These two patches re-work pygrub to use a generic C library for reading
filesystem images. An immediate benefit is that there's no longer any
external dependencies, so all the filesystems always 'just work'.
Slightly longer term, providing a stable C API for filesystem support
adds a considerable amount of future proofing. It's easy for FS
developers to add support for their FS in a stable manner. Additionally,
it's worth noting that the facility provided by libfsimage is a generic
one, and could prove useful in circumstances outside pygrub.
It also adds Solaris UFS support. The current plugins are all based on
slightly modified Grub legacy code (so far it's taken about an hour to
get a new FS from the grub code working).
Ideally, this library would be provided by the OS vendors themselves, as
there's nothing Xen-specific to it, but for now I've provided it as a
patch to xen-unstable.
Comments?
regards,
john
# HG changeset patch
# User john.levon@xxxxxxx
# Date 1161961324 25200
# Node ID 5ff4ddea38e2cd18a10cb66040879ab7eca6797a
# Parent eff2bb014e46dd9745ff30e43a5c35cf908d8c58
Add libfsimage, a C library for reading files from filesystem images. Initial
support is provided for Solaris UFS, ext2, and reiserfs.
Signed-off-by: John Levon <john.levon@xxxxxxx>
diff --git a/tools/Makefile b/tools/Makefile
--- a/tools/Makefile
+++ b/tools/Makefile
@@ -18,6 +18,7 @@ SUBDIRS-y += xenstat
SUBDIRS-y += xenstat
SUBDIRS-$(CONFIG_Linux) += libaio
SUBDIRS-$(CONFIG_Linux) += blktap
+SUBDIRS-y += libfsimage
# These don't cross-compile
ifeq ($(XEN_COMPILE_ARCH),$(XEN_TARGET_ARCH))
diff --git a/tools/libfsimage/Makefile b/tools/libfsimage/Makefile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/Makefile
@@ -0,0 +1,12 @@
+XEN_ROOT = ../..
+include $(XEN_ROOT)/tools/Rules.mk
+
+SUBDIRS-y = common ufs ext2fs reiserfs
+
+.PHONY: all
+all install clean:
+ @set -e; for subdir in $(SUBDIRS-y); do \
+ $(MAKE) -C $$subdir $@; \
+ done
+
+distclean: clean
diff --git a/tools/libfsimage/Rules.mk b/tools/libfsimage/Rules.mk
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/Rules.mk
@@ -0,0 +1,32 @@
+include $(XEN_ROOT)/tools/Rules.mk
+
+DEPS = .*.d
+
+CFLAGS += -I$(XEN_ROOT)/tools/libfsimage/common/ -Werror -D_LARGEFILE_SOURCE
-D_FILE_OFFSET_BITS=64 -Wp,-MD,.$(@F).d
+LDFLAGS += -L../common/
+
+PIC_OBJS := $(patsubst %.c,%.opic,$(LIB_SRCS-y))
+
+FSDIR-$(CONFIG_Linux) = $(LIBDIR)/fs/$(FS)
+FSDIR-$(CONFIG_SunOS)-x86_64 = lib/fs/$(FS)/64
+FSDIR-$(CONFIG_SunOS)-x86_32 = lib/fs/$(FS)/
+FSDIR-$(CONFIG_SunOS) = $(FSDIR-$(CONFIG_SunOS)-$(XEN_TARGET_ARCH))
+FSDIR = $(FSDIR-y)
+
+FSLIB = fsimage.so
+
+.PHONY: fs-all
+fs-all: $(FSLIB)
+
+.PHONY: fs-install
+fs-install: fs-all
+ $(INSTALL_DIR) $(DESTDIR)/usr/$(FSDIR)
+ $(INSTALL_PROG) $(FSLIB) $(DESTDIR)/usr/$(FSDIR)
+
+$(FSLIB): $(PIC_OBJS)
+ $(CC) $(CFLAGS) $(LDFLAGS) $(SHLIB_CFLAGS) -o $@ $^ -lfsimage
+
+clean distclean:
+ rm -f $(PIC_OBJS) $(FSLIB)
+
+-include $(DEPS)
diff --git a/tools/libfsimage/common/Makefile b/tools/libfsimage/common/Makefile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/Makefile
@@ -0,0 +1,45 @@
+XEN_ROOT = ../../..
+include $(XEN_ROOT)/tools/Rules.mk
+
+MAJOR = 1.0
+MINOR = 0
+
+CFLAGS += -Werror -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -Wp,-MD,.$(@F).d
+DEPS = .*.d
+
+LDFLAGS-$(CONFIG_SunOS) = -Wl,-M -Wl,mapfile
+LDFLAGS = $(LDFLAGS-y)
+
+LIB_SRCS-y = fsimage.c fsimage_plugin.c fsimage_grub.c
+
+PIC_OBJS := $(patsubst %.c,%.opic,$(LIB_SRCS-y))
+
+LIB = libfsimage.so libfsimage.so.$(MAJOR) libfsimage.so.$(MAJOR).$(MINOR)
+
+.PHONY: all
+all: $(LIB)
+
+.PHONY: install
+install: all
+ [ -d $(DESTDIR)/usr/$(LIBDIR) ] || $(INSTALL_DIR)
$(DESTDIR)/usr/$(LIBDIR)
+ [ -d $(DESTDIR)/usr/include ] || $(INSTALL_DIR) $(DESTDIR)/usr/include
+ $(INSTALL_PROG) libfsimage.so.$(MAJOR).$(MINOR) $(DESTDIR)/usr/$(LIBDIR)
+ ln -sf libfsimage.so.$(MAJOR).$(MINOR)
$(DESTDIR)/usr/$(LIBDIR)/libfsimage.so.$(MAJOR)
+ ln -sf libfsimage.so.$(MAJOR) $(DESTDIR)/usr/$(LIBDIR)/libfsimage.so
+ $(INSTALL_DATA) fsimage.h $(DESTDIR)/usr/include
+ $(INSTALL_DATA) fsimage_plugin.h $(DESTDIR)/usr/include
+ $(INSTALL_DATA) fsimage_grub.h $(DESTDIR)/usr/include
+
+clean distclean:
+ rm -f $(PIC_OBJS) $(LIB)
+
+libfsimage.so: libfsimage.so.$(MAJOR)
+ ln -sf $< $@
+libfsimage.so.$(MAJOR): libfsimage.so.$(MAJOR).$(MINOR)
+ ln -sf $< $@
+
+libfsimage.so.$(MAJOR).$(MINOR): $(PIC_OBJS)
+ $(CC) $(CFLAGS) $(LDFLAGS) -Wl,$(SONAME_LDFLAG)
-Wl,libfsimage.so.$(MAJOR) $(SHLIB_CFLAGS) -o $@ $^ -lpthread
+
+-include $(DEPS)
+
diff --git a/tools/libfsimage/common/fsimage.c
b/tools/libfsimage/common/fsimage.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage.c
@@ -0,0 +1,140 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <strings.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <pthread.h>
+
+#include "fsimage_plugin.h"
+#include "fsimage_priv.h"
+
+static pthread_mutex_t fsi_lock = PTHREAD_MUTEX_INITIALIZER;
+
+fsi_t *fsi_open_fsimage(const char *path, uint64_t off)
+{
+ fsi_t *fsi = NULL;
+ int fd;
+ int err;
+
+ if ((fd = open(path, O_RDONLY)) == -1)
+ goto fail;
+
+ if ((fsi = malloc(sizeof(*fsi))) == NULL)
+ goto fail;
+
+ fsi->f_fd = fd;
+ fsi->f_off = off;
+ fsi->f_data = NULL;
+
+ pthread_mutex_lock(&fsi_lock);
+ err = find_plugin(fsi);
+ pthread_mutex_unlock(&fsi_lock);
+ if (err != 0)
+ goto fail;
+
+ return (fsi);
+
+fail:
+ err = errno;
+ if (fd != -1)
+ (void) close(fd);
+ free(fsi);
+ errno = err;
+ return (NULL);
+}
+
+void fsi_close_fsimage(fsi_t *fsi)
+{
+ pthread_mutex_lock(&fsi_lock);
+ fsip_fs_free(fsi);
+ pthread_mutex_unlock(&fsi_lock);
+}
+
+int fsi_file_exists(fsi_t *fsi, const char *path)
+{
+ fsi_file_t *ffi;
+
+ if ((ffi = fsi_open_file(fsi, path)) == NULL)
+ return (0);
+
+ fsi_close_file(ffi);
+ return (1);
+}
+
+fsi_file_t *fsi_open_file(fsi_t *fsi, const char *path)
+{
+ fsi_plugin_ops_t *ops;
+ fsi_file_t *ffi;
+
+ pthread_mutex_lock(&fsi_lock);
+ ops = fsi->f_plugin->fp_ops;
+ ffi = ops->fpo_open(fsi, path);
+ pthread_mutex_unlock(&fsi_lock);
+
+ return (ffi);
+}
+
+int fsi_close_file(fsi_file_t *ffi)
+{
+ fsi_plugin_ops_t *ops;
+ int err;
+
+ pthread_mutex_lock(&fsi_lock);
+ ops = ffi->ff_fsi->f_plugin->fp_ops;
+ err = ops->fpo_close(ffi);
+ pthread_mutex_unlock(&fsi_lock);
+
+ return (err);
+}
+
+ssize_t fsi_read_file(fsi_file_t *ffi, void *buf, size_t nbytes)
+{
+ fsi_plugin_ops_t *ops;
+ ssize_t ret;
+
+ pthread_mutex_lock(&fsi_lock);
+ ops = ffi->ff_fsi->f_plugin->fp_ops;
+ ret = ops->fpo_read(ffi, buf, nbytes);
+ pthread_mutex_unlock(&fsi_lock);
+
+ return (ret);
+}
+
+ssize_t fsi_pread_file(fsi_file_t *ffi, void *buf, size_t nbytes, uint64_t off)
+{
+ fsi_plugin_ops_t *ops;
+ ssize_t ret;
+
+ pthread_mutex_lock(&fsi_lock);
+ ops = ffi->ff_fsi->f_plugin->fp_ops;
+ ret = ops->fpo_pread(ffi, buf, nbytes, off);
+ pthread_mutex_unlock(&fsi_lock);
+
+ return (ret);
+}
diff --git a/tools/libfsimage/common/fsimage.h
