Index: head/lib/libkvm/kvm.h =================================================================== --- head/lib/libkvm/kvm.h (revision 354542) +++ head/lib/libkvm/kvm.h (revision 354543) @@ -1,132 +1,132 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 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. * * @(#)kvm.h 8.1 (Berkeley) 6/2/93 * $FreeBSD$ */ #ifndef _KVM_H_ #define _KVM_H_ #include #include #include /* * Including vm/vm.h causes namespace pollution issues. For the * most part, only things using kvm_walk_pages() need to #include it. */ #ifndef VM_H typedef u_char vm_prot_t; #endif /* Default version symbol. */ #define VRS_SYM "_version" #define VRS_KEY "VERSION" #ifndef _SIZE_T_DECLARED typedef __size_t size_t; #define _SIZE_T_DECLARED #endif #ifndef _SSIZE_T_DECLARED typedef __ssize_t ssize_t; #define _SSIZE_T_DECLARED #endif struct kvm_nlist { const char *n_name; unsigned char n_type; kvaddr_t n_value; }; typedef struct __kvm kvm_t; struct kinfo_proc; struct proc; struct kvm_swap { char ksw_devname[32]; u_int ksw_used; u_int ksw_total; int ksw_flags; u_int ksw_reserved1; u_int ksw_reserved2; }; struct kvm_page { - unsigned int version; - u_long paddr; - u_long kmap_vaddr; - u_long dmap_vaddr; - vm_prot_t prot; - u_long offset; - size_t len; - /* end of version 1 */ + u_int kp_version; + kpaddr_t kp_paddr; + kvaddr_t kp_kmap_vaddr; + kvaddr_t kp_dmap_vaddr; + vm_prot_t kp_prot; + off_t kp_offset; + size_t kp_len; + /* end of version 2 */ }; #define SWIF_DEV_PREFIX 0x0002 -#define LIBKVM_WALK_PAGES_VERSION 1 +#define LIBKVM_WALK_PAGES_VERSION 2 __BEGIN_DECLS int kvm_close(kvm_t *); int kvm_dpcpu_setcpu(kvm_t *, unsigned int); char **kvm_getargv(kvm_t *, const struct kinfo_proc *, int); int kvm_getcptime(kvm_t *, long *); char **kvm_getenvv(kvm_t *, const struct kinfo_proc *, int); char *kvm_geterr(kvm_t *); int kvm_getloadavg(kvm_t *, double [], int); int kvm_getmaxcpu(kvm_t *); int kvm_getncpus(kvm_t *); void *kvm_getpcpu(kvm_t *, int); uint64_t kvm_counter_u64_fetch(kvm_t *, u_long); struct kinfo_proc * kvm_getprocs(kvm_t *, int, int, int *); int kvm_getswapinfo(kvm_t *, struct kvm_swap *, int, int); int kvm_native(kvm_t *); int kvm_nlist(kvm_t *, struct nlist *); int kvm_nlist2(kvm_t *, struct kvm_nlist *); kvm_t *kvm_open (const char *, const char *, const char *, int, const char *); kvm_t *kvm_openfiles (const char *, const char *, const char *, int, char *); kvm_t *kvm_open2 (const char *, const char *, int, char *, int (*)(const char *, kvaddr_t *)); ssize_t kvm_read(kvm_t *, unsigned long, void *, size_t); ssize_t kvm_read_zpcpu(kvm_t *, unsigned long, void *, size_t, int); ssize_t kvm_read2(kvm_t *, kvaddr_t, void *, size_t); ssize_t kvm_write(kvm_t *, unsigned long, const void *, size_t); typedef int kvm_walk_pages_cb_t(struct kvm_page *, void *); int kvm_walk_pages(kvm_t *, kvm_walk_pages_cb_t *, void *); __END_DECLS #endif /* !_KVM_H_ */ Index: head/lib/libkvm/kvm_private.c =================================================================== --- head/lib/libkvm/kvm_private.c (revision 354542) +++ head/lib/libkvm/kvm_private.c (revision 354543) @@ -1,768 +1,768 @@ /*- * Copyright (c) 1989, 1992, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software developed by the Computer Systems * Engineering group at Lawrence Berkeley Laboratory under DARPA contract * BG 91-66 and contributed to Berkeley. * * 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. */ #include __FBSDID("$FreeBSD$"); #include #include #define _WANT_VNET #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "kvm_private.h" /* * Routines private to libkvm. */ /* from src/lib/libc/gen/nlist.c */ int __fdnlist(int, struct nlist *); /* * Report an error using printf style arguments. "program" is kd->program * on hard errors, and 0 on soft errors, so that under sun error emulation, * only hard errors are printed out (otherwise, programs like gdb