Index: head/sys/dev/amd_ecc_inject/ecc_inject.c =================================================================== --- head/sys/dev/amd_ecc_inject/ecc_inject.c (revision 338029) +++ head/sys/dev/amd_ecc_inject/ecc_inject.c (revision 338030) @@ -1,243 +1,243 @@ /*- * Copyright (c) 2017 Andriy Gapon * 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. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * See BKDG for AMD Family 15h Models 00h-0Fh Processors * (publication 42301 Rev 3.08 - March 12, 2012): * - 2.13.3.1 DRAM Error Injection * - D18F3xB8 NB Array Address * - D18F3xBC NB Array Data Port * - D18F3xBC_x8 DRAM ECC */ #define NB_MCA_CFG 0x44 #define DRAM_ECC_EN (1 << 22) #define NB_MCA_EXTCFG 0x180 #define ECC_SYMB_SZ (1 << 25) #define NB_ARRAY_ADDR 0xb8 #define DRAM_ECC_SEL (0x8 << 28) #define QUADRANT_SHIFT 1 #define QUADRANT_MASK 0x3 #define NB_ARRAY_PORT 0xbc #define INJ_WORD_SHIFT 20 #define INJ_WORD_MASK 0x1ff #define DRAM_ERR_EN (1 << 18) #define DRAM_WR_REQ (1 << 17) #define DRAM_RD_REQ (1 << 16) #define INJ_VECTOR_MASK 0xffff static void ecc_ei_inject(int); static device_t nbdev; static int delay_ms = 0; static int quadrant = 0; /* 0 - 3 */ static int word_mask = 0x001; /* 9 bits: 8 + 1 for ECC */ static int bit_mask = 0x0001; /* 16 bits */ static int sysctl_int_with_max(SYSCTL_HANDLER_ARGS) { u_int value; int error; value = *(u_int *)arg1; error = sysctl_handle_int(oidp, &value, 0, req); if (error || req->newptr == NULL) return (error); if (value > arg2) return (EINVAL); *(u_int *)arg1 = value; return (0); } static int sysctl_nonzero_int_with_max(SYSCTL_HANDLER_ARGS) { u_int value; int error; value = *(u_int *)arg1; error = sysctl_int_with_max(oidp, &value, arg2, req); if (error || req->newptr == NULL) return (error); if (value == 0) return (EINVAL); *(u_int *)arg1 = value; return (0); } static int sysctl_proc_inject(SYSCTL_HANDLER_ARGS) { int error; int i; i = 0; error = sysctl_handle_int(oidp, &i, 0, req); if (error) return (error); if (i != 0) ecc_ei_inject(i); return (0); } static SYSCTL_NODE(_hw, OID_AUTO, error_injection, CTLFLAG_RD, NULL, "Hardware error injection"); static SYSCTL_NODE(_hw_error_injection, OID_AUTO, dram_ecc, CTLFLAG_RD, NULL, "DRAM ECC error injection"); SYSCTL_UINT(_hw_error_injection_dram_ecc, OID_AUTO, delay, CTLTYPE_UINT | CTLFLAG_RW, &delay_ms, 0, "Delay in milliseconds between error injections"); SYSCTL_PROC(_hw_error_injection_dram_ecc, OID_AUTO, quadrant, CTLTYPE_UINT | CTLFLAG_RW, &quadrant, QUADRANT_MASK, sysctl_int_with_max, "IU", "Index of 16-byte quadrant within 64-byte line where errors " "should be injected"); SYSCTL_PROC(_hw_error_injection_dram_ecc, OID_AUTO, word_mask, CTLTYPE_UINT | CTLFLAG_RW, &word_mask, INJ_WORD_MASK, sysctl_nonzero_int_with_max, "IU", "9-bit mask of words where errors should be injected (8 data + 1 ECC)"); SYSCTL_PROC(_hw_error_injection_dram_ecc, OID_AUTO, bit_mask, CTLTYPE_UINT | CTLFLAG_RW, &bit_mask, INJ_VECTOR_MASK, sysctl_nonzero_int_with_max, "IU", "16-bit mask of bits within each selected word where errors " "should be injected"); SYSCTL_PROC(_hw_error_injection_dram_ecc, OID_AUTO, inject, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, 0, sysctl_proc_inject, "I", "Inject a number of errors according to configured parameters"); static void ecc_ei_inject_one(void *arg, size_t size) { volatile uint64_t *memory = arg; uint32_t val; int i; val = DRAM_ECC_SEL | (quadrant << QUADRANT_SHIFT); pci_write_config(nbdev, NB_ARRAY_ADDR, val, 4); val = (word_mask << INJ_WORD_SHIFT) | DRAM_WR_REQ | bit_mask; pci_write_config(nbdev, NB_ARRAY_PORT, val, 4); for (i = 0; i < size / sizeof(uint64_t); i++) { memory[i] = 0; val = pci_read_config(nbdev, NB_ARRAY_PORT, 4); if ((val & DRAM_WR_REQ) == 0) break; } for (i = 0; i < size / sizeof(uint64_t); i++) memory[0] = memory[i]; } static void ecc_ei_inject(int count) { vm_offset_t memory; int injected; KASSERT((quadrant & ~QUADRANT_MASK) == 0, ("quadrant value is outside of range: %u", quadrant)); KASSERT(word_mask != 0 && (word_mask & ~INJ_WORD_MASK) == 0, ("word mask value is outside of range: 0x%x", word_mask)); KASSERT(bit_mask != 0 && (bit_mask & ~INJ_VECTOR_MASK) == 0, ("bit mask value is outside of range: 0x%x", bit_mask)); - memory = kmem_alloc_attr(kernel_arena, PAGE_SIZE, M_WAITOK, 0, ~0, + memory = kmem_alloc_attr(PAGE_SIZE, M_WAITOK, 0, ~0, VM_MEMATTR_UNCACHEABLE); for (injected = 0; injected < count; injected++) { ecc_ei_inject_one((void*)memory, PAGE_SIZE); if (delay_ms != 0 && injected != count - 1) pause_sbt("ecc_ei_inject", delay_ms * SBT_1MS, 0, 0); } kmem_free(kernel_arena, memory, PAGE_SIZE); } static int ecc_ei_load(void) { uint32_t val; if (cpu_vendor_id != CPU_VENDOR_AMD || CPUID_TO_FAMILY(cpu_id) < 0x10) { printf("DRAM ECC error injection is not supported\n"); return (ENXIO); } nbdev = pci_find_bsf(0, 24, 3); if (nbdev == NULL) { printf("Couldn't