u-boot-brain/fs/fs.c
Heinrich Schuchardt 5f59518a7b efi_loader: setting boot device
Up to now the bootefi command used the last file loaded to determine the
boot partition. This has led to errors when the fdt had been loaded from
another partition after the EFI binary.

Before setting the boot device from a loaded file check if it is a PE-COFF
image or a FIT image.

For a PE-COFF image remember address and size, boot device and path.

For a FIT image remember boot device and path.

If the PE-COFF image is overwritten by loading another file, forget it.

Do not allow to start an image via bootefi which is not the last loaded
PE-COFF image.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
2021-01-13 02:38:00 +01:00

956 lines
19 KiB
C

// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2012, NVIDIA CORPORATION. All rights reserved.
*/
#define LOG_CATEGORY LOGC_CORE
#include <command.h>
#include <config.h>
#include <errno.h>
#include <common.h>
#include <env.h>
#include <lmb.h>
#include <log.h>
#include <mapmem.h>
#include <part.h>
#include <ext4fs.h>
#include <fat.h>
#include <fs.h>
#include <sandboxfs.h>
#include <ubifs_uboot.h>
#include <btrfs.h>
#include <asm/io.h>
#include <div64.h>
#include <linux/math64.h>
#include <efi_loader.h>
#include <squashfs.h>
DECLARE_GLOBAL_DATA_PTR;
static struct blk_desc *fs_dev_desc;
static int fs_dev_part;
static struct disk_partition fs_partition;
static int fs_type = FS_TYPE_ANY;
static inline int fs_probe_unsupported(struct blk_desc *fs_dev_desc,
struct disk_partition *fs_partition)
{
log_err("** Unrecognized filesystem type **\n");
return -1;
}
static inline int fs_ls_unsupported(const char *dirname)
{
return -1;
}
/* generic implementation of ls in terms of opendir/readdir/closedir */
__maybe_unused
static int fs_ls_generic(const char *dirname)
{
struct fs_dir_stream *dirs;
struct fs_dirent *dent;
int nfiles = 0, ndirs = 0;
dirs = fs_opendir(dirname);
if (!dirs)
return -errno;
while ((dent = fs_readdir(dirs))) {
if (dent->type == FS_DT_DIR) {
printf(" %s/\n", dent->name);
ndirs++;
} else if (dent->type == FS_DT_LNK) {
printf(" <SYM> %s\n", dent->name);
nfiles++;
} else {
printf(" %8lld %s\n", dent->size, dent->name);
nfiles++;
}
}
fs_closedir(dirs);
printf("\n%d file(s), %d dir(s)\n\n", nfiles, ndirs);
return 0;
}
static inline int fs_exists_unsupported(const char *filename)
{
return 0;
}
static inline int fs_size_unsupported(const char *filename, loff_t *size)
{
return -1;
}
static inline int fs_read_unsupported(const char *filename, void *buf,
loff_t offset, loff_t len,
loff_t *actread)
{
return -1;
}
static inline int fs_write_unsupported(const char *filename, void *buf,
loff_t offset, loff_t len,
loff_t *actwrite)
{
return -1;
}
static inline int fs_ln_unsupported(const char *filename, const char *target)
{
return -1;
}
static inline void fs_close_unsupported(void)
{
}
static inline int fs_uuid_unsupported(char *uuid_str)
{
return -1;
}
static inline int fs_opendir_unsupported(const char *filename,
struct fs_dir_stream **dirs)
{
return -EACCES;
}
static inline int fs_unlink_unsupported(const char *filename)
{
return -1;
}
static inline int fs_mkdir_unsupported(const char *dirname)
{
return -1;
}
struct fstype_info {
int fstype;
char *name;
/*
* Is it legal to pass NULL as .probe()'s fs_dev_desc parameter? This
* should be false in most cases. For "virtual" filesystems which
* aren't based on a U-Boot block device (e.g. sandbox), this can be
* set to true. This should also be true for the dummy entry at the end
* of fstypes[], since that is essentially a "virtual" (non-existent)
* filesystem.
*/
bool null_dev_desc_ok;
int (*probe)(struct blk_desc *fs_dev_desc,
struct disk_partition *fs_partition);
int (*ls)(const char *dirname);
int (*exists)(const char *filename);
int (*size)(const char *filename, loff_t *size);
int (*read)(const char *filename, void *buf, loff_t offset,
loff_t len, loff_t *actread);
int (*write)(const char *filename, void *buf, loff_t offset,
loff_t len, loff_t *actwrite);
void (*close)(void);
int (*uuid)(char *uuid_str);
/*
* Open a directory stream. On success return 0 and directory
* stream pointer via 'dirsp'. On error, return -errno. See
* fs_opendir().