will * generate tons of error messages when trying to access bogus pointers). */ void _kvm_err(kvm_t *kd, const char *program, const char *fmt, ...) { va_list ap; va_start(ap, fmt); if (program != NULL) { (void)fprintf(stderr, "%s: ", program); (void)vfprintf(stderr, fmt, ap); (void)fputc('\n', stderr); } else (void)vsnprintf(kd->errbuf, sizeof(kd->errbuf), fmt, ap); va_end(ap); } void _kvm_syserr(kvm_t *kd, const char *program, const char *fmt, ...) { va_list ap; int n; va_start(ap, fmt); if (program != NULL) { (void)fprintf(stderr, "%s: ", program); (void)vfprintf(stderr, fmt, ap); (void)fprintf(stderr, ": %s\n", strerror(errno)); } else { char *cp = kd->errbuf; (void)vsnprintf(cp, sizeof(kd->errbuf), fmt, ap); n = strlen(cp); (void)snprintf(&cp[n], sizeof(kd->errbuf) - n, ": %s", strerror(errno)); } va_end(ap); } void * _kvm_malloc(kvm_t *kd, size_t n) { void *p; if ((p = calloc(n, sizeof(char))) == NULL) _kvm_err(kd, kd->program, "can't allocate %zu bytes: %s", n, strerror(errno)); return (p); } int _kvm_probe_elf_kernel(kvm_t *kd, int class, int machine) { return (kd->nlehdr.e_ident[EI_CLASS] == class && ((machine == EM_PPC || machine == EM_PPC64) ? kd->nlehdr.e_type == ET_DYN : kd->nlehdr.e_type == ET_EXEC) && kd->nlehdr.e_machine == machine); } int _kvm_is_minidump(kvm_t *kd) { char minihdr[8]; if (kd->rawdump) return (0); if (pread(kd->pmfd, &minihdr, 8, 0) == 8 && memcmp(&minihdr, "minidump", 8) == 0) return (1); return (0); } /* * The powerpc backend has a hack to strip a leading kerneldump * header from the core before treating it as an ELF header. * * We can add that here if we can get a change to libelf to support * an initial offset into the file. Alternatively we could patch * savecore to extract cores from a regular file instead. */ int _kvm_read_core_phdrs(kvm_t *kd, size_t *phnump, GElf_Phdr **phdrp) { GElf_Ehdr ehdr; GElf_Phdr *phdr; Elf *elf; size_t i, phnum; elf = elf_begin(kd->pmfd, ELF_C_READ, NULL); if (elf == NULL) { _kvm_err(kd, kd->program, "%s", elf_errmsg(0)); return (-1); } if (elf_kind(elf) != ELF_K_ELF) { _kvm_err(kd, kd->program, "invalid core"); goto bad; } if (gelf_getclass(elf) != kd->nlehdr.e_ident[EI_CLASS]) { _kvm_err(kd, kd->program, "invalid core"); goto bad; } if (gelf_getehdr(elf, &ehdr) == NULL) { _kvm_err(kd, kd->program, "%s", elf_errmsg(0)); goto bad; } if (ehdr.e_type != ET_CORE) { _kvm_err(kd, kd->program, "invalid core"); goto bad; } if (ehdr.e_machine != kd->nlehdr.e_machine) { _kvm_err(kd, kd->program, "invalid core"); goto bad; } if (elf_getphdrnum(elf, &phnum) == -1) { _kvm_err(kd, kd->program, "%s", elf_errmsg(0)); goto bad; } phdr = calloc(phnum, sizeof(*phdr)); if (phdr == NULL) { _kvm_err(kd, kd->program, "failed to allocate phdrs"); goto bad; } for (i = 0; i < phnum; i++) { if (gelf_getphdr(elf, i, &phdr[i]) == NULL) { free(phdr); _kvm_err(kd, kd->program, "%s", elf_errmsg(0)); goto bad; } } elf_end(elf); *phnump = phnum; *phdrp = phdr; return (0); bad: elf_end(elf); return (-1); } /* * Transform v such that only bits [bit0, bitN) may be set. Generates a * bitmask covering the number of bits, then shifts so +bit0+ is the first. */ static uint64_t bitmask_range(uint64_t v, uint64_t bit0, uint64_t bitN) { if (bit0 == 0 && bitN == BITS_IN(v)) return (v); return (v & (((1ULL << (bitN - bit0)) - 1ULL) << bit0)); } /* * Returns the number of bits in a given byte array range starting at a * given base, from bit0 to bitN. bit0 may be non-zero in the case of * counting backwards from bitN. */ static uint64_t popcount_bytes(uint64_t *addr, uint32_t bit0, uint32_t bitN) { uint32_t res = bitN - bit0; uint64_t count = 0; uint32_t bound; /* Align to 64-bit boundary on the left side if needed. */ if ((bit0 % BITS_IN(*addr)) != 0) { bound = MIN(bitN, roundup2(bit0, BITS_IN(*addr))); count += __bitcount64(bitmask_range(*addr, bit0, bound)); res -= (bound - bit0); addr++; } while (res > 0) { bound = MIN(res, BITS_IN(*addr)); count += __bitcount64(bitmask_range(*addr, 0, bound)); res -= bound; addr++; } return (count); } void * _kvm_pmap_get(kvm_t *kd, u_long idx, size_t len) { uintptr_t off = idx * len; if ((off_t)off >= kd->pt_sparse_off) return (NULL); return (void *)((uintptr_t)kd->page_map + off); } void * _kvm_map_get(kvm_t *kd, u_long pa, unsigned int page_size) { off_t off; uintptr_t addr; off = _kvm_pt_find(kd, pa, page_size); if (off == -1) return NULL; addr = (uintptr_t)kd->page_map + off; if (off >= kd->pt_sparse_off) addr = (uintptr_t)kd->sparse_map + (off - kd->pt_sparse_off); return (void *)addr; } int _kvm_pt_init(kvm_t *kd, size_t map_len, off_t map_off, off_t sparse_off, int page_size, int word_size) { uint64_t *addr; uint32_t *popcount_bin; int bin_popcounts = 0; uint64_t pc_bins, res; ssize_t rd; /* * Map the bitmap specified by the arguments. */ kd->pt_map = _kvm_malloc(kd, map_len); if (kd->pt_map == NULL) { _kvm_err(kd, kd->program, "cannot allocate %zu bytes for bitmap", map_len); return (-1); } rd = pread(kd->pmfd, kd->pt_map, map_len, map_off); if (rd < 0 || rd != (ssize_t)map_len) { _kvm_err(kd, kd->program, "cannot read %zu bytes for bitmap", map_len); return (-1); } kd->pt_map_size = map_len; /* * Generate a popcount cache for every POPCOUNT_BITS in the bitmap, * so lookups only have to calculate the number of bits set between * a cache point and their bit. This reduces lookups to O(1), * without significantly increasing memory requirements. * * Round up the number of bins so that 'upper half' lookups work for * the final bin, if needed. The first popcount is 0, since no bits * precede bit 0, so add 1 for that also. Without this, extra work * would be needed to handle the first PTEs in _kvm_pt_find(). */ addr = kd->pt_map; res = map_len; pc_bins = 1 + (res * NBBY + POPCOUNT_BITS / 2) / POPCOUNT_BITS; kd->pt_popcounts = calloc(pc_bins, sizeof(uint32_t)); if (kd->pt_popcounts == NULL) { _kvm_err(kd, kd->program, "cannot allocate popcount bins"); return (-1); } for (popcount_bin = &kd->pt_popcounts[1]; res > 0; addr++, res -= sizeof(*addr)) { *popcount_bin += popcount_bytes(addr, 0, MIN(res * NBBY, BITS_IN(*addr))); if (++bin_popcounts == POPCOUNTS_IN(*addr)) { popcount_bin++; *popcount_bin = *(popcount_bin - 1); bin_popcounts = 0; } } assert(pc_bins * sizeof(*popcount_bin) == ((uintptr_t)popcount_bin - (uintptr_t)kd->pt_popcounts)); kd->pt_sparse_off = sparse_off; kd->pt_sparse_size = (uint64_t)*popcount_bin * page_size; kd->pt_page_size = page_size; kd->pt_word_size = word_size; /* * Map the sparse page array. This is useful for performing point * lookups of specific pages, e.g. for kvm_walk_pages. Generally, * this is much larger than is reasonable to read in up front, so * mmap it in instead. */ kd->sparse_map = mmap(NULL, kd->pt_sparse_size, PROT_READ, MAP_PRIVATE, kd->pmfd, kd->pt_sparse_off); if (kd->sparse_map == MAP_FAILED) { _kvm_err(kd, kd->program, "cannot map %" PRIu64 " bytes from fd %d offset %jd for sparse map: %s", kd->pt_sparse_size, kd->pmfd, (intmax_t)kd->pt_sparse_off, strerror(errno)); return (-1); } return (0); } int _kvm_pmap_init(kvm_t *kd, uint32_t pmap_size, off_t pmap_off) { ssize_t exp_len = pmap_size; kd->page_map_size = pmap_size; kd->page_map_off = pmap_off; kd->page_map = _kvm_malloc(kd, pmap_size); if (kd->page_map == NULL) { _kvm_err(kd, kd->program, "cannot allocate %u bytes " "for page map", pmap_size); return (-1); } if (pread(kd->pmfd, kd->page_map, pmap_size, pmap_off) != exp_len) { _kvm_err(kd, kd->program, "cannot read %d bytes from " "offset %jd for page map", pmap_size, (intmax_t)pmap_off); return (-1); } return (0); } /* * Find the offset for the given physical page address; returns -1 otherwise. * * A page's offset is represented by the sparse page base offset plus the * number of bits set before its bit multiplied by page size. This means * that if a page exists in the