find NB PCI device\n"); return (ENXIO); } val = pci_read_config(nbdev, NB_MCA_CFG, 4); if ((val & DRAM_ECC_EN) == 0) { printf("DRAM ECC is not supported or disabled\n"); return (ENXIO); } printf("DRAM ECC error injection support loaded\n"); return (0); } static int tsc_modevent(module_t mod __unused, int type, void *data __unused) { int error; error = 0; switch (type) { case MOD_LOAD: error = ecc_ei_load(); break; case MOD_UNLOAD: case MOD_SHUTDOWN: break; default: return (EOPNOTSUPP); } return (0); } DEV_MODULE(tsc, tsc_modevent, NULL); Index: head/sys/dev/drm/drm_scatter.c =================================================================== --- head/sys/dev/drm/drm_scatter.c (revision 338029) +++ head/sys/dev/drm/drm_scatter.c (revision 338030) @@ -1,129 +1,129 @@ /*- * Copyright (c) 2009 Robert C. Noland III * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ #include __FBSDID("$FreeBSD$"); /** @file drm_scatter.c * Allocation of memory for scatter-gather mappings by the graphics chip. * The memory allocated here is then made into an aperture in the card * by mapping the pages into the GART. */ #include "dev/drm/drmP.h" int drm_sg_alloc(struct drm_device *dev, struct drm_scatter_gather *request) { struct drm_sg_mem *entry; vm_size_t size; vm_pindex_t pindex; if (dev->sg) return EINVAL; DRM_DEBUG("request size=%ld\n", request->size); entry = malloc(sizeof(*entry), DRM_MEM_DRIVER, M_WAITOK | M_ZERO); size = round_page(request->size); entry->pages = atop(size); entry->busaddr = malloc(entry->pages * sizeof(*entry->busaddr), DRM_MEM_SGLISTS, M_WAITOK | M_ZERO); - entry->vaddr = kmem_alloc_attr(kernel_arena, size, M_WAITOK | M_ZERO, - 0, BUS_SPACE_MAXADDR_32BIT, VM_MEMATTR_WRITE_COMBINING); + entry->vaddr = kmem_alloc_attr(size, M_WAITOK | M_ZERO, 0, + BUS_SPACE_MAXADDR_32BIT, VM_MEMATTR_WRITE_COMBINING); if (entry->vaddr == 0) { drm_sg_cleanup(entry); return (ENOMEM); } for(pindex = 0; pindex < entry->pages; pindex++) { entry->busaddr[pindex] = vtophys(entry->vaddr + IDX_TO_OFF(pindex)); } DRM_LOCK(); if (dev->sg) { DRM_UNLOCK(); drm_sg_cleanup(entry); return (EINVAL); } dev->sg = entry; DRM_UNLOCK(); request->handle = entry->vaddr; DRM_DEBUG("allocated %ju pages @ 0x%08zx, contents=%08lx\n", entry->pages, entry->vaddr, *(unsigned long *)entry->vaddr); return (0); } int drm_sg_alloc_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_scatter_gather *request = data; DRM_DEBUG("\n"); return (drm_sg_alloc(dev, request)); } void drm_sg_cleanup(struct drm_sg_mem *entry) { if (entry == NULL) return; if (entry->vaddr != 0) kmem_free(kernel_arena, entry->vaddr, IDX_TO_OFF(entry->pages)); free(entry->busaddr, DRM_MEM_SGLISTS); free(entry, DRM_MEM_DRIVER); return; } int drm_sg_free(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_scatter_gather *request = data; struct drm_sg_mem *entry; DRM_LOCK(); entry = dev->sg; dev->sg = NULL; DRM_UNLOCK(); if (!entry || entry->vaddr != request->handle) return (EINVAL); DRM_DEBUG("free 0x%zx\n", entry->vaddr); drm_sg_cleanup(entry); return (0); } Index: head/sys/dev/drm2/drm_scatter.c =================================================================== --- head/sys/dev/drm2/drm_scatter.c (revision 338029) +++ head/sys/dev/drm2/drm_scatter.c (revision 338030) @@ -1,136 +1,136 @@ /*- * Copyright (c) 2009 Robert C. Noland III * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ #include __FBSDID("$FreeBSD$"); /** @file drm_scatter.c * Allocation of memory for scatter-gather mappings by the graphics chip. * The memory allocated here is then made into an aperture in the card * by mapping the pages into the GART. */ #include #define DEBUG_SCATTER 0 static inline vm_offset_t drm_vmalloc_dma(vm_size_t size) { - return kmem_alloc_attr(kernel_arena, size, M_NOWAIT | M_ZERO, - 0, BUS_SPACE_MAXADDR_32BIT, VM_MEMATTR_WRITE_COMBINING); + return kmem_alloc_attr(size, M_NOWAIT | M_ZERO, 0, + BUS_SPACE_MAXADDR_32BIT, VM_MEMATTR_WRITE_COMBINING); } void drm_sg_cleanup(struct drm_sg_mem * entry) { if (entry == NULL) return; if (entry->vaddr != 0) kmem_free(kernel_arena, entry->vaddr, IDX_TO_OFF(entry->pages)); free(entry->busaddr, DRM_MEM_SGLISTS); free(entry, DRM_MEM_DRIVER); } int drm_sg_alloc(struct drm_device *dev, struct drm_scatter_gather * request) { struct drm_sg_mem *entry; vm_size_t size; vm_pindex_t pindex; DRM_DEBUG("\n"); if (!drm_core_check_feature(dev, DRIVER_SG)) return -EINVAL; if (dev->sg) return -EINVAL; entry = malloc(sizeof(*entry), DRM_MEM_DRIVER, M_NOWAIT | M_ZERO); if (!entry) return -ENOMEM; DRM_DEBUG("request size=%ld\n", request->size); size = round_page(request->size); entry->pages = atop(size); entry->busaddr = malloc(entry->pages * sizeof(*entry->busaddr), DRM_MEM_SGLISTS, M_NOWAIT | M_ZERO); if (!entry->busaddr) { free(entry, DRM_MEM_DRIVER); return -ENOMEM; } entry->vaddr = drm_vmalloc_dma(size); if (entry->vaddr == 