*/
int (*opendir)(const char *filename, struct fs_dir_stream **dirsp);
/*
* Read next entry from directory stream. On success return 0
* and directory entry pointer via 'dentp'. On error return
* -errno. See fs_readdir().
*/
int (*readdir)(struct fs_dir_stream *dirs, struct fs_dirent **dentp);
/* see fs_closedir() */
void (*closedir)(struct fs_dir_stream *dirs);
int (*unlink)(const char *filename);
int (*mkdir)(const char *dirname);
int (*ln)(const char *filename, const char *target);
};
static struct fstype_info fstypes[] = {
#ifdef CONFIG_FS_FAT
{
.fstype = FS_TYPE_FAT,
.name = "fat",
.null_dev_desc_ok = false,
.probe = fat_set_blk_dev,
.close = fat_close,
.ls = fs_ls_generic,
.exists = fat_exists,
.size = fat_size,
.read = fat_read_file,
#if CONFIG_IS_ENABLED(FAT_WRITE)
.write = file_fat_write,
.unlink = fat_unlink,
.mkdir = fat_mkdir,
#else
.write = fs_write_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
#endif
.uuid = fat_uuid,
.opendir = fat_opendir,
.readdir = fat_readdir,
.closedir = fat_closedir,
.ln = fs_ln_unsupported,
},
#endif
#if CONFIG_IS_ENABLED(FS_EXT4)
{
.fstype = FS_TYPE_EXT,
.name = "ext4",
.null_dev_desc_ok = false,
.probe = ext4fs_probe,
.close = ext4fs_close,
.ls = ext4fs_ls,
.exists = ext4fs_exists,
.size = ext4fs_size,
.read = ext4_read_file,
#ifdef CONFIG_CMD_EXT4_WRITE
.write = ext4_write_file,
.ln = ext4fs_create_link,
#else
.write = fs_write_unsupported,
.ln = fs_ln_unsupported,
#endif
.uuid = ext4fs_uuid,
.opendir = fs_opendir_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
},
#endif
#ifdef CONFIG_SANDBOX
{
.fstype = FS_TYPE_SANDBOX,
.name = "sandbox",
.null_dev_desc_ok = true,
.probe = sandbox_fs_set_blk_dev,
.close = sandbox_fs_close,
.ls = sandbox_fs_ls,
.exists = sandbox_fs_exists,
.size = sandbox_fs_size,
.read = fs_read_sandbox,
.write = fs_write_sandbox,
.uuid = fs_uuid_unsupported,
.opendir = fs_opendir_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
.ln = fs_ln_unsupported,
},
#endif
#ifdef CONFIG_CMD_UBIFS
{
.fstype = FS_TYPE_UBIFS,
.name = "ubifs",
.null_dev_desc_ok = true,
.probe = ubifs_set_blk_dev,
.close = ubifs_close,
.ls = ubifs_ls,
.exists = ubifs_exists,
.size = ubifs_size,
.read = ubifs_read,
.write = fs_write_unsupported,
.uuid = fs_uuid_unsupported,
.opendir = fs_opendir_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
.ln = fs_ln_unsupported,
},
#endif
#ifdef CONFIG_FS_BTRFS
{
.fstype = FS_TYPE_BTRFS,
.name = "btrfs",
.null_dev_desc_ok = false,
.probe = btrfs_probe,
.close = btrfs_close,
.ls = btrfs_ls,
.exists = btrfs_exists,
.size = btrfs_size,
.read = btrfs_read,
.write = fs_write_unsupported,
.uuid = btrfs_uuid,
.opendir = fs_opendir_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
.ln = fs_ln_unsupported,
},
#endif
#if IS_ENABLED(CONFIG_FS_SQUASHFS)
{
.fstype = FS_TYPE_SQUASHFS,
.name = "squashfs",
.null_dev_desc_ok = false,
.probe = sqfs_probe,
.opendir = sqfs_opendir,
.readdir = sqfs_readdir,
.ls = fs_ls_generic,
.read = sqfs_read,
.size = sqfs_size,
.close = sqfs_close,
.closedir = sqfs_closedir,
.exists = sqfs_exists,
.uuid = fs_uuid_unsupported,
.write = fs_write_unsupported,
.ln = fs_ln_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
},
#endif
{
.fstype = FS_TYPE_ANY,
.name = "unsupported",
.null_dev_desc_ok = true,
.probe = fs_probe_unsupported,
.close = fs_close_unsupported,
.ls = fs_ls_unsupported,
.exists = fs_exists_unsupported,
.size = fs_size_unsupported,
.read = fs_read_unsupported,
.write = fs_write_unsupported,
.uuid = fs_uuid_unsupported,
.opendir = fs_opendir_unsupported,
.unlink = fs_unlink_unsupported,
.mkdir = fs_mkdir_unsupported,
.ln = fs_ln_unsupported,
},
};
static struct fstype_info *fs_get_info(int fstype)
{
struct fstype_info *info;
int i;
for (i = 0, info = fstypes; i < ARRAY_SIZE(fstypes) - 1; i++, info++) {
if (fstype == info->fstype)
return info;
}
/* Return the 'unsupported' sentinel */
return info;
}
/**
* fs_get_type() - Get type of current filesystem
*
* Return: filesystem type
*
* Returns filesystem type representing the current filesystem, or
* FS_TYPE_ANY for any unrecognised filesystem.