dump, it's necessary to know how many pages * in the dump precede it. Reduce this O(n) counting to O(1) by caching the * number of bits set at POPCOUNT_BITS intervals. * * Then to find the number of pages before the requested address, simply * index into the cache and count the number of bits set between that cache * bin and the page's bit. Halve the number of bytes that have to be * checked by also counting down from the next higher bin if it's closer. */ off_t _kvm_pt_find(kvm_t *kd, uint64_t pa, unsigned int page_size) { uint64_t *bitmap = kd->pt_map; uint64_t pte_bit_id = pa / page_size; uint64_t pte_u64 = pte_bit_id / BITS_IN(*bitmap); uint64_t popcount_id = pte_bit_id / POPCOUNT_BITS; uint64_t pte_mask = 1ULL << (pte_bit_id % BITS_IN(*bitmap)); uint64_t bitN; uint32_t count; /* Check whether the page address requested is in the dump. */ if (pte_bit_id >= (kd->pt_map_size * NBBY) || (bitmap[pte_u64] & pte_mask) == 0) return (-1); /* * Add/sub popcounts from the bitmap until the PTE's bit is reached. * For bits that are in the upper half between the calculated * popcount id and the next one, use the next one and subtract to * minimize the number of popcounts required. */ if ((pte_bit_id % POPCOUNT_BITS) < (POPCOUNT_BITS / 2)) { count = kd->pt_popcounts[popcount_id] + popcount_bytes( bitmap + popcount_id * POPCOUNTS_IN(*bitmap), 0, pte_bit_id - popcount_id * POPCOUNT_BITS); } else { /* * Counting in reverse is trickier, since we must avoid * reading from bytes that are not in range, and invert. */ uint64_t pte_u64_bit_off = pte_u64 * BITS_IN(*bitmap); popcount_id++; bitN = MIN(popcount_id * POPCOUNT_BITS, kd->pt_map_size * BITS_IN(uint8_t)); count = kd->pt_popcounts[popcount_id] - popcount_bytes( bitmap + pte_u64, pte_bit_id - pte_u64_bit_off, bitN - pte_u64_bit_off); } /* * This can only happen if the core is truncated. Treat these * entries as if they don't exist, since their backing doesn't. */ if (count >= (kd->pt_sparse_size / page_size)) return (-1); return (kd->pt_sparse_off + (uint64_t)count * page_size); } static int kvm_fdnlist(kvm_t *kd, struct kvm_nlist *list) { kvaddr_t addr; int error, nfail; if (kd->resolve_symbol == NULL) { struct nlist *nl; int count, i; for (count = 0; list[count].n_name != NULL && list[count].n_name[0] != '\0'; count++) ; nl = calloc(count + 1, sizeof(*nl)); for (i = 0; i < count; i++) nl[i].n_name = list[i].n_name; nfail = __fdnlist(kd->nlfd, nl); for (i = 0; i < count; i++) { list[i].n_type = nl[i].n_type; list[i].n_value = nl[i].n_value; } free(nl); return (nfail); } nfail = 0; while (list->n_name != NULL && list->n_name[0] != '\0') { error = kd->resolve_symbol(list->n_name, &addr); if (error != 0) { nfail++; list->n_value = 0; list->n_type = 0; } else { list->n_value = addr; list->n_type = N_DATA | N_EXT; } list++; } return (nfail); } /* * Walk the list of unresolved symbols, generate a new list and prefix the * symbol names, try again, and merge back what we could resolve. */ static int kvm_fdnlist_prefix(kvm_t *kd, struct kvm_nlist *nl, int missing, const char *prefix, kvaddr_t (*validate_fn)(kvm_t *, kvaddr_t)) { struct kvm_nlist *n, *np, *p; char *cp, *ce; const char *ccp; size_t len; int slen, unresolved; /* * Calculate the space we need to malloc for nlist and names. * We are going to store the name twice for later lookups: once * with the prefix and once the unmodified name delmited by \0. */ len = 0; unresolved = 0; for (p = nl; p->n_name && p->n_name[0]; ++p) { if (p->n_type != N_UNDF) continue; len += sizeof(struct kvm_nlist) + strlen(prefix) + 2 * (strlen(p->n_name) + 1); unresolved++; } if (unresolved == 0) return (unresolved); /* Add space for the terminating nlist entry. */ len += sizeof(struct kvm_nlist); unresolved++; /* Alloc one chunk for (nlist, [names]) and setup pointers. */ n = np = malloc(len); bzero(n, len); if (n == NULL) return (missing); cp = ce = (char *)np; cp += unresolved * sizeof(struct kvm_nlist); ce += len; /* Generate shortened nlist with special prefix. */ unresolved = 0; for (p = nl; p->n_name && p->n_name[0]; ++p) { if (p->n_type != N_UNDF) continue; *np = *p; /* Save the new\0orig. name so we can later match it again. */ slen = snprintf(cp, ce - cp, "%s%s%c%s", prefix, (prefix[0] != '\0' && p->n_name[0] == '_') ? (p->n_name + 1) : p->n_name, '\0', p->n_name); if (slen < 0 || slen >= ce - cp) continue; np->n_name = cp; cp += slen + 1; np++; unresolved++; } /* Do lookup on the reduced list. */ np = n; unresolved = kvm_fdnlist(kd, np); /* Check if we could resolve further symbols and update the list. */ if (unresolved >= 0 && unresolved < missing) { /* Find the first freshly resolved entry. */ for (; np->n_name && np->n_name[0]; np++) if (np->n_type != N_UNDF) break; /* * The lists are both in the same order, * so we can walk them in parallel. */ for (p = nl; np->n_name && np->n_name[0] && p->n_name && p->n_name[0]; ++p) { if (p->n_type != N_UNDF) continue; /* Skip expanded name and compare to orig. one. */ ccp = np->n_name + strlen(np->n_name) + 1; if (strcmp(ccp, p->n_name) != 0) continue; /* Update nlist with new, translated results. */ p->n_type = np->n_type; if (validate_fn) p->n_value = (*validate_fn)(kd, np->n_value); else p->n_value = np->n_value; missing--; /* Find next freshly resolved entry. */ for (np++; np->n_name && np->n_name[0]; np++) if (np->n_type != N_UNDF) break; } } /* We could assert missing = unresolved here. */ free(n); return (unresolved); } int _kvm_nlist(kvm_t *kd, struct kvm_nlist *nl, int initialize) { struct kvm_nlist *p; int nvalid; struct kld_sym_lookup lookup; int error; const char *prefix = ""; char symname[1024]; /* XXX-BZ symbol name length limit? */ int tried_vnet, tried_dpcpu; /* * If we can't use the kld symbol lookup, revert to the * slow library call. */ if (!ISALIVE(kd)) { error = kvm_fdnlist(kd, nl); if (error <= 0) /* Hard error or success. */ return (error); if (_kvm_vnet_initialized(kd, initialize)) error = kvm_fdnlist_prefix(kd, nl, error, VNET_SYMPREFIX, _kvm_vnet_validaddr); if (error > 0 && _kvm_dpcpu_initialized(kd, initialize)) error = kvm_fdnlist_prefix(kd, nl, error, DPCPU_SYMPREFIX, _kvm_dpcpu_validaddr); return (error); } /* * We can use the kld lookup syscall. Go through each nlist entry * and look it up with a kldsym(2) syscall. */ nvalid = 0; tried_vnet = 0; tried_dpcpu = 0; again: for (p = nl; p->n_name && p->n_name[0]; ++p) { if (p->n_type != N_UNDF) continue; lookup.version = sizeof(lookup); lookup.symvalue = 0; lookup.symsize = 0; error = snprintf(symname, sizeof(symname), "%s%s", prefix, (prefix[0] != '\0' && p->n_name[0] == '_') ? (p->n_name + 1) : p->n_name); if (error < 0 || error >= (int)sizeof(symname)) continue; lookup.symname = symname; if (lookup.symname[0] == '_') lookup.symname++; if (kldsym(0, KLDSYM_LOOKUP, &lookup) != -1) { p->n_type = N_TEXT; if (_kvm_vnet_initialized(kd, initialize) && strcmp(prefix, VNET_SYMPREFIX) == 0) p->n_value = _kvm_vnet_validaddr(kd, lookup.symvalue); else if (_kvm_dpcpu_initialized(kd, initialize) && strcmp(prefix, DPCPU_SYMPREFIX) == 0) p->n_value = _kvm_dpcpu_validaddr(kd, lookup.symvalue); else p->n_value = lookup.symvalue; ++nvalid; /* lookup.symsize */ } } /* * Check the number of entries that weren't found. If they exist, * try again with a prefix for virtualized or DPCPU symbol names. */ error = ((p - nl) - nvalid); if (error && _kvm_vnet_initialized(kd, initialize) && !tried_vnet) { tried_vnet = 1; prefix = VNET_SYMPREFIX; goto again; } if (error && _kvm_dpcpu_initialized(kd, initialize) && !tried_dpcpu) { tried_dpcpu = 1; prefix = DPCPU_SYMPREFIX; goto again; } /* * Return the number of entries that weren't found. If they exist, * also fill internal error buffer. */ error = ((p - nl) - nvalid); if (error) _kvm_syserr(kd, kd->program, "kvm_nlist"); return (error); } int _kvm_bitmap_init(struct