == 0) vm_page_flash(m); } /* * Keep page from being freed by the page daemon * much of the same effect as wiring, except much lower * overhead and should be used only for *very* temporary * holding ("wiring"). */ void vm_page_hold(vm_page_t mem) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); mem->hold_count++; } void vm_page_unhold(vm_page_t mem) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); --mem->hold_count; KASSERT(mem->hold_count >= 0, ("vm_page_unhold: hold count < 0!!!")); if (mem->hold_count == 0 && VM_PAGE_INQUEUE2(mem, PQ_HOLD)) vm_page_free_toq(mem); } /* * vm_page_free: * * Free a page. */ void vm_page_free(vm_page_t m) { m->flags &= ~PG_ZERO; vm_page_free_toq(m); } /* * vm_page_free_zero: * * Free a page to the zerod-pages queue */ void vm_page_free_zero(vm_page_t m) { m->flags |= PG_ZERO; vm_page_free_toq(m); } /* * vm_page_sleep: * * Sleep and release the page queues lock. * * The object containing the given page must be locked. */ void vm_page_sleep(vm_page_t m, const char *msg) { VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (!mtx_owned(&vm_page_queue_mtx)) vm_page_lock_queues(); vm_page_flag_set(m, PG_REFERENCED); vm_page_unlock_queues(); /* * It's possible that while we sleep, the page will get * unbusied and freed. If we are holding the object * lock, we will assume we hold a reference to the object * such that even if m->object changes, we can re-lock * it. */ m->oflags |= VPO_WANTED; msleep(m, VM_OBJECT_MTX(m->object), PVM, msg, 0); } /* * vm_page_dirty: * * make page all dirty */ void vm_page_dirty(vm_page_t m) { KASSERT(VM_PAGE_GETKNOWNQUEUE1(m) != PQ_CACHE, ("vm_page_dirty: page in cache!")); KASSERT(!VM_PAGE_IS_FREE(m), ("vm_page_dirty: page is free!")); m->dirty = VM_PAGE_BITS_ALL; } /* * vm_page_splay: * * Implements Sleator and Tarjan's top-down splay algorithm. Returns * the vm_page containing the given pindex. If, however, that * pindex is not found in the vm_object, returns a vm_page that is * adjacent to the pindex, coming before or after it. */ vm_page_t vm_page_splay(vm_pindex_t pindex, vm_page_t root) { struct vm_page dummy; vm_page_t lefttreemax, righttreemin, y; if (root == NULL) return (root); lefttreemax = righttreemin = &dummy; for (;; root = y) { if (pindex < root->pindex) { if ((y = root->left) == NULL) break; if (pindex < y->pindex) { /* Rotate right. */ root->left = y->right; y->right = root; root = y; if ((y = root->left) == NULL) break; } /* Link into the new root's right tree. */ righttreemin->left = root; righttreemin = root; } else if (pindex > root->pindex) { if ((y = root->right) == NULL) break; if (pindex > y->pindex) { /* Rotate left. */ root->right = y->left; y->left = root; root = y; if ((y = root->right) == NULL) break; } /* Link into the new root's left tree. */ lefttreemax->right = root; lefttreemax = root; } else break; } /* Assemble the new root. */ lefttreemax->right = root->left; righttreemin->left = root->right; root->left = dummy.right; root->right = dummy.left; return (root); } /* * vm_page_insert: [ internal use only ] * * Inserts the given mem entry into the object and object list. * * The pagetables are not updated but will presumably fault the page * in if necessary, or if a kernel page the caller will at some point * enter the page into the kernel's pmap. We are not allowed to block * here so we *can't* do this anyway. * * The object and page must be locked. * This routine may not block. */ void vm_page_insert(vm_page_t m, vm_object_t object, vm_pindex_t pindex) { vm_page_t root; VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); if (m->object != NULL) panic("vm_page_insert: page already inserted"); /* * Record the object/offset pair in this page */ m->object = object; m->pindex = pindex; /* * Now link into the object's ordered list of backed pages. */ root = object->root; if (root == NULL) { m->left = NULL; m->right = NULL; TAILQ_INSERT_TAIL(&object->memq, m, listq); } else { root = vm_page_splay(pindex, root); if (pindex < root->pindex) { m->left = root->left; m->right = root; root->left = NULL; TAILQ_INSERT_BEFORE(root, m, listq); } else if (pindex == root->pindex) panic("vm_page_insert: offset already allocated"); else { m->right = root->right; m->left = root; root->right = NULL; TAILQ_INSERT_AFTER(&object->memq, root, m, listq); } } object->root = m; object->generation++; /* * show that the object has one more resident page. */ object->resident_page_count++; /* * Hold the vnode until the last page is released. */ if (object->resident_page_count == 1 && object->type == OBJT_VNODE) vhold((struct vnode *)object->handle); /* * Since we are inserting a new and possibly dirty page, * update the object's OBJ_MIGHTBEDIRTY flag. */ if (m->flags & PG_WRITEABLE) vm_object_set_writeable_dirty(object); } /* * vm_page_remove: * NOTE: used by device pager as well -wfj * * Removes the given mem entry from the object/offset-page * table and the object page list, but do not invalidate/terminate * the backing store. * * The object and page must be locked. * The underlying pmap entry (if any) is NOT removed here. * This routine may not block. */ void vm_page_remove(vm_page_t m) { vm_object_t object; vm_page_t root; if ((object = m->object) == NULL) return; VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); if (m->oflags & VPO_BUSY) { m->oflags &= ~VPO_BUSY; vm_page_flash(m); } mtx_assert(&vm_page_queue_mtx, MA_OWNED); /* * Now remove from the object's list of backed pages. */ if (m != object->root) vm_page_splay(m->pindex, object->root); if (m->left == NULL) root = m->right; else { root = vm_page_splay(m->pindex, m->left); root->right = m->right; } object->root = root; TAILQ_REMOVE(&object->memq, m, listq); /* * And show that the object has one fewer resident page. */ object->resident_page_count--; object->generation++; /* * The vnode may now be recycled. */ if (object->resident_page_count == 0 && object->type == OBJT_VNODE) vdrop((struct vnode *)object->handle); m->object = NULL; } /* * vm_page_lookup: * * Returns the page associated with the object/offset * pair specified; if none is found, NULL is returned. * * The object must be locked. * This routine may not block. * This is a critical path routine */ vm_page_t vm_page_lookup(vm_object_t object, vm_pindex_t pindex) { vm_page_t m; VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); if ((m = object->root) != NULL && m->pindex != pindex) { m = vm_page_splay(pindex, m); if ((object->root = m)->pindex != pindex) m = NULL; } return (m); } /* * vm_page_rename: * * Move the given memory entry from its * current object to the specified target object/offset. * * The object must be locked. * This routine may not block. * * Note: swap associated with the page must be invalidated by the move. We * have to do this for several reasons: (1) we aren't freeing the * page, (2) we are dirtying the page, (3) the VM system is probably * moving the page from object A to B, and will then later move * the backing store from A to B