b/tools/libfsimage/common/fsimage.h
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage.h
@@ -0,0 +1,52 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _FSIMAGE_H
+#define _FSIMAGE_H
+
+#ifdef __cplusplus
+extern C {
+#endif
+
+#include <sys/types.h>
+#include <stdint.h>
+#include <unistd.h>
+
+typedef struct fsi fsi_t;
+typedef struct fsi_file fsi_file_t;
+
+fsi_t *fsi_open_fsimage(const char *, uint64_t);
+void fsi_close_fsimage(fsi_t *);
+
+int fsi_file_exists(fsi_t *, const char *);
+fsi_file_t *fsi_open_file(fsi_t *, const char *);
+int fsi_close_file(fsi_file_t *);
+
+ssize_t fsi_read_file(fsi_file_t *, void *, size_t);
+ssize_t fsi_pread_file(fsi_file_t *, void *, size_t, uint64_t);
+
+#ifdef __cplusplus
+};
+#endif
+
+#endif /* _FSIMAGE_H */
diff --git a/tools/libfsimage/common/fsimage_grub.c
b/tools/libfsimage/common/fsimage_grub.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage_grub.c
@@ -0,0 +1,277 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef __sun__
+#define _XOPEN_SOURCE 500
+#endif
+#include <stdlib.h>
+#include <strings.h>
+#include <errno.h>
+
+#include "fsimage_grub.h"
+#include "fsimage_priv.h"
+
+static char *disk_read_junk;
+
+typedef struct fsig_data {
+ char fd_buf[FSYS_BUFLEN];
+} fsig_data_t;
+
+typedef struct fsig_file_data {
+ char ffd_buf[FSYS_BUFLEN];
+ uint64_t ffd_curpos;
+ uint64_t ffd_filepos;
+ uint64_t ffd_filemax;
+ int ffd_int1;
+ int ffd_int2;
+ int ffd_errnum;
+} fsig_file_data_t;
+
+fsi_file_t *
+fsig_file_alloc(fsi_t *fsi)
+{
+ fsi_file_t *ffi;
+ fsig_file_data_t *data = malloc(sizeof (fsig_file_data_t));
+
+ if (data == NULL)
+ return (NULL);
+
+ bzero(data, sizeof (fsig_file_data_t));
+ bcopy(fsig_fs_buf(fsi), data->ffd_buf, FSYS_BUFLEN);
+
+ if ((ffi = fsip_file_alloc(fsi, data)) == NULL) {
+ free(data);
+ return (NULL);
+ }
+
+ return (ffi);
+}
+
+void *
+fsig_fs_buf(fsi_t *fsi)
+{
+ fsig_data_t *data = fsip_fs_data(fsi);
+ return ((void *)data->fd_buf);
+}
+
+void *
+fsig_file_buf(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return ((void *)data->ffd_buf);
+}
+
+uint64_t *
+fsig_filepos(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return (&data->ffd_filepos);
+}
+
+uint64_t *
+fsig_filemax(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return (&data->ffd_filemax);
+}
+
+int *
+fsig_int1(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return (&data->ffd_int1);
+}
+
+int *
+fsig_int2(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return (&data->ffd_int2);
+}
+
+int *
+fsig_errnum(fsi_file_t *ffi)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ return (&data->ffd_errnum);
+}
+
+char **
+fsig_disk_read_junk(void)
+{
+ return (&disk_read_junk);
+}
+
+int
+fsig_devread(fsi_file_t *ffi, unsigned int sector, unsigned int offset,
+ unsigned int bufsize, char *buf)
+{
+ uint64_t off = ffi->ff_fsi->f_off + ((uint64_t)(sector * 512)) + offset;
+ ssize_t bytes_read = 0;
+
+ while (bufsize) {
+ ssize_t ret = pread(ffi->ff_fsi->f_fd, buf + bytes_read,
+ bufsize, (off_t)off);
+ if (ret == -1)
+ return (0);
+ if (ret == 0)
+ return (0);
+
+ bytes_read += ret;
+ bufsize -= ret;
+ }
+
+ return (1);
+}
+
+int
+fsig_substring(const char *s1, const char *s2)
+{
+ while (*s1 == *s2) {
+ if (*s1 == '\0')
+ return (0);
+ s1++;
+ s2++;
+ }
+
+ if (*s1 == '\0')
+ return (-1);
+
+ return (1);
+}
+
+static int
+fsig_mount(fsi_t *fsi)
+{
+ fsig_plugin_ops_t *ops = fsi->f_plugin->fp_data;
+ fsi_file_t *ffi;
+ fsi->f_data = malloc(sizeof (fsig_data_t));
+
+ if (fsi->f_data == NULL)
+ return (-1);
+
+ if ((ffi = fsig_file_alloc(fsi)) == NULL) {
+ free(fsi->f_data);
+ fsi->f_data = NULL;
+ return (-1);
+ }
+
+ bzero(fsi->f_data, sizeof (fsig_data_t));
+
+ if (!ops->fpo_mount(ffi)) {
+ fsip_file_free(ffi);
+ free(fsi->f_data);
+ fsi->f_data = NULL;
+ return (-1);
+ }
+
+ bcopy(fsig_file_buf(ffi), fsig_fs_buf(fsi), FSYS_BUFLEN);
+ fsip_file_free(ffi);
+ return (0);
+}
+
+static int
+fsig_umount(fsi_t *fsi)
+{
+ fsip_fs_free(fsi);
+ return (0);
+}
+
+static fsi_file_t *
+fsig_open(fsi_t *fsi, const char *name)
+{
+ fsig_plugin_ops_t *ops = fsi->f_plugin->fp_data;
+ char *path = strdup(name);
+ fsi_file_t *ffi = NULL;
+
+ if (path == NULL || (ffi = fsig_file_alloc(fsi)) == NULL)
+ goto out;
+
+ if (ops->fpo_dir(ffi, path) == 0) {
+ fsip_file_free(ffi);
+ ffi = NULL;
+ errno = ENOENT;
+ }
+
+out:
+ free(path);
+ return (ffi);
+}
+
+static ssize_t
+fsig_pread(fsi_file_t *ffi, void *buf, size_t nbytes, uint64_t off)
+{
+ fsig_plugin_ops_t *ops = ffi->ff_fsi->f_plugin->fp_data;
+ fsig_file_data_t *data = fsip_file_data(ffi);
+
+ data->ffd_filepos = off;
+
+ if (data->ffd_filepos >= data->ffd_filemax)
+ return (0);
+
+ /* FIXME: check */
+ if (data->ffd_filepos + nbytes > data->ffd_filemax)
+ nbytes = data->ffd_filemax - data->ffd_filepos;
+
+ errnum = 0;
+ return (ops->fpo_read(ffi, buf, nbytes));
+}
+
+static ssize_t
+fsig_read(fsi_file_t *ffi, void *buf, size_t nbytes)
+{
+ fsig_file_data_t *data = fsip_file_data(ffi);
+ ssize_t ret;
+
+ ret = fsig_pread(ffi, buf, nbytes, data->ffd_curpos);
+ data->ffd_curpos = data->ffd_filepos;
+ return (ret);
+}
+
+static int
+fsig_close(fsi_file_t *ffi)
+{
+ fsip_file_free(ffi);
+ return (0);
+}
+
+static fsi_plugin_ops_t fsig_grub_ops = {
+ .fpo_version = FSIMAGE_PLUGIN_VERSION,
+ .fpo_mount = fsig_mount,
+ .fpo_umount = fsig_umount,
+ .fpo_open = fsig_open,
+ .fpo_read = fsig_read,
+ .fpo_pread = fsig_pread,
+ .fpo_close = fsig_close
+};
+
+fsi_plugin_ops_t *
+fsig_init(fsi_plugin_t *plugin, fsig_plugin_ops_t *ops)
+{
+ if (ops->fpo_version > FSIMAGE_PLUGIN_VERSION)
+ return (NULL);
+
+ plugin->fp_data = ops;
+
+ return (&fsig_grub_ops);
+}
diff --git a/tools/libfsimage/common/fsimage_grub.h
b/tools/libfsimage/common/fsimage_grub.h
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage_grub.h
@@ -0,0 +1,92 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _FSIMAGE_GRUB_H
+#define _FSIMAGE_GRUB_H
+
+#ifdef __cplusplus
+extern C {
+#endif
+
+#include <sys/types.h>
+#include <string.h>
+#include <stdlib.h>
+#include <ctype.h>
+
+#include "fsimage.h"
+#include "fsimage_plugin.h"
+
+typedef struct fsig_plugin_ops {
+ int fpo_version;
+ int (*fpo_mount)(fsi_file_t *);
+ int (*fpo_dir)(fsi_file_t *, char *);
+ int (*fpo_read)(fsi_file_t *, char *, int);
+} fsig_plugin_ops_t;
+
+#define STAGE1_5
+#define FSYS_BUFLEN 0x8000
+#define SECTOR_BITS 9
+#define SECTOR_SIZE 0x200
+
+#define FSYS_BUF (fsig_file_buf(ffi))
+#define filepos (*fsig_filepos(ffi))
+#define filemax (*fsig_filemax(ffi))
+#define devread fsig_devread
+#define substring fsig_substring
+#define errnum (*fsig_errnum(ffi))
+#define disk_read_func (*fsig_disk_read_junk())
+#define disk_read_hook (*fsig_disk_read_junk())
+#define print_possibilities 0
+
+#define grub_memset memset
+#define grub_memmove memmove
+
+extern char **fsig_disk_read_junk(void);
+
+#define ERR_FSYS_CORRUPT 1
