/* Copyright (C) 1994, 1995 Free Software Foundation This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ #include #include #include #include #include #include #include #include #include "ext2fs.h" #include "error.h" /* ---------------------------------------------------------------- */ int diskfs_link_max = EXT2_LINK_MAX; int diskfs_maxsymlinks = 8; int diskfs_shortcut_symlink = 1; int diskfs_shortcut_chrdev = 1; int diskfs_shortcut_blkdev = 1; int diskfs_shortcut_fifo = 1; int diskfs_shortcut_ifsock = 1; char *diskfs_server_name = "ext2fs"; int diskfs_major_version = 0; int diskfs_minor_version = 0; int diskfs_edit_version = 0; int diskfs_synchronous = 0; int diskfs_readonly = 0; /* ---------------------------------------------------------------- */ struct node *diskfs_root_node; /* Set diskfs_root_node to the root inode. */ static void warp_root (void) { error_t err; err = iget (EXT2_ROOT_INO, &diskfs_root_node); assert (!err); mutex_unlock (&diskfs_root_node->lock); } /* ---------------------------------------------------------------- */ /* XXX */ struct mutex printf_lock; int printf (const char *fmt, ...) { va_list arg; int done; va_start (arg, fmt); mutex_lock (&printf_lock); done = vprintf (fmt, arg); mutex_unlock (&printf_lock); va_end (arg); return done; } static char error_buf[1024]; void ext2_error (const char * function, const char * fmt, ...) { va_list args; va_start (args, fmt); vsprintf (error_buf, fmt, args); va_end (args); mutex_lock(&printf_lock); fprintf (stderr, "ext2fs: %s: %s: %s\n", device_name, function, error_buf); mutex_unlock(&printf_lock); } void ext2_panic (const char * function, const char * fmt, ...) { va_list args; va_start (args, fmt); vsprintf (error_buf, fmt, args); va_end (args); mutex_lock(&printf_lock); fprintf(stderr, "ext2fs: %s: panic: %s: %s\n", device_name, function, error_buf); mutex_unlock(&printf_lock); exit (0); } void ext2_warning (const char * function, const char * fmt, ...) { va_list args; va_start (args, fmt); vsprintf (error_buf, fmt, args); va_end (args); mutex_lock(&printf_lock); fprintf (stderr, "ext2fs: %s: %s: %s\n", device_name, function, error_buf); mutex_unlock(&printf_lock); } /* ---------------------------------------------------------------- */ #define USAGE "Usage: %s [OPTION...] DEVICE\n" static void usage(int status) { if (status != 0) fprintf(stderr, "Try `%s --help' for more information.\n", program_invocation_name); else { printf(USAGE, program_invocation_name); printf("\ \n\ -r, --readonly disable writing to DEVICE\n\ -s, --synchronous write data out immediately\n\ --help display this help and exit\n\ --version output version information and exit\n\ "); } exit (status); } #define SHORT_OPTIONS "rs?" static struct option options[] = { {"readonly", no_argument, 0, 'r'}, {"synchronous", no_argument, 0, 's'}, {"help", no_argument, 0, '?'}, {"version", no_argument, 0, 'V'}, {0, 0, 0, 0} }; /* ---------------------------------------------------------------- */ int check_string = 1; void main (int argc, char **argv) { error_t err; mach_port_t bootstrap; int sizes[DEV_GET_SIZE_COUNT]; unsigned sizescnt = 2; mutex_init (&printf_lock); /* XXX */ if (getpid () > 0) { int opt; int fd = open ("/dev/console", O_RDWR); /* Make errors go somewhere reasonable. */ while (fd >= 0 && fd < 2) fd = dup(fd); if (fd > 2) close (fd); while ((opt = getopt_long(argc, argv, SHORT_OPTIONS, options, 0)) != EOF) switch (opt) { case 'r': diskfs_readonly = 1; break; case 's': diskfs_synchronous = 1; break; case 'V': printf("%s %d.%d.%d\n", diskfs_server_name, diskfs_major_version, diskfs_minor_version, diskfs_edit_version); exit(0); case '?': usage(0); default: usage(1); } if (argc - optind != 1) { fprintf (stderr, USAGE, program_invocation_name); usage (1); } device_name = argv[optind]; } else /* We are the bootstrap filesystem. */ device_name = diskfs_parse_bootargs (argc, argv); task_get_bootstrap_port (mach_task_self (), &bootstrap); /* Initialize the diskfs library. This must come before any other diskfs call. */ diskfs_init_diskfs (); do { char *line = 0; size_t linesz = 0; ssize_t len; err = device_open (diskfs_master_device, (diskfs_readonly ? 