and we can't have a conflict. * * Note: we *always* dirty the page. It is necessary both for the * fact that we moved it, and because we may be invalidating * swap. If the page is on the cache, we have to deactivate it * or vm_page_dirty() will panic. Dirty pages are not allowed * on the cache. */ void vm_page_rename(vm_page_t m, vm_object_t new_object, vm_pindex_t new_pindex) { vm_page_remove(m); vm_page_insert(m, new_object, new_pindex); if (VM_PAGE_INQUEUE1(m, PQ_CACHE)) vm_page_deactivate(m); vm_page_dirty(m); } /* * vm_page_select_cache: * * Move a page of the given color from the cache queue to the free * queue. As pages might be found, but are not applicable, they are * deactivated. * * This routine may not block. */ vm_page_t vm_page_select_cache(void) { vm_object_t object; vm_page_t m; boolean_t was_trylocked; mtx_assert(&vm_page_queue_mtx, MA_OWNED); while ((m = TAILQ_FIRST(&vm_page_queues[PQ_CACHE].pl)) != NULL) { KASSERT(m->dirty == 0, ("Found dirty cache page %p", m)); KASSERT(!pmap_page_is_mapped(m), ("Found mapped cache page %p", m)); KASSERT((m->flags & PG_UNMANAGED) == 0, ("Found unmanaged cache page %p", m)); KASSERT(m->wire_count == 0, ("Found wired cache page %p", m)); if (m->hold_count == 0 && (object = m->object, (was_trylocked = VM_OBJECT_TRYLOCK(object)) || VM_OBJECT_LOCKED(object))) { KASSERT((m->oflags & VPO_BUSY) == 0 && m->busy == 0, ("Found busy cache page %p", m)); vm_page_free(m); if (was_trylocked) VM_OBJECT_UNLOCK(object); break; } vm_page_deactivate(m); } return (m); } /* * vm_page_alloc: * * Allocate and return a memory cell associated * with this VM object/offset pair. * * page_req classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * VM_ALLOC_ZERO zero page * * This routine may not block. * * Additional special handling is required when called from an * interrupt (VM_ALLOC_INTERRUPT). We are not allowed to mess with * the page cache in this case. */ vm_page_t vm_page_alloc(vm_object_t object, vm_pindex_t pindex, int req) { vm_page_t m = NULL; int flags, page_req; page_req = req & VM_ALLOC_CLASS_MASK; KASSERT(curthread->td_intr_nesting_level == 0 || page_req == VM_ALLOC_INTERRUPT, ("vm_page_alloc(NORMAL|SYSTEM) in interrupt context")); if ((req & VM_ALLOC_NOOBJ) == 0) { KASSERT(object != NULL, ("vm_page_alloc: NULL object.")); VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); } /* * The pager is allowed to eat deeper into the free page list. */ if ((curproc == pageproc) && (page_req != VM_ALLOC_INTERRUPT)) { page_req = VM_ALLOC_SYSTEM; }; loop: mtx_lock(&vm_page_queue_free_mtx); if (cnt.v_free_count > cnt.v_free_reserved || (page_req == VM_ALLOC_SYSTEM && cnt.v_cache_count == 0 && cnt.v_free_count > cnt.v_interrupt_free_min) || (page_req == VM_ALLOC_INTERRUPT && cnt.v_free_count > 0)) { /* * Allocate from the free queue if the number of free pages * exceeds the minimum for the request class. */ m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); } else if (page_req != VM_ALLOC_INTERRUPT) { mtx_unlock(&vm_page_queue_free_mtx); /* * Allocatable from cache (non-interrupt only). On success, * we must free the page and try again, thus ensuring that * cnt.v_*_free_min counters are replenished. */ vm_page_lock_queues(); if ((m = vm_page_select_cache()) == NULL) { KASSERT(cnt.v_cache_count == 0, ("vm_page_alloc: cache queue is missing %d pages", cnt.v_cache_count)); vm_page_unlock_queues(); atomic_add_int(&vm_pageout_deficit, 1); pagedaemon_wakeup(); if (page_req != VM_ALLOC_SYSTEM) return (NULL); mtx_lock(&vm_page_queue_free_mtx); if (cnt.v_free_count <= cnt.v_interrupt_free_min) { mtx_unlock(&vm_page_queue_free_mtx); return (NULL); } m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); } else { vm_page_unlock_queues(); goto loop; } } else { /* * Not allocatable from cache from interrupt, give up. */ mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, 1); pagedaemon_wakeup(); return (NULL); } /* * At this point we had better have found a good page. */ KASSERT( m != NULL, ("vm_page_alloc(): missing page on free queue") ); KASSERT(VM_PAGE_IS_FREE(m), ("vm_page_alloc: page %p is not free", m)); /* * Initialize structure. Only the PG_ZERO flag is inherited. */ flags = 0; if (m->flags & PG_ZERO) { vm_page_zero_count--; if (req & VM_ALLOC_ZERO) flags = PG_ZERO; } if (object != NULL && object->type == OBJT_PHYS) flags |= PG_UNMANAGED; m->flags = flags; if (req & (VM_ALLOC_NOBUSY | VM_ALLOC_NOOBJ)) m->oflags = 0; else m->oflags = VPO_BUSY; if (req & VM_ALLOC_WIRED) { atomic_add_int(&cnt.v_wire_count, 1); m->wire_count = 1; } else m->wire_count = 0; m->hold_count = 0; m->act_count = 0; m->busy = 0; m->valid = 0; KASSERT(m->dirty == 0, ("vm_page_alloc: free/cache page %p was dirty", m)); mtx_unlock(&vm_page_queue_free_mtx); if ((req & VM_ALLOC_NOOBJ) == 0) vm_page_insert(m, object, pindex); else m->pindex = pindex; /* * Don't wakeup too often - wakeup the pageout daemon when * we would be nearly out of memory. */ if (vm_paging_needed()) pagedaemon_wakeup(); return (m); } /* * vm_wait: (also see VM_WAIT macro) * * Block until free pages are available for allocation * - Called in various places before memory allocations. */ void vm_wait(void) { mtx_lock(&vm_page_queue_free_mtx); if (curproc == pageproc) { vm_pageout_pages_needed = 1; msleep(&vm_pageout_pages_needed, &vm_page_queue_free_mtx, PDROP | PSWP, "VMWait", 0); } else { if (!vm_pages_needed) { vm_pages_needed = 1; wakeup(&vm_pages_needed); } msleep(&cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PVM, "vmwait", 0); } } /* * vm_waitpfault: (also see VM_WAITPFAULT macro) * * Block until free pages are available for allocation * - Called only in vm_fault so that processes page faulting * can be easily tracked. * - Sleeps at a lower priority than vm_wait() so that vm_wait()ing * processes will be able to grab memory first. Do not change * this balance without careful testing first. */ void vm_waitpfault(void) { mtx_lock(&vm_page_queue_free_mtx); if (!vm_pages_needed) { vm_pages_needed = 1; wakeup(&vm_pages_needed); } msleep(&cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PUSER, "pfault", 0); } /* * vm_page_activate: * * Put the specified page on the active list (if appropriate). * Ensure that act_count is at least ACT_INIT but do not otherwise * mess with it. * * The page queues must be locked. * This routine may not block. */ void vm_page_activate(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (VM_PAGE_GETKNOWNQUEUE2(m) != PQ_ACTIVE) { if (VM_PAGE_INQUEUE1(m, PQ_CACHE)) cnt.v_reactivated++; vm_pageq_remove(m); if (m->wire_count == 0 && (m->flags & PG_UNMANAGED) == 0) { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; vm_pageq_enqueue(PQ_ACTIVE, m); } } else { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; } } /* * vm_page_free_wakeup: * * Helper routine for vm_page_free_toq() and vm_page_cache(). This * routine is called when a page has been added to the cache or free * queues. * * The page queues must be locked. * This routine may