0) { free(entry->busaddr, DRM_MEM_DRIVER); free(entry, DRM_MEM_DRIVER); return -ENOMEM; } for (pindex = 0; pindex < entry->pages; pindex++) { entry->busaddr[pindex] = vtophys(entry->vaddr + IDX_TO_OFF(pindex)); } request->handle = entry->vaddr; dev->sg = entry; DRM_DEBUG("allocated %ju pages @ 0x%08zx, contents=%08lx\n", entry->pages, entry->vaddr, *(unsigned long *)entry->vaddr); return 0; } int drm_sg_alloc_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_scatter_gather *request = data; return drm_sg_alloc(dev, request); } int drm_sg_free(struct drm_device *dev, void *data, struct drm_file *file_priv) { struct drm_scatter_gather *request = data; struct drm_sg_mem *entry; if (!drm_core_check_feature(dev, DRIVER_SG)) return -EINVAL; entry = dev->sg; dev->sg = NULL; if (!entry || entry->vaddr != request->handle) return -EINVAL; DRM_DEBUG("free 0x%zx\n", entry->vaddr); drm_sg_cleanup(entry); return 0; } Index: head/sys/vm/vm_extern.h =================================================================== --- head/sys/vm/vm_extern.h (revision 338029) +++ head/sys/vm/vm_extern.h (revision 338030) @@ -1,131 +1,131 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1992, 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. 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_extern.h 8.2 (Berkeley) 1/12/94 * $FreeBSD$ */ #ifndef _VM_EXTERN_H_ #define _VM_EXTERN_H_ struct pmap; struct proc; struct vmspace; struct vnode; struct vmem; #ifdef _KERNEL struct cdev; struct cdevsw; /* These operate on kernel virtual addresses only. */ vm_offset_t kva_alloc(vm_size_t); void kva_free(vm_offset_t, vm_size_t); /* These operate on pageable virtual addresses. */ vm_offset_t kmap_alloc_wait(vm_map_t, vm_size_t); void kmap_free_wakeup(vm_map_t, vm_offset_t, vm_size_t); /* These operate on virtual addresses backed by memory. */ -vm_offset_t kmem_alloc_attr(struct vmem *, vm_size_t size, int flags, +vm_offset_t kmem_alloc_attr(vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, vm_memattr_t memattr); vm_offset_t kmem_alloc_attr_domain(int domain, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, vm_memattr_t memattr); vm_offset_t kmem_alloc_contig(struct vmem *, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr); vm_offset_t kmem_alloc_contig_domain(int domain, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr); vm_offset_t kmem_malloc(struct vmem *, vm_size_t size, int flags); vm_offset_t kmem_malloc_domain(int domain, vm_size_t size, int flags); void kmem_free(struct vmem *, vm_offset_t, vm_size_t); /* This provides memory for previously allocated address space. */ int kmem_back(vm_object_t, vm_offset_t, vm_size_t, int); int kmem_back_domain(int, vm_object_t, vm_offset_t, vm_size_t, int); void kmem_unback(vm_object_t, vm_offset_t, vm_size_t); /* Bootstrapping. */ void kmem_bootstrap_free(vm_offset_t, vm_size_t); vm_map_t kmem_suballoc(vm_map_t, vm_offset_t *, vm_offset_t *, vm_size_t, boolean_t); void kmem_init(vm_offset_t, vm_offset_t); void kmem_init_zero_region(void); void kmeminit(void); int kernacc(void *, int, int); int useracc(void *, int, int); int vm_fault(vm_map_t, vm_offset_t, vm_prot_t, int); void vm_fault_copy_entry(vm_map_t, vm_map_t, vm_map_entry_t, vm_map_entry_t, vm_ooffset_t *); int vm_fault_disable_pagefaults(void); void vm_fault_enable_pagefaults(int save); int vm_fault_hold(vm_map_t map, vm_offset_t vaddr, vm_prot_t fault_type, int fault_flags, vm_page_t *m_hold); int vm_fault_quick_hold_pages(vm_map_t map, vm_offset_t addr, vm_size_t len, vm_prot_t prot, vm_page_t *ma, int max_count); int vm_forkproc(struct thread *, struct proc *, struct thread *, struct vmspace *, int); void vm_waitproc(struct proc *); int vm_mmap(vm_map_t, vm_offset_t *, vm_size_t, vm_prot_t, vm_prot_t, int, objtype_t, void *, vm_ooffset_t); int vm_mmap_object(vm_map_t, vm_offset_t *, vm_size_t, vm_prot_t, vm_prot_t, int, vm_object_t, vm_ooffset_t, boolean_t, struct thread *); int vm_mmap_to_errno(int rv); int vm_mmap_cdev(struct thread *, vm_size_t, vm_prot_t, vm_prot_t *, int *, struct cdev *, struct cdevsw *, vm_ooffset_t *, vm_object_t *); int vm_mmap_vnode(struct thread *, vm_size_t, vm_prot_t, vm_prot_t *, int *, struct vnode *, vm_ooffset_t *, vm_object_t *, boolean_t *); void vm_set_page_size(void); void vm_sync_icache(vm_map_t, vm_offset_t, vm_size_t); typedef int (*pmap_pinit_t)(struct pmap *pmap); struct vmspace *vmspace_alloc(vm_offset_t, vm_offset_t, pmap_pinit_t); struct vmspace *vmspace_fork(struct vmspace *, vm_ooffset_t *); int vmspace_exec(struct proc *, vm_offset_t, vm_offset_t); int vmspace_unshare(struct proc *); void vmspace_exit(struct thread *); struct vmspace *vmspace_acquire_ref(struct proc *); void vmspace_free(struct vmspace *); void vmspace_exitfree(struct proc *); void vmspace_switch_aio(struct vmspace *); void vnode_pager_setsize(struct vnode *, vm_ooffset_t); int vslock(void *, size_t); void vsunlock(void *, size_t); struct sf_buf *vm_imgact_map_page(vm_object_t object, vm_ooffset_t offset); void vm_imgact_unmap_page(struct sf_buf *sf); void vm_thread_dispose(struct thread *td); int vm_thread_new(struct thread *td, int pages); u_int vm_active_count(void); u_int vm_inactive_count(void); u_int vm_laundry_count(void); u_int vm_wait_count(void); #endif /* _KERNEL */ #endif /* !