+#define ERR_SYMLINK_LOOP 1
+#define ERR_FILELENGTH 1
+#define ERR_BAD_FILETYPE 1
+#define ERR_BAD_FILETYPE 1
+#define ERR_FILE_NOT_FOUND 1
+
+fsi_plugin_ops_t *fsig_init(fsi_plugin_t *, fsig_plugin_ops_t *);
+
+int fsig_devread(fsi_file_t *, unsigned int, unsigned int, unsigned int, char
*);
+int fsig_substring(const char *, const char *);
+
+void *fsig_fs_buf(fsi_t *);
+
+fsi_file_t *fsig_file_alloc(fsi_t *);
+void *fsig_file_buf(fsi_file_t *);
+uint64_t *fsig_filepos(fsi_file_t *);
+uint64_t *fsig_filemax(fsi_file_t *);
+int *fsig_int1(fsi_file_t *);
+int *fsig_int2(fsi_file_t *);
+int *fsig_errnum(fsi_file_t *);
+
+#ifdef __cplusplus
+};
+#endif
+
+#endif /* _FSIMAGE_GRUB_H */
diff --git a/tools/libfsimage/common/fsimage_plugin.c
b/tools/libfsimage/common/fsimage_plugin.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage_plugin.c
@@ -0,0 +1,208 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+#include <dirent.h>
+#include <dlfcn.h>
+#include <errno.h>
+
+#include "fsimage_plugin.h"
+#include "fsimage_priv.h"
+
+static fsi_plugin_t *plugins;
+
+void
+fsip_fs_set_data(fsi_t *fsi, void *data)
+{
+ fsi->f_data = data;
+}
+
+void
+fsip_fs_free(fsi_t *fsi)
+{
+ free(fsi->f_data);
+ free(fsi);
+}
+
+fsi_file_t *
+fsip_file_alloc(fsi_t *fsi, void *data)
+{
+ fsi_file_t *ffi = malloc(sizeof (fsi_file_t));
+ if (ffi == NULL)
+ return (NULL);
+
+ bzero(ffi, sizeof (fsi_file_t));
+
+ ffi->ff_fsi = fsi;
+ ffi->ff_data = data;
+ return (ffi);
+}
+
+void
+fsip_file_free(fsi_file_t *ffi)
+{
+ free(ffi->ff_data);
+ free(ffi);
+}
+
+fsi_t *
+fsip_fs(fsi_file_t *ffi)
+{
+ return (ffi->ff_fsi);
+}
+
+void *
+fsip_fs_data(fsi_t *fsi)
+{
+ return (fsi->f_data);
+}
+
+void *
+fsip_file_data(fsi_file_t *ffi)
+{
+ return (ffi->ff_data);
+}
+
+static int init_plugin(const char *lib)
+{
+ fsi_plugin_init_t init;
+ fsi_plugin_t *fp = malloc(sizeof (fsi_plugin_t));
+
+ if (fp == NULL)
+ return (-1);
+
+ bzero(fp, sizeof (fsi_plugin_t));
+
+ if ((fp->fp_dlh = dlopen(lib, RTLD_LAZY | RTLD_LOCAL)) == NULL) {
+ free(fp);
+ return (0);
+ }
+
+ init = dlsym(fp->fp_dlh, "fsi_init_plugin");
+
+ if (init == NULL)
+ goto fail;
+
+ fp->fp_ops = init(FSIMAGE_PLUGIN_VERSION, fp, &fp->fp_name);
+ if (fp->fp_ops == NULL ||
+ fp->fp_ops->fpo_version > FSIMAGE_PLUGIN_VERSION)
+ goto fail;
+
+ fp->fp_next = plugins;
+ plugins = fp;
+
+ return (0);
+fail:
+ (void) dlclose(fp->fp_dlh);
+ free(fp);
+ return (-1);
+}
+
+static int load_plugins(void)
+{
+ const char *fsdir = getenv("FSIMAGE_FSDIR");
+ const char *isadir = "";
+ struct dirent *dp = NULL;
+ struct dirent *dpp;
+ DIR *dir = NULL;
+ char *tmp = NULL;
+ size_t name_max;
+ int err;
+ int ret = -1;
+
+#ifdef __sun__
+ if (fsdir == NULL)
+ fsdir = "/usr/lib/fs";
+
+ if (sizeof(void *) == 8)
+ isadir = "64/";
+#else
+ if (fsdir == NULL) {
+ if (sizeof(void *) == 8)
+ fsdir = "/usr/lib64/fs";
+ else
+ fsdir = "/usr/lib/fs";
+ }
+#endif
+
+ if ((name_max = pathconf(fsdir, _PC_NAME_MAX)) == -1)
+ goto fail;
+
+ if ((tmp = malloc(name_max + 1)) == NULL)
+ goto fail;
+
+ if ((dp = malloc(sizeof (struct dirent) + name_max + 1)) == NULL)
+ goto fail;
+
+ if ((dir = opendir(fsdir)) == NULL)
+ goto fail;
+
+ bzero(dp, sizeof (struct dirent) + name_max + 1);
+
+ while (readdir_r(dir, dp, &dpp) == 0 && dpp != NULL) {
+ if (strcmp(dpp->d_name, ".") == 0)
+ continue;
+ if (strcmp(dpp->d_name, "..") == 0)
+ continue;
+
+ (void) snprintf(tmp, name_max, "%s/%s/%sfsimage.so", fsdir,
+ dpp->d_name, isadir);
+
+ if (init_plugin(tmp) != 0)
+ goto fail;
+ }
+
+ ret = 0;
+
+fail:
+ err = errno;
+ if (dir != NULL)
+ (void) closedir(dir);
+ free(tmp);
+ free(dp);
+ errno = err;
+ return (ret);
+}
+
+int find_plugin(fsi_t *fsi)
+{
+ fsi_plugin_t *fp;
+ int ret = 0;
+
+ if (plugins == NULL && (ret = load_plugins()) != 0)
+ goto out;
+
+ for (fp = plugins; fp != NULL; fp = fp->fp_next) {
+ fsi->f_plugin = fp;
+ if (fp->fp_ops->fpo_mount(fsi) == 0)
+ goto out;
+ }
+
+ ret = -1;
+ errno = ENOTSUP;
+out:
+ return (ret);
+}
diff --git a/tools/libfsimage/common/fsimage_plugin.h
b/tools/libfsimage/common/fsimage_plugin.h
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage_plugin.h
@@ -0,0 +1,64 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _FSIMAGE_PLUGIN_H
+#define _FSIMAGE_PLUGIN_H
+
+#ifdef __cplusplus
+extern C {
+#endif
+
+#include <sys/types.h>
+
+#include "fsimage.h"
+
+#define FSIMAGE_PLUGIN_VERSION 1
+
+typedef struct fsi_plugin fsi_plugin_t;
+
+typedef struct fsi_plugin_ops {
+ int fpo_version;
+ int (*fpo_mount)(fsi_t *);
+ int (*fpo_umount)(fsi_t *);
+ fsi_file_t *(*fpo_open)(fsi_t *, const char *);
+ ssize_t (*fpo_read)(fsi_file_t *, void *, size_t);
+ ssize_t (*fpo_pread)(fsi_file_t *, void *, size_t, uint64_t);
+ int (*fpo_close)(fsi_file_t *);
+} fsi_plugin_ops_t;
+
+typedef fsi_plugin_ops_t *
+ (*fsi_plugin_init_t)(int, fsi_plugin_t *, const char **);
+
+void fsip_fs_set_data(fsi_t *, void *);
+void fsip_fs_free(fsi_t *);
+fsi_file_t *fsip_file_alloc(fsi_t *, void *);
+void fsip_file_free(fsi_file_t *);
+fsi_t * fsip_fs(fsi_file_t *ffi);
+void *fsip_fs_data(fsi_t *);
+void *fsip_file_data(fsi_file_t *);
+
+#ifdef __cplusplus
+};
+#endif
+
+#endif /* _FSIMAGE_PLUGIN_H */
diff --git a/tools/libfsimage/common/fsimage_priv.h
b/tools/libfsimage/common/fsimage_priv.h
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/fsimage_priv.h
@@ -0,0 +1,62 @@
+/*
+ * 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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _FSIMAGE_PRIV_H
+#define _FSIMAGE_PRIV_H
+
+#ifdef __cplusplus
+extern C {
+#endif
+
+#include <sys/types.h>
+
+#include "fsimage.h"
+#include "fsimage_plugin.h"
+
+struct fsi_plugin {
+ const char *fp_name;
+ void *fp_dlh;
+ fsi_plugin_ops_t *fp_ops;
+ struct fsi_plugin *fp_next;
+ void *fp_data;
+};
+
+struct fsi {
+ int f_fd;
+ uint64_t f_off;
+ void *f_data;
+ fsi_plugin_t *f_plugin;
+};
+
+struct fsi_file {
+ fsi_t *ff_fsi;
+ void *ff_data;
+};
+
+int find_plugin(fsi_t *);
+
+#ifdef __cplusplus
+};
+#endif
+
+#endif /* _FSIMAGE_PRIV_H */
diff --git a/tools/libfsimage/common/mapfile b/tools/libfsimage/common/mapfile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/common/mapfile
@@ -0,0 +1,34 @@
+libfsimage.so.1.1 {
+ global:
+ fsi_open_fsimage;
+ fsi_close_fsimage;
+ fsi_file_exists;
+ fsi_open_file;
+ fsi_close_file;
+ fsi_read_file;
+ fsi_pread_file;
+
+ fsip_fs_set_data;
+ fsip_fs_free;
+ fsip_file_alloc;
+ fsip_file_free;
+ fsip_fs;
+ fsip_fs_data;
+ fsip_file_data;
+
+ fsig_init;
+ fsig_devread;
+ fsig_substring;
+ fsig_fs_buf;
+ fsig_file_alloc;
+ fsig_file_buf;
+ fsig_filepos;
+ fsig_filemax;
+ fsig_int1;
+ fsig_int2;
+ fsig_errnum;
+ fsig_disk_read_junk;
+
+ local:
+ *;
+};
diff --git a/tools/libfsimage/ext2fs/Makefile b/tools/libfsimage/ext2fs/Makefile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/ext2fs/Makefile
@@ -0,0 +1,13 @@
+XEN_ROOT = ../../..
