/* * Copyright (c) 2003-2022 The DragonFly Project. All rights reserved. * * This code is derived from software contributed to The DragonFly Project * by Matthew Dillon <dillon@backplane.com> * * 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. Neither the name of The DragonFly Project 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 COPYRIGHT HOLDERS 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 * COPYRIGHT HOLDERS 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. * * --- * * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * Copyright (c) 1994 John S. Dyson * All rights reserved. * Copyright (c) 1994 David Greenman * 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. * 3. 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. * * --- * * 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. */ /* * Page fault handling module. */ #include "opt_vm.h" #include <sys/param.h> #include <sys/systm.h> #include <sys/kernel.h> #include <sys/proc.h> #include <sys/vnode.h> #include <sys/resourcevar.h> #include <sys/vmmeter.h> #include <sys/vkernel.h> #include <sys/lock.h> #include <sys/sysctl.h> #include <cpu/lwbuf.h> #include <vm/vm.h> #include <vm/vm_param.h> #include <vm/pmap.h> #include <vm/vm_map.h> #include <vm/vm_object.h> #include <vm/vm_page.h> #include <vm/vm_pageout.h> #include <vm/vm_kern.h> #include <vm/vm_pager.h> #include <vm/vnode_pager.h> #include <vm/swap_pager.h> #include <vm/vm_extern.h> #include <vm/vm_page2.h> #define VM_FAULT_MAX_QUICK 16 struct faultstate { vm_page_t mary[VM_FAULT_MAX_QUICK]; vm_map_backing_t ba; vm_prot_t prot; vm_page_t first_m; vm_map_backing_t first_ba; vm_prot_t first_prot; vm_map_t map; vm_map_entry_t entry; int lookup_still_valid; /* 0=inv 1=valid/rel -1=valid/atomic */ int hardfault; int fault_flags; int shared; int msoftonly; int first_shared; int wflags; int first_ba_held; /* 0=unlocked 1=locked/rel -1=lock/atomic */ struct vnode *vp; }; __read_mostly static int debug_fault = 0; SYSCTL_INT(_vm, OID_AUTO, debug_fault, CTLFLAG_RW, &debug_fault, 0, ""); __read_mostly static int debug_cluster = 0; SYSCTL_INT(_vm, OID_AUTO, debug_cluster, CTLFLAG_RW, &debug_cluster, 0, ""); /* VPAGETABLE debugging - counts and optional verbose output */ static long vpagetable_fault_count = 0; SYSCTL_LONG(_vm, OID_AUTO, vpagetable_faults, CTLFLAG_RW, &vpagetable_fault_count, 0, "Number of VPAGETABLE faults"); __read_mostly int debug_vpagetable = 0; SYSCTL_INT(_vm, OID_AUTO, debug_vpagetable, CTLFLAG_RW, &debug_vpagetable, 0, "Debug VPAGETABLE operations"); #if 0 static int virtual_copy_enable = 1; SYSCTL_INT(_vm, OID_AUTO, virtual_copy_enable, CTLFLAG_RW, &virtual_copy_enable, 0, ""); #endif __read_mostly int vm_shared_fault = 1; TUNABLE_INT("vm.shared_fault", &vm_shared_fault); SYSCTL_INT(_vm, OID_AUTO, shared_fault, CTLFLAG_RW, &vm_shared_fault, 0, "Allow shared token on vm_object"); __read_mostly static int vm_fault_bypass_count = 1; TUNABLE_INT("vm.fault_bypass", &vm_fault_bypass_count); SYSCTL_INT(_vm, OID_AUTO, fault_bypass, CTLFLAG_RW, &vm_fault_bypass_count, 0, "Allow fast vm_fault shortcut"); /* * Define here for debugging ioctls. Note that these are globals, so * they were cause a ton of cache line bouncing. Only use for debugging * purposes. */ /*#define VM_FAULT_QUICK_DEBUG */ #ifdef VM_FAULT_QUICK_DEBUG static long vm_fault_bypass_success_count = 0; SYSCTL_LONG(_vm, OID_AUTO, fault_bypass_success_count, CTLFLAG_RW, &vm_fault_bypass_success_count, 0, ""); static long vm_fault_bypass_failure_count1 = 0; SYSCTL_LONG(_vm, OID_AUTO, fault_bypass_failure_count1, CTLFLAG_RW, &vm_fault_bypass_failure_count1, 0, ""); static long vm_fault_bypass_failure_count2 = 0; SYSCTL_LONG(_vm, OID_AUTO, fault_bypass_failure_count2, CTLFLAG_RW, &vm_fault_bypass_failure_count2, 0, ""); static long vm_fault_bypass_failure_count3 = 0; SYSCTL_LONG(_vm, OID_AUTO, fault_bypass_failure_count3, CTLFLAG_RW, &vm_fault_bypass_failure_count3, 0, ""); static long vm_fault_bypass_failure_count4 = 0; SYSCTL_LONG(_vm, OID_AUTO, fault_bypass_failure_count4, CTLFLAG_RW, &vm_fault_bypass_failure_count4, 0, ""); #endif static int vm_fault_bypass(struct faultstate *fs, vm_pindex_t first_pindex, vm_pindex_t first_count, int *mextcountp, vm_prot_t fault_type); static int vm_fault_object(struct faultstate *, vm_pindex_t, vm_prot_t, int); static int vm_fault_vpagetable(struct faultstate *, vm_pindex_t *, vpte_t, int, int); static void vm_set_nosync(vm_page_t m, vm_map_entry_t entry); static void vm_prefault(pmap_t pmap, vm_offset_t addra, vm_map_entry_t entry, int prot, int fault_flags); static void vm_prefault_quick(pmap_t pmap, vm_offset_t addra, vm_map_entry_t entry, int prot, int fault_flags); #if 0 static struct krate vkrate = { 1 }; #endif static __inline void release_page(struct faultstate *fs) { vm_page_deactivate(fs->mary[0]); vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; } static __inline void unlock_map(struct faultstate *fs) { if (fs->ba != fs->first_ba) vm_object_drop(fs->ba->object); if (fs->first_ba && fs->first_ba_held == 1) { vm_object_drop(fs->first_ba->object); fs->first_ba_held = 0; fs->first_ba = NULL; } fs->ba = NULL; /* * NOTE: If lookup_still_valid == -1 the map is assumed to be locked * and caller expects it to remain locked atomically. */ if (fs->lookup_still_valid == 1 && fs->map) { vm_map_lookup_done(fs->map, fs->entry, 0); fs->lookup_still_valid = 0; fs->entry = NULL; } } /* * Clean up after a successful call to vm_fault_object() so another call * to vm_fault_object() can be made. */ static void cleanup_fault(struct faultstate *fs) { /* * We allocated a junk page for a COW operation that did * not occur, the page must be freed. */ if (fs->ba != fs->first_ba) { KKASSERT(fs->first_shared == 0); /* * first_m could be completely valid and we got here * because of a PG_RAM, don't mistakenly free it! */ if ((fs->first_m->valid & VM_PAGE_BITS_ALL) == VM_PAGE_BITS_ALL) { vm_page_wakeup(fs->first_m); } else { vm_page_free(fs->first_m); } vm_object_pip_wakeup(fs->ba->object); fs->first_m = NULL; /* * Reset fs->ba without calling unlock_map(), so we need a * little duplication. */ vm_object_drop(fs->ba->object); fs->ba = fs->first_ba; } } static void unlock_things(struct faultstate *fs) { cleanup_fault(fs); unlock_map(fs); if (fs->vp != NULL) { vput(fs->vp); fs->vp = NULL; } } #if 0 /* * Virtual copy tests. Used by the fault code to determine if a * page can be moved from an orphan vm_object into its shadow * instead of copying its contents. */ static __inline int virtual_copy_test(struct faultstate *fs) { /* * Must be holding exclusive locks */ if (fs->first_shared || fs->shared || virtual_copy_enable == 0) return 0; /* * Map, if present, has not changed */ if (fs->map && fs->map_generation != fs->map->timestamp) return 0; /* * No refs, except us */ if (fs->ba->object->ref_count != 1) return 0; /* * No one else can look this object up */ if (fs->ba->object->handle != NULL) return 0; /* * No other ways to look the object up */ if (fs->ba->object->type != OBJT_DEFAULT && fs->ba->object->type != OBJT_SWAP) return 0; /* * We don't chase down the shadow chain */ if (fs->ba != fs->first_ba->backing_ba) return 0; return 1; } static __inline int virtual_copy_ok(struct faultstate *fs) { if (virtual_copy_test(fs)) { /* * Grab the lock and re-test changeable items. */ if (fs->lookup_still_valid == 0 && fs->map) { if (lockmgr(&fs->map->lock, LK_EXCLUSIVE|LK_NOWAIT)) return 0; fs->lookup_still_valid = 1; if (virtual_copy_test(fs)) { fs->map_generation = ++fs->map->timestamp; return 1; } fs->lookup_still_valid = 0; lockmgr(&fs->map->lock, LK_RELEASE); } } return 0; } #endif /* * TRYPAGER * * Determine if the pager for the current object *might* contain the page. * * We only need to try the pager if this is not a default object (default * objects are zero-fill and have no real pager), and if we are not taking * a wiring fault forcing backing operation. * * Regular faults on wired areas do not force pager operation. */ #define TRYPAGER(fs) \ (fs->ba->object->type != OBJT_DEFAULT && \ (((fs->fault_flags & VM_FAULT_WIRE_MASK) == 0))) /* * vm_fault: * * Handle a page fault occuring at the given address, requiring the given * permissions, in the map specified. If successful, the page is inserted * into the associated physical map. * * NOTE: The given address should be truncated to the proper page address. * * KERN_SUCCESS is returned if the page fault is handled; otherwise, * a standard error specifying why the fault is fatal is returned. * * The map in question must be referenced, and remains so. * The caller may hold no locks. * No other requirements. */ int vm_fault(vm_map_t map, vm_offset_t vaddr, vm_prot_t fault_type, int fault_flags) { vm_pindex_t first_pindex; vm_pindex_t first_count; struct faultstate fs; struct lwp *lp; #if !defined(NO_SWAPPING) struct proc *p; #endif thread_t td; struct vm_map_ilock ilock; int mextcount; int didilock; int growstack; int retry = 0; int inherit_prot; int result; int n; inherit_prot = fault_type & VM_PROT_NOSYNC; fs.hardfault = 0; fs.fault_flags = fault_flags; fs.vp = NULL; fs.shared = vm_shared_fault; fs.first_shared = vm_shared_fault; growstack = 1; /* * vm_map interactions */ td = curthread; if ((lp = td->td_lwp) != NULL) lp->lwp_flags |= LWP_PAGING; RetryFault: /* * vm_fault_bypass() can shortcut us. */ fs.msoftonly = 0; fs.first_ba_held = 0; mextcount = 1; /* * Find the vm_map_entry representing the backing store and resolve * the top level object and page index. This may have the side * effect of executing a copy-on-write on the map entry, * creating a shadow object, or splitting an anonymous entry for * performance, but will not COW any actual VM pages. * * On success fs.map is left read-locked and various other fields * are initialized but not otherwise referenced or locked. * * NOTE! vm_map_lookup will try to upgrade the fault_type to * VM_FAULT_WRITE if the map entry is a virtual page table * and also writable, so we can set the 'A'accessed bit in * the virtual page table entry. */ fs.map = map; result = vm_map_lookup(&fs.map, vaddr, fault_type, &fs.entry, &fs.first_ba, &first_pindex, &first_count, &fs.first_prot, &fs.wflags); /* * If the lookup failed or the map protections are incompatible, * the fault generally fails. * * The failure could be due to TDF_NOFAULT if vm_map_lookup() * tried to do a COW fault. * * If the caller is trying to do a user wiring we have more work * to do. */ if (result != KERN_SUCCESS) { if (result == KERN_FAILURE_NOFAULT) { result = KERN_FAILURE; goto