_VM_EXTERN_H_ */ Index: head/sys/vm/vm_init.c =================================================================== --- head/sys/vm/vm_init.c (revision 338029) +++ head/sys/vm/vm_init.c (revision 338030) @@ -1,352 +1,351 @@ /*- * SPDX-License-Identifier: (BSD-3-Clause AND MIT-CMU) * * Copyright (c) 1991, 1993 * The 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. * 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. * * from: @(#)vm_init.c 8.1 (Berkeley) 6/11/93 * * * 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. */ /* * Initialize the Virtual Memory subsystem. */ #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 extern void uma_startup1(void); extern void uma_startup2(void); extern void vm_radix_reserve_kva(void); #if VM_NRESERVLEVEL > 0 #define KVA_QUANTUM (1 << (VM_LEVEL_0_ORDER + PAGE_SHIFT)) #else /* On non-superpage architectures want large import sizes. */ #define KVA_QUANTUM (PAGE_SIZE * 1024) #endif long physmem; /* * System initialization */ static void vm_mem_init(void *); SYSINIT(vm_mem, SI_SUB_VM, SI_ORDER_FIRST, vm_mem_init, NULL); /* * Import kva into the kernel arena. */ static int kva_import(void *unused, vmem_size_t size, int flags, vmem_addr_t *addrp) { vm_offset_t addr; int result; KASSERT((size % KVA_QUANTUM) == 0, ("kva_import: Size %jd is not a multiple of %d", (intmax_t)size, (int)KVA_QUANTUM)); addr = vm_map_min(kernel_map); result = vm_map_find(kernel_map, NULL, 0, &addr, size, 0, VMFS_SUPER_SPACE, VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT); if (result != KERN_SUCCESS) return (ENOMEM); *addrp = addr; return (0); } #if VM_NRESERVLEVEL > 0 /* * Import a superpage from the normal kernel arena into the special * arena for allocations with different permissions. */ static int kernel_rwx_alloc(void *arena, vmem_size_t size, int flags, vmem_addr_t *addrp) { KASSERT((size % KVA_QUANTUM) == 0, ("kernel_rwx_alloc: Size %jd is not a multiple of %d", (intmax_t)size, (int)KVA_QUANTUM)); return (vmem_xalloc(arena, size, KVA_QUANTUM, 0, 0, VMEM_ADDR_MIN, VMEM_ADDR_MAX, flags, addrp)); } #endif /* * vm_init initializes the virtual memory system. * This is done only by the first cpu up. * * The start and end address of physical memory is passed in. */ /* ARGSUSED*/ static void vm_mem_init(dummy) void *dummy; { int domain; /* * Initializes resident memory structures. From here on, all physical * memory is accounted for, and we use only virtual addresses. */ vm_set_page_size(); virtual_avail = vm_page_startup(virtual_avail); #ifdef UMA_MD_SMALL_ALLOC /* Announce page availability to UMA. */ uma_startup1(); #endif /* * Initialize other VM packages */ vmem_startup(); vm_object_init(); vm_map_startup(); kmem_init(virtual_avail, virtual_end); /* * Initialize the kernel_arena. This can grow on demand. */ vmem_init(kernel_arena, "kernel arena", 0, 0, PAGE_SIZE, 0, 0); vmem_set_import(kernel_arena, kva_import, NULL, NULL, KVA_QUANTUM); #if VM_NRESERVLEVEL > 0 /* * In an architecture with superpages, maintain a separate arena * for allocations with permissions that differ from the "standard" * read/write permissions used for memory in the kernel_arena. */ kernel_rwx_arena = vmem_create("kernel rwx arena", 0, 0, PAGE_SIZE, 0, M_WAITOK); vmem_set_import(kernel_rwx_arena, kernel_rwx_alloc, (vmem_release_t *)vmem_xfree, kernel_arena, KVA_QUANTUM); #endif for (domain = 0; domain < vm_ndomains; domain++) { vm_dom[domain].vmd_kernel_arena = vmem_create( "kernel arena domain", 0, 0, PAGE_SIZE, 0, M_WAITOK); vmem_set_import(vm_dom[domain].vmd_kernel_arena, (vmem_import_t *)vmem_alloc, NULL, kernel_arena, KVA_QUANTUM); #if VM_NRESERVLEVEL > 0 vm_dom[domain].vmd_kernel_rwx_arena = vmem_create( "kernel rwx arena domain", 0, 0, PAGE_SIZE, 0, M_WAITOK); vmem_set_import(vm_dom[domain].vmd_kernel_rwx_arena, kernel_rwx_alloc, (vmem_release_t *)vmem_xfree, vm_dom[domain].vmd_kernel_arena, KVA_QUANTUM); #endif } #ifndef UMA_MD_SMALL_ALLOC /* Set up radix zone to use noobj_alloc. */ vm_radix_reserve_kva(); #endif /* Announce full page availability to UMA. */ uma_startup2(); kmem_init_zero_region(); pmap_init(); vm_pager_init(); } void vm_ksubmap_init(struct kva_md_info *kmi) { vm_offset_t firstaddr; caddr_t v; vm_size_t size = 0; long physmem_est; vm_offset_t minaddr; vm_offset_t maxaddr; /* * Allocate space for system data structures. * The first available kernel virtual address is in "v". * As pages of kernel virtual memory are allocated, "v" is incremented. * As pages of memory are allocated and