*/
int fs_get_type(void)
{
return fs_type;
}
/**
* fs_get_type_name() - Get type of current filesystem
*
* Return: Pointer to filesystem name
*
* Returns a string describing the current filesystem, or the sentinel
* "unsupported" for any unrecognised filesystem.
*/
const char *fs_get_type_name(void)
{
return fs_get_info(fs_type)->name;
}
int fs_set_blk_dev(const char *ifname, const char *dev_part_str, int fstype)
{
struct fstype_info *info;
int part, i;
#ifdef CONFIG_NEEDS_MANUAL_RELOC
static int relocated;
if (!relocated) {
for (i = 0, info = fstypes; i < ARRAY_SIZE(fstypes);
i++, info++) {
info->name += gd->reloc_off;
info->probe += gd->reloc_off;
info->close += gd->reloc_off;
info->ls += gd->reloc_off;
info->read += gd->reloc_off;
info->write += gd->reloc_off;
}
relocated = 1;
}
#endif
part = blk_get_device_part_str(ifname, dev_part_str, &fs_dev_desc,
&fs_partition, 1);
if (part < 0)
return -1;
for (i = 0, info = fstypes; i < ARRAY_SIZE(fstypes); i++, info++) {
if (fstype != FS_TYPE_ANY && info->fstype != FS_TYPE_ANY &&
fstype != info->fstype)
continue;
if (!fs_dev_desc && !info->null_dev_desc_ok)
continue;
if (!info->probe(fs_dev_desc, &fs_partition)) {
fs_type = info->fstype;
fs_dev_part = part;
return 0;
}
}
return -1;
}
/* set current blk device w/ blk_desc + partition # */
int fs_set_blk_dev_with_part(struct blk_desc *desc, int part)
{
struct fstype_info *info;
int ret, i;
if (part >= 1)
ret = part_get_info(desc, part, &fs_partition);
else
ret = part_get_info_whole_disk(desc, &fs_partition);
if (ret)
return ret;
fs_dev_desc = desc;
for (i = 0, info = fstypes; i < ARRAY_SIZE(fstypes); i++, info++) {
if (!info->probe(fs_dev_desc, &fs_partition)) {
fs_type = info->fstype;
fs_dev_part = part;
return 0;
}
}
return -1;
}
void fs_close(void)
{
struct fstype_info *info = fs_get_info(fs_type);
info->close();
fs_type = FS_TYPE_ANY;
}
int fs_uuid(char *uuid_str)
{
struct fstype_info *info = fs_get_info(fs_type);
return info->uuid(uuid_str);
}
int fs_ls(const char *dirname)
{
int ret;
struct fstype_info *info = fs_get_info(fs_type);
ret = info->ls(dirname);
fs_close();
return ret;
}
int fs_exists(const char *filename)
{
int ret;
struct fstype_info *info = fs_get_info(fs_type);
ret = info->exists(filename);
fs_close();
return ret;
}
int fs_size(const char *filename, loff_t *size)
{
int ret;
struct fstype_info *info = fs_get_info(fs_type);
ret = info->size(filename, size);
fs_close();
return ret;
}
#ifdef CONFIG_LMB
/* Check if a file may be read to the given address */
static int fs_read_lmb_check(const char *filename, ulong addr, loff_t offset,
loff_t len, struct fstype_info *info)
{
struct lmb lmb;
int ret;
loff_t size;
loff_t read_len;
/* get the actual size of the file */
ret = info->size(filename, &size);
if (ret)
return ret;
if (offset >= size) {
/* offset >= EOF, no bytes will be written */
return 0;
}
read_len = size - offset;
/* limit to 'len' if it is smaller */
if (len && len < read_len)
read_len = len;
lmb_init_and_reserve(&lmb, gd->bd, (void *)gd->fdt_blob);
lmb_dump_all(&lmb);
if (lmb_alloc_addr(&lmb, addr, read_len) == addr)
return 0;
log_err("** Reading file would overwrite reserved memory **\n");
return -ENOSPC;
}
#endif
static int _fs_read(const char *filename, ulong addr, loff_t offset, loff_t len,
int do_lmb_check, loff_t *actread)
{
struct fstype_info *info = fs_get_info(fs_type);
void *buf;
int ret;
#ifdef CONFIG_LMB
if (do_lmb_check) {
ret = fs_read_lmb_check(filename, addr, offset, len, info);
if (ret)
return ret;
}
#endif
/*
* We don't actually know how many bytes are being read, since len==0
* means read the whole file.