done; } if (result != KERN_PROTECTION_FAILURE || (fs.fault_flags & VM_FAULT_WIRE_MASK) != VM_FAULT_USER_WIRE) { if (result == KERN_INVALID_ADDRESS && growstack && map != kernel_map && curproc != NULL) { result = vm_map_growstack(map, vaddr); if (result == KERN_SUCCESS) { growstack = 0; ++retry; goto RetryFault; } result = KERN_FAILURE; } goto done; } /* * If we are user-wiring a r/w segment, and it is COW, then * we need to do the COW operation. Note that we don't * currently COW RO sections now, because it is NOT desirable * to COW .text. We simply keep .text from ever being COW'ed * and take the heat that one cannot debug wired .text sections. * * XXX Try to allow the above by specifying OVERRIDE_WRITE. */ result = vm_map_lookup(&fs.map, vaddr, VM_PROT_READ | VM_PROT_WRITE | VM_PROT_OVERRIDE_WRITE, &fs.entry, &fs.first_ba, &first_pindex, &first_count, &fs.first_prot, &fs.wflags); if (result != KERN_SUCCESS) { /* could also be KERN_FAILURE_NOFAULT */ result = KERN_FAILURE; goto done; } /* * If we don't COW now, on a user wire, the user will never * be able to write to the mapping. If we don't make this * restriction, the bookkeeping would be nearly impossible. * * XXX We have a shared lock, this will have a MP race but * I don't see how it can hurt anything. */ if ((fs.first_prot & VM_PROT_WRITE) == 0) { atomic_clear_char(&fs.entry->max_protection, VM_PROT_WRITE); } } /* * fs.map is read-locked * * Misc checks. Save the map generation number to detect races. */ fs.lookup_still_valid = 1; fs.first_m = NULL; fs.ba = fs.first_ba; /* so unlock_things() works */ fs.prot = fs.first_prot; /* default (used by uksmap) */ if (fs.entry->eflags & (MAP_ENTRY_NOFAULT | MAP_ENTRY_KSTACK)) { if (fs.entry->eflags & MAP_ENTRY_NOFAULT) { panic("vm_fault: fault on nofault entry, addr: %p", (void *)vaddr); } if ((fs.entry->eflags & MAP_ENTRY_KSTACK) && vaddr >= fs.entry->ba.start && vaddr < fs.entry->ba.start + PAGE_SIZE) { panic("vm_fault: fault on stack guard, addr: %p", (void *)vaddr); } } /* * A user-kernel shared map has no VM object and bypasses * everything. We execute the uksmap function with a temporary * fictitious vm_page. The address is directly mapped with no * management. */ if (fs.entry->maptype == VM_MAPTYPE_UKSMAP) { struct vm_page fakem; bzero(&fakem, sizeof(fakem)); fakem.pindex = first_pindex; fakem.flags = PG_FICTITIOUS | PG_UNQUEUED; fakem.busy_count = PBUSY_LOCKED; fakem.valid = VM_PAGE_BITS_ALL; fakem.pat_mode = VM_MEMATTR_DEFAULT; if (fs.entry->ba.uksmap(&fs.entry->ba, UKSMAPOP_FAULT, fs.entry->aux.dev, &fakem)) { result = KERN_FAILURE; unlock_things(&fs); goto done2; } pmap_enter(fs.map->pmap, vaddr, &fakem, fs.prot | inherit_prot, (fs.wflags & FW_WIRED), fs.entry); goto done_success; } /* * A system map entry may return a NULL object. No object means * no pager means an unrecoverable kernel fault. */ if (fs.first_ba == NULL) { panic("vm_fault: unrecoverable fault at %p in entry %p", (void *)vaddr, fs.entry); } /* * Fail here if not a trivial anonymous page fault and TDF_NOFAULT * is set. * * Unfortunately a deadlock can occur if we are forced to page-in * from swap, but diving all the way into the vm_pager_get_page() * function to find out is too much. Just check the object type. * * The deadlock is a CAM deadlock on a busy VM page when trying * to finish an I/O if another process gets stuck in * vop_helper_read_shortcut() due to a swap fault. */ if ((td->td_flags & TDF_NOFAULT) && (retry || fs.first_ba->object->type == OBJT_VNODE || fs.first_ba->object->type == OBJT_SWAP || fs.first_ba->backing_ba)) { result = KERN_FAILURE; unlock_things(&fs); goto done2; } #if 0 /* * REMOVED - The wiring flag does change fault behaviors */ /* * If the entry is wired the page protection level is limited to * what the vm_map_lookup() allowed us. * * XXX it is unclear if this code is still needed as vm_map_lookup() * no longer prevents protection changes on locked memory. REMOVE * IF WE DETERMINE THAT THIS CODE IS NO LONGER NEEDED. */ if (fs.wflags & FW_WIRED) fault_type = fs.first_prot; #endif /* * We generally want to avoid unnecessary exclusive modes on backing * and terminal objects because this can seriously interfere with * heavily fork()'d processes (particularly /bin/sh scripts). * * However, we also want to avoid unnecessary retries due to needed * shared->exclusive promotion for common faults. Exclusive mode is * always needed if any page insertion, rename, or free occurs in an * object (and also indirectly if any I/O is done). * * The main issue here is going to be fs.first_shared. If the * first_object has a backing object which isn't shadowed and the * process is single-threaded we might as well use an exclusive * lock/chain right off the bat. */ #if 0 /* WORK IN PROGRESS, CODE REMOVED */ if (fs.first_shared && fs.first_object->backing_object && LIST_EMPTY(&fs.first_object->shadow_head) && td->td_proc && td->td_proc->p_nthreads == 1) { fs.first_shared = 0; } #endif /* * VM_FAULT_UNSWAP - swap_pager_unswapped() needs an exclusive object * VM_FAULT_DIRTY - may require swap_pager_unswapped() later, but * we can try shared first. */ if (fault_flags & VM_FAULT_UNSWAP) fs.first_shared = 0; /* * Try to shortcut the entire mess and run the fault lockless. * This will burst in multiple pages via fs->mary[]. */ if (vm_fault_bypass_count && vm_fault_bypass(&fs, first_pindex, first_count, &mextcount, fault_type) == KERN_SUCCESS) { didilock = 0; fault_flags &= ~VM_FAULT_BURST; goto success; } /* * Exclusive heuristic (alloc page vs page exists) */ if (fs.first_ba->flags & VM_MAP_BACK_EXCL_HEUR) fs.first_shared = 0; /* * Obtain a top-level object lock, shared or exclusive depending * on fs.first_shared. If a shared lock winds up being insufficient * we will retry with an exclusive lock. * * The vnode pager lock is always shared. */ if (fs.first_shared) vm_object_hold_shared(fs.first_ba->object); else vm_object_hold(fs.first_ba->object); if (fs.vp == NULL) fs.vp = vnode_pager_lock(fs.first_ba); fs.first_ba_held = 1; /* * The page we want is at (first_object, first_pindex), but if the * vm_map_entry is VM_MAPTYPE_VPAGETABLE we have to traverse the * page table to figure out the actual pindex. * * NOTE! DEVELOPMENT IN PROGRESS, THIS IS AN INITIAL IMPLEMENTATION * ONLY */ didilock = 0; if (fs.entry->maptype == VM_MAPTYPE_VPAGETABLE) { ++vpagetable_fault_count; if (debug_vpagetable) { kprintf("VPAGETABLE fault: vaddr=%lx pde=%lx type=%02x pid=%d\n", vaddr, fs.entry->aux.master_pde, fault_type, (curproc ? curproc->p_pid : -1)); } vm_map_interlock(fs.map, &ilock, vaddr, vaddr + PAGE_SIZE); didilock = 1; result = vm_fault_vpagetable(&fs, &first_pindex, fs.entry->aux.master_pde, fault_type, 1); #if 0 krateprintf(&vkrate, "VKF va=%016jx m=%p res=%d\n", vaddr, fs.mary[0], result); #endif if (result == KERN_TRY_AGAIN) { vm_map_deinterlock(fs.map, &ilock); ++retry; goto RetryFault; } if (result != KERN_SUCCESS) { vm_map_deinterlock(fs.map, &ilock); goto done; } } /* * Now we have the actual (object, pindex), fault in the page. If * vm_fault_object() fails it will unlock and deallocate the FS * data. If it succeeds everything remains locked and fs->ba->object * will have an additional PIP count if fs->ba != fs->first_ba. * * vm_fault_object will set fs->prot for the pmap operation. It is * allowed to set VM_PROT_WRITE if fault_type == VM_PROT_READ if the * page can be safely written. However, it will force a read-only * mapping for a read fault if the memory is managed by a virtual * page table. * * If the fault code uses the shared object lock shortcut * we must not try to burst (we can't allocate VM pages). */ result = vm_fault_object(&fs, first_pindex, fault_type, 1); if (debug_fault > 0) { --debug_fault; kprintf("VM_FAULT result %d addr=%jx type=%02x flags=%02x " "fs.m=%p fs.prot=%02x fs.wflags=%02x fs.entry=%p\n", result, (intmax_t)vaddr, fault_type, fault_flags, fs.mary[0], fs.prot, fs.wflags, fs.entry); } if (result == KERN_TRY_AGAIN) { if (didilock) vm_map_deinterlock(fs.map, &ilock); ++retry; goto RetryFault; } if (result != KERN_SUCCESS) { if (didilock) vm_map_deinterlock(fs.map, &ilock); goto done; } success: /* * On success vm_fault_object() does not unlock or deallocate, and fs.m * will contain a busied page. It does drop fs->ba if appropriate. * * Enter the page into the pmap and do pmap-related adjustments. * * WARNING! Soft-busied fs.m's can only be manipulated in limited * ways. */ KKASSERT(fs.lookup_still_valid != 0); vm_page_flag_set(fs.mary[0], PG_REFERENCED); #if 0 /* * Mark pages mapped via VPAGETABLE so the pmap layer knows * that the backing_list scan won't find these mappings. * The vkernel is responsible for calling MADV_INVAL when * it modifies its page tables. */ if (fs.entry->maptype == VM_MAPTYPE_VPAGETABLE) { for (n = 0; n < mextcount; ++n) vm_page_flag_set(fs.mary[n], PG_VPTMAPPED); } #endif for (n = 0; n < mextcount; ++n) { pmap_enter(fs.map->pmap, vaddr + (n << PAGE_SHIFT), fs.mary[n], fs.prot | inherit_prot, fs.wflags & FW_WIRED, fs.entry); } if (didilock) vm_map_deinterlock(fs.map, &ilock); /* * If the page is not wired down, then put it where the pageout daemon * can find it. * * NOTE: We cannot safely wire, unwire, or adjust queues for a * soft-busied page. */ for (n = 0; n < mextcount; ++n) { if (fs.msoftonly) { KKASSERT(fs.mary[n]->busy_count & PBUSY_MASK); KKASSERT((fs.fault_flags & VM_FAULT_WIRE_MASK) == 0); vm_page_sbusy_drop(fs.mary[n]); } else { if (fs.fault_flags & VM_FAULT_WIRE_MASK) { #if 0 /* now handled by pmap_enter */ if (fs.wflags & FW_WIRED) vm_page_wire(fs.mary[n]); else vm_page_unwire(fs.mary[n], 1); #endif } else { vm_page_activate(fs.mary[n]); } KKASSERT(fs.mary[n]->busy_count & PBUSY_LOCKED); vm_page_wakeup(fs.mary[n]); } } /* * Burst in a few more pages if possible. The fs.map should still * be locked. To avoid interlocking against a vnode->getblk * operation we had to be sure to unbusy our primary vm_page above * first. * * A normal burst can continue down backing store, only execute * if we are holding an exclusive lock, otherwise the exclusive * locks the burst code gets might cause excessive SMP collisions. * * A quick burst can be utilized when there is no backing object * (i.e. a shared file mmap). */ if ((fault_flags & VM_FAULT_BURST) && (fs.fault_flags & VM_FAULT_WIRE_MASK) == 0 && (fs.wflags & FW_WIRED) == 0) { if (fs.first_shared == 0 && fs.shared == 0) { vm_prefault(fs.map->pmap, vaddr, fs.entry, fs.prot, fault_flags); } else { vm_prefault_quick(fs.map->pmap, vaddr, fs.entry, fs.prot, fault_flags); } } done_success: /* * Unlock everything, and return */ unlock_things(&fs); mycpu->gd_cnt.v_vm_faults++; if (td->td_lwp) { if (fs.hardfault) { ++td->td_lwp->lwp_ru.ru_majflt; } else { ++td->td_lwp->lwp_ru.ru_minflt; } } /*vm_object_deallocate(fs.first_ba->object);*/ /*fs.m = NULL; */ result = KERN_SUCCESS; done: if (fs.first_ba && fs.first_ba->object && fs.first_ba_held == 1) { vm_object_drop(fs.first_ba->object); fs.first_ba_held = 0; } done2: if (lp) lp->lwp_flags &= ~LWP_PAGING; #if !defined(NO_SWAPPING) /* * Check the process RSS limit and force deactivation and * (asynchronous) paging if necessary. This is a complex operation, * only do it for direct user-mode faults, for now. * * To reduce overhead implement approximately a ~16MB hysteresis. */ p = td->td_proc; if ((fault_flags & VM_FAULT_USERMODE) && lp && p->p_limit && map->pmap && vm_pageout_memuse_mode >= 1 && map != kernel_map) { vm_pindex_t limit; vm_pindex_t size; limit = OFF_TO_IDX(qmin(p->p_rlimit[RLIMIT_RSS].rlim_cur, p->p_rlimit[RLIMIT_RSS].rlim_max)); size = pmap_resident_tlnw_count(map->pmap); if (limit >= 0 && size > 4096 && size - 4096 >= limit) { vm_pageout_map_deactivate_pages(map, limit); } } #endif if (result != KERN_SUCCESS && debug_fault < 0) { kprintf("VM_FAULT %d:%d (%s) result %d " "addr=%jx type=%02x flags=%02x " "fs.m=%p fs.prot=%02x fs.wflags=%02x fs.entry=%p\n", (curthread->td_proc ? curthread->td_proc->p_pid : -1), (curthread->td_lwp ? curthread->td_lwp->lwp_tid : -1), curthread->td_comm, result, (intmax_t)vaddr, fault_type, fault_flags, fs.mary[0], fs.prot, fs.wflags, fs.entry); while (debug_fault < 0 && (debug_fault & 1)) tsleep(&debug_fault, 0, "DEBUG", hz); } return (result); } /* * Attempt a lockless vm_fault() shortcut. The stars have to align for this * to work. But if it does we can get our page only soft-busied and not * have to touch the vm_object or vnode locks at all. */ static int vm_fault_bypass(struct faultstate *fs, vm_pindex_t first_pindex, vm_pindex_t first_count, int *mextcountp, vm_prot_t fault_type) { vm_page_t m; vm_object_t obj; /* NOT LOCKED */ int n; int nlim; /* * Don't waste time if the object is only being used by one vm_map. */ obj = fs->first_ba->object; #if 0 if (obj->flags & OBJ_ONEMAPPING) return KERN_FAILURE; #endif /* * This will try to wire/unwire a page, which can't be done with * a soft-busied page. */ if (fs->fault_flags & VM_FAULT_WIRE_MASK) return KERN_FAILURE; /* * Can't handle VPAGETABLE - requires vm_fault_vpagetable() to * translate the pindex. */ if (fs->entry->maptype == VM_MAPTYPE_VPAGETABLE) { #ifdef VM_FAULT_QUICK_DEBUG ++vm_fault_bypass_failure_count1; #endif return KERN_FAILURE; } /* * Ok, try to get the vm_page quickly via the hash table. The * page will be soft-busied on success (NOT hard-busied). */ m = vm_page_hash_get(obj, first_pindex); if (m == NULL) { #ifdef VM_FAULT_QUICK_DEBUG ++vm_fault_bypass_failure_count2; #endif return KERN_FAILURE; } if ((obj->flags & OBJ_DEAD) || m->valid != VM_PAGE_BITS_ALL || m->queue - m->pc != PQ_ACTIVE || (m->flags & PG_SWAPPED)) { vm_page_sbusy_drop(m); #ifdef VM_FAULT_QUICK_DEBUG ++vm_fault_bypass_failure_count3; #endif return KERN_FAILURE; } /* * The page is already fully valid, ACTIVE, and is not PG_SWAPPED. * * Don't map the page writable when emulating the dirty bit, a * fault must be taken for proper emulation (vkernel). */ if (curthread->td_lwp && curthread->td_lwp->lwp_vmspace && pmap_emulate_ad_bits(&curthread->td_lwp->lwp_vmspace->vm_pmap)) { if ((fault_type & VM_PROT_WRITE) == 0) fs->prot &= ~VM_PROT_WRITE; } /* * If this is a write fault the object and the page must already * be writable. Since we don't hold an object lock and only a * soft-busy on the page, we cannot manipulate the object or * the page state (other than the page queue). */ if (fs->prot & VM_PROT_WRITE) { if ((obj->flags & (OBJ_WRITEABLE | OBJ_MIGHTBEDIRTY)) != (OBJ_WRITEABLE | OBJ_MIGHTBEDIRTY) || m->dirty != VM_PAGE_BITS_ALL) { vm_page_sbusy_drop(m); #ifdef VM_FAULT_QUICK_DEBUG ++vm_fault_bypass_failure_count4; #endif return KERN_FAILURE; } vm_set_nosync(m, fs->entry); } /* * Set page and potentially burst in more * * Even though we are only soft-busied we can still move pages * around in the normal queue(s). The soft-busy prevents the * page from being removed from the object, etc (normal operation). * * However, in this fast path it is excessively important to avoid * any hard locks, so we use a special passive version of activate. */ fs->msoftonly = 1; fs->mary[0] = m; vm_page_soft_activate(m); if (vm_fault_bypass_count > 1) { nlim = vm_fault_bypass_count; if (nlim > VM_FAULT_MAX_QUICK) /* array limit(+1) */ nlim = VM_FAULT_MAX_QUICK; if (nlim > first_count) /* user limit */ nlim = first_count; for (n = 1; n < nlim; ++n) { m = vm_page_hash_get(obj, first_pindex + n); if (m == NULL) break; if (m->valid != VM_PAGE_BITS_ALL || m->queue - m->pc != PQ_ACTIVE || (m->flags & PG_SWAPPED)) { vm_page_sbusy_drop(m); break; } if (fs->prot & VM_PROT_WRITE) { if ((obj->flags & (OBJ_WRITEABLE | OBJ_MIGHTBEDIRTY)) != (OBJ_WRITEABLE | OBJ_MIGHTBEDIRTY) || m->dirty != VM_PAGE_BITS_ALL) { vm_page_sbusy_drop(m); break; } } vm_page_soft_activate(m); fs->mary[n] = m; } *mextcountp = n; } #ifdef VM_FAULT_QUICK_DEBUG ++vm_fault_bypass_success_count; #endif return KERN_SUCCESS; } /* * Fault in the specified virtual address in the current process map, * returning a held VM page or NULL. See vm_fault_page() for more * information. * * No requirements. */ vm_page_t vm_fault_page_quick(vm_offset_t va, vm_prot_t fault_type, int *errorp, int *busyp) { struct lwp *lp = curthread->td_lwp; vm_page_t m; m = vm_fault_page(&lp->lwp_vmspace->vm_map, va, fault_type, VM_FAULT_NORMAL, errorp, busyp); return(m); } /* * Fault in the specified virtual address in the specified map, doing all * necessary manipulation of the object store and all necessary I/O. Return * a held VM page or NULL, and set *errorp. The related pmap is not * updated. * * If busyp is not NULL then *busyp will be set to TRUE if this routine * decides to return a busied page (aka VM_PROT_WRITE), or FALSE if it * does not (VM_PROT_WRITE not specified or busyp is NULL). If busyp is * NULL the returned page is only held. * * If the caller has no intention of writing to the page's contents, busyp * can be passed as NULL along with VM_PROT_WRITE to force a COW operation * without busying the page. * * The returned page will also be marked PG_REFERENCED. * * If the page cannot be faulted writable and VM_PROT_WRITE was specified, an * error will be returned. * * No requirements. */ vm_page_t vm_fault_page(vm_map_t map, vm_offset_t vaddr, vm_prot_t fault_type, int fault_flags, int *errorp, int *busyp) { vm_pindex_t first_pindex; vm_pindex_t first_count; struct faultstate fs; int result; int retry; int growstack; int didcow; vm_prot_t orig_fault_type = fault_type; retry = 0; didcow = 0; fs.hardfault = 0; fs.fault_flags = fault_flags; KKASSERT((fault_flags & VM_FAULT_WIRE_MASK) == 0); /* * Dive the pmap (concurrency possible). If we find the * appropriate page we can terminate early and quickly. * * This works great for normal programs but will always return * NULL for host lookups of vkernel maps in VMM mode. * * NOTE: pmap_fault_page_quick() might not busy the page. If * VM_PROT_WRITE is set in fault_type and pmap_fault_page_quick() * returns non-NULL, it will safely dirty the returned vm_page_t * for us. We cannot safely dirty it here (it might not be * busy). */ fs.mary[0] = pmap_fault_page_quick(map->pmap, vaddr, fault_type, busyp); if (fs.mary[0]) { *errorp = 0; return(fs.mary[0]); } /* * Otherwise take a concurrency hit and do a formal page * fault. */ fs.vp = NULL; fs.shared = vm_shared_fault; fs.first_shared = vm_shared_fault; fs.msoftonly = 0; growstack = 1; /* * VM_FAULT_UNSWAP - swap_pager_unswapped() needs an exclusive object * VM_FAULT_DIRTY - may require swap_pager_unswapped() later, but * we can try shared first. */ if (fault_flags & VM_FAULT_UNSWAP) { fs.first_shared = 0; } RetryFault: /* * Find the vm_map_entry representing the backing store and resolve * the top level object and page index. This may have the side * effect of executing a copy-on-write on the map entry and/or * creating a shadow object, but will not COW any actual VM pages. * * On success fs.map is left read-locked and various other fields * are initialized but not otherwise referenced or locked. * * NOTE! vm_map_lookup will upgrade the fault_type to VM_FAULT_WRITE * if the map entry is a virtual page table and also writable, * so we can set the 'A'accessed bit in the virtual page table * entry. */ fs.map = map; fs.first_ba_held = 0; result = vm_map_lookup(&fs.map, vaddr, fault_type, &fs.entry, &fs.first_ba, &first_pindex, &first_count, &fs.first_prot, &fs.wflags); if (result != KERN_SUCCESS) { if (result == KERN_FAILURE_NOFAULT) { *errorp = KERN_FAILURE; fs.mary[0] = NULL; goto done; } if (result != KERN_PROTECTION_FAILURE || (fs.fault_flags & VM_FAULT_WIRE_MASK) != VM_FAULT_USER_WIRE) { if (result == KERN_INVALID_ADDRESS && growstack && map != kernel_map && curproc != NULL) { result = vm_map_growstack(map, vaddr); if (result == KERN_SUCCESS) { growstack = 0; ++retry; goto RetryFault; } result = KERN_FAILURE; } fs.mary[0] = NULL; *errorp = result; goto done; } /* * If we are user-wiring a r/w segment, and it is COW, then * we need to do the COW operation. Note that we don't * currently COW RO sections now, because it is NOT desirable * to COW .text. We simply keep .text from ever being COW'ed * and take the heat that one cannot debug wired .text sections. */ result = vm_map_lookup(&fs.map, vaddr, VM_PROT_READ | VM_PROT_WRITE | VM_PROT_OVERRIDE_WRITE, &fs.entry, &fs.first_ba, &first_pindex, &first_count, &fs.first_prot, &fs.wflags); if (result != KERN_SUCCESS) { /* could also be KERN_FAILURE_NOFAULT */ *errorp = KERN_FAILURE; fs.mary[0] = NULL; goto done; } /* * If we don't COW now, on a user wire, the user will never * be able to write to the mapping. If we don't make this * restriction, the bookkeeping would be nearly impossible. * * XXX We have a shared lock, this will have a MP race but * I don't see how it can hurt anything. */ if ((fs.first_prot & VM_PROT_WRITE) == 0) { atomic_clear_char(&fs.entry->max_protection, VM_PROT_WRITE); } } /* * fs.map is read-locked * * Misc checks. Save the map generation number to detect races. */ fs.lookup_still_valid = 1; fs.first_m = NULL; fs.ba = fs.first_ba; if (fs.entry->eflags & MAP_ENTRY_NOFAULT) { panic("vm_fault: fault on nofault