not block. */ static inline void vm_page_free_wakeup(void) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); /* * if pageout daemon needs pages, then tell it that there are * some free. */ if (vm_pageout_pages_needed && cnt.v_cache_count + cnt.v_free_count >= cnt.v_pageout_free_min) { wakeup(&vm_pageout_pages_needed); vm_pageout_pages_needed = 0; } /* * wakeup processes that are waiting on memory if we hit a * high water mark. And wakeup scheduler process if we have * lots of memory. this process will swapin processes. */ if (vm_pages_needed && !vm_page_count_min()) { vm_pages_needed = 0; wakeup(&cnt.v_free_count); } } /* * vm_page_free_toq: * * Returns the given page to the free list, * disassociating it with any VM object. * * Object and page must be locked prior to entry. * This routine may not block. */ void vm_page_free_toq(vm_page_t m) { if (VM_PAGE_GETQUEUE(m) != PQ_NONE) mtx_assert(&vm_page_queue_mtx, MA_OWNED); KASSERT(!pmap_page_is_mapped(m), ("vm_page_free_toq: freeing mapped page %p", m)); PCPU_INC(cnt.v_tfree); if (m->busy || VM_PAGE_IS_FREE(m)) { printf( "vm_page_free: pindex(%lu), busy(%d), VPO_BUSY(%d), hold(%d)\n", (u_long)m->pindex, m->busy, (m->oflags & VPO_BUSY) ? 1 : 0, m->hold_count); if (VM_PAGE_IS_FREE(m)) panic("vm_page_free: freeing free page"); else panic("vm_page_free: freeing busy page"); } /* * unqueue, then remove page. Note that we cannot destroy * the page here because we do not want to call the pager's * callback routine until after we've put the page on the * appropriate free queue. */ vm_pageq_remove_nowakeup(m); vm_page_remove(m); /* * If fictitious remove object association and * return, otherwise delay object association removal. */ if ((m->flags & PG_FICTITIOUS) != 0) { return; } m->valid = 0; vm_page_undirty(m); if (m->wire_count != 0) { if (m->wire_count > 1) { panic("vm_page_free: invalid wire count (%d), pindex: 0x%lx", m->wire_count, (long)m->pindex); } panic("vm_page_free: freeing wired page"); } if (m->hold_count != 0) { m->flags &= ~PG_ZERO; vm_pageq_enqueue(PQ_HOLD, m); } else { m->flags |= PG_FREE; mtx_lock(&vm_page_queue_free_mtx); if ((m->flags & PG_ZERO) != 0) { vm_phys_free_pages(m, 0); ++vm_page_zero_count; } else { vm_phys_free_pages(m, 0); vm_page_zero_idle_wakeup(); } vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } } /* * vm_page_wire: * * Mark this page as wired down by yet * another map, removing it from paging queues * as necessary. * * The page queues must be locked. * This routine may not block. */ void vm_page_wire(vm_page_t m) { /* * Only bump the wire statistics if the page is not already wired, * and only unqueue the page if it is on some queue (if it is unmanaged * it is already off the queues). */ mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (m->flags & PG_FICTITIOUS) return; if (m->wire_count == 0) { if ((m->flags & PG_UNMANAGED) == 0) vm_pageq_remove(m); atomic_add_int(&cnt.v_wire_count, 1); } m->wire_count++; KASSERT(m->wire_count != 0, ("vm_page_wire: wire_count overflow m=%p", m)); } /* * vm_page_unwire: * * Release one wiring of this page, potentially * enabling it to be paged again. * * Many pages placed on the inactive queue should actually go * into the cache, but it is difficult to figure out which. What * we do instead, if the inactive target is well met, is to put * clean pages at the head of the inactive queue instead of the tail. * This will cause them to be moved to the cache more quickly and * if not actively re-referenced, freed more quickly. If we just * stick these pages at the end of the inactive queue, heavy filesystem * meta-data accesses can cause an unnecessary paging load on memory bound * processes. This optimization causes one-time-use metadata to be * reused more quickly. * * BUT, if we are in a low-memory situation we have no choice but to * put clean pages on the cache queue. * * A number of routines use vm_page_unwire() to guarantee that the page * will go into either the inactive or active queues, and will NEVER * be placed in the cache - for example, just after dirtying a page. * dirty pages in the cache are not allowed. * * The page queues must be locked. * This routine may not block. */ void vm_page_unwire(vm_page_t m, int activate) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (m->flags & PG_FICTITIOUS) return; if (m->wire_count > 0) { m->wire_count--; if (m->wire_count == 0) { atomic_subtract_int(&cnt.v_wire_count, 1); if (m->flags & PG_UNMANAGED) { ; } else if (activate) vm_pageq_enqueue(PQ_ACTIVE, m); else { vm_page_flag_clear(m, PG_WINATCFLS); vm_pageq_enqueue(PQ_INACTIVE, m); } } } else { panic("vm_page_unwire: invalid wire count: %d", m->wire_count); } } /* * Move the specified page to the inactive queue. If the page has * any associated swap, the swap is deallocated. * * Normally athead is 0 resulting in LRU operation. athead is set * to 1 if we want this page to be 'as if it were placed in the cache', * except without unmapping it from the process address space. * * This routine may not block. */ static inline void _vm_page_deactivate(vm_page_t m, int athead) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); /* * Ignore if already inactive. */ if (VM_PAGE_INQUEUE2(m, PQ_INACTIVE)) return; if (m->wire_count == 0 && (m->flags & PG_UNMANAGED) == 0) { if (VM_PAGE_INQUEUE1(m, PQ_CACHE)) cnt.v_reactivated++; vm_page_flag_clear(m, PG_WINATCFLS); vm_pageq_remove(m); if (athead) TAILQ_INSERT_HEAD(&vm_page_queues[PQ_INACTIVE].pl, m, pageq); else TAILQ_INSERT_TAIL(&vm_page_queues[PQ_INACTIVE].pl, m, pageq); VM_PAGE_SETQUEUE2(m, PQ_INACTIVE); cnt.v_inactive_count++; } } void vm_page_deactivate(vm_page_t m) { _vm_page_deactivate(m, 0); } /* * vm_page_try_to_cache: * * Returns 0 on failure, 1 on success */ int vm_page_try_to_cache(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (m->dirty || m->hold_count || m->busy || m->wire_count || (m->oflags & VPO_BUSY) || (m->flags & PG_UNMANAGED)) { return (0); } pmap_remove_all(m); if (m->dirty) return (0); vm_page_cache(m); return (1); } /* * vm_page_try_to_free() * * Attempt to free the page. If we cannot free it, we do nothing. * 1 is returned on success, 0 on failure. */ int vm_page_try_to_free(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (m->object != NULL) VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (m->dirty || m->hold_count || m->busy || m->wire_count || (m->oflags & VPO_BUSY) || (m->flags & PG_UNMANAGED)) { return (0); } pmap_remove_all(m); if (m->dirty) return (0); vm_page_free(m); return (1); } /* * vm_page_cache * * Put the specified page onto the page cache queue (if appropriate). * * This routine may not block. */ void vm_page_cache(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if ((m->flags & PG_UNMANAGED) || (m->oflags & VPO_BUSY) || m->busy || m->hold_count || m->wire_count) { panic("vm_page_cache: attempting to cache busy page"); } if (VM_PAGE_INQUEUE1(m, PQ_CACHE)) return; + cnt.v_tcached++; /* * Remove all pmaps and indicate that the page is