+
+LIB_SRCS-y = fsys_ext2fs.c
+
+FS = ext2fs
+
+.PHONY: all
+all: fs-all
+
+.PHONY: install
+install: fs-install
+
+include $(XEN_ROOT)/tools/libfsimage/Rules.mk
diff --git a/tools/libfsimage/ext2fs/fsys_ext2fs.c
b/tools/libfsimage/ext2fs/fsys_ext2fs.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/ext2fs/fsys_ext2fs.c
@@ -0,0 +1,804 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 1999, 2001, 2003 Free Software Foundation, Inc.
+ *
+ * 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 of the License, 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 <fsimage_grub.h>
+
+#define mapblock1 (*fsig_int1(ffi))
+#define mapblock2 (*fsig_int2(ffi))
+
+/* sizes are always in bytes, BLOCK values are always in DEV_BSIZE (sectors) */
+#define DEV_BSIZE 512
+
+/* include/linux/fs.h */
+#define BLOCK_SIZE 1024 /* initial block size for superblock
read */
+/* made up, defaults to 1 but can be passed via mount_opts */
+#define WHICH_SUPER 1
+/* kind of from fs/ext2/super.c */
+#define SBLOCK (WHICH_SUPER * BLOCK_SIZE / DEV_BSIZE) /* = 2 */
+
+/* include/asm-i386/types.h */
+typedef __signed__ char __s8;
+typedef unsigned char __u8;
+typedef __signed__ short __s16;
+typedef unsigned short __u16;
+typedef __signed__ int __s32;
+typedef unsigned int __u32;
+
+/*
+ * Constants relative to the data blocks, from ext2_fs.h
+ */
+#define EXT2_NDIR_BLOCKS 12
+#define EXT2_IND_BLOCK EXT2_NDIR_BLOCKS
+#define EXT2_DIND_BLOCK (EXT2_IND_BLOCK + 1)
+#define EXT2_TIND_BLOCK (EXT2_DIND_BLOCK + 1)
+#define EXT2_N_BLOCKS (EXT2_TIND_BLOCK + 1)
+
+/* include/linux/ext2_fs.h */
+struct ext2_super_block
+ {
+ __u32 s_inodes_count; /* Inodes count */
+ __u32 s_blocks_count; /* Blocks count */
+ __u32 s_r_blocks_count; /* Reserved blocks count */
+ __u32 s_free_blocks_count; /* Free blocks count */
+ __u32 s_free_inodes_count; /* Free inodes count */
+ __u32 s_first_data_block; /* First Data Block */
+ __u32 s_log_block_size; /* Block size */
+ __s32 s_log_frag_size; /* Fragment size */
+ __u32 s_blocks_per_group; /* # Blocks per group */
+ __u32 s_frags_per_group; /* # Fragments per group */
+ __u32 s_inodes_per_group; /* # Inodes per group */
+ __u32 s_mtime; /* Mount time */
+ __u32 s_wtime; /* Write time */
+ __u16 s_mnt_count; /* Mount count */
+ __s16 s_max_mnt_count; /* Maximal mount count */
+ __u16 s_magic; /* Magic signature */
+ __u16 s_state; /* File system state */
+ __u16 s_errors; /* Behaviour when detecting errors */
+ __u16 s_pad;
+ __u32 s_lastcheck; /* time of last check */
+ __u32 s_checkinterval; /* max. time between checks */
+ __u32 s_creator_os; /* OS */
+ __u32 s_rev_level; /* Revision level */
+ __u16 s_def_resuid; /* Default uid for reserved blocks */
+ __u16 s_def_resgid; /* Default gid for reserved blocks */
+ __u32 s_reserved[235]; /* Padding to the end of the block */
+ };
+
+struct ext2_group_desc
+ {
+ __u32 bg_block_bitmap; /* Blocks bitmap block */
+ __u32 bg_inode_bitmap; /* Inodes bitmap block */
+ __u32 bg_inode_table; /* Inodes table block */
+ __u16 bg_free_blocks_count; /* Free blocks count */
+ __u16 bg_free_inodes_count; /* Free inodes count */
+ __u16 bg_used_dirs_count; /* Directories count */
+ __u16 bg_pad;
+ __u32 bg_reserved[3];
+ };
+
+struct ext2_inode
+ {
+ __u16 i_mode; /* File mode */
+ __u16 i_uid; /* Owner Uid */
+ __u32 i_size; /* 4: Size in bytes */
+ __u32 i_atime; /* Access time */
+ __u32 i_ctime; /* 12: Creation time */
+ __u32 i_mtime; /* Modification time */
+ __u32 i_dtime; /* 20: Deletion Time */
+ __u16 i_gid; /* Group Id */
+ __u16 i_links_count; /* 24: Links count */
+ __u32 i_blocks; /* Blocks count */
+ __u32 i_flags; /* 32: File flags */
+ union
+ {
+ struct
+ {
+ __u32 l_i_reserved1;
+ }
+ linux1;
+ struct
+ {
+ __u32 h_i_translator;
+ }
+ hurd1;
+ struct
+ {
+ __u32 m_i_reserved1;
+ }
+ masix1;
+ }
+ osd1; /* OS dependent 1 */
+ __u32 i_block[EXT2_N_BLOCKS]; /* 40: Pointers to blocks */
+ __u32 i_version; /* File version (for NFS) */
+ __u32 i_file_acl; /* File ACL */
+ __u32 i_dir_acl; /* Directory ACL */
+ __u32 i_faddr; /* Fragment address */
+ union
+ {
+ struct
+ {
+ __u8 l_i_frag; /* Fragment number */
+ __u8 l_i_fsize; /* Fragment size */
+ __u16 i_pad1;
+ __u32 l_i_reserved2[2];
+ }
+ linux2;
+ struct
+ {
+ __u8 h_i_frag; /* Fragment number */
+ __u8 h_i_fsize; /* Fragment size */
+ __u16 h_i_mode_high;
+ __u16 h_i_uid_high;
+ __u16 h_i_gid_high;
+ __u32 h_i_author;
+ }
+ hurd2;
+ struct
+ {
+ __u8 m_i_frag; /* Fragment number */
+ __u8 m_i_fsize; /* Fragment size */
+ __u16 m_pad1;
+ __u32 m_i_reserved2[2];
+ }
+ masix2;
+ }
+ osd2; /* OS dependent 2 */
+ };
+
+/* linux/limits.h */
+#define NAME_MAX 255 /* # chars in a file name */
+
+/* linux/posix_type.h */
+typedef long linux_off_t;
+
+/* linux/ext2fs.h */
+#define EXT2_NAME_LEN 255
+struct ext2_dir_entry
+ {
+ __u32 inode; /* Inode number */
+ __u16 rec_len; /* Directory entry length */
+ __u8 name_len; /* Name length */
+ __u8 file_type;
+ char name[EXT2_NAME_LEN]; /* File name */
+ };
+
+/* linux/ext2fs.h */
+/*
+ * EXT2_DIR_PAD defines the directory entries boundaries
+ *
+ * NOTE: It must be a multiple of 4
+ */
+#define EXT2_DIR_PAD 4
+#define EXT2_DIR_ROUND (EXT2_DIR_PAD - 1)
+#define EXT2_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT2_DIR_ROUND) & \
+ ~EXT2_DIR_ROUND)
+
+
+/* ext2/super.c */
+#define log2(n) ffz(~(n))
+
+#define EXT2_SUPER_MAGIC 0xEF53 /* include/linux/ext2_fs.h */
+#define EXT2_ROOT_INO 2 /* include/linux/ext2_fs.h */
+#define PATH_MAX 1024 /* include/linux/limits.h */
+#define MAX_LINK_COUNT 5 /* number of symbolic links to follow */
+
+/* made up, these are pointers into FSYS_BUF */
+/* read once, always stays there: */
+#define SUPERBLOCK \
+ ((struct ext2_super_block *)(FSYS_BUF))
+#define GROUP_DESC \
+ ((struct ext2_group_desc *) \