entry, addr: %lx", (u_long)vaddr); } /* * A user-kernel shared map has no VM object and bypasses * everything. We execute the uksmap function with a temporary * fictitious vm_page. The address is directly mapped with no * management. */ if (fs.entry->maptype == VM_MAPTYPE_UKSMAP) { struct vm_page fakem; bzero(&fakem, sizeof(fakem)); fakem.pindex = first_pindex; fakem.flags = PG_FICTITIOUS | PG_UNQUEUED; fakem.busy_count = PBUSY_LOCKED; fakem.valid = VM_PAGE_BITS_ALL; fakem.pat_mode = VM_MEMATTR_DEFAULT; if (fs.entry->ba.uksmap(&fs.entry->ba, UKSMAPOP_FAULT, fs.entry->aux.dev, &fakem)) { *errorp = KERN_FAILURE; fs.mary[0] = NULL; unlock_things(&fs); goto done2; } fs.mary[0] = PHYS_TO_VM_PAGE(fakem.phys_addr); vm_page_hold(fs.mary[0]); if (busyp) *busyp = 0; /* don't need to busy R or W */ unlock_things(&fs); *errorp = 0; goto done; } /* * A system map entry may return a NULL object. No object means * no pager means an unrecoverable kernel fault. */ if (fs.first_ba == NULL) { panic("vm_fault: unrecoverable fault at %p in entry %p", (void *)vaddr, fs.entry); } /* * Fail here if not a trivial anonymous page fault and TDF_NOFAULT * is set. * * Unfortunately a deadlock can occur if we are forced to page-in * from swap, but diving all the way into the vm_pager_get_page() * function to find out is too much. Just check the object type. */ if ((curthread->td_flags & TDF_NOFAULT) && (retry || fs.first_ba->object->type == OBJT_VNODE || fs.first_ba->object->type == OBJT_SWAP || fs.first_ba->backing_ba)) { *errorp = KERN_FAILURE; unlock_things(&fs); fs.mary[0] = NULL; goto done2; } #if 0 /* * REMOVED - The wiring flag does change fault behaviors */ /* * If the entry is wired the page protection level is limited to * what the vm_map_lookup() allowed us. * * XXX it is unclear if this code is still needed as vm_map_lookup() * no longer prevents protection changes on locked memory. REMOVE * IF WE DETERMINE THAT THIS CODE IS NO LONGER NEEDED. */ if (fs.wflags & FW_WIRED) fault_type = fs.first_prot; #endif /* * Make a reference to this object to prevent its disposal while we * are messing with it. Once we have the reference, the map is free * to be diddled. Since objects reference their shadows (and copies), * they will stay around as well. * * The reference should also prevent an unexpected collapse of the * parent that might move pages from the current object into the * parent unexpectedly, resulting in corruption. * * Bump the paging-in-progress count to prevent size changes (e.g. * truncation operations) during I/O. This must be done after * obtaining the vnode lock in order to avoid possible deadlocks. */ if (fs.first_ba->flags & VM_MAP_BACK_EXCL_HEUR) fs.first_shared = 0; if (fs.first_shared) vm_object_hold_shared(fs.first_ba->object); else vm_object_hold(fs.first_ba->object); fs.first_ba_held = 1; if (fs.vp == NULL) fs.vp = vnode_pager_lock(fs.first_ba); /* shared */ /* * The page we want is at (first_object, first_pindex), but if the * vm_map_entry is VM_MAPTYPE_VPAGETABLE we have to traverse the * page table to figure out the actual pindex. * * NOTE! DEVELOPMENT IN PROGRESS, THIS IS AN INITIAL IMPLEMENTATION * ONLY */ if (fs.entry->maptype == VM_MAPTYPE_VPAGETABLE) { result = vm_fault_vpagetable(&fs, &first_pindex, fs.entry->aux.master_pde, fault_type, 1); first_count = 1; if (result == KERN_TRY_AGAIN) { ++retry; goto RetryFault; } if (result != KERN_SUCCESS) { *errorp = result; fs.mary[0] = NULL; goto done; } } /* * Now we have the actual (object, pindex), fault in the page. If * vm_fault_object() fails it will unlock and deallocate the FS * data. If it succeeds everything remains locked and fs->ba->object * will have an additinal PIP count if fs->ba != fs->first_ba. */ fs.mary[0] = NULL; result = vm_fault_object(&fs, first_pindex, fault_type, 1); if (result == KERN_TRY_AGAIN) { KKASSERT(fs.first_ba_held == 0); ++retry; didcow |= fs.wflags & FW_DIDCOW; goto RetryFault; } if (result != KERN_SUCCESS) { *errorp = result; fs.mary[0] = NULL; goto done; } if ((orig_fault_type & VM_PROT_WRITE) && (fs.prot & VM_PROT_WRITE) == 0) { *errorp = KERN_PROTECTION_FAILURE; unlock_things(&fs); fs.mary[0] = NULL; goto done; } /* * Generally speaking we don't want to update the pmap because * this routine can be called many times for situations that do * not require updating the pmap, not to mention the page might * already be in the pmap. * * However, if our vm_map_lookup() results in a COW, we need to * at least remove the pte from the pmap to guarantee proper * visibility of modifications made to the process. For example, * modifications made by vkernel uiocopy/related routines and * modifications made by ptrace(). */ vm_page_flag_set(fs.mary[0], PG_REFERENCED); #if 0 /* * Mark pages mapped via VPAGETABLE so the pmap layer knows * that the backing_list scan won't find these mappings. */ if (fs.entry->maptype == VM_MAPTYPE_VPAGETABLE) vm_page_flag_set(fs.mary[0], PG_VPTMAPPED); #endif #if 0 pmap_enter(fs.map->pmap, vaddr, fs.mary[0], fs.prot, fs.wflags & FW_WIRED, NULL); mycpu->gd_cnt.v_vm_faults++; if (curthread->td_lwp) ++curthread->td_lwp->lwp_ru.ru_minflt; #endif if ((fs.wflags | didcow) & FW_DIDCOW) { pmap_remove(fs.map->pmap, vaddr & ~PAGE_MASK, (vaddr & ~PAGE_MASK) + PAGE_SIZE); #ifdef _KERNEL_VIRTUAL /* * For the vkernel, we must also call pmap_enter() to install * the new page in the software page table (VPTE) after COW. * The native kernel doesn't need this because the hardware * MMU will fault again, but the vkernel writes via DMAP and * the guest reads via the VPTE, so the VPTE must be updated * immediately. */ pmap_enter(fs.map->pmap, vaddr, fs.mary[0], fs.prot, fs.wflags & FW_WIRED, NULL); #endif } /* * On success vm_fault_object() does not unlock or deallocate, and * fs.mary[0] will contain a busied page. So we must unlock here * after having messed with the pmap. */ unlock_things(&fs); /* * Return a held page. We are not doing any pmap manipulation so do * not set PG_MAPPED. However, adjust the page flags according to * the fault type because the caller may not use a managed pmapping * (so we don't want to lose the fact that the page will be dirtied * if a write fault was specified). */ if (fault_type & VM_PROT_WRITE) vm_page_dirty(fs.mary[0]); vm_page_activate(fs.mary[0]); if (curthread->td_lwp) { if (fs.hardfault) { curthread->td_lwp->lwp_ru.ru_majflt++; } else { curthread->td_lwp->lwp_ru.ru_minflt++; } } /* * Unlock everything, and return the held or busied page. */ if (busyp) { if (fault_type & VM_PROT_WRITE) { vm_page_dirty(fs.mary[0]); *busyp = 1; } else { *busyp = 0; vm_page_hold(fs.mary[0]); vm_page_wakeup(fs.mary[0]); } } else { vm_page_hold(fs.mary[0]); vm_page_wakeup(fs.mary[0]); } /*vm_object_deallocate(fs.first_ba->object);*/ *errorp = 0; done: KKASSERT(fs.first_ba_held == 0); done2: return(fs.mary[0]); } /* * Fault in the specified (object,offset), dirty the returned page as * needed. If the requested fault_type cannot be done NULL and an * error is returned. * * A held (but not busied) page is returned. * * The passed in object must be held as specified by the shared * argument. */ vm_page_t vm_fault_object_page(vm_object_t object, vm_ooffset_t offset, vm_prot_t fault_type, int fault_flags, int *sharedp, int *errorp) { int result; vm_pindex_t first_pindex; vm_pindex_t first_count; struct faultstate fs; struct vm_map_entry entry; /* * Since we aren't actually faulting the page into a * pmap we can just fake the entry.ba. */ ASSERT_LWKT_TOKEN_HELD(vm_object_token(object)); bzero(&entry, sizeof(entry)); entry.maptype = VM_MAPTYPE_NORMAL; entry.protection = entry.max_protection = fault_type; entry.ba.backing_ba = NULL; entry.ba.object = object; entry.ba.offset = 0; fs.hardfault = 0; fs.fault_flags = fault_flags; fs.map = NULL; fs.shared = vm_shared_fault; fs.first_shared = *sharedp; fs.msoftonly = 0; fs.vp = NULL; fs.first_ba_held = -1; /* object held across call, prevent drop */ KKASSERT((fault_flags & VM_FAULT_WIRE_MASK) == 0); /* * VM_FAULT_UNSWAP - swap_pager_unswapped() needs an exclusive object * VM_FAULT_DIRTY - may require swap_pager_unswapped() later, but * we can try shared first. */ if (fs.first_shared && (fault_flags & VM_FAULT_UNSWAP)) { fs.first_shared = 0; vm_object_upgrade(object); } /* * Retry loop as needed (typically for shared->exclusive transitions) */ RetryFault: *sharedp = fs.first_shared; first_pindex = OFF_TO_IDX(offset); first_count = 1; fs.first_ba = &entry.ba; fs.ba = fs.first_ba; fs.entry = &entry; fs.first_prot = fault_type; fs.wflags = 0; /* * Make a reference to this object to prevent its disposal while we * are messing with it. Once we have the reference, the map is free * to be diddled. Since objects reference their shadows (and copies), * they will stay around as well. * * The reference should also prevent an unexpected collapse of the * parent that might move pages from the current object into the * parent unexpectedly, resulting in corruption. * * Bump the paging-in-progress count to prevent size changes (e.g. * truncation operations) during I/O. This must be done after * obtaining the vnode lock in order to avoid possible deadlocks. */ if (fs.vp == NULL) fs.vp = vnode_pager_lock(fs.first_ba); fs.lookup_still_valid = 1; fs.first_m = NULL; /* * Now we have the actual (object, pindex), fault in the page. If * vm_fault_object() fails it will unlock and deallocate the FS * data. If it succeeds everything remains locked and fs->ba->object * will have an additinal PIP count if fs->ba != fs->first_ba. * * On KERN_TRY_AGAIN vm_fault_object() leaves fs.first_ba intact. * We may have to upgrade its lock to handle the requested fault. */ result = vm_fault_object(&fs, first_pindex, fault_type, 0); if (result == KERN_TRY_AGAIN) { if (fs.first_shared == 0 && *sharedp) vm_object_upgrade(object); goto RetryFault; } if (result != KERN_SUCCESS) { *errorp = result; return(NULL); } if ((fault_type & VM_PROT_WRITE) && (fs.prot & VM_PROT_WRITE) == 0) { *errorp = KERN_PROTECTION_FAILURE; unlock_things(&fs); return(NULL); } /* * On success vm_fault_object() does not unlock or deallocate, so we * do it here. Note that the returned fs.m will be busied. */ unlock_things(&fs); /* * Return a held page. We are not doing any pmap manipulation so do * not set PG_MAPPED. However, adjust the page flags according to * the fault type because the caller may not use a managed pmapping * (so we don't want to lose the fact that the page will be dirtied * if a write fault was specified). */ vm_page_hold(fs.mary[0]); vm_page_activate(fs.mary[0]); if ((fault_type & VM_PROT_WRITE) || (fault_flags & VM_FAULT_DIRTY)) vm_page_dirty(fs.mary[0]); if (fault_flags & VM_FAULT_UNSWAP) swap_pager_unswapped(fs.mary[0]); /* * Indicate that the page was accessed. */ vm_page_flag_set(fs.mary[0], PG_REFERENCED); if (curthread->td_lwp) { if (fs.hardfault) { curthread->td_lwp->lwp_ru.ru_majflt++; } else { curthread->td_lwp->lwp_ru.ru_minflt++; } } /* * Unlock everything, and return the held page. */ vm_page_wakeup(fs.mary[0]); /*vm_object_deallocate(fs.first_ba->object);*/ *errorp = 0; return(fs.mary[0]); } /* * Translate the virtual page number (first_pindex) that is relative * to the address space into a logical page number that is relative to the * backing object. Use the virtual page table pointed to by (vpte). * * Possibly downgrade the protection based on the vpte bits. * * This implements an N-level page table. Any level can terminate the * scan by setting VPTE_PS. A linear mapping is accomplished by setting * VPTE_PS in the master page directory entry set via mcontrol(MADV_SETMAP). */ static int vm_fault_vpagetable(struct faultstate *fs, vm_pindex_t *pindex, vpte_t vpte, int fault_type, int allow_nofault) { struct lwbuf *lwb; struct lwbuf lwb_cache; int vshift = VPTE_FRAME_END - PAGE_SHIFT; /* index bits remaining */ int result; vpte_t *ptep; ASSERT_LWKT_TOKEN_HELD(vm_object_token(fs->first_ba->object)); for (;;) { /* * We cannot proceed if the vpte is not valid, not readable * for a read fault, not writable for a write fault, or * not executable for an instruction execution fault. */ if ((vpte & VPTE_V) == 0) { unlock_things(fs); return (KERN_FAILURE); } if ((fault_type & VM_PROT_WRITE) && (vpte & VPTE_RW) == 0) { unlock_things(fs); return (KERN_FAILURE); } if ((fault_type & VM_PROT_EXECUTE) && (vpte & VPTE_NX)) { unlock_things(fs); return (KERN_FAILURE); } if ((vpte & VPTE_PS) || vshift == 0) break; /* * Get the page table page. Nominally we only read the page * table, but since we are actively setting VPTE_M and VPTE_A, * tell vm_fault_object() that we are writing it. * * There is currently no real need to optimize this. */ result = vm_fault_object(fs, (vpte & VPTE_FRAME) >> PAGE_SHIFT, VM_PROT_READ|VM_PROT_WRITE, allow_nofault); if (result != KERN_SUCCESS) return (result); /* * Process the returned fs.mary[0] and look up the page table * entry in the page table page. */ vshift -= VPTE_PAGE_BITS; lwb = lwbuf_alloc(fs->mary[0], &lwb_cache); ptep = ((vpte_t *)lwbuf_kva(lwb) + ((*pindex >> vshift) & VPTE_PAGE_MASK)); vm_page_activate(fs->mary[0]); /* * Page table write-back - entire operation including * validation of the pte must be atomic to avoid races * against the vkernel changing the pte. * * If the vpte is valid for the* requested operation, do * a write-back to the page table. * * XXX VPTE_M is not set properly for page directory pages. * It doesn't get set in the page directory if the page table * is modified during a read access. */ for (;;) { vpte_t nvpte; /* * Reload for the cmpset, but make sure the pte is * still valid. */ vpte = *ptep; cpu_ccfence(); nvpte = vpte; if ((vpte & VPTE_V) == 0) break; if ((fault_type & VM_PROT_WRITE) && (vpte & VPTE_RW)) nvpte |= VPTE_M | VPTE_A; if (fault_type & (VM_PROT_READ | VM_PROT_EXECUTE)) nvpte |= VPTE_A; if (vpte == nvpte) break; if (atomic_cmpset_long(ptep, vpte, nvpte)) { vm_page_dirty(fs->mary[0]); break; } } lwbuf_free(lwb); vm_page_flag_set(fs->mary[0], PG_REFERENCED); vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; cleanup_fault(fs); } /* * When the vkernel sets VPTE_RW it expects the real kernel to * reflect VPTE_M back when the page is modified via the mapping. * In order to accomplish this the real kernel must map the page * read-only for read faults and use write faults to reflect VPTE_M * back. * * Once VPTE_M has been set, the real kernel's pte allows writing. * If the vkernel clears VPTE_M the vkernel must be sure to * MADV_INVAL the real kernel's mappings to force the real kernel * to re-fault on the next write so oit can set VPTE_M again. */ if ((fault_type & VM_PROT_WRITE) == 0 && (vpte & (VPTE_RW | VPTE_M)) != (VPTE_RW | VPTE_M)) { fs->first_prot &= ~VM_PROT_WRITE; } /* * Disable EXECUTE perms if NX bit is set. */ if (vpte & VPTE_NX) fs->first_prot &= ~VM_PROT_EXECUTE; /* * Combine remaining address bits with the vpte. */ *pindex = ((vpte & VPTE_FRAME) >> PAGE_SHIFT) + (*pindex & ((1L << vshift) - 1)); return (KERN_SUCCESS); } /* * This is the core of the vm_fault code. * * Do all operations required to fault-in (fs.first_ba->object, pindex). * Run through the backing store as necessary and do required COW or virtual * copy operations. The caller has already fully resolved the vm_map_entry * and, if appropriate, has created a copy-on-write layer. All we need to * do is iterate the object chain. * * On failure (fs) is unlocked and deallocated and the caller may return or * retry depending on the failure code. On success (fs) is NOT unlocked or * deallocated, fs.mary[0] will contained a resolved, busied page, and fs.ba's * object will have an additional PIP count if it is not equal to * fs.first_ba. * * If locks based on fs->first_shared or fs->shared are insufficient, * clear the appropriate field(s) and return RETRY. COWs require that * first_shared be 0, while page allocations (or frees) require that * shared be 0. Renames require that both be 0. * * NOTE! fs->[first_]shared might be set with VM_FAULT_DIRTY also set. * we will have to retry with it exclusive if the vm_page is * PG_SWAPPED. * * fs->first_ba->object must be held on call. */ static int vm_fault_object(struct faultstate *fs, vm_pindex_t first_pindex, vm_prot_t fault_type, int allow_nofault) { vm_map_backing_t next_ba; vm_pindex_t pindex; int error; ASSERT_LWKT_TOKEN_HELD(vm_object_token(fs->first_ba->object)); fs->prot = fs->first_prot; pindex = first_pindex; KKASSERT(fs->ba == fs->first_ba); vm_object_pip_add(fs->first_ba->object, 1); /* * If a read fault occurs we try to upgrade the page protection * and make it also writable if possible. There are three cases * where we cannot make the page mapping writable: * * (1) The mapping is read-only or the VM object is read-only, * fs->prot above will simply not have VM_PROT_WRITE set. * * (2) If the VM page is read-only or copy-on-write, upgrading would * just result in an unnecessary COW fault. * * (3) If the pmap specifically requests A/M bit emulation, downgrade * here. */ if (curthread->td_lwp && curthread->td_lwp->lwp_vmspace && pmap_emulate_ad_bits(&curthread->td_lwp->lwp_vmspace->vm_pmap)) { if ((fault_type & VM_PROT_WRITE) == 0) fs->prot &= ~VM_PROT_WRITE; } /* vm_object_hold(fs->ba->object); implied b/c ba == first_ba */ for (;;) { /* * If the object is dead, we stop here */ if (fs->ba->object->flags & OBJ_DEAD) { vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_PROTECTION_FAILURE); } /* * See if the page is resident. Wait/Retry if the page is * busy (lots of stuff may have changed so we can't continue * in that case). * * We can theoretically allow the soft-busy case on a read * fault if the page is marked valid, but since such * pages are typically already pmap'd, putting that * special case in might be more effort then it is * worth. We cannot under any circumstances mess * around with a vm_page_t->busy page except, perhaps, * to pmap it. */ fs->mary[0] = vm_page_lookup_busy_try(fs->ba->object, pindex, TRUE, &error); if (error) { vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); vm_page_sleep_busy(fs->mary[0], TRUE, "vmpfw"); mycpu->gd_cnt.v_intrans++; fs->mary[0] = NULL; return (KERN_TRY_AGAIN); } if (fs->mary[0]) { /* * The page is busied for us. * * If reactivating a page from PQ_CACHE we may have * to rate-limit. */ int queue = fs->mary[0]->queue; vm_page_unqueue_nowakeup(fs->mary[0]); if ((queue - fs->mary[0]->pc) == PQ_CACHE && vm_paging_severe()) { vm_page_activate(fs->mary[0]); vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); if (allow_nofault == 0 || (curthread->td_flags & TDF_NOFAULT) == 0) { thread_t td; vm_wait_pfault(); td = curthread; if (td->td_proc && (td->td_proc->p_flags & P_LOWMEMKILL)) return (KERN_PROTECTION_FAILURE); } return (KERN_TRY_AGAIN); } /* * If it still isn't completely valid (readable), * or if a read-ahead-mark is set on the VM page, * jump to readrest, else we found the page and * can return. * * We can release the spl once we have marked the * page busy. */ if (fs->mary[0]->object != kernel_object) { if ((fs->mary[0]->valid & VM_PAGE_BITS_ALL) != VM_PAGE_BITS_ALL) { goto readrest; } if (fs->mary[0]->flags & PG_RAM) { if (debug_cluster) kprintf("R"); vm_page_flag_clear(fs->mary[0], PG_RAM); goto readrest; } } atomic_clear_int(&fs->first_ba->flags, VM_MAP_BACK_EXCL_HEUR); break; /* break to PAGE HAS BEEN FOUND */ } /* * Page is not resident, If this is the search termination * or the pager might contain the page, allocate a new page. */ if (TRYPAGER(fs) || fs->ba == fs->first_ba) { /* * If this is a SWAP object we can use the shared * lock to check existence of a swap block. If * there isn't one we can skip to the next object. * * However, if this is the first object we allocate * a page now just in case we need to copy to it * later. */ if (fs->ba != fs->first_ba && fs->ba->object->type == OBJT_SWAP) { if (swap_pager_haspage_locked(fs->ba->object, pindex) == 0) { goto next; } } /* * Allocating, must be exclusive. */ atomic_set_int(&fs->first_ba->flags, VM_MAP_BACK_EXCL_HEUR); if (fs->ba == fs->first_ba && fs->first_shared) { fs->first_shared = 0; vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } if (fs->ba != fs->first_ba && fs->shared) { fs->first_shared = 0; fs->shared = 0; vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } /* * If the page is beyond the object size we fail */ if (pindex >= fs->ba->object->size) { vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_PROTECTION_FAILURE); } /* * Allocate a new page for this object/offset pair. * * It is possible for the allocation to race, so * handle the case. * * Does not apply to OBJT_MGTDEVICE (e.g. gpu / drm * subsystem). For OBJT_MGTDEVICE the pages are not * indexed in the VM object at all but instead directly * entered into the pmap. */ fs->mary[0] = NULL; if (fs->ba->object->type == OBJT_MGTDEVICE) goto readrest; if (!vm_paging_severe()) { fs->mary[0] = vm_page_alloc(fs->ba->object, pindex, ((fs->vp || fs->ba->backing_ba) ? VM_ALLOC_NULL_OK | VM_ALLOC_NORMAL : VM_ALLOC_NULL_OK | VM_ALLOC_NORMAL | VM_ALLOC_USE_GD | VM_ALLOC_ZERO)); } if (fs->mary[0] == NULL) { vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); if (allow_nofault == 0 || (curthread->td_flags & TDF_NOFAULT) == 0) { thread_t td; vm_wait_pfault(); td = curthread; if (td->td_proc && (td->td_proc->p_flags & P_LOWMEMKILL)) return (KERN_PROTECTION_FAILURE); } return (KERN_TRY_AGAIN); } /* * Fall through to readrest. We have a new page which * will have to be paged (since m->valid will be 0). */ } readrest: /* * We have found an invalid or partially valid page, a * page with a read-ahead mark which might be partially or * fully valid (and maybe dirty too), or we have allocated * a new page. * * Attempt to fault-in the page if there is a chance that the * pager has it, and potentially fault in additional pages * at the same time. * * If TRYPAGER is true then fs.mary[0] will be non-NULL and * busied for us. */ if (TRYPAGER(fs)) { u_char behavior = vm_map_entry_behavior(fs->entry); vm_object_t object; vm_page_t first_m; int seqaccess; int rv; if (behavior == MAP_ENTRY_BEHAV_RANDOM) seqaccess = 0; else seqaccess = -1; /* * Doing I/O may synchronously insert additional * pages so we can't be shared at this point either. * * NOTE: We can't free fs->mary[0] here in the * allocated case (fs->ba != fs->first_ba) as * this would require an exclusively locked * VM object. */ if (fs->ba == fs->first_ba && fs->first_shared) { if (fs->mary[0]) { vm_page_deactivate(fs->mary[0]); vm_page_wakeup(fs->mary[0]); fs->mary[0]= NULL; } fs->first_shared = 0; vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } if (fs->ba != fs->first_ba && fs->shared) { if (fs->mary[0]) { vm_page_deactivate(fs->mary[0]); vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; } fs->first_shared = 0; fs->shared = 0; vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } object = fs->ba->object; first_m = NULL; /* object is held, no more access to entry or ba's */ /* * Acquire the page data. We still hold object * and the page has been BUSY's. * * We own the page, but we must re-issue the lookup * because the pager may have replaced it (for example, * in order to enter a fictitious page into the * object). In this situation the pager will have * cleaned up the old page and left the new one * busy for us. * * If we got here through a PG_RAM read-ahead * mark the page may be partially dirty and thus * not freeable. Don't bother checking to see * if the pager has the page because we can't free * it anyway. We have to depend on the get_page * operation filling in any gaps whether there is * backing store or not. * * We must dispose of the page (fs->mary[0]) and also * possibly first_m (the fronting layer). If * this is a write fault leave the page intact * because we will probably have to copy fs->mary[0] * to fs->first_m on the retry. If this is a * read fault we probably won't need the page. * * For OBJT_MGTDEVICE (and eventually all types), * fs->mary[0] is not pre-allocated and may be set * to a vm_page (busied for us) without being inserted * into the object. In this case we want to return * the vm_page directly so the caller can issue the * pmap_enter(). */ rv = vm_pager_get_page(object, pindex, &fs->mary[0], seqaccess); if (rv == VM_PAGER_OK) { ++fs->hardfault; if (object->type == OBJT_MGTDEVICE) { break; } fs->mary[0] = vm_page_lookup(object, pindex); if (fs->mary[0]) { vm_page_activate(fs->mary[0]); vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; } if (fs->mary[0]) { /* NOT REACHED */ /* have page */ break; } vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } /* * If the pager doesn't have the page, continue on * to the next object. Retain the vm_page if this * is the first object, we may need to copy into * it later. */ if (rv == VM_PAGER_FAIL) { if (fs->ba != fs->first_ba) { if (fs->mary[0]) { vm_page_free(fs->mary[0]); fs->mary[0] = NULL; } } goto next; } /* * Remove the bogus page (which does not exist at this * object/offset). * * Also wake up any other process that may want to bring * in this page. * * If this is the top-level object, we must leave the * busy page to prevent another process from rushing * past us, and inserting the page in that object at * the same time that we are. */ if (rv == VM_PAGER_ERROR) { if (curproc) { kprintf("vm_fault: pager read error, " "pid %d (%s)\n", curproc->p_pid, curproc->p_comm); } else { kprintf("vm_fault: pager read error, " "thread %p (%s)\n", curthread, curthread->td_comm); } } /* * I/O error or data outside pager's range. */ if (fs->mary[0]) { vnode_pager_freepage(fs->mary[0]); fs->mary[0] = NULL; } if (first_m) { vm_page_free(first_m); first_m = NULL; /* safety */ } vm_object_pip_wakeup(object); unlock_things(fs); switch(rv) { case VM_PAGER_ERROR: return (KERN_FAILURE); case VM_PAGER_BAD: return (KERN_PROTECTION_FAILURE); default: return (KERN_PROTECTION_FAILURE); } #if 0 /* * Data outside the range of the pager or an I/O error * * The page may have been wired during the pagein, * e.g. by the buffer cache, and cannot simply be * freed. Call vnode_pager_freepage() to deal with it. * * The object is not held shared so we can safely * free the page. */ if (fs->ba != fs->first_ba) { /* * XXX - we cannot just fall out at this * point, m has been freed and is invalid! */ } /* * XXX - the check for kernel_map is a kludge to work * around having the machine panic on a kernel space * fault w/ I/O error. */ if (((fs->map != kernel_map) && (rv == VM_PAGER_ERROR)) || (rv == VM_PAGER_BAD)) { if (fs->m) { /* from just above */ KKASSERT(fs->first_shared == 0); vnode_pager_freepage(fs->m); fs->m = NULL; } /* NOT REACHED */ } #endif } next: /* * We get here if the object has a default pager (or unwiring) * or the pager doesn't have the page. * * fs->first_m will be used for the COW unless we find a * deeper page to be mapped read-only, in which case the * unlock*(fs) will free first_m. */ if (fs->ba == fs->first_ba) fs->first_m = fs->mary[0]; /* * Move on to the next object. The chain lock should prevent * the backing_object from getting ripped out from under us. * * The object lock for the next object is governed by * fs->shared. */ next_ba = fs->ba->backing_ba; if (next_ba == NULL) { /* * If there's no object left, fill the page in the top * object with zeros. */ if (fs->ba != fs->first_ba) { vm_object_pip_wakeup(fs->ba->object); vm_object_drop(fs->ba->object); fs->ba = fs->first_ba; pindex = first_pindex; fs->mary[0] = fs->first_m; } fs->first_m = NULL; /* * Zero the page and mark it valid. */ vm_page_zero_fill(fs->mary[0]); mycpu->gd_cnt.v_zfod++; fs->mary[0]->valid = VM_PAGE_BITS_ALL; break; /* break to PAGE HAS BEEN FOUND */ } if (fs->shared) vm_object_hold_shared(next_ba->object); else vm_object_hold(next_ba->object); KKASSERT(next_ba == fs->ba->backing_ba); pindex -= OFF_TO_IDX(fs->ba->offset); pindex += OFF_TO_IDX(next_ba->offset); if (fs->ba != fs->first_ba) { vm_object_pip_wakeup(fs->ba->object); vm_object_lock_swap(); /* flip ba/next_ba */ vm_object_drop(fs->ba->object); } fs->ba = next_ba; vm_object_pip_add(next_ba->object, 1); } /* * PAGE HAS BEEN FOUND. [Loop invariant still holds -- the object lock * is held.] * * object still held. * vm_map may not be locked (determined by fs->lookup_still_valid) * * local shared variable may be different from fs->shared. * * If the page is being written, but isn't already owned by the * top-level object, we have to copy it into a new page owned by the * top-level object. */ KASSERT((fs->mary[0]->busy_count & PBUSY_LOCKED) != 0, ("vm_fault: not busy after main loop")); if (fs->ba != fs->first_ba) { /* * We only really need to copy if we want to write it. */ if (fault_type & VM_PROT_WRITE) { #if 0 /* CODE REFACTOR IN PROGRESS, REMOVE OPTIMIZATION */ /* * This allows pages to be virtually copied from a * backing_object into the first_object, where the * backing object has no other refs to it, and cannot * gain any more refs. Instead of a bcopy, we just * move the page from the backing object to the * first object. Note that we must mark the page * dirty in the first object so that it will go out * to swap when needed. */ if (virtual_copy_ok(fs)) { /* * (first_m) and (m) are both busied. We have * move (m) into (first_m)'s object/pindex * in an atomic fashion, then free (first_m). * * first_object is held so second remove * followed by the rename should wind * up being atomic. vm_page_free() might * block so we don't do it until after the * rename. */ vm_page_protect(fs->first_m, VM_PROT_NONE); vm_page_remove(fs->first_m); vm_page_rename(fs->mary[0], fs->first_ba->object, first_pindex); vm_page_free(fs->first_m); fs->first_m = fs->mary[0]; fs->mary[0] = NULL; mycpu->gd_cnt.v_cow_optim++; } else #endif { /* * Oh, well, lets copy it. * * We used to unmap the original page here * because vm_fault_page() didn't and this * would cause havoc for the umtx*() code * and the procfs code. * * This is no longer necessary. The * vm_fault_page() routine will now unmap the * page after a COW, and the umtx code will * recover on its own. */ /* * NOTE: Since fs->mary[0] is a backing page, * it is read-only, so there isn't any * copy race vs writers. */ KKASSERT(fs->first_shared == 0); vm_page_copy(fs->mary[0], fs->first_m); /* pmap_remove_specific( &curthread->td_lwp->lwp_vmspace->vm_pmap, fs->mary[0]); */ } /* * We no longer need the old page or object. */ if (fs->mary[0]) release_page(fs); /* * fs->ba != fs->first_ba due to above conditional */ vm_object_pip_wakeup(fs->ba->object); vm_object_drop(fs->ba->object); fs->ba = fs->first_ba; /* * Only use the new page below... * * Set FW_DIDCOW so callers (vm_fault_page) know that * page-level COW occurred and can update the pmap * accordingly (required for vkernel VPTE sync). */ mycpu->gd_cnt.v_cow_faults++; fs->wflags |= FW_DIDCOW; fs->mary[0] = fs->first_m; pindex = first_pindex; } else { /* * If it wasn't a write fault avoid having to copy * the page by mapping it read-only from backing * store. The process is not allowed to modify * backing pages. */ fs->prot &= ~VM_PROT_WRITE; } } /* * Relock the map if necessary, then check the generation count. * relock_map() will update fs->timestamp to account for the * relocking if necessary. * * If the count has changed after relocking then all sorts of * crap may have happened and we have to retry. * * NOTE: The relock_map() can fail due to a deadlock against * the vm_page we are holding BUSY. */ KKASSERT(fs->lookup_still_valid != 0); #if 0 if (fs->lookup_still_valid == 0 && fs->map) { if (relock_map(fs) || fs->map->timestamp != fs->map_generation) { release_page(fs); vm_object_pip_wakeup(fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } } #endif /* * If the fault is a write, we know that this page is being * written NOW so dirty it explicitly to save on pmap_is_modified() * calls later. * * If this is a NOSYNC mmap we do not want to set PG_NOSYNC * if the page is already dirty to prevent data written with * the expectation of being synced from not being synced. * Likewise if this entry does not request NOSYNC then make * sure the page isn't marked NOSYNC. Applications sharing * data should use the same flags to avoid ping ponging. * * Also tell the backing pager, if any, that it should remove * any swap backing since the page is now dirty. */ vm_page_activate(fs->mary[0]); if (fs->prot & VM_PROT_WRITE) { vm_object_set_writeable_dirty(fs->first_ba->object); vm_set_nosync(fs->mary[0], fs->entry); if (fs->fault_flags & VM_FAULT_DIRTY) { vm_page_dirty(fs->mary[0]); if (fs->mary[0]->flags & PG_SWAPPED) { /* * If the page is swapped out we have to call * swap_pager_unswapped() which requires an * exclusive object lock. If we are shared, * we must clear the shared flag and retry. */ if ((fs->ba == fs->first_ba && fs->first_shared) || (fs->ba != fs->first_ba && fs->shared)) { vm_page_wakeup(fs->mary[0]); fs->mary[0] = NULL; if (fs->ba == fs->first_ba) fs->first_shared = 0; else fs->shared = 0; vm_object_pip_wakeup( fs->first_ba->object); unlock_things(fs); return (KERN_TRY_AGAIN); } swap_pager_unswapped(fs->mary[0]); } } } /* * We found our page at backing layer ba. Leave the layer state * intact. */ vm_object_pip_wakeup(fs->first_ba->object); #if 0 if (fs->ba != fs->first_ba) vm_object_drop(fs->ba->object); #endif /* * Page had better still be busy. We are still locked up and * fs->ba->object will have another PIP reference for the case * where fs->ba != fs->first_ba. */ KASSERT(fs->mary[0]->busy_count & PBUSY_LOCKED, ("vm_fault: page %p not busy!", fs->mary[0])); /* * Sanity check: page must be completely valid or it is not fit to * map into user space. vm_pager_get_pages() ensures this. */ if (fs->mary[0]->valid != VM_PAGE_BITS_ALL) { vm_page_zero_invalid(fs->mary[0], TRUE); kprintf("Warning: page %p partially invalid on fault\n", fs->mary[0]); } return (KERN_SUCCESS); } /* * Wire down a range of virtual addresses in a map. The entry in question * should be marked in-transition and the map must be locked. We must * release the map temporarily while faulting-in the page to avoid a * deadlock. Note that the entry may be clipped while we are blocked but * will never be freed. * * map must be locked on entry. */ int vm_fault_wire(vm_map_t map, vm_map_entry_t entry, boolean_t user_wire, int kmflags) { boolean_t fictitious; vm_offset_t start; vm_offset_t end; vm_offset_t va; pmap_t pmap; int rv; int wire_prot; int fault_flags; vm_page_t m; if (user_wire) { wire_prot = VM_PROT_READ; fault_flags = VM_FAULT_USER_WIRE; } else { wire_prot = VM_PROT_READ | VM_PROT_WRITE; fault_flags = VM_FAULT_CHANGE_WIRING; } if (kmflags & KM_NOTLBSYNC) wire_prot |= VM_PROT_NOSYNC; pmap = vm_map_pmap(map); start = entry->ba.start; end = entry->ba.end; switch(entry->maptype) { case VM_MAPTYPE_NORMAL: case VM_MAPTYPE_VPAGETABLE: fictitious = entry->ba.object && ((entry->ba.object->type == OBJT_DEVICE) || (entry->ba.object->type == OBJT_MGTDEVICE)); break; case VM_MAPTYPE_UKSMAP: fictitious = TRUE; break; default: fictitious = FALSE; break; } if (entry->eflags & MAP_ENTRY_KSTACK) start += PAGE_SIZE; map->timestamp++; vm_map_unlock(map); /* * We simulate a fault to get the page and enter it in the physical * map. */ for (va = start; va < end; va += PAGE_SIZE) { rv = vm_fault(map, va, wire_prot, fault_flags); if (rv) { vm_offset_t rva = start; while (rva < va) { m = pmap_unwire(pmap, &rva); if (m && !fictitious) { vm_page_busy_wait(m, FALSE, "vmwrpg"); vm_page_unwire(m, 1); vm_page_wakeup(m); } } goto done; } } rv = KERN_SUCCESS; done: vm_map_lock(map); return (rv); } /* * Unwire a range of virtual addresses in a map. The map should be * locked. */ void vm_fault_unwire(vm_map_t map, vm_map_entry_t entry) { boolean_t fictitious; vm_offset_t start; vm_offset_t end; vm_offset_t va; pmap_t pmap; vm_page_t m; /* * We only actually unwire stuff once the entry's wired_count * reaches 0. More than one frontend might have been wiring the * entry. */ KKASSERT(entry->wired_count); if (--entry->wired_count > 0) return; pmap = vm_map_pmap(map); start = entry->ba.start; end = entry->ba.end; fictitious = entry->ba.object && ((entry->ba.object->type == OBJT_DEVICE) || (entry->ba.object->type == OBJT_MGTDEVICE)); if (entry->eflags & MAP_ENTRY_KSTACK) start += PAGE_SIZE; /* * Unwire any wired pages found. Not all the pages in the * range will necessarily be wired. */ va = start; while (va < end) { m = pmap_unwire(pmap, &va); if (m && !fictitious) { vm_page_busy_wait(m, FALSE, "vmwrpg"); vm_page_unwire(m, 1); vm_page_wakeup(m); } } } /* * Simulate write faults to bring all data into the head object, return * KERN_SUCCESS on success (which should be always unless the system runs * out of memory). * * The caller will handle destroying the backing_ba's. */ int vm_fault_collapse(vm_map_t map, vm_map_entry_t entry) { struct faultstate fs; vm_ooffset_t scan; vm_pindex_t pindex; vm_object_t object; int rv; int all_shadowed; bzero(&fs, sizeof(fs)); object = entry->ba.object; fs.first_prot = entry->max_protection | /* optional VM_PROT_EXECUTE */ VM_PROT_READ | VM_PROT_WRITE | VM_PROT_OVERRIDE_WRITE; fs.fault_flags = VM_FAULT_NORMAL; fs.map = map; fs.entry = entry; fs.lookup_still_valid = -1; /* leave map atomically locked */ fs.first_ba = &entry->ba; fs.first_ba_held = -1; /* leave object held */ /* fs.hardfault */ vm_object_hold(object); rv = KERN_SUCCESS; scan = entry->ba.start; all_shadowed = 1; while (scan < entry->ba.end) { pindex = OFF_TO_IDX(entry->ba.offset + (scan - entry->ba.start)); if (vm_page_lookup(object, pindex)) { scan += PAGE_SIZE; continue; } all_shadowed = 0; fs.ba = fs.first_ba; fs.prot = fs.first_prot; rv = vm_fault_object(&fs, pindex, fs.first_prot, 1); if (rv == KERN_TRY_AGAIN) continue; if (rv != KERN_SUCCESS) break; vm_page_flag_set(fs.mary[0], PG_REFERENCED); vm_page_activate(fs.mary[0]); vm_page_wakeup(fs.mary[0]); scan += PAGE_SIZE; } KKASSERT(entry->ba.object == object); vm_object_drop(object); /* * If the fronting object did not have every page we have to clear * the pmap range due to the pages being changed so we can fault-in * the proper pages. */ if (all_shadowed == 0) pmap_remove(map->pmap, entry->ba.start, entry->ba.end); return rv; } /* * Copy all of the pages from one map entry to another. If the source * is wired down we just use vm_page_lookup(). If not we use * vm_fault_object(). * * The source and destination maps must be locked for write. * The source and destination maps token must be held * * No other requirements. * * XXX do segment optimization */ void vm_fault_copy_entry(vm_map_t dst_map, vm_map_t src_map, vm_map_entry_t dst_entry, vm_map_entry_t src_entry) { vm_object_t dst_object; vm_object_t src_object; vm_ooffset_t dst_offset; vm_ooffset_t src_offset; vm_prot_t prot; vm_offset_t vaddr; vm_page_t dst_m; vm_page_t src_m; src_object = src_entry->ba.object; src_offset = src_entry->ba.offset; /* * Create the top-level object for the destination entry. (Doesn't * actually shadow anything - we copy the pages directly.) */ vm_map_entry_allocate_object(dst_entry); dst_object = dst_entry->ba.object; prot = dst_entry->max_protection; /* * Loop through all of the pages in the entry's range, copying each * one from the source object (it should be there) to the destination * object. */ vm_object_hold(src_object); vm_object_hold(dst_object); for (vaddr = dst_entry->ba.start, dst_offset = 0; vaddr < dst_entry->ba.end; vaddr += PAGE_SIZE, dst_offset += PAGE_SIZE) { /* * Allocate a page in the destination object */ do { dst_m = vm_page_alloc(dst_object, OFF_TO_IDX(dst_offset), VM_ALLOC_NORMAL); if (dst_m == NULL) { vm_wait(0); } } while (dst_m == NULL); /* * Find the page in the source object, and copy it in. * (Because the source is wired down, the page will be in * memory.) */ src_m = vm_page_lookup(src_object, OFF_TO_IDX(dst_offset + src_offset)); if (src_m == NULL) panic("vm_fault_copy_wired: page missing"); vm_page_copy(src_m, dst_m); /* * Enter it in the pmap... */ pmap_enter(dst_map->pmap, vaddr, dst_m, prot, FALSE, dst_entry); /* * Mark it no longer busy, and put it on the active list. */ vm_page_activate(dst_m); vm_page_wakeup(dst_m); } vm_object_drop(dst_object); vm_object_drop(src_object); } /* * vm_prefault() provides a quick way of clustering pagefaults into a * processes address space. It is a "cousin" of pmap_object_init_pt, * except it runs at page fault time instead of mmap time. * * vm.fast_fault Enables pre-faulting zero-fill pages * * vm.prefault_pages Number of pages (1/2 negative, 1/2 positive) to * prefault. Scan stops in either direction when * a page is found to already exist. * * This code used to be per-platform pmap_prefault(). It is now * machine-independent and enhanced to also pre-fault zero-fill pages * (see vm.fast_fault) as well as make them writable, which greatly * reduces the number of page faults programs incur. * * Application performance when pre-faulting zero-fill pages is heavily * dependent on the application. Very tiny applications like /bin/echo * lose a little performance while applications of any appreciable size * gain performance. Prefaulting multiple pages also reduces SMP * congestion and can improve SMP performance significantly. * * NOTE! prot may allow writing but this only applies to the top level * object. If we wind up mapping a page extracted from a backing * object we have to make sure it is read-only. * * NOTE! The caller has already handled any COW operations on the * vm_map_entry via the normal fault code. Do NOT call this * shortcut unless the normal fault code has run on this entry. * * The related map must be locked. * No other requirements. */ __read_mostly static int vm_prefault_pages = 8; SYSCTL_INT(_vm, OID_AUTO, prefault_pages, CTLFLAG_RW, &vm_prefault_pages, 0, "Maximum number of pages to pre-fault"); __read_mostly static int vm_fast_fault = 1; SYSCTL_INT(_vm, OID_AUTO, fast_fault, CTLFLAG_RW, &vm_fast_fault, 0, "Burst fault zero-fill regions"); /* * Set PG_NOSYNC if the map entry indicates so, but only if the page * is not already dirty by other means. This will prevent passive * filesystem syncing as well as 'sync' from writing out the page. */ static void vm_set_nosync(vm_page_t m, vm_map_entry_t entry) { if (entry->eflags & MAP_ENTRY_NOSYNC) { if (m->dirty == 0) vm_page_flag_set(m, PG_NOSYNC); } else { vm_page_flag_clear(m, PG_NOSYNC); } } static void vm_prefault(pmap_t pmap, vm_offset_t addra, vm_map_entry_t entry, int prot, int fault_flags) { vm_map_backing_t ba; /* first ba */ struct lwp *lp; vm_page_t m; vm_offset_t addr; vm_pindex_t index; vm_pindex_t pindex; vm_object_t object; int pprot; int i; int noneg; int nopos; int maxpages; /* * Get stable max count value, disabled if set to 0 */ maxpages = vm_prefault_pages; cpu_ccfence(); if (maxpages <= 0) return; /* * We do not currently prefault mappings that use virtual page * tables. We do not prefault foreign pmaps. */ if (entry->maptype != VM_MAPTYPE_NORMAL) return; lp = curthread->td_lwp; if (lp == NULL || (pmap != vmspace_pmap(lp->lwp_vmspace))) return; /* * Limit pre-fault count to 1024 pages. */ if (maxpages > 1024) maxpages = 1024; ba = &entry->ba; object = entry->ba.object; KKASSERT(object != NULL); /* * NOTE: VM_FAULT_DIRTY allowed later so must hold object exclusively * now (or do something more complex XXX). */ vm_object_hold(object); noneg = 0; nopos = 0; for (i = 0; i < maxpages; ++i) { vm_object_t lobject; vm_object_t nobject; vm_map_backing_t last_ba; /* last ba */ vm_map_backing_t next_ba; /* last ba */ int allocated = 0; int error; /* * This can eat a lot of time on a heavily contended * machine so yield on the tick if needed. */ if ((i & 7) == 7) lwkt_yield(); /* * Calculate the page to pre-fault, stopping the scan in * each direction separately if the limit is reached. */ if (i & 1) { if (noneg) continue; addr = addra - ((i + 1) >> 1) * PAGE_SIZE; } else { if (nopos) continue; addr = addra + ((i + 2) >> 1) * PAGE_SIZE; } if (addr < entry->ba.start) { noneg = 1; if (noneg && nopos) break; continue; } if (addr >= entry->ba.end) { nopos = 1; if (noneg && nopos) break; continue; } /* * Skip pages already mapped, and stop scanning in that * direction. When the scan terminates in both directions * we are done. */ if (pmap_prefault_ok(pmap, addr) == 0) { if (i & 1) noneg = 1; else nopos = 1; if (noneg && nopos) break; continue; } /* * Follow the backing layers to obtain the page to be mapped * into the pmap. * * If we reach the terminal object without finding a page * and we determine it would be advantageous, then allocate * a zero-fill page for the base object. The base object * is guaranteed to be OBJT_DEFAULT for this case. * * In order to not have to check the pager via *haspage*() * we stop if any non-default object is encountered. e.g. * a vnode or swap object would stop the loop. */ index = ((addr - entry->ba.start) + entry->ba.offset) >> PAGE_SHIFT; last_ba = ba; lobject = object; pindex = index; pprot = prot; /*vm_object_hold(lobject); implied */ while ((m = vm_page_lookup_busy_try(lobject, pindex, TRUE, &error)) == NULL) { if (lobject->type != OBJT_DEFAULT) break; if ((next_ba = last_ba->backing_ba) == NULL) { if (vm_fast_fault == 0) break; if ((prot & VM_PROT_WRITE) == 0 || vm_paging_min()) { break; } /* * NOTE: Allocated from base object */ m = vm_page_alloc(object, index, VM_ALLOC_NORMAL | VM_ALLOC_ZERO | VM_ALLOC_USE_GD | VM_ALLOC_NULL_OK); if (m == NULL) break; allocated = 1; pprot = prot; /* lobject = object .. not needed */ break; } if (next_ba->offset & PAGE_MASK) break; nobject = next_ba->object; vm_object_hold(nobject); pindex -= last_ba->offset >> PAGE_SHIFT; pindex += next_ba->offset >> PAGE_SHIFT; if (last_ba != ba) { vm_object_lock_swap(); vm_object_drop(lobject); } lobject = nobject; last_ba = next_ba; pprot &= ~VM_PROT_WRITE; } /* * NOTE: A non-NULL (m) will be associated with lobject if * it was found there, otherwise it is probably a * zero-fill page associated with the base object. * * Give-up if no page is available. */ if (m == NULL) { if (last_ba != ba) vm_object_drop(lobject); break; } /* * The object must be marked dirty if we are mapping a * writable page. Note that (m) does not have to be * entered into the object, so use lobject or object * as appropriate instead of m->object. * * Do this before we potentially drop the object. */ if (pprot & VM_PROT_WRITE) { vm_object_set_writeable_dirty( (allocated ? object : lobject)); } /* * Do not conditionalize on PG_RAM. If pages are present in * the VM system we assume optimal caching. If caching is * not optimal the I/O gravy train will be restarted when we * hit an unavailable page. We do not want to try to restart * the gravy train now because we really don't know how much * of the object has been cached. The cost for restarting * the gravy train should be low (since accesses will likely * be I/O bound anyway). */ if (last_ba != ba) vm_object_drop(lobject); /* * Enter the page into the pmap if appropriate. If we had * allocated the page we have to place it on a queue. If not * we just have to make sure it isn't on the cache queue * (pages on the cache queue are not allowed to be mapped). * * When allocated is TRUE, m corresponds to object, * not lobject. */ if (allocated) { /* * Page must be zerod. */ vm_page_zero_fill(m); mycpu->gd_cnt.v_zfod++; m->valid = VM_PAGE_BITS_ALL; /* * Handle dirty page case */ if (pprot & VM_PROT_WRITE) vm_set_nosync(m, entry); pmap_enter(pmap, addr, m, pprot, 0, entry); #if 0 /* REMOVE ME, a burst counts as one fault */ mycpu->gd_cnt.v_vm_faults++; if (curthread->td_lwp) ++curthread->td_lwp->lwp_ru.ru_minflt; #endif vm_page_deactivate(m); if (pprot & VM_PROT_WRITE) { /*vm_object_set_writeable_dirty(object);*/ vm_set_nosync(m, entry); if (fault_flags & VM_FAULT_DIRTY) { vm_page_dirty(m); /*XXX*/ swap_pager_unswapped(m); } } vm_page_wakeup(m); } else if (error) { /* couldn't busy page, no wakeup */ } else if ( ((m->valid & VM_PAGE_BITS_ALL) == VM_PAGE_BITS_ALL) && (m->flags & PG_FICTITIOUS) == 0) { /* * A fully valid page not undergoing soft I/O can * be immediately entered into the pmap. * * When allocated is false, m corresponds to lobject. */ if ((m->queue - m->pc) == PQ_CACHE) vm_page_deactivate(m); if (pprot & VM_PROT_WRITE) { /*vm_object_set_writeable_dirty(lobject);*/ vm_set_nosync(m, entry); if (fault_flags & VM_FAULT_DIRTY) { vm_page_dirty(m); /*XXX*/ swap_pager_unswapped(m); } } if (pprot & VM_PROT_WRITE) vm_set_nosync(m, entry); pmap_enter(pmap, addr, m, pprot, 0, entry); #if 0 /* REMOVE ME, a burst counts as one fault */ mycpu->gd_cnt.v_vm_faults++; if (curthread->td_lwp) ++curthread->td_lwp->lwp_ru.ru_minflt; #endif vm_page_wakeup(m); } else { vm_page_wakeup(m); } } vm_object_drop(object); } /* * Object can be held shared */ static void vm_prefault_quick(pmap_t pmap, vm_offset_t addra, vm_map_entry_t entry, int prot, int fault_flags) { struct lwp *lp; vm_page_t m; vm_offset_t addr; vm_pindex_t pindex; vm_object_t object; int i; int noneg; int nopos; int maxpages; /* * Get stable max count value, disabled if set to 0 */ maxpages = vm_prefault_pages; cpu_ccfence(); if (maxpages <= 0) return; /* * We do not currently prefault mappings that use virtual page * tables. We do not prefault foreign pmaps. */ if (entry->maptype != VM_MAPTYPE_NORMAL) return; lp = curthread->td_lwp; if (lp == NULL || (pmap != vmspace_pmap(lp->lwp_vmspace))) return; object = entry->ba.object; if (entry->ba.backing_ba != NULL) return; ASSERT_LWKT_TOKEN_HELD(vm_object_token(object)); /* * Limit pre-fault count to 1024 pages. */ if (maxpages > 1024) maxpages = 1024; noneg = 0; nopos = 0; for (i = 0; i < maxpages; ++i) { int error; /* * Calculate the page to pre-fault, stopping the scan in * each direction separately if the limit is reached. */ if (i & 1) { if (noneg) continue; addr = addra - ((i + 1) >> 1) * PAGE_SIZE; } else { if (nopos) continue; addr = addra + ((i + 2) >> 1) * PAGE_SIZE; } if (addr < entry->ba.start) { noneg = 1; if (noneg && nopos) break; continue; } if (addr >= entry->ba.end) { nopos = 1; if (noneg && nopos) break; continue; } /* * Follow the VM object chain to obtain the page to be mapped * into the pmap. This version of the prefault code only * works with terminal objects. * * The page must already exist. If we encounter a problem * we stop here. * * WARNING! We cannot call swap_pager_unswapped() or insert * a new vm_page with a shared token. */ pindex = ((addr - entry->ba.start) + entry->ba.offset) >> PAGE_SHIFT; /* * Skip pages already mapped, and stop scanning in that * direction. When the scan terminates in both directions * we are done. */ if (pmap_prefault_ok(pmap, addr) == 0) { if (i & 1) noneg = 1; else nopos = 1; if (noneg && nopos) break; continue; } /* * Shortcut the read-only mapping case using the far more * efficient vm_page_lookup_sbusy_try() function. This * allows us to acquire the page soft-busied only which * is especially nice for concurrent execs of the same * program. * * The lookup function also validates page suitability * (all valid bits set, and not fictitious). * * If the page is in PQ_CACHE we have to fall-through * and hard-busy it so we can move it out of PQ_CACHE. */ if ((prot & VM_PROT_WRITE) == 0) { m = vm_page_lookup_sbusy_try(object, pindex, 0, PAGE_SIZE); if (m == NULL) break; if ((m->queue - m->pc) != PQ_CACHE) { pmap_enter(pmap, addr, m, prot, 0, entry); #if 0 /* REMOVE ME, a burst counts as one fault */ mycpu->gd_cnt.v_vm_faults++; if (curthread->td_lwp) ++curthread->td_lwp->lwp_ru.ru_minflt; #endif vm_page_sbusy_drop(m); continue; } vm_page_sbusy_drop(m); } /* * Fallback to normal vm_page lookup code. This code * hard-busies the page. Not only that, but the page * can remain in that state for a significant period * time due to pmap_enter()'s overhead. */ m = vm_page_lookup_busy_try(object, pindex, TRUE, &error); if (m == NULL || error) break; /* * Stop if the page cannot be trivially entered into the * pmap. */ if (((m->valid & VM_PAGE_BITS_ALL) != VM_PAGE_BITS_ALL) || (m->flags & PG_FICTITIOUS) || ((m->flags & PG_SWAPPED) && (prot & VM_PROT_WRITE) && (fault_flags & VM_FAULT_DIRTY))) { vm_page_wakeup(m); break; } /* * Enter the page into the pmap. The object might be held * shared so we can't do any (serious) modifying operation * on it. */ if ((m->queue - m->pc) == PQ_CACHE) vm_page_deactivate(m); if (prot & VM_PROT_WRITE) { vm_object_set_writeable_dirty(m->object); vm_set_nosync(m, entry); if (fault_flags & VM_FAULT_DIRTY) { vm_page_dirty(m); /* can't happeen due to conditional above */ /* swap_pager_unswapped(m); */ } } pmap_enter(pmap, addr, m, prot, 0, entry); #if 0 /* REMOVE ME, a burst counts as one fault */ mycpu->gd_cnt.v_vm_faults++; if (curthread->td_lwp) ++curthread->td_lwp->lwp_ru.ru_minflt; #endif vm_page_wakeup(m); } }