not * writeable or mapped. */ pmap_remove_all(m); if (m->dirty != 0) { panic("vm_page_cache: caching a dirty page, pindex: %ld", (long)m->pindex); } vm_pageq_remove_nowakeup(m); vm_pageq_enqueue(PQ_CACHE, m); mtx_lock(&vm_page_queue_free_mtx); vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } /* * vm_page_dontneed * * Cache, deactivate, or do nothing as appropriate. This routine * is typically used by madvise() MADV_DONTNEED. * * Generally speaking we want to move the page into the cache so * it gets reused quickly. However, this can result in a silly syndrome * due to the page recycling too quickly. Small objects will not be * fully cached. On the otherhand, if we move the page to the inactive * queue we wind up with a problem whereby very large objects * unnecessarily blow away our inactive and cache queues. * * The solution is to move the pages based on a fixed weighting. We * either leave them alone, deactivate them, or move them to the cache, * where moving them to the cache has the highest weighting. * By forcing some pages into other queues we eventually force the * system to balance the queues, potentially recovering other unrelated * space from active. The idea is to not force this to happen too * often. */ void vm_page_dontneed(vm_page_t m) { static int dnweight; int dnw; int head; mtx_assert(&vm_page_queue_mtx, MA_OWNED); dnw = ++dnweight; /* * occassionally leave the page alone */ if ((dnw & 0x01F0) == 0 || VM_PAGE_INQUEUE2(m, PQ_INACTIVE) || VM_PAGE_INQUEUE1(m, PQ_CACHE) ) { if (m->act_count >= ACT_INIT) --m->act_count; return; } if (m->dirty == 0 && pmap_is_modified(m)) vm_page_dirty(m); if (m->dirty || (dnw & 0x0070) == 0) { /* * Deactivate the page 3 times out of 32. */ head = 0; } else { /* * Cache the page 28 times out of every 32. Note that * the page is deactivated instead of cached, but placed * at the head of the queue instead of the tail. */ head = 1; } _vm_page_deactivate(m, head); } /* * Grab a page, waiting until we are waken up due to the page * changing state. We keep on waiting, if the page continues * to be in the object. If the page doesn't exist, first allocate it * and then conditionally zero it. * * This routine may block. */ vm_page_t vm_page_grab(vm_object_t object, vm_pindex_t pindex, int allocflags) { vm_page_t m; VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); retrylookup: if ((m = vm_page_lookup(object, pindex)) != NULL) { if (vm_page_sleep_if_busy(m, TRUE, "pgrbwt")) { if ((allocflags & VM_ALLOC_RETRY) == 0) return (NULL); goto retrylookup; } else { if ((allocflags & VM_ALLOC_WIRED) != 0) { vm_page_lock_queues(); vm_page_wire(m); vm_page_unlock_queues(); } if ((allocflags & VM_ALLOC_NOBUSY) == 0) vm_page_busy(m); return (m); } } m = vm_page_alloc(object, pindex, allocflags & ~VM_ALLOC_RETRY); if (m == NULL) { VM_OBJECT_UNLOCK(object); VM_WAIT; VM_OBJECT_LOCK(object); if ((allocflags & VM_ALLOC_RETRY) == 0) return (NULL); goto retrylookup; } if (allocflags & VM_ALLOC_ZERO && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); return (m); } /* * Mapping function for valid bits or for dirty bits in * a page. May not block. * * Inputs are required to range within a page. */ int vm_page_bits(int base, int size) { int first_bit; int last_bit; KASSERT( base + size <= PAGE_SIZE, ("vm_page_bits: illegal base/size %d/%d", base, size) ); if (size == 0) /* handle degenerate case */ return (0); first_bit = base >> DEV_BSHIFT; last_bit = (base + size - 1) >> DEV_BSHIFT; return ((2 << last_bit) - (1 << first_bit)); } /* * vm_page_set_validclean: * * Sets portions of a page valid and clean. The arguments are expected * to be DEV_BSIZE aligned but if they aren't the bitmap is inclusive * of any partial chunks touched by the range. The invalid portion of * such chunks will be zero'd. * * This routine may not block. * * (base + size) must be less then or equal to PAGE_SIZE. */ void vm_page_set_validclean(vm_page_t m, int base, int size) { int pagebits; int frag; int endoff; mtx_assert(&vm_page_queue_mtx, MA_OWNED); VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (size == 0) /* handle degenerate case */ return; /* * If the base is not DEV_BSIZE aligned and the valid * bit is clear, we have to zero out a portion of the * first block. */ if ((frag = base & ~(DEV_BSIZE - 1)) != base && (m->valid & (1 << (base >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, frag, base - frag); /* * If the ending offset is not DEV_BSIZE aligned and the * valid bit is clear, we have to zero out a portion of * the last block. */ endoff = base + size; if ((frag = endoff & ~(DEV_BSIZE - 1)) != endoff && (m->valid & (1 << (endoff >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, endoff, DEV_BSIZE - (endoff & (DEV_BSIZE - 1))); /* * Set valid, clear dirty bits. If validating the entire * page we can safely clear the pmap modify bit. We also * use this opportunity to clear the VPO_NOSYNC flag. If a process * takes a write fault on a MAP_NOSYNC memory area the flag will * be set again. * * We set valid bits inclusive of any overlap, but we can only * clear dirty bits for DEV_BSIZE chunks that are fully within * the range. */ pagebits = vm_page_bits(base, size); m->valid |= pagebits; #if 0 /* NOT YET */ if ((frag = base & (DEV_BSIZE - 1)) != 0) { frag = DEV_BSIZE - frag; base += frag; size -= frag; if (size < 0) size = 0; } pagebits = vm_page_bits(base, size & (DEV_BSIZE - 1)); #endif m->dirty &= ~pagebits; if (base == 0 && size == PAGE_SIZE) { pmap_clear_modify(m); m->oflags &= ~VPO_NOSYNC; } } void vm_page_clear_dirty(vm_page_t m, int base, int size) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); m->dirty &= ~vm_page_bits(base, size); } /* * vm_page_set_invalid: * * Invalidates DEV_BSIZE'd chunks within a page. Both the * valid and dirty bits for the effected areas are cleared. * * May not block. */ void vm_page_set_invalid(vm_page_t m, int base, int size) { int bits; VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); bits = vm_page_bits(base, size); mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (m->valid == VM_PAGE_BITS_ALL && bits != 0) pmap_remove_all(m); m->valid &= ~bits; m->dirty &= ~bits; m->object->generation++; } /* * vm_page_zero_invalid() * * The kernel assumes that the invalid portions of a page contain * garbage, but such pages can be mapped into memory by user code. * When this occurs, we must zero out the non-valid portions of the * page so user code sees what it expects. * * Pages are most often semi-valid when the end of a file is mapped * into memory and the file's size is not page aligned. */ void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid) { int b; int i; VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); /* * Scan the valid bits looking for invalid sections that * must be zerod. Invalid sub-DEV_BSIZE'd areas ( where the * valid bit may be set ) have already been zerod by * vm_page_set_validclean(). */ for (b = i = 0; i <= PAGE_SIZE / DEV_BSIZE; ++i) { if (i == (PAGE_SIZE / DEV_BSIZE) || (m->valid & (1 << i)) ) { if (i > b) { pmap_zero_page_area(m, b << DEV_BSHIFT, (i - b) << DEV_BSHIFT); } b = i + 1; } } /* * setvalid is TRUE when we can safely set the zero'd areas * as being valid. We can do this if there are no cache consistancy * issues. e.g. it is ok to do with UFS, but not ok to do with NFS. */ if (setvalid) m->valid = VM_PAGE_BITS_ALL; } /* * vm_page_is_valid: * * Is (partial) page valid? Note that the case where size == 0 * will return FALSE in the degenerate case where the page is * entirely invalid, and TRUE otherwise. * * May not block. */ int vm_page_is_valid(vm_page_t m, int base, int size) { int bits = vm_page_bits(base, size); VM_OBJECT_LOCK_ASSERT(m->object, MA_OWNED); if (m->valid && ((m->valid & bits) == bits)) return 1; else return 0; } /* * update dirty bits from pmap/mmu. May not block. */ void vm_page_test_dirty(vm_page_t m) { if ((m->dirty != VM_PAGE_BITS_ALL) && pmap_is_modified(m)) { vm_page_dirty(m); } } int so_zerocp_fullpage = 0; /* * Replace the given page with a copy. The copied page assumes * the portion of the given page's "wire_count" that is not the * responsibility of this copy-on-write mechanism. * * The object containing the given page must have a non-zero * paging-in-progress count and be locked. */ void vm_page_cowfault(vm_page_t m) { vm_page_t mnew; vm_object_t object; vm_pindex_t pindex; object = m->object; VM_OBJECT_LOCK_ASSERT(object, MA_OWNED); KASSERT(object->paging_in_progress != 0, ("vm_page_cowfault: object %p's paging-in-progress count is zero.", object)); pindex = m->pindex; retry_alloc: pmap_remove_all(m); vm_page_remove(m); mnew = vm_page_alloc(object, pindex, VM_ALLOC_NORMAL | VM_ALLOC_NOBUSY); if (mnew == NULL) { vm_page_insert(m, object, pindex); vm_page_unlock_queues(); VM_OBJECT_UNLOCK(object); VM_WAIT; VM_OBJECT_LOCK(object); if (m == vm_page_lookup(object, pindex)) { vm_page_lock_queues(); goto retry_alloc; } else { /* * Page disappeared during the wait. */ vm_page_lock_queues(); return; } } if (m->cow == 0) { /* * check to see if we raced with an xmit complete when * waiting to allocate a page. If so, put things back * the way they were */ vm_page_free(mnew); vm_page_insert(m, object, pindex); } else { /* clear COW & copy page */ if (!so_zerocp_fullpage) pmap_copy_page(m, mnew); mnew->valid = VM_PAGE_BITS_ALL; vm_page_dirty(mnew); mnew->wire_count = m->wire_count - m->cow; m->wire_count = m->cow; } } void vm_page_cowclear(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); if (m->cow) { m->cow--; /* * let vm_fault add back write permission lazily */ } /* * sf_buf_free() will free the page, so we needn't do it here */ } void vm_page_cowsetup(vm_page_t m) { mtx_assert(&vm_page_queue_mtx, MA_OWNED); m->cow++; pmap_remove_write(m); } #include "opt_ddb.h" #ifdef DDB #include #include DB_SHOW_COMMAND(page, vm_page_print_page_info) { db_printf("cnt.v_free_count: %d\n", cnt.v_free_count); db_printf("cnt.v_cache_count: %d\n", cnt.v_cache_count); db_printf("cnt.v_inactive_count: %d\n", cnt.v_inactive_count); db_printf("cnt.v_active_count: %d\n", cnt.v_active_count); db_printf("cnt.v_wire_count: %d\n", cnt.v_wire_count); db_printf("cnt.v_free_reserved: %d\n", cnt.v_free_reserved); db_printf("cnt.v_free_min: %d\n", cnt.v_free_min); db_printf("cnt.v_free_target: %d\n", cnt.v_free_target); db_printf("cnt.v_cache_min: %d\n", cnt.v_cache_min); db_printf("cnt.v_inactive_target: %d\n", cnt.v_inactive_target); } DB_SHOW_COMMAND(pageq, vm_page_print_pageq_info) { db_printf("PQ_FREE:"); db_printf(" %d", cnt.v_free_count); db_printf("\n"); db_printf("PQ_CACHE:"); db_printf(" %d", *vm_page_queues[PQ_CACHE].cnt); db_printf("\n"); db_printf("PQ_ACTIVE: %d, PQ_INACTIVE: %d\n", *vm_page_queues[PQ_ACTIVE].cnt, *vm_page_queues[PQ_INACTIVE].cnt); } #endif /* DDB */ Index: head/usr.bin/vmstat/vmstat.c =================================================================== --- head/usr.bin/vmstat/vmstat.c (revision 171632) +++ head/usr.bin/vmstat/vmstat.c (revision 171633) @@ -1,1042 +1,1044 @@ /* * Copyright (c) 1980, 1986, 1991, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1980, 1986, 1991, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #if 0 #ifndef lint static char sccsid[] = "@(#)vmstat.c 8.1 (Berkeley) 6/6/93"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static char da[] = "da"; static struct nlist namelist[] = { #define X_CPTIME 0 { "_cp_time" }, #define X_SUM 1 { "_cnt" }, #define X_BOOTTIME 2 { "_boottime" }, #define X_HZ 3 { "_hz" }, #define X_STATHZ 4 { "_stathz" }, #define X_NCHSTATS 5 { "_nchstats" }, #define X_INTRNAMES 6 { "_intrnames" }, #define X_EINTRNAMES 7 { "_eintrnames" }, #define X_INTRCNT 8 { "_intrcnt" }, #define X_EINTRCNT 9 { "_eintrcnt" }, #define X_KMEMSTATS 10 { "_kmemstatistics" }, #define X_KMEMZONES 11 { "_kmemzones" }, #ifdef notyet #define X_DEFICIT XXX { "_deficit" }, #define X_REC XXX { "_rectime" }, #define X_PGIN XXX { "_pgintime" }, #define X_XSTATS XXX { "_xstats" }, #define X_END XXX #else #define X_END 12 #endif { "" }, }; static struct statinfo cur, last; static int num_devices, maxshowdevs; static long generation; static struct device_selection *dev_select; static int num_selected; static struct devstat_match *matches; static int num_matches = 0; static int num_devices_specified, num_selections; static long select_generation; static char **specified_devices; static devstat_select_mode select_mode; static struct vmmeter sum, osum; static int winlines = 20; static int aflag; static int nflag; static kvm_t *kd; #define FORKSTAT 0x01 #define INTRSTAT 0x02 #define MEMSTAT 0x04 #define SUMSTAT 0x08 #define TIMESTAT 0x10 #define VMSTAT 0x20 #define ZMEMSTAT 0x40 static void cpustats(void); static void devstats(void); static void doforkst(void); static void dointr(void); static void dosum(void); static void dovmstat(unsigned int, int); static void domemstat_malloc(void); static void domemstat_zone(void); static void kread(int, void *, size_t); static void kreado(int, void *, size_t, size_t); static char *kgetstr(const char *); static void needhdr(int); static void printhdr(void); static void usage(void); static long pct(long, long); static long getuptime(void); static char **getdrivedata(char **); int main(int argc, char *argv[]) { int c, todo; unsigned int interval; int reps; char *memf, *nlistf; char errbuf[_POSIX2_LINE_MAX]; memf = nlistf = NULL; interval = reps = todo = 0; maxshowdevs = 2; while ((c = getopt(argc, argv, "ac:fiM:mN:n:p:stw:z")) != -1) { switch (c) { case 'a': aflag++; break; case 'c': reps = atoi(optarg); break; case 'f': todo |= FORKSTAT; break; case 'i': todo |= INTRSTAT; break; case 'M': memf = optarg; break; case 'm': todo |= MEMSTAT; break; case 'N': nlistf = optarg; break; case 'n': nflag = 1; maxshowdevs = atoi(optarg); if (maxshowdevs < 0) errx(1, "number of devices %d is < 0", maxshowdevs); break; case 'p': if (devstat_buildmatch(optarg, &matches, &num_matches) != 0) errx(1, "%s", devstat_errbuf); break; case 