+ ((char *)SUPERBLOCK + sizeof(struct ext2_super_block)))
+#define INODE \
+ ((struct ext2_inode *)((caddr_t)GROUP_DESC + EXT2_BLOCK_SIZE(SUPERBLOCK)))
+#define DATABLOCK1 \
+ ((char *)((caddr_t)INODE + sizeof(struct ext2_inode)))
+#define DATABLOCK2 \
+ ((char *)((caddr_t)DATABLOCK1 + EXT2_BLOCK_SIZE(SUPERBLOCK)))
+
+/* linux/ext2_fs.h */
+#define EXT2_ADDR_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof (__u32))
+#define EXT2_ADDR_PER_BLOCK_BITS(s) (log2(EXT2_ADDR_PER_BLOCK(s)))
+
+/* linux/ext2_fs.h */
+#define EXT2_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
+/* kind of from ext2/super.c */
+#define EXT2_BLOCK_SIZE(s) (1 << EXT2_BLOCK_SIZE_BITS(s))
+/* linux/ext2fs.h */
+#define EXT2_DESC_PER_BLOCK(s) \
+ (EXT2_BLOCK_SIZE(s) / sizeof (struct ext2_group_desc))
+/* linux/stat.h */
+#define S_IFMT 00170000
+#define S_IFLNK 0120000
+#define S_IFREG 0100000
+#define S_IFDIR 0040000
+#define S_ISLNK(m) (((m) & S_IFMT) == S_IFLNK)
+#define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
+#define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
+
+/* include/asm-i386/bitops.h */
+/*
+ * ffz = Find First Zero in word. Undefined if no zero exists,
+ * so code should check against ~0UL first..
+ */
+#ifdef __amd64
+#define BSF "bsfq"
+#else
+#define BSF "bsfl"
+#endif
+static __inline__ unsigned long
+ffz (unsigned long word)
+{
+ __asm__ (BSF " %1,%0"
+: "=r" (word)
+: "r" (~word));
+ return word;
+}
+
+/* check filesystem types and read superblock into memory buffer */
+int
+ext2fs_mount (fsi_file_t *ffi)
+{
+ int retval = 1;
+
+ if (/*(((current_drive & 0x80) || (current_slice != 0))
+ && (current_slice != PC_SLICE_TYPE_EXT2FS)
+ && (current_slice != PC_SLICE_TYPE_LINUX_RAID)
+ && (! IS_PC_SLICE_TYPE_BSD_WITH_FS (current_slice, FS_EXT2FS))
+ && (! IS_PC_SLICE_TYPE_BSD_WITH_FS (current_slice, FS_OTHER)))
+ || part_length < (SBLOCK + (sizeof (struct ext2_super_block) /
DEV_BSIZE))
+ || */ !devread (ffi, SBLOCK, 0, sizeof (struct ext2_super_block),
+ (char *) SUPERBLOCK)
+ || SUPERBLOCK->s_magic != EXT2_SUPER_MAGIC)
+ retval = 0;
+
+ return retval;
+}
+
+/* Takes a file system block number and reads it into BUFFER. */
+static int
+ext2_rdfsb (fsi_file_t *ffi, int fsblock, char *buffer)
+{
+#ifdef E2DEBUG
+ printf ("fsblock %d buffer %d\n", fsblock, buffer);
+#endif /* E2DEBUG */
+ return devread (ffi, fsblock * (EXT2_BLOCK_SIZE (SUPERBLOCK) / DEV_BSIZE), 0,
+ EXT2_BLOCK_SIZE (SUPERBLOCK), (char *) buffer);
+}
+
+/* from
+ ext2/inode.c:ext2_bmap()
+*/
+/* Maps LOGICAL_BLOCK (the file offset divided by the blocksize) into
+ a physical block (the location in the file system) via an inode. */
+static int
+ext2fs_block_map (fsi_file_t *ffi, int logical_block)
+{
+
+#ifdef E2DEBUG
+ unsigned char *i;
+ for (i = (unsigned char *) INODE;
+ i < ((unsigned char *) INODE + sizeof (struct ext2_inode));
+ i++)
+ {
+ printf ("%c", "0123456789abcdef"[*i >> 4]);
+ printf ("%c", "0123456789abcdef"[*i % 16]);
+ if (!((i + 1 - (unsigned char *) INODE) % 16))
+ {
+ printf ("\n");
+ }
+ else
+ {
+ printf (" ");
+ }
+ }
+ printf ("logical block %d\n", logical_block);
+#endif /* E2DEBUG */
+
+ /* if it is directly pointed to by the inode, return that physical addr */
+ if (logical_block < EXT2_NDIR_BLOCKS)
+ {
+#ifdef E2DEBUG
+ printf ("returning %d\n", (unsigned char *)
(INODE->i_block[logical_block]));
+ printf ("returning %d\n", INODE->i_block[logical_block]);
+#endif /* E2DEBUG */
+ return INODE->i_block[logical_block];
+ }
+ /* else */
+ logical_block -= EXT2_NDIR_BLOCKS;
+ /* try the indirect block */
+ if (logical_block < EXT2_ADDR_PER_BLOCK (SUPERBLOCK))
+ {
+ if (mapblock1 != 1
+ && !ext2_rdfsb (ffi, INODE->i_block[EXT2_IND_BLOCK], DATABLOCK1))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ mapblock1 = 1;
+ return ((__u32 *) DATABLOCK1)[logical_block];
+ }
+ /* else */
+ logical_block -= EXT2_ADDR_PER_BLOCK (SUPERBLOCK);
+ /* now try the double indirect block */
+ if (logical_block < (1 << (EXT2_ADDR_PER_BLOCK_BITS (SUPERBLOCK) * 2)))
+ {
+ int bnum;
+ if (mapblock1 != 2
+ && !ext2_rdfsb (ffi, INODE->i_block[EXT2_DIND_BLOCK], DATABLOCK1))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ mapblock1 = 2;
+ if ((bnum = (((__u32 *) DATABLOCK1)
+ [logical_block >> EXT2_ADDR_PER_BLOCK_BITS (SUPERBLOCK)]))
+ != mapblock2
+ && !ext2_rdfsb (ffi, bnum, DATABLOCK2))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ mapblock2 = bnum;
+ return ((__u32 *) DATABLOCK2)
+ [logical_block & (EXT2_ADDR_PER_BLOCK (SUPERBLOCK) - 1)];
+ }
+ /* else */
+ mapblock2 = -1;
+ logical_block -= (1 << (EXT2_ADDR_PER_BLOCK_BITS (SUPERBLOCK) * 2));
+ if (mapblock1 != 3
+ && !ext2_rdfsb (ffi, INODE->i_block[EXT2_TIND_BLOCK], DATABLOCK1))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ mapblock1 = 3;
+ if (!ext2_rdfsb (ffi, ((__u32 *) DATABLOCK1)
+ [logical_block >> (EXT2_ADDR_PER_BLOCK_BITS (SUPERBLOCK)
+ * 2)],
+ DATABLOCK2))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ if (!ext2_rdfsb (ffi, ((__u32 *) DATABLOCK2)
+ [(logical_block >> EXT2_ADDR_PER_BLOCK_BITS (SUPERBLOCK))
+ & (EXT2_ADDR_PER_BLOCK (SUPERBLOCK) - 1)],
+ DATABLOCK2))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return -1;
+ }
+ return ((__u32 *) DATABLOCK2)
+ [logical_block & (EXT2_ADDR_PER_BLOCK (SUPERBLOCK) - 1)];
+}
+
+/* preconditions: all preconds of ext2fs_block_map */
+int
+ext2fs_read (fsi_file_t *ffi, char *buf, int len)
+{
+ int logical_block;
+ int offset;
+ int map;
+ int ret = 0;
+ int size = 0;
+
+#ifdef E2DEBUG
+ static char hexdigit[] = "0123456789abcdef";
+ unsigned char *i;
+ for (i = (unsigned char *) INODE;
+ i < ((unsigned char *) INODE + sizeof (struct ext2_inode));
+ i++)
+ {
+ printf ("%c", hexdigit[*i >> 4]);
+ printf ("%c", hexdigit[*i % 16]);
+ if (!((i + 1 - (unsigned char *) INODE) % 16))
+ {