*/
buf = map_sysmem(addr, len);
ret = info->read(filename, buf, offset, len, actread);
unmap_sysmem(buf);
/* If we requested a specific number of bytes, check we got it */
if (ret == 0 && len && *actread != len)
log_debug("** %s shorter than offset + len **\n", filename);
fs_close();
return ret;
}
int fs_read(const char *filename, ulong addr, loff_t offset, loff_t len,
loff_t *actread)
{
return _fs_read(filename, addr, offset, len, 0, actread);
}
int fs_write(const char *filename, ulong addr, loff_t offset, loff_t len,
loff_t *actwrite)
{
struct fstype_info *info = fs_get_info(fs_type);
void *buf;
int ret;
buf = map_sysmem(addr, len);
ret = info->write(filename, buf, offset, len, actwrite);
unmap_sysmem(buf);
if (ret < 0 && len != *actwrite) {
log_err("** Unable to write file %s **\n", filename);
ret = -1;
}
fs_close();
return ret;
}
struct fs_dir_stream *fs_opendir(const char *filename)
{
struct fstype_info *info = fs_get_info(fs_type);
struct fs_dir_stream *dirs = NULL;
int ret;
ret = info->opendir(filename, &dirs);
fs_close();
if (ret) {
errno = -ret;
return NULL;
}
dirs->desc = fs_dev_desc;
dirs->part = fs_dev_part;
return dirs;
}
struct fs_dirent *fs_readdir(struct fs_dir_stream *dirs)
{
struct fstype_info *info;
struct fs_dirent *dirent;
int ret;
fs_set_blk_dev_with_part(dirs->desc, dirs->part);
info = fs_get_info(fs_type);
ret = info->readdir(dirs, &dirent);
fs_close();
if (ret) {
errno = -ret;
return NULL;
}
return dirent;
}
void fs_closedir(struct fs_dir_stream *dirs)
{
struct fstype_info *info;
if (!dirs)
return;
fs_set_blk_dev_with_part(dirs->desc, dirs->part);
info = fs_get_info(fs_type);
info->closedir(dirs);
fs_close();
}
int fs_unlink(const char *filename)
{
int ret;
struct fstype_info *info = fs_get_info(fs_type);
ret = info->unlink(filename);
fs_close();
return ret;
}
int fs_mkdir(const char *dirname)
{
int ret;
struct fstype_info *info = fs_get_info(fs_type);
ret = info->mkdir(dirname);
fs_close();
return ret;
}
int fs_ln(const char *fname, const char *target)
{
struct fstype_info *info = fs_get_info(fs_type);
int ret;
ret = info->ln(fname, target);
if (ret < 0) {
log_err("** Unable to create link %s -> %s **\n", fname, target);
ret = -1;
}
fs_close();
return ret;
}
int do_size(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
loff_t size;
if (argc != 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
if (fs_size(argv[3], &size) < 0)
return CMD_RET_FAILURE;
env_set_hex("filesize", size);
return 0;
}
int do_load(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
unsigned long addr;
const char *addr_str;
const char *filename;
loff_t bytes;
loff_t pos;
loff_t len_read;
int ret;
unsigned long time;
char *ep;
if (argc < 2)
return CMD_RET_USAGE;
if (argc > 7)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], (argc >= 3) ? argv[2] : NULL, fstype))
return 1;
if (argc >= 4) {
addr = simple_strtoul(argv[3], &ep, 16);
if (ep == argv[3] || *ep != '\0')
return CMD_RET_USAGE;
} else {
addr_str = env_get("loadaddr");
if (addr_str != NULL)
addr = simple_strtoul(addr_str, NULL, 16);
else
addr = CONFIG_SYS_LOAD_ADDR;
}
if (argc >= 5) {
filename = argv[4];
} else {
filename = env_get("bootfile");
if (!filename) {
puts("** No boot file defined **\n");
return 1;
}
}
if (argc >= 6)
bytes = simple_strtoul(argv[5], NULL, 16);
else
bytes = 0;
if (argc >= 7)
pos = simple_strtoul(argv[6], NULL, 16);
else
pos = 0;
time = get_timer(0);
ret = _fs_read(filename, addr, pos, bytes, 1, &len_read);