's': todo |= SUMSTAT; break; case 't': #ifdef notyet todo |= TIMESTAT; #else errx(EX_USAGE, "sorry, -t is not (re)implemented yet"); #endif break; case 'w': interval = atoi(optarg); break; case 'z': todo |= ZMEMSTAT; break; case '?': default: usage(); } } argc -= optind; argv += optind; if (todo == 0) todo = VMSTAT; if (memf != NULL) { kd = kvm_openfiles(nlistf, memf, NULL, O_RDONLY, errbuf); if (kd == NULL) errx(1, "kvm_openfiles: %s", errbuf); } if (kd != NULL && (c = kvm_nlist(kd, namelist)) != 0) { if (c > 0) { warnx("undefined symbols:"); for (c = 0; c < (int)(sizeof(namelist)/sizeof(namelist[0])); c++) if (namelist[c].n_type == 0) (void)fprintf(stderr, " %s", namelist[c].n_name); (void)fputc('\n', stderr); } else warnx("kvm_nlist: %s", kvm_geterr(kd)); exit(1); } if (todo & VMSTAT) { struct winsize winsize; /* * Make sure that the userland devstat version matches the * kernel devstat version. If not, exit and print a * message informing the user of his mistake. */ if (devstat_checkversion(NULL) < 0) errx(1, "%s", devstat_errbuf); argv = getdrivedata(argv); winsize.ws_row = 0; (void) ioctl(STDOUT_FILENO, TIOCGWINSZ, (char *)&winsize); if (winsize.ws_row > 0) winlines = winsize.ws_row; } #define BACKWARD_COMPATIBILITY #ifdef BACKWARD_COMPATIBILITY if (*argv) { interval = atoi(*argv); if (*++argv) reps = atoi(*argv); } #endif if (interval) { if (!reps) reps = -1; } else if (reps) interval = 1; if (todo & FORKSTAT) doforkst(); if (todo & MEMSTAT) domemstat_malloc(); if (todo & ZMEMSTAT) domemstat_zone(); if (todo & SUMSTAT) dosum(); #ifdef notyet if (todo & TIMESTAT) dotimes(); #endif if (todo & INTRSTAT) dointr(); if (todo & VMSTAT) dovmstat(interval, reps); exit(0); } static int mysysctl(const char *name, void *oldp, size_t *oldlenp, void *newp, size_t newlen) { int error; error = sysctlbyname(name, oldp, oldlenp, newp, newlen); if (error != 0 && errno != ENOMEM) err(1, "sysctl(%s)", name); return (error); } static char ** getdrivedata(char **argv) { if ((num_devices = devstat_getnumdevs(NULL)) < 0) errx(1, "%s", devstat_errbuf); cur.dinfo = (struct devinfo *)malloc(sizeof(struct devinfo)); last.dinfo = (struct devinfo *)malloc(sizeof(struct devinfo)); bzero(cur.dinfo, sizeof(struct devinfo)); bzero(last.dinfo, sizeof(struct devinfo)); if (devstat_getdevs(NULL, &cur) == -1) errx(1, "%s", devstat_errbuf); num_devices = cur.dinfo->numdevs; generation = cur.dinfo->generation; specified_devices = (char **)malloc(sizeof(char *)); for (num_devices_specified = 0; *argv; ++argv) { if (isdigit(**argv)) break; num_devices_specified++; specified_devices = (char **)realloc(specified_devices, sizeof(char *) * num_devices_specified); specified_devices[num_devices_specified - 1] = *argv; } dev_select = NULL; if (nflag == 0 && maxshowdevs < num_devices_specified) maxshowdevs = num_devices_specified; /* * People are generally only interested in disk statistics when * they're running vmstat. So, that's what we're going to give * them if they don't specify anything by default. We'll also give * them any other random devices in the system so that we get to * maxshowdevs devices, if that many devices exist. If the user * specifies devices on the command line, either through a pattern * match or by naming them explicitly, we will give the user only * those devices. */ if ((num_devices_specified == 0) && (num_matches == 0)) { if (devstat_buildmatch(da, &matches, &num_matches) != 0) errx(1, "%s", devstat_errbuf); select_mode = DS_SELECT_ADD; } else select_mode = DS_SELECT_ONLY; /* * At this point, selectdevs will almost surely indicate that the * device list has changed, so we don't look for return values of 0 * or 1. If we get back -1, though, there is an error. */ if (devstat_selectdevs(&dev_select, &num_selected, &num_selections, &select_generation, generation, cur.dinfo->devices, num_devices, matches, num_matches, specified_devices, num_devices_specified, select_mode, maxshowdevs, 0) == -1) errx(1, "%s", devstat_errbuf); return(argv); } static long getuptime(void) { struct timespec sp; time_t uptime; (void)clock_gettime(CLOCK_MONOTONIC, &sp); uptime = sp.tv_sec; if (uptime <= 0 || uptime > 60*60*24*365*10) errx(1, "time makes no sense; namelist must be wrong"); return(uptime); } static void fill_vmmeter(struct vmmeter *vmmp) { if (kd != NULL) { kread(X_SUM, vmmp, sizeof(*vmmp)); } else { size_t size = sizeof(unsigned int); #define GET_VM_STATS(cat, name) \ mysysctl("vm.stats." #cat "." #name, &vmmp->name, &size, NULL, 0) /* sys */ GET_VM_STATS(sys, v_swtch); GET_VM_STATS(sys, v_trap); GET_VM_STATS(sys, v_syscall); GET_VM_STATS(sys, v_intr); GET_VM_STATS(sys, v_soft); /* vm */ GET_VM_STATS(vm, v_vm_faults); GET_VM_STATS(vm, v_cow_faults); GET_VM_STATS(vm, v_cow_optim); GET_VM_STATS(vm, v_zfod); GET_VM_STATS(vm, v_ozfod); GET_VM_STATS(vm, v_swapin); GET_VM_STATS(vm, v_swapout); GET_VM_STATS(vm, v_swappgsin); GET_VM_STATS(vm, v_swappgsout); GET_VM_STATS(vm, v_vnodein); GET_VM_STATS(vm, v_vnodeout); GET_VM_STATS(vm, v_vnodepgsin); GET_VM_STATS(vm, v_vnodepgsout); GET_VM_STATS(vm, v_intrans); GET_VM_STATS(vm, v_reactivated); GET_VM_STATS(vm, v_pdwakeups); GET_VM_STATS(vm, v_pdpages); + GET_VM_STATS(vm, v_tcached); GET_VM_STATS(vm, v_dfree); GET_VM_STATS(vm, v_pfree); GET_VM_STATS(vm, v_tfree); GET_VM_STATS(vm, v_page_size); GET_VM_STATS(vm, v_page_count); GET_VM_STATS(vm, v_free_reserved); GET_VM_STATS(vm, v_free_target); GET_VM_STATS(vm, v_free_min); GET_VM_STATS(vm, v_free_count); GET_VM_STATS(vm, v_wire_count); GET_VM_STATS(vm, v_active_count); GET_VM_STATS(vm, v_inactive_target); GET_VM_STATS(vm, v_inactive_count); GET_VM_STATS(vm, v_cache_count); GET_VM_STATS(vm, v_cache_min); GET_VM_STATS(vm, v_cache_max); GET_VM_STATS(vm, v_pageout_free_min); GET_VM_STATS(vm, v_interrupt_free_min); /*GET_VM_STATS(vm, v_free_severe);*/ GET_VM_STATS(vm, v_forks); GET_VM_STATS(vm, v_vforks); GET_VM_STATS(vm, v_rforks); GET_VM_STATS(vm, v_kthreads); GET_VM_STATS(vm, v_forkpages); GET_VM_STATS(vm, v_vforkpages); GET_VM_STATS(vm, v_rforkpages); GET_VM_STATS(vm, v_kthreadpages); #undef GET_VM_STATS } } static void fill_vmtotal(struct vmtotal *vmtp) { if (kd != NULL) { /* XXX fill vmtp */ errx(1, "not implemented"); } else { size_t size = sizeof(*vmtp); mysysctl("vm.vmtotal", vmtp, &size, NULL, 0); if (size != sizeof(*vmtp)) errx(1, "vm.total size mismatch"); } } static int hz, hdrcnt; static void dovmstat(unsigned int interval, int reps) { struct vmtotal total; time_t uptime, halfuptime; struct devinfo *tmp_dinfo; size_t size; uptime = getuptime(); halfuptime = uptime / 2; (void)signal(SIGCONT, needhdr); if (kd != NULL) { if (namelist[X_STATHZ].n_type != 0 && namelist[X_STATHZ].n_value != 0) kread(X_STATHZ, &hz, sizeof(hz)); if (!hz) kread(X_HZ, &hz, sizeof(hz)); } else { struct clockinfo clockrate; size = sizeof(clockrate); mysysctl("kern.clockrate", &clockrate, &size, NULL, 0); if (size != sizeof(clockrate)) errx(1, "clockrate size mismatch"); hz = clockrate.hz; } for (hdrcnt = 1;;) { if (!