/* $NetBSD: kernel.c,v 1.5 2010/12/28 13:36:09 haad Exp $ */ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2008 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #pragma ident "%Z%%M% %I% %E% SMI" #include __RCSID("$NetBSD: kernel.c,v 1.5 2010/12/28 13:36:09 haad Exp $"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Emulation of kernel services in userland. */ #ifdef XXXNETBSD int hz = 119; /* frequency when using gethrtime() >> 23 for lbolt */ #endif int aok; uint64_t physmem; vnode_t *rootdir = (vnode_t *)0xabcd1234; char hw_serial[11]; size_t pgsize; struct utsname utsname = { "userland" }; /* this only exists to have its address taken */ struct proc p0; /* * ========================================================================= * threads * ========================================================================= */ /*ARGSUSED*/ kthread_t * zk_thread_create(void (*func)(), void *arg) { thread_t tid; VERIFY(thr_create(0, 0, (void *(*)(void *))func, arg, THR_DETACHED, &tid) == 0); return ((void *)(uintptr_t)tid); } /* * ========================================================================= * kstats * ========================================================================= */ /*ARGSUSED*/ kstat_t * kstat_create(char *module, int instance, char *name, char *class, uchar_t type, ulong_t ndata, uchar_t ks_flag) { return (NULL); } /*ARGSUSED*/ void kstat_install(kstat_t *ksp) {} /*ARGSUSED*/ void kstat_delete(kstat_t *ksp) {} /* * ========================================================================= * vnode operations * ========================================================================= */ /* * Note: for the xxxat() versions of these functions, we assume that the * starting vp is always rootdir (which is true for spa_directory.c, the only * ZFS consumer of these interfaces). We assert this is true, and then emulate * them by adding '/' in front of the path. */ /*ARGSUSED*/ int vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3) { int fd; vnode_t *vp; int old_umask; char realpath[MAXPATHLEN]; struct stat64 st; /* * If we're accessing a real disk from userland, we need to use * the character interface to avoid caching. This is particularly * important if we're trying to look at a real in-kernel storage * pool from userland, e.g. via zdb, because otherwise we won't * see the changes occurring under the segmap cache. * On the other hand, the stupid character device returns zero * for its size. So -- gag -- we open the block device to get * its size, and remember it for subsequent VOP_GETATTR(). */ if (strncmp(path, "/dev/", 5) == 0) { char *dsk; fd = open64(path, O_RDONLY); if (fd == -1) return (errno); if (fstat64(fd, &st) == -1) { close(fd); return (errno); } close(fd); (void) sprintf(realpath, "%s", path); dsk = strstr(path, "/dsk/"); if (dsk != NULL) (void) sprintf(realpath + (dsk - path) + 1, "r%s", dsk + 1); } else { (void) sprintf(realpath, "%s", path); if (!(flags & FCREAT) && stat64(realpath, &st) == -1) return (errno); } if (flags & FCREAT) old_umask = umask(0); /* * The construct 'flags - FREAD' conveniently maps combinations of * FREAD and FWRITE to the corresponding O_RDONLY, O_WRONLY, and O_RDWR. */ fd = open64(realpath, flags - FREAD, mode); if (flags & FCREAT) (void) umask(old_umask); if (fd == -1) return (errno); if (fstat64(fd, &st) == -1) { close(fd); return (errno); } (void) fcntl(fd, F_SETFD, FD_CLOEXEC); *vpp = vp = umem_zalloc(sizeof (vnode_t), UMEM_NOFAIL); vp->v_fd = fd; if (S_ISCHR(st.st_mode)) { #ifdef XXXAD ioctl(fd, DIOCGMEDIASIZE, &vp->v_size); #endif } else vp->v_size = st.st_size; vp->v_path = spa_strdup(path); return (0); } int vn_openat(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3, vnode_t *startvp, int fd) { char *realpath = umem_alloc(strlen(path) + 2, UMEM_NOFAIL); int ret; ASSERT(startvp == rootdir); (void) sprintf(realpath, "/%s", path); ret = vn_open(realpath, x1, flags, mode, vpp, x2, x3); umem_free(realpath, strlen(path) + 2); return (ret); } int vn_getattr(vnode_t *vp, vattr_t *va) { int fd; struct stat64 st; fd = vp->v_fd; if (fstat64(fd, &st) == -1) return (errno); vp->v_size = st.st_size; va->va_size = st.st_size; return 0; } /*ARGSUSED*/ int vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset, int x1, int x2, rlim64_t x3, void *x4, ssize_t *residp) { ssize_t iolen, split; if (uio == UIO_READ) { iolen = pread64(vp->v_fd, addr, len, offset); } else { /* * To simulate partial disk writes, we split writes into two * system calls so that the process