+ printf ("\n");
+ }
+ else
+ {
+ printf (" ");
+ }
+ }
+#endif /* E2DEBUG */
+ while (len > 0)
+ {
+ /* find the (logical) block component of our location */
+ logical_block = filepos >> EXT2_BLOCK_SIZE_BITS (SUPERBLOCK);
+ offset = filepos & (EXT2_BLOCK_SIZE (SUPERBLOCK) - 1);
+ map = ext2fs_block_map (ffi, logical_block);
+#ifdef E2DEBUG
+ printf ("map=%d\n", map);
+#endif /* E2DEBUG */
+ if (map < 0)
+ break;
+
+ size = EXT2_BLOCK_SIZE (SUPERBLOCK);
+ size -= offset;
+ if (size > len)
+ size = len;
+
+ if (map == 0) {
+ memset ((char *) buf, 0, size);
+ } else {
+ disk_read_func = disk_read_hook;
+
+ devread (ffi, map * (EXT2_BLOCK_SIZE (SUPERBLOCK) / DEV_BSIZE),
+ offset, size, buf);
+
+ disk_read_func = NULL;
+ }
+
+ buf += size;
+ len -= size;
+ filepos += size;
+ ret += size;
+ }
+
+ if (errnum)
+ ret = 0;
+
+ return ret;
+}
+
+
+/* Based on:
+ def_blk_fops points to
+ blkdev_open, which calls (I think):
+ sys_open()
+ do_open()
+ open_namei()
+ dir_namei() which accesses current->fs->root
+ fs->root was set during original mount:
+ (something)... which calls (I think):
+ ext2_read_super()
+ iget()
+ __iget()
+ read_inode()
+ ext2_read_inode()
+ uses desc_per_block_bits, which is set in ext2_read_super()
+ also uses group descriptors loaded during ext2_read_super()
+ lookup()
+ ext2_lookup()
+ ext2_find_entry()
+ ext2_getblk()
+
+*/
+
+static inline
+int ext2_is_fast_symlink (fsi_file_t *ffi)
+{
+ int ea_blocks;
+ ea_blocks = INODE->i_file_acl ? EXT2_BLOCK_SIZE (SUPERBLOCK) / DEV_BSIZE : 0;
+ return INODE->i_blocks == ea_blocks;
+}
+
+/* preconditions: ext2fs_mount already executed, therefore supblk in buffer
+ * known as SUPERBLOCK
+ * returns: 0 if error, nonzero iff we were able to find the file successfully
+ * postconditions: on a nonzero return, buffer known as INODE contains the
+ * inode of the file we were trying to look up
+ * side effects: messes up GROUP_DESC buffer area
+ */
+int
+ext2fs_dir (fsi_file_t *ffi, char *dirname)
+{
+ int current_ino = EXT2_ROOT_INO; /* start at the root */
+ int updir_ino = current_ino; /* the parent of the current directory */
+ int group_id; /* which group the inode is in */
+ int group_desc; /* fs pointer to that group */
+ int desc; /* index within that group */
+ int ino_blk; /* fs pointer of the inode's information */
+ int str_chk = 0; /* used to hold the results of a string compare
*/
+ struct ext2_group_desc *gdp;
+ struct ext2_inode *raw_inode; /* inode info corresponding to
current_ino */
+
+ char linkbuf[PATH_MAX]; /* buffer for following symbolic links */
+ int link_count = 0;
+
+ char *rest;
+ char ch; /* temp char holder */
+
+ int off; /* offset within block of directory entry (off
mod blocksize) */
+ int loc; /* location within a directory */
+ int blk; /* which data blk within dir entry (off div
blocksize) */
+ long map; /* fs pointer of a particular block from dir
entry */
+ struct ext2_dir_entry *dp; /* pointer to directory entry */
+#ifdef E2DEBUG
+ unsigned char *i;
+#endif /* E2DEBUG */
+
+ /* loop invariants:
+ current_ino = inode to lookup
+ dirname = pointer to filename component we are cur looking up within
+ the directory known pointed to by current_ino (if any)
+ */
+
+ while (1)
+ {
+#ifdef E2DEBUG
+ printf ("inode %d\n", current_ino);
+ printf ("dirname=%s\n", dirname);
+#endif /* E2DEBUG */
+
+ /* look up an inode */
+ group_id = (current_ino - 1) / (SUPERBLOCK->s_inodes_per_group);
+ group_desc = group_id >> log2 (EXT2_DESC_PER_BLOCK (SUPERBLOCK));
+ desc = group_id & (EXT2_DESC_PER_BLOCK (SUPERBLOCK) - 1);
+#ifdef E2DEBUG
+ printf ("ipg=%d, dpb=%d\n", SUPERBLOCK->s_inodes_per_group,
+ EXT2_DESC_PER_BLOCK (SUPERBLOCK));
+ printf ("group_id=%d group_desc=%d desc=%d\n", group_id, group_desc,
desc);
+#endif /* E2DEBUG */
+ if (!ext2_rdfsb (ffi,
+ (WHICH_SUPER + group_desc +
SUPERBLOCK->s_first_data_block),
+ (char *)GROUP_DESC))
+ {
+ return 0;
+ }
+ gdp = GROUP_DESC;
+ ino_blk = gdp[desc].bg_inode_table +
+ (((current_ino - 1) % (SUPERBLOCK->s_inodes_per_group))
+ >> log2 (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode)));
+#ifdef E2DEBUG
+ printf ("inode table fsblock=%d\n", ino_blk);
+#endif /* E2DEBUG */
+ if (!ext2_rdfsb (ffi, ino_blk, (char *)INODE))
+ {
+ return 0;
+ }
+
+ /* reset indirect blocks! */
+ mapblock2 = mapblock1 = -1;
+
+ raw_inode = INODE +
+ ((current_ino - 1)
+ & (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode) - 1));
+#ifdef E2DEBUG
+ printf ("ipb=%d, sizeof(inode)=%d\n",
+ (EXT2_BLOCK_SIZE (SUPERBLOCK) / sizeof (struct ext2_inode)),
+ sizeof (struct ext2_inode));
+ printf ("inode=%x, raw_inode=%x\n", INODE, raw_inode);
+ printf ("offset into inode table block=%d\n", (int) raw_inode - (int)
INODE);
+ for (i = (unsigned char *) INODE; i <= (unsigned char *) raw_inode;
+ i++)
+ {
+ printf ("%c", "0123456789abcdef"[*i >> 4]);
+ printf ("%c", "0123456789abcdef"[*i % 16]);
+ if (!((i + 1 - (unsigned char *) INODE) % 16))
+ {
+ printf ("\n");
+ }
+ else
+ {
+ printf (" ");
+ }
+ }
+ printf ("first word=%x\n", *((int *) raw_inode));
+#endif /* E2DEBUG */
+
+ /* copy inode to fixed location */
+ memmove ((void *) INODE, (void *) raw_inode, sizeof (struct ext2_inode));
+
+#ifdef E2DEBUG
+ printf ("first word=%x\n", *((int *) INODE));
+#endif /* E2DEBUG */
+
+ /* If we've got a symbolic link, then chase it. */
+ if (S_ISLNK (INODE->i_mode))
+ {
+ int len;
+ if (++link_count > MAX_LINK_COUNT)
+ {
+ errnum = ERR_SYMLINK_LOOP;
+ return 0;
+ }
+
+ /* Find out how long our remaining name is. */
+ len = 0;
+ while (dirname[len] && !isspace (dirname[len]))
+ len++;
+
+ /* Get the symlink size. */
+ filemax = (INODE->i_size);
+ if (filemax + len > sizeof (linkbuf) - 2)
+ {
+ errnum = ERR_FILELENGTH;
+ return 0;
+ }
+
+ if (len)
+ {
+ /* Copy the remaining name to the end of the symlink data.