cleared, * "firstaddr" is incremented. */ /* * Make two passes. The first pass calculates how much memory is * needed and allocates it. The second pass assigns virtual * addresses to the various data structures. */ firstaddr = 0; again: v = (caddr_t)firstaddr; /* * Discount the physical memory larger than the size of kernel_map * to avoid eating up all of KVA space. */ physmem_est = lmin(physmem, btoc(kernel_map->max_offset - kernel_map->min_offset)); v = kern_vfs_bio_buffer_alloc(v, physmem_est); /* * End of first pass, size has been calculated so allocate memory */ if (firstaddr == 0) { size = (vm_size_t)v; #ifdef VM_FREELIST_DMA32 /* * Try to protect 32-bit DMAable memory from the largest * early alloc of wired mem. */ - firstaddr = kmem_alloc_attr(kernel_arena, size, - M_ZERO | M_NOWAIT, (vm_paddr_t)1 << 32, - ~(vm_paddr_t)0, VM_MEMATTR_DEFAULT); + firstaddr = kmem_alloc_attr(size, M_ZERO | M_NOWAIT, + (vm_paddr_t)1 << 32, ~(vm_paddr_t)0, VM_MEMATTR_DEFAULT); if (firstaddr == 0) #endif firstaddr = kmem_malloc(kernel_arena, size, M_ZERO | M_WAITOK); if (firstaddr == 0) panic("startup: no room for tables"); goto again; } /* * End of second pass, addresses have been assigned */ if ((vm_size_t)((char *)v - firstaddr) != size) panic("startup: table size inconsistency"); /* * Allocate the clean map to hold all of the paging and I/O virtual * memory. */ size = (long)nbuf * BKVASIZE + (long)nswbuf * MAXPHYS + (long)bio_transient_maxcnt * MAXPHYS; kmi->clean_sva = firstaddr = kva_alloc(size); kmi->clean_eva = firstaddr + size; /* * Allocate the buffer arena. * * Enable the quantum cache if we have more than 4 cpus. This * avoids lock contention at the expense of some fragmentation. */ size = (long)nbuf * BKVASIZE; kmi->buffer_sva = firstaddr; kmi->buffer_eva = kmi->buffer_sva + size; vmem_init(buffer_arena, "buffer arena", kmi->buffer_sva, size, PAGE_SIZE, (mp_ncpus > 4) ? BKVASIZE * 8 : 0, 0); firstaddr += size; /* * Now swap kva. */ swapbkva = firstaddr; size = (long)nswbuf * MAXPHYS; firstaddr += size; /* * And optionally transient bio space. */ if (bio_transient_maxcnt != 0) { size = (long)bio_transient_maxcnt * MAXPHYS; vmem_init(transient_arena, "transient arena", firstaddr, size, PAGE_SIZE, 0, 0); firstaddr += size; } if (firstaddr != kmi->clean_eva) panic("Clean map calculation incorrect"); /* * Allocate the pageable submaps. We may cache an exec map entry per * CPU, so we therefore need to reserve space for at least ncpu+1 * entries to avoid deadlock. The exec map is also used by some image * activators, so we leave a fixed number of pages for their use. */ #ifdef __LP64__ exec_map_entries = 8 * mp_ncpus; #else exec_map_entries = 2 * mp_ncpus + 4; #endif exec_map_entry_size = round_page(PATH_MAX + ARG_MAX); exec_map = kmem_suballoc(kernel_map, &minaddr, &maxaddr, exec_map_entries * exec_map_entry_size + 64 * PAGE_SIZE, FALSE); pipe_map = kmem_suballoc(kernel_map, &minaddr, &maxaddr, maxpipekva, FALSE); } Index: head/sys/vm/vm_kern.c =================================================================== --- head/sys/vm/vm_kern.c (revision 338029) +++ head/sys/vm/vm_kern.c (revision 338030) @@ -1,746 +1,743 @@ /*- * SPDX-License-Identifier: (BSD-3-Clause AND MIT-CMU) * * Copyright (c) 1991, 1993 * The 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. * 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. * * from: @(#)vm_kern.c 8.3 (Berkeley) 1/12/94 * * * 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. */ /* * Kernel memory management. */ #include __FBSDID("$FreeBSD$"); #include "opt_vm.h" #include #include #include /* for ticks and hz */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include vm_map_t kernel_map; vm_map_t exec_map; vm_map_t pipe_map; const void *zero_region; CTASSERT((ZERO_REGION_SIZE & PAGE_MASK) == 0); /* NB: Used by kernel debuggers. */ const u_long vm_maxuser_address = VM_MAXUSER_ADDRESS; u_int exec_map_entry_size; u_int exec_map_entries; SYSCTL_ULONG(_vm, OID_AUTO, min_kernel_address, CTLFLAG_RD, SYSCTL_NULL_ULONG_PTR, VM_MIN_KERNEL_ADDRESS, "Min kernel address"); SYSCTL_ULONG(_vm, OID_AUTO, max_kernel_address, CTLFLAG_RD, #if defined(__arm__) || defined(__sparc64__) &vm_max_kernel_address, 0, #else SYSCTL_NULL_ULONG_PTR, VM_MAX_KERNEL_ADDRESS, #endif "Max kernel address"); /* * kva_alloc: * * Allocate a virtual address range with no underlying object and * no initial mapping to physical memory. Any mapping from this * range to physical memory must be explicitly created prior to * its use, typically with pmap_qenter(). Any attempt to create * a mapping on demand through vm_fault() will result in a panic. */ vm_offset_t kva_alloc(vm_size_t size) { vm_offset_t addr; size = round_page(size); if (vmem_alloc(kernel_arena, size, M_BESTFIT | M_NOWAIT, &addr)) return (0); return (addr); } /* * kva_free: * * Release a region of kernel virtual memory allocated * with