kvm_bitmap *bm, u_long bitmapsize, u_long *idx) { *idx = ULONG_MAX; bm->map = calloc(bitmapsize, sizeof *bm->map); if (bm->map == NULL) return (0); bm->size = bitmapsize; return (1); } void _kvm_bitmap_set(struct kvm_bitmap *bm, u_long pa, unsigned int page_size) { u_long bm_index = pa / page_size; uint8_t *byte = &bm->map[bm_index / 8]; *byte |= (1UL << (bm_index % 8)); } int _kvm_bitmap_next(struct kvm_bitmap *bm, u_long *idx) { u_long first_invalid = bm->size * CHAR_BIT; if (*idx == ULONG_MAX) *idx = 0; else (*idx)++; /* Find the next valid idx. */ for (; *idx < first_invalid; (*idx)++) { unsigned int mask = *idx % CHAR_BIT; if ((bm->map[*idx * CHAR_BIT] & mask) == 0) break; } return (*idx < first_invalid); } void _kvm_bitmap_deinit(struct kvm_bitmap *bm) { free(bm->map); } int _kvm_visit_cb(kvm_t *kd, kvm_walk_pages_cb_t *cb, void *arg, u_long pa, u_long kmap_vaddr, u_long dmap_vaddr, vm_prot_t prot, size_t len, unsigned int page_size) { unsigned int pgsz = page_size ? page_size : len; struct kvm_page p = { - .version = LIBKVM_WALK_PAGES_VERSION, - .paddr = pa, - .kmap_vaddr = kmap_vaddr, - .dmap_vaddr = dmap_vaddr, - .prot = prot, - .offset = _kvm_pt_find(kd, pa, pgsz), - .len = len, + .kp_version = LIBKVM_WALK_PAGES_VERSION, + .kp_paddr = pa, + .kp_kmap_vaddr = kmap_vaddr, + .kp_dmap_vaddr = dmap_vaddr, + .kp_prot = prot, + .kp_offset = _kvm_pt_find(kd, pa, pgsz), + .kp_len = len, }; return cb(&p, arg); } Index: head/sys/sys/types.h =================================================================== --- head/sys/sys/types.h (revision 354542) +++ head/sys/sys/types.h (revision 354543) @@ -1,437 +1,438 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1982, 1986, 1991, 1993, 1994 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * 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. * * @(#)types.h 8.6 (Berkeley) 2/19/95 * $FreeBSD$ */ #ifndef _SYS_TYPES_H_ #define _SYS_TYPES_H_ #include /* Machine type dependent parameters. */ #include #include #include #if __BSD_VISIBLE typedef unsigned char u_char; typedef unsigned short u_short; typedef unsigned int u_int; typedef unsigned long u_long; #ifndef _KERNEL typedef unsigned short ushort; /* Sys V compatibility */ typedef unsigned int uint; /* Sys V compatibility */ #endif #endif /* * XXX POSIX sized integrals that should appear only in . */ #include typedef __uint8_t u_int8_t; /* unsigned integrals (deprecated) */ typedef __uint16_t u_int16_t; typedef __uint32_t u_int32_t; typedef __uint64_t u_int64_t; typedef __uint64_t u_quad_t; /* quads (deprecated) */ typedef __int64_t quad_t; typedef quad_t * qaddr_t; typedef char * caddr_t; /* core address */ typedef const char * c_caddr_t; /* core address, pointer to const */ #ifndef _BLKSIZE_T_DECLARED typedef __blksize_t blksize_t; #define _BLKSIZE_T_DECLARED #endif typedef __cpuwhich_t cpuwhich_t; typedef __cpulevel_t cpulevel_t; typedef __cpusetid_t cpusetid_t; #ifndef _BLKCNT_T_DECLARED typedef __blkcnt_t blkcnt_t; #define _BLKCNT_T_DECLARED #endif #ifndef _CLOCK_T_DECLARED typedef __clock_t clock_t; #define _CLOCK_T_DECLARED #endif #ifndef _CLOCKID_T_DECLARED typedef __clockid_t clockid_t; #define _CLOCKID_T_DECLARED #endif typedef __critical_t critical_t; /* Critical section value */ typedef __daddr_t daddr_t; /* disk address */ #ifndef _DEV_T_DECLARED typedef __dev_t dev_t; /* device number or struct cdev */ #define _DEV_T_DECLARED #endif #ifndef _FFLAGS_T_DECLARED typedef __fflags_t fflags_t; /* file flags */ #define _FFLAGS_T_DECLARED #endif typedef __fixpt_t fixpt_t; /* fixed point number */ #ifndef _FSBLKCNT_T_DECLARED /* for statvfs() */ typedef __fsblkcnt_t fsblkcnt_t; typedef __fsfilcnt_t fsfilcnt_t; #define _FSBLKCNT_T_DECLARED #endif #ifndef _GID_T_DECLARED typedef __gid_t gid_t; /* group id */ #define _GID_T_DECLARED #endif #ifndef _IN_ADDR_T_DECLARED typedef __uint32_t in_addr_t; /* base type for internet address */ #define _IN_ADDR_T_DECLARED #endif #ifndef _IN_PORT_T_DECLARED typedef __uint16_t in_port_t; #define _IN_PORT_T_DECLARED #endif #ifndef _ID_T_DECLARED typedef __id_t id_t; /* can hold a uid_t or pid_t */ #define _ID_T_DECLARED #endif #ifndef _INO_T_DECLARED typedef __ino_t ino_t; /* inode number */ #define _INO_T_DECLARED #endif #ifndef _KEY_T_DECLARED typedef __key_t key_t; /* IPC key (for Sys V IPC) */ #define _KEY_T_DECLARED #endif #ifndef _LWPID_T_DECLARED typedef __lwpid_t lwpid_t; /* Thread ID (a.k.a. LWP) */ #define _LWPID_T_DECLARED #endif #ifndef _MODE_T_DECLARED typedef __mode_t mode_t; /* permissions */ #define _MODE_T_DECLARED #endif #ifndef _ACCMODE_T_DECLARED typedef __accmode_t accmode_t; /* access permissions */ #define _ACCMODE_T_DECLARED #endif #ifndef _NLINK_T_DECLARED typedef __nlink_t nlink_t; /* link count */ #define _NLINK_T_DECLARED #endif #ifndef _OFF_T_DECLARED typedef __off_t off_t; /* file offset */ #define _OFF_T_DECLARED #endif #ifndef _OFF64_T_DECLARED typedef __off64_t off64_t; /* file offset (alias) */ #define _OFF64_T_DECLARED #endif #ifndef _PID_T_DECLARED typedef __pid_t pid_t; /* process id */ #define _PID_T_DECLARED #endif typedef __register_t register_t; #ifndef _RLIM_T_DECLARED typedef __rlim_t rlim_t; /* resource limit */ #define _RLIM_T_DECLARED #endif typedef __int64_t sbintime_t; typedef __segsz_t segsz_t; /* segment size (in pages) */ #ifndef _SIZE_T_DECLARED typedef __size_t size_t; #define _SIZE_T_DECLARED #endif #ifndef _SSIZE_T_DECLARED typedef __ssize_t ssize_t; #define _SSIZE_T_DECLARED #endif #ifndef _SUSECONDS_T_DECLARED typedef __suseconds_t suseconds_t; /* microseconds (signed) */ #define _SUSECONDS_T_DECLARED #endif #ifndef _TIME_T_DECLARED typedef __time_t time_t; #define _TIME_T_DECLARED #endif #ifndef _TIMER_T_DECLARED typedef __timer_t timer_t; #define _TIMER_T_DECLARED #endif #ifndef _MQD_T_DECLARED typedef __mqd_t mqd_t; #define _MQD_T_DECLARED #endif typedef __u_register_t u_register_t; #ifndef _UID_T_DECLARED typedef __uid_t uid_t; /* user id */ #define _UID_T_DECLARED #endif #ifndef _USECONDS_T_DECLARED typedef __useconds_t useconds_t; /* microseconds (unsigned) */ #define _USECONDS_T_DECLARED #endif #ifndef _CAP_IOCTL_T_DECLARED #define _CAP_IOCTL_T_DECLARED typedef unsigned long cap_ioctl_t; #endif #ifndef _CAP_RIGHTS_T_DECLARED #define _CAP_RIGHTS_T_DECLARED struct cap_rights; typedef struct cap_rights cap_rights_t; #endif /* - * Types suitable for exporting size and pointers (as virtual addresses) - * from the kernel independent of native word size. These should be - * used in place of size_t and (u)intptr_t in structs which contain such - * types that are shared with userspace. + * Types suitable for exporting physical addresses, virtual addresses + * (pointers), and memory object sizes from the kernel independent of native + * word size. These should be used in place of vm_paddr_t, (u)intptr_t, and + * size_t in structs which contain such types that are shared with userspace. */ +typedef __uint64_t kpaddr_t; typedef __uint64_t kvaddr_t; typedef __uint64_t ksize_t; typedef __vm_offset_t vm_offset_t; typedef __uint64_t vm_ooffset_t; typedef __vm_paddr_t vm_paddr_t; typedef __uint64_t vm_pindex_t; typedef __vm_size_t vm_size_t; typedef __rman_res_t rman_res_t; #ifdef _KERNEL typedef int boolean_t; typedef struct device *device_t; typedef __intfptr_t intfptr_t; /* * XXX this is fixed width for historical reasons. It should have had type * __int_fast32_t. Fixed-width types should not be used unless binary * compatibility is essential. Least-width types should be used even less * since they provide smaller benefits. * * XXX should be MD. * * XXX this is bogus in -current, but still used for spl*(). */ typedef __uint32_t intrmask_t; /* Interrupt mask (spl, xxx_imask...) */ typedef __uintfptr_t uintfptr_t; typedef __uint64_t uoff_t; typedef char vm_memattr_t; /* memory attribute codes */ typedef struct vm_page *vm_page_t; #if !defined(__bool_true_false_are_defined) && !defined(__cplusplus) #define __bool_true_false_are_defined 1 #define false 0 #define true 1 #if __STDC_VERSION__ < 199901L && __GNUC__ < 3 && !defined(__INTEL_COMPILER) typedef int _Bool; #endif typedef _Bool bool; #endif /* !