+ Note that DIRNAME and LINKBUF may overlap! */
+ memmove (linkbuf + filemax, dirname, len);
+ }
+ linkbuf[filemax + len] = '\0';
+
+ /* Read the symlink data. */
+ if (! ext2_is_fast_symlink (ffi))
+ {
+ /* Read the necessary blocks, and reset the file pointer. */
+ len = ext2fs_read (ffi, linkbuf, filemax);
+ filepos = 0;
+ if (!len)
+ return 0;
+ }
+ else
+ {
+ /* Copy the data directly from the inode. */
+ len = filemax;
+ memmove (linkbuf, (char *) INODE->i_block, len);
+ }
+
+#ifdef E2DEBUG
+ printf ("symlink=%s\n", linkbuf);
+#endif
+
+ dirname = linkbuf;
+ if (*dirname == '/')
+ {
+ /* It's an absolute link, so look it up in root. */
+ current_ino = EXT2_ROOT_INO;
+ updir_ino = current_ino;
+ }
+ else
+ {
+ /* Relative, so look it up in our parent directory. */
+ current_ino = updir_ino;
+ }
+
+ /* Try again using the new name. */
+ continue;
+ }
+
+ /* if end of filename, INODE points to the file's inode */
+ if (!*dirname || isspace (*dirname))
+ {
+ if (!S_ISREG (INODE->i_mode))
+ {
+ errnum = ERR_BAD_FILETYPE;
+ return 0;
+ }
+
+ filemax = (INODE->i_size);
+ return 1;
+ }
+
+ /* else we have to traverse a directory */
+ updir_ino = current_ino;
+
+ /* skip over slashes */
+ while (*dirname == '/')
+ dirname++;
+
+ /* if this isn't a directory of sufficient size to hold our file, abort
*/
+ if (!(INODE->i_size) || !S_ISDIR (INODE->i_mode))
+ {
+ errnum = ERR_BAD_FILETYPE;
+ return 0;
+ }
+
+ /* skip to next slash or end of filename (space) */
+ for (rest = dirname; (ch = *rest) && !isspace (ch) && ch != '/';
+ rest++);
+
+ /* look through this directory and find the next filename component */
+ /* invariant: rest points to slash after the next filename component */
+ *rest = 0;
+ loc = 0;
+
+ do
+ {
+
+#ifdef E2DEBUG
+ printf ("dirname=%s, rest=%s, loc=%d\n", dirname, rest, loc);
+#endif /* E2DEBUG */
+
+ /* if our location/byte offset into the directory exceeds the size,
+ give up */
+ if (loc >= INODE->i_size)
+ {
+ if (print_possibilities < 0)
+ {
+# if 0
+ putchar ('\n');
+# endif
+ }
+ else
+ {
+ errnum = ERR_FILE_NOT_FOUND;
+ *rest = ch;
+ }
+ return (print_possibilities < 0);
+ }
+
+ /* else, find the (logical) block component of our location */
+ blk = loc >> EXT2_BLOCK_SIZE_BITS (SUPERBLOCK);
+
+ /* we know which logical block of the directory entry we are looking
+ for, now we have to translate that to the physical (fs) block on
+ the disk */
+ map = ext2fs_block_map (ffi, blk);
+#ifdef E2DEBUG
+ printf ("fs block=%d\n", map);
+#endif /* E2DEBUG */
+ mapblock2 = -1;
+ if ((map < 0) || !ext2_rdfsb (ffi, map, DATABLOCK2))
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ *rest = ch;
+ return 0;
+ }
+ off = loc & (EXT2_BLOCK_SIZE (SUPERBLOCK) - 1);
+ dp = (struct ext2_dir_entry *) (DATABLOCK2 + off);
+ /* advance loc prematurely to next on-disk directory entry */
+ loc += dp->rec_len;
+
+ /* NOTE: ext2fs filenames are NOT null-terminated */
+
+#ifdef E2DEBUG
+ printf ("directory entry ino=%d\n", dp->inode);
+ if (dp->inode)
+ printf ("entry=%s\n", dp->name);
+#endif /* E2DEBUG */
+
+ if (dp->inode)
+ {
+ int saved_c = dp->name[dp->name_len];
+
+ dp->name[dp->name_len] = 0;
+ str_chk = substring (dirname, dp->name);
+
+# ifndef STAGE1_5
+ if (print_possibilities && ch != '/'
+ && (!*dirname || str_chk <= 0))
+ {
+ if (print_possibilities > 0)
+ print_possibilities = -print_possibilities;
+ print_a_completion (dp->name);
+ }
+# endif
+
+ dp->name[dp->name_len] = saved_c;
+ }
+
+ }
+ while (!dp->inode || (str_chk || (print_possibilities && ch != '/')));
+
+ current_ino = dp->inode;
+ *(dirname = rest) = ch;
+ }
+ /* never get here */
+}
+
+fsi_plugin_ops_t *
+fsi_init_plugin(int version, fsi_plugin_t *fp, const char **name)
+{
+ static fsig_plugin_ops_t ops = {
+ FSIMAGE_PLUGIN_VERSION,
+ .fpo_mount = ext2fs_mount,
+ .fpo_dir = ext2fs_dir,
+ .fpo_read = ext2fs_read
+ };
+
+ *name = "ext2fs";
+ return (fsig_init(fp, &ops));
+}
diff --git a/tools/libfsimage/reiserfs/Makefile
b/tools/libfsimage/reiserfs/Makefile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/reiserfs/Makefile
@@ -0,0 +1,13 @@
+XEN_ROOT = ../../..
+
+LIB_SRCS-y = fsys_reiserfs.c
+
+FS = reiserfs
+
+.PHONY: all
+all: fs-all
+
+.PHONY: install
+install: fs-install
+
+include $(XEN_ROOT)/tools/libfsimage/Rules.mk
diff --git a/tools/libfsimage/reiserfs/fsys_reiserfs.c
b/tools/libfsimage/reiserfs/fsys_reiserfs.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/reiserfs/fsys_reiserfs.c
@@ -0,0 +1,1254 @@
+/* fsys_reiserfs.c - an implementation for the ReiserFS filesystem */
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 2000, 2001 Free Software Foundation, Inc.
+ *
+ * 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 of the License, 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 <fsimage_grub.h>
+
+#undef REISERDEBUG
+
+/* Some parts of this code (mainly the structures and defines) are
+ * from the original reiser fs code, as found in the linux kernel.
+ */
+
+/* include/asm-i386/types.h */
+typedef __signed__ char __s8;
+typedef unsigned char __u8;
+typedef __signed__ short __s16;
+typedef unsigned short __u16;
+typedef __signed__ int __s32;
+typedef unsigned int __u32;
+typedef unsigned long long __u64;
+
+/* linux/posix_type.h */
+typedef long linux_off_t;
+
+/* linux/little_endian.h */
+#define __cpu_to_le64(x) ((__u64) (x))
+#define __le64_to_cpu(x) ((__u64) (x))
+#define __cpu_to_le32(x) ((__u32) (x))
+#define __le32_to_cpu(x) ((__u32) (x))
+#define __cpu_to_le16(x) ((__u16) (x))
+#define __le16_to_cpu(x) ((__u16) (x))
+
+/* include/linux/reiser_fs.h */
+/* This is the new super block of a journaling reiserfs system */
+struct reiserfs_super_block
+{
+ __u32 s_block_count; /* blocks count */
+ __u32 s_free_blocks; /* free blocks count */
+ __u32 s_root_block; /* root block number */
+ __u32 s_journal_block; /* journal block number */
+ __u32 s_journal_dev; /* journal device number */
+ __u32 s_journal_size; /* size of the journal on FS creation.
used to make sure they don't overflow it */
+ __u32 s_journal_trans_max; /* max number of blocks in a
transaction. */
+ __u32 s_journal_magic; /* random value made on fs creation */
+ __u32 s_journal_max_batch; /* max number of blocks to batch into
a trans */
+ __u32 s_journal_max_commit_age; /* in seconds, how old can an async
commit be */
+ __u32 s_journal_max_trans_age; /* in seconds, how old can a
transaction be */
+ __u16 s_blocksize; /* block size */
+ __u16 s_oid_maxsize; /* max size of object id array */
+ __u16 s_oid_cursize; /* current size of object id array */
+ __u16 s_state; /* valid or error */
+ char s_magic[16]; /* reiserfs magic string indicates
that file system is reiserfs */
+ __u16 s_tree_height; /* height of disk tree */
+ __u16 s_bmap_nr; /* amount of bitmap blocks needed to
address each block of file system */
+ __u16 s_version;
+ char s_unused[128]; /* zero filled by mkreiserfs */
+};
+
+#define REISERFS_MAX_SUPPORTED_VERSION 2
+#define REISERFS_SUPER_MAGIC_STRING "ReIsErFs"
+#define REISER2FS_SUPER_MAGIC_STRING "ReIsEr2Fs"
+#define REISER3FS_SUPER_MAGIC_STRING "ReIsEr3Fs"
+
+#define MAX_HEIGHT 7
+
+/* must be correct to keep the desc and commit structs at 4k */
+#define JOURNAL_TRANS_HALF 1018
+
+/* first block written in a commit. */
+struct reiserfs_journal_desc {
+ __u32 j_trans_id; /* id of commit */
+ __u32 j_len; /* length of commit. len +1 is the
commit block */
+ __u32 j_mount_id; /* mount id of this trans*/
+ __u32 j_realblock[JOURNAL_TRANS_HALF]; /* real locations for the first
blocks */
+ char j_magic[12];
+};
+
+/* last block written in a commit */
+struct reiserfs_journal_commit {
+ __u32 j_trans_id; /* must match j_trans_id from the desc
block */
+ __u32 j_len; /* ditto */
+ __u32 j_realblock[JOURNAL_TRANS_HALF]; /* real locations for the last blocks
*/
+ char j_digest[16]; /* md5 sum of all the blocks involved,
including desc and commit. not used, kill it */
+};
+
+/* this header block gets written whenever a transaction is considered
+ fully flushed, and is more recent than the last fully flushed
+ transaction.
+ fully flushed means all the log blocks and all the real blocks are
+ on disk, and this transaction does not need to be replayed.
+*/
+struct reiserfs_journal_header {
+ /* id of last fully flushed transaction */
+ __u32 j_last_flush_trans_id;
+ /* offset in the log of where to start replay after a crash */
+ __u32 j_first_unflushed_offset;
+ /* mount id to detect very old transactions */
+ __u32 j_mount_id;
+};
+
+/* magic string to find desc blocks in the journal */
+#define JOURNAL_DESC_MAGIC "ReIsErLB"
+
+
+/*
+ * directories use this key as well as old files
+ */
+struct offset_v1
+{
+ /*
+ * for regular files this is the offset to the first byte of the
+ * body, contained in the object-item, as measured from the start of
+ * the entire body of the object.