--hdrcnt) printhdr(); if (kd != NULL) { kread(X_CPTIME, cur.cp_time, sizeof(cur.cp_time)); } else { size = sizeof(cur.cp_time); mysysctl("kern.cp_time", &cur.cp_time, &size, NULL, 0); if (size != sizeof(cur.cp_time)) errx(1, "cp_time size mismatch"); } tmp_dinfo = last.dinfo; last.dinfo = cur.dinfo; cur.dinfo = tmp_dinfo; last.snap_time = cur.snap_time; /* * Here what we want to do is refresh our device stats. * getdevs() returns 1 when the device list has changed. * If the device list has changed, we want to go through * the selection process again, in case a device that we * were previously displaying has gone away. */ switch (devstat_getdevs(NULL, &cur)) { case -1: errx(1, "%s", devstat_errbuf); break; case 1: { int retval; num_devices = cur.dinfo->numdevs; generation = cur.dinfo->generation; retval = devstat_selectdevs(&dev_select, &num_selected, &num_selections, &select_generation, generation, cur.dinfo->devices, num_devices, matches, num_matches, specified_devices, num_devices_specified, select_mode, maxshowdevs, 0); switch (retval) { case -1: errx(1, "%s", devstat_errbuf); break; case 1: printhdr(); break; default: break; } } default: break; } fill_vmmeter(&sum); fill_vmtotal(&total); (void)printf("%2d %1d %1d", total.t_rq - 1, total.t_dw + total.t_pw, total.t_sw); #define vmstat_pgtok(a) ((a) * (sum.v_page_size >> 10)) #define rate(x) (((x) + halfuptime) / uptime) /* round */ (void)printf(" %7d %6d ", vmstat_pgtok(total.t_avm), vmstat_pgtok(total.t_free)); (void)printf("%5lu ", (unsigned long)rate(sum.v_vm_faults - osum.v_vm_faults)); (void)printf("%3lu ", (unsigned long)rate(sum.v_reactivated - osum.v_reactivated)); (void)printf("%3lu ", (unsigned long)rate(sum.v_swapin + sum.v_vnodein - (osum.v_swapin + osum.v_vnodein))); (void)printf("%3lu ", (unsigned long)rate(sum.v_swapout + sum.v_vnodeout - (osum.v_swapout + osum.v_vnodeout))); (void)printf("%5lu ", (unsigned long)rate(sum.v_tfree - osum.v_tfree)); (void)printf("%3lu ", (unsigned long)rate(sum.v_pdpages - osum.v_pdpages)); devstats(); (void)printf("%4lu %4lu %4lu ", (unsigned long)rate(sum.v_intr - osum.v_intr), (unsigned long)rate(sum.v_syscall - osum.v_syscall), (unsigned long)rate(sum.v_swtch - osum.v_swtch)); cpustats(); (void)printf("\n"); (void)fflush(stdout); if (reps >= 0 && --reps <= 0) break; osum = sum; uptime = interval; /* * We round upward to avoid losing low-frequency events * (i.e., >= 1 per interval but < 1 per second). */ if (interval != 1) halfuptime = (uptime + 1) / 2; else halfuptime = 0; (void)sleep(interval); } } static void printhdr(void) { int i, num_shown; num_shown = (num_selected < maxshowdevs) ? num_selected : maxshowdevs; (void)printf(" procs memory page%*s", 19, ""); if (num_shown > 1) (void)printf(" disks %*s", num_shown * 4 - 7, ""); else if (num_shown == 1) (void)printf("disk"); (void)printf(" faults cpu\n"); (void)printf(" r b w avm fre flt re pi po fr sr "); for (i = 0; i < num_devices; i++) if ((dev_select[i].selected) && (dev_select[i].selected <= maxshowdevs)) (void)printf("%c%c%d ", dev_select[i].device_name[0], dev_select[i].device_name[1], dev_select[i].unit_number); (void)printf(" in sy cs us sy id\n"); hdrcnt = winlines - 2; } /* * Force a header to be prepended to the next output. */ static void needhdr(int dummy __unused) { hdrcnt = 1; } #ifdef notyet static void dotimes(void) { unsigned int pgintime, rectime; kread(X_REC, &rectime, sizeof(rectime)); kread(X_PGIN, &pgintime, sizeof(pgintime)); kread(X_SUM, &sum, sizeof(sum)); (void)printf("%u reclaims, %u total time (usec)\n", sum.v_pgrec, rectime); (void)printf("average: %u usec / reclaim\n", rectime / sum.v_pgrec); (void)printf("\n"); (void)printf("%u page ins, %u total time (msec)\n", sum.v_pgin, pgintime / 10); (void)printf("average: %8.1f msec / page in\n", pgintime / (sum.v_pgin * 10.0)); } #endif static long pct(long top, long bot) { long ans; if (bot == 0) return(0); ans = (quad_t)top * 100 / bot; return (ans); } #define PCT(top, bot) pct((long)(top), (long)(bot)) static void dosum(void) { struct nchstats lnchstats; long nchtotal; fill_vmmeter(&sum); (void)printf("%9u cpu context switches\n", sum.v_swtch); (void)printf("%9u device interrupts\n", sum.v_intr); (void)printf("%9u software interrupts\n", sum.v_soft); (void)printf("%9u traps\n", sum.v_trap); (void)printf("%9u system calls\n", sum.v_syscall); (void)printf("%9u kernel threads created\n", sum.v_kthreads); (void)printf("%9u fork() calls\n", sum.v_forks); (void)printf("%9u vfork() calls\n", sum.v_vforks); (void)printf("%9u rfork() calls\n", sum.v_rforks); (void)printf("%9u swap pager pageins\n", sum.v_swapin); (void)printf("%9u swap pager pages paged in\n", sum.v_swappgsin); (void)printf("%9u swap pager pageouts\n", sum.v_swapout); (void)printf("%9u swap pager pages paged out\n", sum.v_swappgsout); (void)printf("%9u vnode pager pageins\n", sum.v_vnodein); (void)printf("%9u vnode pager pages paged in\n", sum.v_vnodepgsin); (void)printf("%9u vnode pager pageouts\n", sum.v_vnodeout); (void)printf("%9u vnode pager pages paged out\n", sum.v_vnodepgsout); (void)printf("%9u page daemon wakeups\n", sum.v_pdwakeups); (void)printf("%9u pages examined by the page daemon\n", sum.v_pdpages); (void)printf("%9u pages reactivated\n", sum.v_reactivated); (void)printf("%9u copy-on-write faults\n", sum.v_cow_faults); (void)printf("%9u copy-on-write optimized faults\n", sum.v_cow_optim); (void)printf("%9u zero fill pages zeroed\n", sum.v_zfod); (void)printf("%9u zero fill pages prezeroed\n", sum.v_ozfod); (void)printf("%9u intransit blocking page faults\n", sum.v_intrans); (void)printf("%9u total VM faults taken\n", sum.v_vm_faults); (void)printf("%9u pages affected by kernel thread creation\n", sum.v_kthreadpages); (void)printf("%9u pages affected by fork()\n", sum.v_forkpages); (void)printf("%9u pages affected by vfork()\n", sum.v_vforkpages); (void)printf("%9u pages affected by rfork()\n", sum.v_rforkpages); + (void)printf("%9u pages cached\n", sum.v_tcached); (void)printf("%9u pages freed\n", sum.v_tfree); (void)printf("%9u pages freed by daemon\n", sum.v_dfree); (void)printf("%9u pages freed by exiting processes\n", sum.v_pfree); (void)printf("%9u pages active\n", sum.v_active_count); (void)printf("%9u pages inactive\n", sum.v_inactive_count); (void)printf("%9u pages in VM cache\n", sum.v_cache_count); (void)printf("%9u pages wired down\n", sum.v_wire_count); (void)printf("%9u pages free\n", sum.v_free_count); (void)printf("%9u bytes per page\n", sum.v_page_size); if (kd != NULL) { kread(X_NCHSTATS, &lnchstats, sizeof(lnchstats)); } else { size_t size = sizeof(lnchstats); mysysctl("vfs.cache.nchstats", &lnchstats, &size, NULL, 0); if (size != sizeof(lnchstats)) errx(1, "vfs.cache.nchstats size mismatch"); } nchtotal = lnchstats.ncs_goodhits + lnchstats.ncs_neghits + lnchstats.ncs_badhits + lnchstats.ncs_falsehits + lnchstats.ncs_miss + lnchstats.ncs_long; (void)printf("%9ld total name lookups\n", nchtotal); (void)printf( "%9s cache hits (%ld%% pos + %ld%% neg) system %ld%% per-directory\n", "", PCT(lnchstats.ncs_goodhits, nchtotal), PCT(lnchstats.ncs_neghits, nchtotal), PCT(lnchstats.ncs_pass2, nchtotal)); (void)printf("%9s deletions %ld%%, falsehits %ld%%, toolong %ld%%\n", "", PCT(lnchstats.ncs_badhits, nchtotal), PCT(lnchstats.ncs_falsehits, nchtotal), PCT(lnchstats.ncs_long, nchtotal)); } static void doforkst(void) { fill_vmmeter(&sum); (void)printf("%u forks, %u pages, average %.2f\n", sum.v_forks, sum.v_forkpages, sum.v_forks == 0 ? 