__bool_true_false_are_defined && !__cplusplus */ #define offsetof(type, field) __offsetof(type, field) #endif /* !_KERNEL */ /* * The following are all things that really shouldn't exist in this header, * since its purpose is to provide typedefs, not miscellaneous doodads. */ #ifdef __POPCNT__ #define __bitcount64(x) __builtin_popcountll((__uint64_t)(x)) #define __bitcount32(x) __builtin_popcount((__uint32_t)(x)) #define __bitcount16(x) __builtin_popcount((__uint16_t)(x)) #define __bitcountl(x) __builtin_popcountl((unsigned long)(x)) #define __bitcount(x) __builtin_popcount((unsigned int)(x)) #else /* * Population count algorithm using SWAR approach * - "SIMD Within A Register". */ static __inline __uint16_t __bitcount16(__uint16_t _x) { _x = (_x & 0x5555) + ((_x & 0xaaaa) >> 1); _x = (_x & 0x3333) + ((_x & 0xcccc) >> 2); _x = (_x + (_x >> 4)) & 0x0f0f; _x = (_x + (_x >> 8)) & 0x00ff; return (_x); } static __inline __uint32_t __bitcount32(__uint32_t _x) { _x = (_x & 0x55555555) + ((_x & 0xaaaaaaaa) >> 1); _x = (_x & 0x33333333) + ((_x & 0xcccccccc) >> 2); _x = (_x + (_x >> 4)) & 0x0f0f0f0f; _x = (_x + (_x >> 8)); _x = (_x + (_x >> 16)) & 0x000000ff; return (_x); } #ifdef __LP64__ static __inline __uint64_t __bitcount64(__uint64_t _x) { _x = (_x & 0x5555555555555555) + ((_x & 0xaaaaaaaaaaaaaaaa) >> 1); _x = (_x & 0x3333333333333333) + ((_x & 0xcccccccccccccccc) >> 2); _x = (_x + (_x >> 4)) & 0x0f0f0f0f0f0f0f0f; _x = (_x + (_x >> 8)); _x = (_x + (_x >> 16)); _x = (_x + (_x >> 32)) & 0x000000ff; return (_x); } #define __bitcountl(x) __bitcount64((unsigned long)(x)) #else static __inline __uint64_t __bitcount64(__uint64_t _x) { return (__bitcount32(_x >> 32) + __bitcount32(_x)); } #define __bitcountl(x) __bitcount32((unsigned long)(x)) #endif #define __bitcount(x) __bitcount32((unsigned int)(x)) #endif #if __BSD_VISIBLE #include /* * The major and minor numbers are encoded in dev_t as MMMmmmMm (where * letters correspond to bytes). The encoding of the lower 4 bytes is * constrained by compatibility with 16-bit and 32-bit dev_t's. The * encoding of of the upper 4 bytes is the least unnatural one consistent * with this and other constraints. Also, the decoding of the m bytes by * minor() is unnatural to maximize compatibility subject to not discarding * bits. The upper m byte is shifted into the position of the lower M byte * instead of shifting 3 upper m bytes to close the gap. Compatibility for * minor() is achieved iff the upper m byte is 0. */ #define major(d) __major(d) static __inline int __major(dev_t _d) { return (((_d >> 32) & 0xffffff00) | ((_d >> 8) & 0xff)); } #define minor(d) __minor(d) static __inline int __minor(dev_t _d) { return (((_d >> 24) & 0xff00) | (_d & 0xffff00ff)); } #define makedev(M, m) __makedev((M), (m)) static __inline dev_t __makedev(int _Major, int _Minor) { return (((dev_t)(_Major & 0xffffff00) << 32) | ((_Major & 0xff) << 8) | ((dev_t)(_Minor & 0xff00) << 24) | (_Minor & 0xffff00ff)); } /* * These declarations belong elsewhere, but are repeated here and in * to give broken programs a better chance of working with * 64-bit off_t's. */ #ifndef _KERNEL __BEGIN_DECLS #ifndef _FTRUNCATE_DECLARED #define _FTRUNCATE_DECLARED int ftruncate(int, off_t); #endif #ifndef _LSEEK_DECLARED #define _LSEEK_DECLARED off_t lseek(int, off_t, int); #endif #ifndef _MMAP_DECLARED #define _MMAP_DECLARED void * mmap(void *, size_t, int, int, int, off_t); #endif #ifndef _TRUNCATE_DECLARED #define _TRUNCATE_DECLARED int truncate(const char *, off_t); #endif __END_DECLS #endif /* !_KERNEL */ #endif /* __BSD_VISIBLE */ #endif /* !_SYS_TYPES_H_ */