kva_alloc, and return the physical pages * associated with that region. * * This routine may not block on kernel maps. */ void kva_free(vm_offset_t addr, vm_size_t size) { size = round_page(size); vmem_free(kernel_arena, addr, size); } /* * Allocates a region from the kernel address map and physical pages * within the specified address range to the kernel object. Creates a * wired mapping from this region to these pages, and returns the * region's starting virtual address. The allocated pages are not * necessarily physically contiguous. If M_ZERO is specified through the * given flags, then the pages are zeroed before they are mapped. */ vm_offset_t kmem_alloc_attr_domain(int domain, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, vm_memattr_t memattr) { vmem_t *vmem; vm_object_t object = kernel_object; vm_offset_t addr, i, offset; vm_page_t m; int pflags, tries; size = round_page(size); vmem = vm_dom[domain].vmd_kernel_arena; if (vmem_alloc(vmem, size, M_BESTFIT | flags, &addr)) return (0); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; pflags &= ~(VM_ALLOC_NOWAIT | VM_ALLOC_WAITOK | VM_ALLOC_WAITFAIL); pflags |= VM_ALLOC_NOWAIT; VM_OBJECT_WLOCK(object); for (i = 0; i < size; i += PAGE_SIZE) { tries = 0; retry: m = vm_page_alloc_contig_domain(object, atop(offset + i), domain, pflags, 1, low, high, PAGE_SIZE, 0, memattr); if (m == NULL) { VM_OBJECT_WUNLOCK(object); if (tries < ((flags & M_NOWAIT) != 0 ? 1 : 3)) { if (!vm_page_reclaim_contig_domain(domain, pflags, 1, low, high, PAGE_SIZE, 0) && (flags & M_WAITOK) != 0) vm_wait_domain(domain); VM_OBJECT_WLOCK(object); tries++; goto retry; } kmem_unback(object, addr, i); vmem_free(vmem, addr, size); return (0); } KASSERT(vm_phys_domain(m) == domain, ("kmem_alloc_attr_domain: Domain mismatch %d != %d", vm_phys_domain(m), domain)); if ((flags & M_ZERO) && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, addr + i, m, VM_PROT_RW, VM_PROT_RW | PMAP_ENTER_WIRED, 0); } VM_OBJECT_WUNLOCK(object); return (addr); } vm_offset_t -kmem_alloc_attr(vmem_t *vmem, vm_size_t size, int flags, vm_paddr_t low, - vm_paddr_t high, vm_memattr_t memattr) +kmem_alloc_attr(vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, + vm_memattr_t memattr) { struct vm_domainset_iter di; vm_offset_t addr; int domain; - - KASSERT(vmem == kernel_arena, - ("kmem_alloc_attr: Only kernel_arena is supported.")); vm_domainset_iter_malloc_init(&di, kernel_object, &domain, &flags); do { addr = kmem_alloc_attr_domain(domain, size, flags, low, high, memattr); if (addr != 0) break; } while (vm_domainset_iter_malloc(&di, &domain, &flags) == 0); return (addr); } /* * Allocates a region from the kernel address map and physically * contiguous pages within the specified address range to the kernel * object. Creates a wired mapping from this region to these pages, and * returns the region's starting virtual address. If M_ZERO is specified * through the given flags, then the pages are zeroed before they are * mapped. */ vm_offset_t kmem_alloc_contig_domain(int domain, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr) { vmem_t *vmem; vm_object_t object = kernel_object; vm_offset_t addr, offset, tmp; vm_page_t end_m, m; u_long npages; int pflags, tries; size = round_page(size); vmem = vm_dom[domain].vmd_kernel_arena; if (vmem_alloc(vmem, size, flags | M_BESTFIT, &addr)) return (0); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; pflags &= ~(VM_ALLOC_NOWAIT | VM_ALLOC_WAITOK | VM_ALLOC_WAITFAIL); pflags |= VM_ALLOC_NOWAIT; npages = atop(size); VM_OBJECT_WLOCK(object); tries = 0; retry: m = vm_page_alloc_contig_domain(object, atop(offset), domain, pflags, npages, low, high, alignment, boundary, memattr); if (m == NULL) { VM_OBJECT_WUNLOCK(object); if (tries < ((flags & M_NOWAIT) != 0 ? 1 : 3)) { if (!vm_page_reclaim_contig_domain(domain, pflags, npages, low, high, alignment, boundary) && (flags & M_WAITOK) != 0) vm_wait_domain(domain); VM_OBJECT_WLOCK(object); tries++; goto retry; } vmem_free(vmem, addr, size); return (0); } KASSERT(vm_phys_domain(m) == domain, ("kmem_alloc_contig_domain: Domain mismatch %d != %d", vm_phys_domain(m), domain)); end_m = m + npages; tmp = addr; for (; m < end_m; m++) { if ((flags & M_ZERO) && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, tmp, m, VM_PROT_RW, VM_PROT_RW | PMAP_ENTER_WIRED, 0); tmp += PAGE_SIZE; } VM_OBJECT_WUNLOCK(object); return (addr); } vm_offset_t kmem_alloc_contig(struct vmem *vmem, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr) { struct vm_domainset_iter di; vm_offset_t addr; int domain; KASSERT(vmem == kernel_arena, ("kmem_alloc_contig: Only kernel_arena is supported.")); vm_domainset_iter_malloc_init(&di, kernel_object, &domain, &flags); do { addr = kmem_alloc_contig_domain(domain, size, flags, low, high, alignment, boundary, memattr); if (addr != 0) break; } while (vm_domainset_iter_malloc(&di, &domain, &flags) == 