0.0 : (double)sum.v_forkpages / sum.v_forks); (void)printf("%u vforks, %u pages, average %.2f\n", sum.v_vforks, sum.v_vforkpages, sum.v_vforks == 0 ? 0.0 : (double)sum.v_vforkpages / sum.v_vforks); (void)printf("%u rforks, %u pages, average %.2f\n", sum.v_rforks, sum.v_rforkpages, sum.v_rforks == 0 ? 0.0 : (double)sum.v_rforkpages / sum.v_rforks); } static void devstats(void) { int dn, state; long double transfers_per_second; long double busy_seconds; long tmp; for (state = 0; state < CPUSTATES; ++state) { tmp = cur.cp_time[state]; cur.cp_time[state] -= last.cp_time[state]; last.cp_time[state] = tmp; } busy_seconds = cur.snap_time - last.snap_time; for (dn = 0; dn < num_devices; dn++) { int di; if ((dev_select[dn].selected == 0) || (dev_select[dn].selected > maxshowdevs)) continue; di = dev_select[dn].position; if (devstat_compute_statistics(&cur.dinfo->devices[di], &last.dinfo->devices[di], busy_seconds, DSM_TRANSFERS_PER_SECOND, &transfers_per_second, DSM_NONE) != 0) errx(1, "%s", devstat_errbuf); (void)printf("%3.0Lf ", transfers_per_second); } } static void cpustats(void) { int state; double lpct, total; total = 0; for (state = 0; state < CPUSTATES; ++state) total += cur.cp_time[state]; if (total) lpct = 100.0 / total; else lpct = 0.0; (void)printf("%2.0f ", (cur.cp_time[CP_USER] + cur.cp_time[CP_NICE]) * lpct); (void)printf("%2.0f ", (cur.cp_time[CP_SYS] + cur.cp_time[CP_INTR]) * lpct); (void)printf("%2.0f", cur.cp_time[CP_IDLE] * lpct); } static void dointr(void) { unsigned long *intrcnt, uptime; uint64_t inttotal; size_t clen, inamlen, intrcntlen, istrnamlen; unsigned int i, nintr; char *intrname, *tintrname; uptime = getuptime(); if (kd != NULL) { intrcntlen = namelist[X_EINTRCNT].n_value - namelist[X_INTRCNT].n_value; inamlen = namelist[X_EINTRNAMES].n_value - namelist[X_INTRNAMES].n_value; if ((intrcnt = malloc(intrcntlen)) == NULL || (intrname = malloc(inamlen)) == NULL) err(1, "malloc()"); kread(X_INTRCNT, intrcnt, intrcntlen); kread(X_INTRNAMES, intrname, inamlen); } else { for (intrcnt = NULL, intrcntlen = 1024; ; intrcntlen *= 2) { if ((intrcnt = reallocf(intrcnt, intrcntlen)) == NULL) err(1, "reallocf()"); if (mysysctl("hw.intrcnt", intrcnt, &intrcntlen, NULL, 0) == 0) break; } for (intrname = NULL, inamlen = 1024; ; inamlen *= 2) { if ((intrname = reallocf(intrname, inamlen)) == NULL) err(1, "reallocf()"); if (mysysctl("hw.intrnames", intrname, &inamlen, NULL, 0) == 0) break; } } nintr = intrcntlen / sizeof(unsigned long); tintrname = intrname; istrnamlen = strlen("interrupt"); for (i = 0; i < nintr; i++) { clen = strlen(tintrname); if (clen > istrnamlen) istrnamlen = clen; tintrname += clen + 1; } (void)printf("%-*s %20s %10s\n", istrnamlen, "interrupt", "total", "rate"); inttotal = 0; for (i = 0; i < nintr; i++) { if (intrname[0] != '\0' && (*intrcnt != 0 || aflag)) (void)printf("%-*s %20lu %10lu\n", istrnamlen, intrname, *intrcnt, *intrcnt / uptime); intrname += strlen(intrname) + 1; inttotal += *intrcnt++; } (void)printf("%-*s %20llu %10llu\n", istrnamlen, "Total", (long long)inttotal, (long long)(inttotal / uptime)); } static void domemstat_malloc(void) { struct memory_type_list *mtlp; struct memory_type *mtp; int error, first, i; mtlp = memstat_mtl_alloc(); if (mtlp == NULL) { warn("memstat_mtl_alloc"); return; } if (kd == NULL) { if (memstat_sysctl_malloc(mtlp, 0) < 0) { warnx("memstat_sysctl_malloc: %s", memstat_strerror(memstat_mtl_geterror(mtlp))); return; } } else { if (memstat_kvm_malloc(mtlp, kd) < 0) { error = memstat_mtl_geterror(mtlp); if (error == MEMSTAT_ERROR_KVM) warnx("memstat_kvm_malloc: %s", kvm_geterr(kd)); else warnx("memstat_kvm_malloc: %s", memstat_strerror(error)); } } printf("%13s %5s %6s %7s %8s Size(s)\n", "Type", "InUse", "MemUse", "HighUse", "Requests"); for (mtp = memstat_mtl_first(mtlp); mtp != NULL; mtp = memstat_mtl_next(mtp)) { if (memstat_get_numallocs(mtp) == 0 && memstat_get_count(mtp) == 0) continue; printf("%13s %5lld %5lldK %7s %8lld ", memstat_get_name(mtp), memstat_get_count(mtp), ((int64_t)memstat_get_bytes(mtp) + 1023) / 1024, "-", memstat_get_numallocs(mtp)); first = 1; for (i = 0; i < 32; i++) { if (memstat_get_sizemask(mtp) & (1 << i)) { if (!first) printf(","); printf("%d", 1 << (i + 4)); first = 0; } } printf("\n"); } memstat_mtl_free(mtlp); } static void domemstat_zone(void) { struct memory_type_list *mtlp; struct memory_type *mtp; char name[MEMTYPE_MAXNAME + 1]; int error; mtlp = memstat_mtl_alloc(); if (mtlp == NULL) { warn("memstat_mtl_alloc"); return; } if (kd == NULL) { if (memstat_sysctl_uma(mtlp, 0) < 0) { warnx("memstat_sysctl_uma: %s", memstat_strerror(memstat_mtl_geterror(mtlp))); return; } } else { if (memstat_kvm_uma(mtlp, kd) < 0) { error = memstat_mtl_geterror(mtlp); if (error == MEMSTAT_ERROR_KVM) warnx("memstat_kvm_uma: %s", kvm_geterr(kd)); else warnx("memstat_kvm_uma: %s", memstat_strerror(error)); } } printf("%-20s %8s %8s %8s %8s %8s %8s\n\n", "ITEM", "SIZE", "LIMIT", "USED", "FREE", "REQUESTS", "FAILURES"); for (mtp = memstat_mtl_first(mtlp); mtp != NULL; mtp = memstat_mtl_next(mtp)) { strlcpy(name, memstat_get_name(mtp), MEMTYPE_MAXNAME); strcat(name, ":"); printf("%-20s %8llu, %8llu, %8llu, %8llu, %8llu, %8llu\n", name, memstat_get_size(mtp), memstat_get_countlimit(mtp), memstat_get_count(mtp), memstat_get_free(mtp), memstat_get_numallocs(mtp), memstat_get_failures(mtp)); } memstat_mtl_free(mtlp); printf("\n"); } /* * kread reads something from the kernel, given its nlist index. */ static void kreado(int nlx, void *addr, size_t size, size_t offset) { const char *sym; if (namelist[nlx].n_type == 0 || namelist[nlx].n_value == 0) { sym = namelist[nlx].n_name; if (*sym == '_') ++sym; errx(1, "symbol %s not defined", sym); } if ((size_t)kvm_read(kd, namelist[nlx].n_value + offset, addr, size) != size) { sym = namelist[nlx].n_name; if (*sym == '_') ++sym; errx(1, "%s: %s", sym, kvm_geterr(kd)); } } static void kread(int nlx, void *addr, size_t size) { kreado(nlx, addr, size, 0); } static char * kgetstr(const char *strp) { int n = 0, size = 1; char *ret = NULL; do { if (size == n + 1) { ret = realloc(ret, size); if (ret == NULL) err(1, "%s: realloc", __func__); size *= 2; } if (kvm_read(kd, (u_long)strp + n, &ret[n], 1) != 1) errx(1, "%s: %s", __func__, kvm_geterr(kd)); } while (ret[n++] != '\0'); return (ret); } static void usage(void) { (void)fprintf(stderr, "%s%s", "usage: vmstat [-afimsz] [-c count] [-M core [-N system]] [-w wait]\n", " [-n devs] [-p type,if,pass] [disks]\n"); exit(1); }