time = get_timer(time);
if (ret < 0) {
log_err("Failed to load '%s'\n", filename);
return 1;
}
if (IS_ENABLED(CONFIG_CMD_BOOTEFI))
efi_set_bootdev(argv[1], (argc > 2) ? argv[2] : "",
(argc > 4) ? argv[4] : "", map_sysmem(addr, 0),
len_read);
printf("%llu bytes read in %lu ms", len_read, time);
if (time > 0) {
puts(" (");
print_size(div_u64(len_read, time) * 1000, "/s");
puts(")");
}
puts("\n");
env_set_hex("fileaddr", addr);
env_set_hex("filesize", len_read);
return 0;
}
int do_ls(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
if (argc < 2)
return CMD_RET_USAGE;
if (argc > 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], (argc >= 3) ? argv[2] : NULL, fstype))
return 1;
if (fs_ls(argc >= 4 ? argv[3] : "/"))
return 1;
return 0;
}
int file_exists(const char *dev_type, const char *dev_part, const char *file,
int fstype)
{
if (fs_set_blk_dev(dev_type, dev_part, fstype))
return 0;
return fs_exists(file);
}
int do_save(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
unsigned long addr;
const char *filename;
loff_t bytes;
loff_t pos;
loff_t len;
int ret;
unsigned long time;
if (argc < 6 || argc > 7)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
addr = simple_strtoul(argv[3], NULL, 16);
filename = argv[4];
bytes = simple_strtoul(argv[5], NULL, 16);
if (argc >= 7)
pos = simple_strtoul(argv[6], NULL, 16);
else
pos = 0;
time = get_timer(0);
ret = fs_write(filename, addr, pos, bytes, &len);
time = get_timer(time);
if (ret < 0)
return 1;
printf("%llu bytes written in %lu ms", len, time);
if (time > 0) {
puts(" (");
print_size(div_u64(len, time) * 1000, "/s");
puts(")");
}
puts("\n");
return 0;
}
int do_fs_uuid(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
int ret;
char uuid[37];
memset(uuid, 0, sizeof(uuid));
if (argc < 3 || argc > 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
ret = fs_uuid(uuid);
if (ret)
return CMD_RET_FAILURE;
if (argc == 4)
env_set(argv[3], uuid);
else
printf("%s\n", uuid);
return CMD_RET_SUCCESS;
}
int do_fs_type(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[])
{
struct fstype_info *info;
if (argc < 3 || argc > 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], FS_TYPE_ANY))
return 1;
info = fs_get_info(fs_type);
if (argc == 4)
env_set(argv[3], info->name);
else
printf("%s\n", info->name);
fs_close();
return CMD_RET_SUCCESS;
}
int do_rm(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
if (argc != 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
if (fs_unlink(argv[3]))
return 1;
return 0;
}
int do_mkdir(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
int ret;
if (argc != 4)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
ret = fs_mkdir(argv[3]);
if (ret) {
log_err("** Unable to create a directory \"%s\" **\n", argv[3]);
return 1;
}
return 0;
}
int do_ln(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[],
int fstype)
{
if (argc != 5)
return CMD_RET_USAGE;
if (fs_set_blk_dev(argv[1], argv[2], fstype))
return 1;
if (fs_ln(argv[3], argv[4]))
return 1;
return 0;
}
int do_fs_types(struct cmd_tbl *cmdtp, int flag, int argc, char * const argv[])
{
struct fstype_info *drv = fstypes;
const int n_ents = ARRAY_SIZE(fstypes);
struct fstype_info *entry;
int i = 0;
puts("Supported filesystems");
for (entry = drv; entry != drv + n_ents; entry++) {
if (entry->fstype != FS_TYPE_ANY) {
printf("%c %s", i ? ',' : ':', entry->name);
i++;
}
}
if (!i)
puts(": <none>");
puts("\n");
return CMD_RET_SUCCESS;
}