0 : D_WRITE) | D_READ, device_name, &device_port); if (err == D_NO_SUCH_DEVICE && getpid () <= 0) { /* Prompt the user to give us another name rather than just crashing */ printf ("Cannot open device %s\n", device_name); printf ("Open instead: "); fflush (stdout); len = getline (&line, &linesz, stdin); if (len > 2) device_name = line; } } while (err && err == D_NO_SUCH_DEVICE && getpid () <= 0); if (err) error(1, errno, "%s", device_name); /* Check to make sure device sector size is reasonable. */ err = device_get_status (device_port, DEV_GET_SIZE, sizes, &sizescnt); assert (sizescnt == DEV_GET_SIZE_COUNT); device_block_size = sizes[DEV_GET_SIZE_RECORD_SIZE]; device_size = sizes[DEV_GET_SIZE_DEVICE_SIZE]; assert (device_size >= SBLOCK_OFFS + SBLOCK_SIZE); /* Map the entire disk. */ create_disk_pager (); /* Start the first request thread, to handle RPCs and page requests. */ diskfs_spawn_first_thread (); err = vm_map (mach_task_self (), (vm_address_t *)&disk_image, device_size, 0, 1, disk_pager_port, 0, 0, VM_PROT_READ | (diskfs_readonly ? 0 : VM_PROT_WRITE), VM_PROT_READ | (diskfs_readonly ? 0 : VM_PROT_WRITE), VM_INHERIT_NONE); if (err) error (2, err, "vm_map"); diskfs_register_memory_fault_area (disk_pager->p, 0, disk_image, device_size); pokel_init (&global_pokel, disk_pager->p, disk_image); err = get_hypermetadata(); if (err) error (3, err, "get_hypermetadata"); if (device_size < sblock->s_blocks_count * block_size) ext2_panic("main", "Disk size (%d) too small (superblock says we need %ld)", sizes[DEV_GET_SIZE_DEVICE_SIZE], sblock->s_blocks_count * block_size); /* A handy source of page-aligned zeros. */ vm_allocate (mach_task_self (), &zeroblock, block_size, 1); if (!diskfs_readonly) { sblock->s_state &= ~EXT2_VALID_FS; sync_super_block (); } inode_init (); /* Find our root node. */ warp_root (); /* Now that we are all set up to handle requests, and diskfs_root_node is set properly, it is safe to export our fsys control port to the outside world. */ (void) diskfs_startup_diskfs (bootstrap); #if 0 if (bootstrap == MACH_PORT_NULL) /* We are the bootstrap filesystem; do special boot-time setup. */ diskfs_start_bootstrap (argv); #endif /* Now become a generic request thread. */ diskfs_main_request_loop (); } void diskfs_init_completed () { string_t version; mach_port_t proc, startup; error_t err; sprintf(version, "%s %d.%d.%d", diskfs_server_name, diskfs_major_version, diskfs_minor_version, diskfs_edit_version); proc = getproc (); proc_register_version (proc, diskfs_host_priv, diskfs_server_name, HURD_RELEASE, version); err = proc_getmsgport (proc, 1, &startup); if (!err) { startup_essential_task (startup, mach_task_self (), MACH_PORT_NULL, diskfs_server_name, diskfs_host_priv); mach_port_deallocate (mach_task_self (), startup); } mach_port_deallocate (mach_task_self (), proc); } /* ---------------------------------------------------------------- */ static spin_lock_t free_page_bufs_lock = SPIN_LOCK_INITIALIZER; static vm_address_t free_page_bufs = 0; /* Returns a single page page-aligned buffer. */ vm_address_t get_page_buf () { vm_address_t buf; spin_lock (&free_page_bufs_lock); buf = free_page_bufs; if (buf == 0) { spin_unlock (&free_page_bufs_lock); vm_allocate (mach_task_self (), &buf, vm_page_size, 1); } else { free_page_bufs = *(vm_address_t *)buf; spin_unlock (&free_page_bufs_lock); } return buf; } /* Frees a block returned by get_page_buf. */ void free_page_buf (vm_address_t buf) { spin_lock (&free_page_bufs_lock); *(vm_address_t *)buf = free_page_bufs; free_page_bufs = buf; spin_unlock (&free_page_bufs_lock); }