0); return (addr); } /* * kmem_suballoc: * * Allocates a map to manage a subrange * of the kernel virtual address space. * * Arguments are as follows: * * parent Map to take range from * min, max Returned endpoints of map * size Size of range to find * superpage_align Request that min is superpage aligned */ vm_map_t kmem_suballoc(vm_map_t parent, vm_offset_t *min, vm_offset_t *max, vm_size_t size, boolean_t superpage_align) { int ret; vm_map_t result; size = round_page(size); *min = vm_map_min(parent); ret = vm_map_find(parent, NULL, 0, min, size, 0, superpage_align ? VMFS_SUPER_SPACE : VMFS_ANY_SPACE, VM_PROT_ALL, VM_PROT_ALL, MAP_ACC_NO_CHARGE); if (ret != KERN_SUCCESS) panic("kmem_suballoc: bad status return of %d", ret); *max = *min + size; result = vm_map_create(vm_map_pmap(parent), *min, *max); if (result == NULL) panic("kmem_suballoc: cannot create submap"); if (vm_map_submap(parent, *min, *max, result) != KERN_SUCCESS) panic("kmem_suballoc: unable to change range to submap"); return (result); } /* * kmem_malloc: * * Allocate wired-down pages in the kernel's address space. */ vm_offset_t kmem_malloc_domain(int domain, vm_size_t size, int flags) { vmem_t *arena; vm_offset_t addr; int rv; #if VM_NRESERVLEVEL > 0 if (__predict_true((flags & M_EXEC) == 0)) arena = vm_dom[domain].vmd_kernel_arena; else arena = vm_dom[domain].vmd_kernel_rwx_arena; #else arena = vm_dom[domain].vmd_kernel_arena; #endif size = round_page(size); if (vmem_alloc(arena, size, flags | M_BESTFIT, &addr)) return (0); rv = kmem_back_domain(domain, kernel_object, addr, size, flags); if (rv != KERN_SUCCESS) { vmem_free(arena, addr, size); return (0); } return (addr); } vm_offset_t kmem_malloc(struct vmem *vmem __unused, vm_size_t size, int flags) { struct vm_domainset_iter di; vm_offset_t addr; int domain; vm_domainset_iter_malloc_init(&di, kernel_object, &domain, &flags); do { addr = kmem_malloc_domain(domain, size, flags); if (addr != 0) break; } while (vm_domainset_iter_malloc(&di, &domain, &flags) == 0); return (addr); } /* * kmem_back: * * Allocate physical pages for the specified virtual address range. */ int kmem_back_domain(int domain, vm_object_t object, vm_offset_t addr, vm_size_t size, int flags) { vm_offset_t offset, i; vm_page_t m, mpred; vm_prot_t prot; int pflags; KASSERT(object == kernel_object, ("kmem_back_domain: only supports kernel object.")); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; pflags &= ~(VM_ALLOC_NOWAIT | VM_ALLOC_WAITOK | VM_ALLOC_WAITFAIL); if (flags & M_WAITOK) pflags |= VM_ALLOC_WAITFAIL; prot = (flags & M_EXEC) != 0 ? VM_PROT_ALL : VM_PROT_RW; i = 0; VM_OBJECT_WLOCK(object); retry: mpred = vm_radix_lookup_le(&object->rtree, atop(offset + i)); for (; i < size; i += PAGE_SIZE, mpred = m) { m = vm_page_alloc_domain_after(object, atop(offset + i), domain, pflags, mpred); /* * Ran out of space, free everything up and return. Don't need * to lock page queues here as we know that the pages we got * aren't on any queues. */ if (m == NULL) { if ((flags & M_NOWAIT) == 0) goto retry; VM_OBJECT_WUNLOCK(object); kmem_unback(object, addr, i); return (KERN_NO_SPACE); } KASSERT(vm_phys_domain(m) == domain, ("kmem_back_domain: Domain mismatch %d != %d", vm_phys_domain(m), domain)); if (flags & M_ZERO && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); KASSERT((m->oflags & VPO_UNMANAGED) != 0, ("kmem_malloc: page %p is managed", m)); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, addr + i, m, prot, prot | PMAP_ENTER_WIRED, 0); } VM_OBJECT_WUNLOCK(object); return (KERN_SUCCESS); } int kmem_back(vm_object_t object, vm_offset_t addr, vm_size_t size, int flags) { struct vm_domainset_iter di; int domain; int ret; KASSERT(object == kernel_object, ("kmem_back: only supports kernel object.")); vm_domainset_iter_malloc_init(&di, kernel_object, &domain, &flags); do { ret = kmem_back_domain(domain, object, addr, size, flags); if (ret == KERN_SUCCESS) break; } while (vm_domainset_iter_malloc(&di, &domain, &flags) == 0); return (ret); } /* * kmem_unback: * * Unmap and free the physical pages underlying the specified virtual * address range. * * A physical page must exist within the specified object at each index * that is being unmapped. */ static int _kmem_unback(vm_object_t object, vm_offset_t addr, vm_size_t size) { vm_page_t m, next; vm_offset_t end, offset; int domain; KASSERT(object == kernel_object, ("kmem_unback: only supports kernel object.")); if (size == 0) return (0); pmap_remove(kernel_pmap, addr, addr + size); offset = addr - VM_MIN_KERNEL_ADDRESS; end = offset + size; VM_OBJECT_WLOCK(object); m = vm_page_lookup(object, atop(offset)); domain = vm_phys_domain(m); for (; offset < end; offset += PAGE_SIZE, m = next) { next = vm_page_next(m); vm_page_unwire(m, PQ_NONE); vm_page_free(m); } VM_OBJECT_WUNLOCK(object); return (domain); } void kmem_unback(vm_object_t object, vm_offset_t