can be killed in between. */ split = (len > 0 ? rand() % len : 0); iolen = pwrite64(vp->v_fd, addr, split, offset); iolen += pwrite64(vp->v_fd, (char *)addr + split, len - split, offset + split); } if (iolen == -1) return (errno); if (residp) *residp = len - iolen; else if (iolen != len) return (EIO); return (0); } void vn_close(vnode_t *vp) { close(vp->v_fd); spa_strfree(vp->v_path); umem_free(vp, sizeof (vnode_t)); } #ifdef ZFS_DEBUG /* * ========================================================================= * Figure out which debugging statements to print * ========================================================================= */ static char *dprintf_string; static int dprintf_print_all; int dprintf_find_string(const char *string) { char *tmp_str = dprintf_string; int len = strlen(string); /* * Find out if this is a string we want to print. * String format: file1.c,function_name1,file2.c,file3.c */ while (tmp_str != NULL) { if (strncmp(tmp_str, string, len) == 0 && (tmp_str[len] == ',' || tmp_str[len] == '\0')) return (1); tmp_str = strchr(tmp_str, ','); if (tmp_str != NULL) tmp_str++; /* Get rid of , */ } return (0); } void dprintf_setup(int *argc, char **argv) { int i, j; /* * Debugging can be specified two ways: by setting the * environment variable ZFS_DEBUG, or by including a * "debug=..." argument on the command line. The command * line setting overrides the environment variable. */ for (i = 1; i < *argc; i++) { int len = strlen("debug="); /* First look for a command line argument */ if (strncmp("debug=", argv[i], len) == 0) { dprintf_string = argv[i] + len; /* Remove from args */ for (j = i; j < *argc; j++) argv[j] = argv[j+1]; argv[j] = NULL; (*argc)--; } } if (dprintf_string == NULL) { /* Look for ZFS_DEBUG environment variable */ dprintf_string = getenv("ZFS_DEBUG"); } /* * Are we just turning on all debugging? */ if (dprintf_find_string("on")) dprintf_print_all = 1; } /* * ========================================================================= * debug printfs * ========================================================================= */ void __dprintf(const char *file, const char *func, int line, const char *fmt, ...) { const char *newfile; va_list adx; /* * Get rid of annoying "../common/" prefix to filename. */ newfile = strrchr(file, '/'); if (newfile != NULL) { newfile = newfile + 1; /* Get rid of leading / */ } else { newfile = file; } if (dprintf_print_all || dprintf_find_string(newfile) || dprintf_find_string(func)) { /* Print out just the function name if requested */ flockfile(stdout); if (dprintf_find_string("pid")) (void) printf("%d ", getpid()); if (dprintf_find_string("tid")) (void) printf("%u ", thr_self()); #if 0 if (dprintf_find_string("cpu")) (void) printf("%u ", getcpuid()); #endif if (dprintf_find_string("time")) (void) printf("%llu ", gethrtime()); if (dprintf_find_string("long")) (void) printf("%s, line %d: ", newfile, line); (void) printf("%s: ", func); va_start(adx, fmt); (void) vprintf(fmt, adx); va_end(adx); funlockfile(stdout); } } #endif /* ZFS_DEBUG */ /* * ========================================================================= * cmn_err() and panic() * ========================================================================= */ static char ce_prefix[CE_IGNORE][10] = { "", "NOTICE: ", "WARNING: ", "" }; static char ce_suffix[CE_IGNORE][2] = { "", "\n", "\n", "" }; void vpanic(const char *fmt, va_list adx) { (void) fprintf(stderr, "error: "); (void) vfprintf(stderr, fmt, adx); (void) fprintf(stderr, "\n"); abort(); /* think of it as a "user-level crash dump" */ } void panic(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vpanic(fmt, adx); va_end(adx); } void vcmn_err(int ce, const char *fmt, va_list adx) { if (ce == CE_PANIC) vpanic(fmt, adx); if (ce != CE_NOTE) { /* suppress noise in userland stress testing */ (void) fprintf(stderr, "%s", ce_prefix[ce]); (void) vfprintf(stderr, fmt, adx); (void) fprintf(stderr, "%s", ce_suffix[ce]); } } /*PRINTFLIKE2*/ void cmn_err(int ce, const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(ce, fmt, adx); va_end(adx); } /* * ========================================================================= * kobj interfaces * ========================================================================= */ struct _buf * kobj_open_file(char *name) { struct _buf *file; vnode_t *vp; /* set vp as the _fd field of the file */ if (vn_openat(name, UIO_SYSSPACE, FREAD, 0, &vp, 0, 0, rootdir, 0) != 0) return ((void *)-1UL); file = umem_zalloc(sizeof (struct _buf), UMEM_NOFAIL); file->_fd = (intptr_t)vp; return (file); } int kobj_read_file(struct _buf *file, char *buf, unsigned size, unsigned off) { ssize_t resid; vn_rdwr(UIO_READ, (vnode_t *)file->_fd, buf, size, (offset_t)off, UIO_SYSSPACE, 0, 0, 0, &resid); return (size - resid); } void kobj_close_file(struct _buf *file) { vn_close((vnode_t *)file->_fd); umem_free(file, sizeof (struct _buf)); } int kobj_get_filesize(struct _buf *file, uint64_t *size) { struct stat64 st; vnode_t *vp = (vnode_t *)file->_fd; if (fstat64(vp->v_fd, &st) == -1) { vn_close(vp); return (errno); } *size = st.st_size; return (0); } /* * ========================================================================= * misc routines * ========================================================================= */ void delay(clock_t ticks) { poll(0, 0, ticks * (1000 / hz)); } #if 0 /* * Find highest one bit set. * Returns bit number + 1 of highest bit that is set, otherwise returns 0. * High order bit is 31 (or 63 in _LP64 kernel). */ int highbit(ulong_t i) { register int h = 1; if (i == 0) return (0); #ifdef _LP64 if (i & 0xffffffff00000000ul) { h += 32; i >>= 32; } #endif if (i & 0xffff0000) { h += 16; i >>= 16; } if (i & 0xff00) { h += 8; i >>= 8; } if (i & 0xf0) { h += 4; i >>= 4; } if (i & 0xc) { h += 2; i >>= 2; } if (i & 0x2) { h += 1; } return (h); } #endif static int random_get_bytes_common(uint8_t *ptr, size_t len, char *devname) { int fd = open(devname, O_RDONLY); size_t resid = len; ssize_t bytes; ASSERT(fd != -1); while (resid != 0) { bytes = read(fd, ptr, resid); ASSERT(bytes >= 0); ptr += bytes; resid -= bytes; } close(fd); return (0); } int random_get_bytes(uint8_t *ptr, size_t len) { return (random_get_bytes_common(ptr, len, "/dev/random")); } int random_get_pseudo_bytes(uint8_t *ptr, size_t len) { return (random_get_bytes_common(ptr, len, "/dev/urandom")); } int ddi_strtoul(const char *hw_serial, char **nptr, int base, unsigned long *result) { char *end; *result = strtoul(hw_serial, &end, base); if (*result == 0) return (errno); return (0); } /* * ========================================================================= * kernel emulation setup & teardown * ========================================================================= */ static int umem_out_of_memory(void) { char errmsg[] = "out of memory -- generating core dump\n"; write(fileno(stderr), errmsg, sizeof (errmsg)); abort(); return (0); } void kernel_init(int mode) { umem_nofail_callback(umem_out_of_memory); uint64_t physmem; size_t len = sizeof(physmem); static int mib[2] = { CTL_HW, HW_USERMEM64 }; if (sysctl(mib, sizeof(mib), &physmem, &len, NULL, 0) != 0) { len = 1048576 * 128; } pgsize = sysconf(_SC_PAGE_SIZE); dprintf("physmem = %llu pages (%.2f GB)\n", physmem / pgsize, (double)physmem / (1ULL << 30)); snprintf(hw_serial, sizeof (hw_serial), "%ld", gethostid()); system_taskq_init(); spa_init(mode); } void kernel_fini(void) { spa_fini(); } int z_uncompress(void *dst, size_t *dstlen, const void *src, size_t srclen) { int ret; uLongf len = *dstlen; if ((ret = uncompress(dst, &len, src, srclen)) == Z_OK) *dstlen = (size_t)len; return (ret); } int z_compress_level(void *dst, size_t *dstlen, const void *src, size_t srclen, int level) { int ret; uLongf len = *dstlen; if ((ret = compress2(dst, &len, src, srclen, level)) == Z_OK) *dstlen = (size_t)len; return (ret); } uid_t crgetuid(cred_t *cr) { return (0); } gid_t crgetgid(cred_t *cr) { return (0); } int crgetngroups(cred_t *cr) { return (0); } gid_t * crgetgroups(cred_t *cr) { return (NULL); } int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr) { return (0); } int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr) { return (0); } int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr) { return (0); } ksiddomain_t * ksid_lookupdomain(const char *dom) { ksiddomain_t *kd; kd = umem_zalloc(sizeof (ksiddomain_t), UMEM_NOFAIL); kd->kd_name = spa_strdup(dom); return (kd); } void ksiddomain_rele(ksiddomain_t *ksid) { spa_strfree(ksid->kd_name); umem_free(ksid, sizeof (ksiddomain_t)); } size_t ptob(size_t npg) { return npg * pgsize; } void print_timestamp(int fmt) { return; } /* * Do not change the length of the returned string; it must be freed * with strfree(). */ char * kmem_asprintf(const char *fmt, ...) { int size; va_list adx; char *buf; va_start(adx, fmt); size = vsnprintf(NULL, 0, fmt, adx) + 1; va_end(adx); buf = kmem_alloc(size, KM_SLEEP); va_start(adx, fmt); size = vsnprintf(buf, size, fmt, adx); va_end(adx); return (buf); }