addr, vm_size_t size) { _kmem_unback(object, addr, size); } /* * kmem_free: * * Free memory allocated with kmem_malloc. The size must match the * original allocation. */ void kmem_free(struct vmem *vmem, vm_offset_t addr, vm_size_t size) { struct vmem *arena; int domain; #if VM_NRESERVLEVEL > 0 KASSERT(vmem == kernel_arena || vmem == kernel_rwx_arena, ("kmem_free: Only kernel_arena or kernel_rwx_arena are supported.")); #else KASSERT(vmem == kernel_arena, ("kmem_free: Only kernel_arena is supported.")); #endif size = round_page(size); domain = _kmem_unback(kernel_object, addr, size); #if VM_NRESERVLEVEL > 0 if (__predict_true(vmem == kernel_arena)) arena = vm_dom[domain].vmd_kernel_arena; else arena = vm_dom[domain].vmd_kernel_rwx_arena; #else arena = vm_dom[domain].vmd_kernel_arena; #endif vmem_free(arena, addr, size); } /* * kmap_alloc_wait: * * Allocates pageable memory from a sub-map of the kernel. If the submap * has no room, the caller sleeps waiting for more memory in the submap. * * This routine may block. */ vm_offset_t kmap_alloc_wait(vm_map_t map, vm_size_t size) { vm_offset_t addr; size = round_page(size); if (!swap_reserve(size)) return (0); for (;;) { /* * To make this work for more than one map, use the map's lock * to lock out sleepers/wakers. */ vm_map_lock(map); if (vm_map_findspace(map, vm_map_min(map), size, &addr) == 0) break; /* no space now; see if we can ever get space */ if (vm_map_max(map) - vm_map_min(map) < size) { vm_map_unlock(map); swap_release(size); return (0); } map->needs_wakeup = TRUE; vm_map_unlock_and_wait(map, 0); } vm_map_insert(map, NULL, 0, addr, addr + size, VM_PROT_ALL, VM_PROT_ALL, MAP_ACC_CHARGED); vm_map_unlock(map); return (addr); } /* * kmap_free_wakeup: * * Returns memory to a submap of the kernel, and wakes up any processes * waiting for memory in that map. */ void kmap_free_wakeup(vm_map_t map, vm_offset_t addr, vm_size_t size) { vm_map_lock(map); (void) vm_map_delete(map, trunc_page(addr), round_page(addr + size)); if (map->needs_wakeup) { map->needs_wakeup = FALSE; vm_map_wakeup(map); } vm_map_unlock(map); } void kmem_init_zero_region(void) { vm_offset_t addr, i; vm_page_t m; /* * Map a single physical page of zeros to a larger virtual range. * This requires less looping in places that want large amounts of * zeros, while not using much more physical resources. */ addr = kva_alloc(ZERO_REGION_SIZE); m = vm_page_alloc(NULL, 0, VM_ALLOC_NORMAL | VM_ALLOC_NOOBJ | VM_ALLOC_WIRED | VM_ALLOC_ZERO); if ((m->flags & PG_ZERO) == 0) pmap_zero_page(m); for (i = 0; i < ZERO_REGION_SIZE; i += PAGE_SIZE) pmap_qenter(addr + i, &m, 1); pmap_protect(kernel_pmap, addr, addr + ZERO_REGION_SIZE, VM_PROT_READ); zero_region = (const void *)addr; } /* * kmem_init: * * Create the kernel map; insert a mapping covering kernel text, * data, bss, and all space allocated thus far (`boostrap' data). The * new map will thus map the range between VM_MIN_KERNEL_ADDRESS and * `start' as allocated, and the range between `start' and `end' as free. */ void kmem_init(vm_offset_t start, vm_offset_t end) { vm_map_t m; m = vm_map_create(kernel_pmap, VM_MIN_KERNEL_ADDRESS, end); m->system_map = 1; vm_map_lock(m); /* N.B.: cannot use kgdb to debug, starting with this assignment ... */ kernel_map = m; (void) vm_map_insert(m, NULL, (vm_ooffset_t) 0, #ifdef __amd64__ KERNBASE, #else VM_MIN_KERNEL_ADDRESS, #endif start, VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT); /* ... and ending with the completion of the above `insert' */ vm_map_unlock(m); } /* * kmem_bootstrap_free: * * Free pages backing preloaded data (e.g., kernel modules) to the * system. Currently only supported on platforms that create a * vm_phys segment for preloaded data. */ void kmem_bootstrap_free(vm_offset_t start, vm_size_t size) { #if defined(__i386__) || defined(__amd64__) struct vm_domain *vmd; vm_offset_t end, va; vm_paddr_t pa; vm_page_t m; end = trunc_page(start + size); start = round_page(start); for (va = start; va < end; va += PAGE_SIZE) { pa = pmap_kextract(va); m = PHYS_TO_VM_PAGE(pa); vmd = vm_pagequeue_domain(m); vm_domain_free_lock(vmd); vm_phys_free_pages(m, 0); vmd->vmd_page_count++; vm_domain_free_unlock(vmd); vm_domain_freecnt_inc(vmd, 1); vm_cnt.v_page_count++; } pmap_remove(kernel_pmap, start, end); (void)vmem_add(kernel_arena, start, end - start, M_WAITOK); #endif } #ifdef DIAGNOSTIC /* * Allow userspace to directly trigger the VM drain routine for testing * purposes. */ static int debug_vm_lowmem(SYSCTL_HANDLER_ARGS) { int error, i; i = 0; error = sysctl_handle_int(oidp, &i, 0, req); if (error) return (error); if ((i & ~(VM_LOW_KMEM | VM_LOW_PAGES)) != 0) return (EINVAL); if (i != 0) EVENTHANDLER_INVOKE(vm_lowmem, i); return (0); } SYSCTL_PROC(_debug, OID_AUTO, vm_lowmem, CTLTYPE_INT | CTLFLAG_RW, 0, 0, debug_vm_lowmem, "I", "set to trigger vm_lowmem event with given flags"); #endif