+ *
+ * for directory entries, k_offset consists of hash derived from
+ * hashing the name and using few bits (23 or more) of the resulting
+ * hash, and generation number that allows distinguishing names with
+ * hash collisions. If number of collisions overflows generation
+ * number, we return EEXIST. High order bit is 0 always
+ */
+ __u32 k_offset;
+ __u32 k_uniqueness;
+};
+
+struct offset_v2
+{
+ /*
+ * for regular files this is the offset to the first byte of the
+ * body, contained in the object-item, as measured from the start of
+ * the entire body of the object.
+ *
+ * for directory entries, k_offset consists of hash derived from
+ * hashing the name and using few bits (23 or more) of the resulting
+ * hash, and generation number that allows distinguishing names with
+ * hash collisions. If number of collisions overflows generation
+ * number, we return EEXIST. High order bit is 0 always
+ */
+ __u64 k_offset:60;
+ __u64 k_type: 4;
+};
+
+
+struct key
+{
+ /* packing locality: by default parent directory object id */
+ __u32 k_dir_id;
+ /* object identifier */
+ __u32 k_objectid;
+ /* the offset and node type (old and new form) */
+ union
+ {
+ struct offset_v1 v1;
+ struct offset_v2 v2;
+ }
+ u;
+};
+
+#define KEY_SIZE (sizeof (struct key))
+
+/* Header of a disk block. More precisely, header of a formatted leaf
+ or internal node, and not the header of an unformatted node. */
+struct block_head
+{
+ __u16 blk_level; /* Level of a block in the tree. */
+ __u16 blk_nr_item; /* Number of keys/items in a block. */
+ __u16 blk_free_space; /* Block free space in bytes. */
+ struct key blk_right_delim_key; /* Right delimiting key for this block
(supported for leaf level nodes
+ only) */
+};
+#define BLKH_SIZE (sizeof (struct block_head))
+#define DISK_LEAF_NODE_LEVEL 1 /* Leaf node level. */
+
+struct item_head
+{
+ struct key ih_key; /* Everything in the tree is found by searching for it
based on its key.*/
+
+ union
+ {
+ __u16 ih_free_space; /* The free space in the last unformatted node of an
indirect item if this
+ is an indirect item. This equals 0xFFFF iff this
is a direct item or
+ stat data item. Note that the key, not this field,
is used to determine
+ the item type, and thus which field this union
contains. */
+ __u16 ih_entry_count; /* Iff this is a directory item, this field equals
the number of directory
+ entries in the directory item. */
+ }
+ u;
+ __u16 ih_item_len; /* total size of the item body
*/
+ __u16 ih_item_location; /* an offset to the item body within the block
*/
+ __u16 ih_version; /* ITEM_VERSION_1 for all old items,
+ ITEM_VERSION_2 for new ones.
+ Highest bit is set by fsck
+ temporary, cleaned after all done */
+};
+/* size of item header */
+#define IH_SIZE (sizeof (struct item_head))
+
+#define ITEM_VERSION_1 0
+#define ITEM_VERSION_2 1
+#define IH_KEY_OFFSET(ih) ((ih)->ih_version == ITEM_VERSION_1 \
+ ? (ih)->ih_key.u.v1.k_offset \
+ : (ih)->ih_key.u.v2.k_offset)
+
+#define IH_KEY_ISTYPE(ih, type) ((ih)->ih_version == ITEM_VERSION_1 \
+ ? (ih)->ih_key.u.v1.k_uniqueness == V1_##type \
+ : (ih)->ih_key.u.v2.k_type == V2_##type)
+
+struct disk_child
+{
+ unsigned long dc_block_number; /* Disk child's block
number. */
+ unsigned short dc_size; /* Disk child's used space.
*/
+};
+
+#define DC_SIZE (sizeof (struct disk_child))
+
+/* Stat Data on disk.
+ *
+ * Note that reiserfs has two different forms of stat data. Luckily
+ * the fields needed by grub are at the same position.
+ */
+struct stat_data
+{
+ __u16 sd_mode; /* file type, permissions */
+ __u16 sd_notused1[3]; /* fields not needed by reiserfs */
+ __u32 sd_size; /* file size */
+ __u32 sd_size_hi; /* file size high 32 bits (since version 2) */
+};
+
+struct reiserfs_de_head
+{
+ __u32 deh_offset; /* third component of the directory entry key */
+ __u32 deh_dir_id; /* objectid of the parent directory of the
+ object, that is referenced by directory entry */
+ __u32 deh_objectid;/* objectid of the object, that is referenced by
+ directory entry */
+ __u16 deh_location;/* offset of name in the whole item */
+ __u16 deh_state; /* whether 1) entry contains stat data (for
+ future), and 2) whether entry is hidden
+ (unlinked) */
+};
+
+#define DEH_SIZE (sizeof (struct reiserfs_de_head))
+
+#define DEH_Statdata (1 << 0) /* not used now */
+#define DEH_Visible (1 << 2)
+
+#define SD_OFFSET 0
+#define SD_UNIQUENESS 0
+#define DOT_OFFSET 1
+#define DOT_DOT_OFFSET 2
+#define DIRENTRY_UNIQUENESS 500
+
+#define V1_TYPE_STAT_DATA 0x0
+#define V1_TYPE_DIRECT 0xffffffff
+#define V1_TYPE_INDIRECT 0xfffffffe
+#define V1_TYPE_DIRECTORY_MAX 0xfffffffd
+#define V2_TYPE_STAT_DATA 0
+#define V2_TYPE_INDIRECT 1
+#define V2_TYPE_DIRECT 2
+#define V2_TYPE_DIRENTRY 3
+
+#define REISERFS_ROOT_OBJECTID 2
+#define REISERFS_ROOT_PARENT_OBJECTID 1
+#define REISERFS_DISK_OFFSET_IN_BYTES (64 * 1024)
+/* the spot for the super in versions 3.5 - 3.5.11 (inclusive) */
+#define REISERFS_OLD_DISK_OFFSET_IN_BYTES (8 * 1024)
+#define REISERFS_OLD_BLOCKSIZE 4096
+
+#define S_ISREG(mode) (((mode) & 0170000) == 0100000)
+#define S_ISDIR(mode) (((mode) & 0170000) == 0040000)
+#define S_ISLNK(mode) (((mode) & 0170000) == 0120000)
+
+#define PATH_MAX 1024 /* include/linux/limits.h */
+#define MAX_LINK_COUNT 5 /* number of symbolic links to follow */
+
+/* The size of the node cache */
+#define FSYSREISER_CACHE_SIZE 24*1024
+#define FSYSREISER_MIN_BLOCKSIZE SECTOR_SIZE
+#define FSYSREISER_MAX_BLOCKSIZE FSYSREISER_CACHE_SIZE / 3
+
+/* Info about currently opened file */
+struct fsys_reiser_fileinfo
+{
+ __u32 k_dir_id;
+ __u32 k_objectid;
+};
+
+/* In memory info about the currently mounted filesystem */
+struct fsys_reiser_info
+{
+ /* The last read item head */
+ struct item_head *current_ih;
+ /* The last read item */
+ char *current_item;
+ /* The information for the currently opened file */
+ struct fsys_reiser_fileinfo fileinfo;
+ /* The start of the journal */
+ __u32 journal_block;
+ /* The size of the journal */
+ __u32 journal_block_count;
+ /* The first valid descriptor block in journal
+ (relative to journal_block) */
+ __u32 journal_first_desc;
+
+ /* The ReiserFS version. */
+ __u16 version;
+ /* The current depth of the reiser tree. */
+ __u16 tree_depth;
+ /* SECTOR_SIZE << blocksize_shift == blocksize. */
+ __u8 blocksize_shift;
+ /* 1 << full_blocksize_shift == blocksize. */
+ __u8 fullblocksize_shift;
+ /* The reiserfs block size (must be a power of 2) */
+ __u16 blocksize;
+ /* The number of cached tree nodes */
+ __u16 cached_slots;
+ /* The number of valid transactions in journal */
+ __u16 journal_transactions;
+
+ unsigned int blocks[MAX_HEIGHT];
+ unsigned int next_key_nr[MAX_HEIGHT];
+};
+
+/* The cached s+tree blocks in FSYS_BUF, see below
+ * for a more detailed description.
+ */
+#define ROOT ((char *) FSYS_BUF)
+#define CACHE(i) (ROOT + ((i) << INFO->fullblocksize_shift))
+#define LEAF CACHE (DISK_LEAF_NODE_LEVEL)
+
+#define BLOCKHEAD(cache) ((struct block_head *) cache)
+#define ITEMHEAD ((struct item_head *) ((char *) LEAF + BLKH_SIZE))
+#define KEY(cache) ((struct key *) ((char *) cache + BLKH_SIZE))
+#define DC(cache) ((struct disk_child *) \
+ ((char *) cache + BLKH_SIZE + KEY_SIZE * nr_item))
+/* The fsys_reiser_info block.
+ */
+#define INFO \
+ ((struct fsys_reiser_info *) ((char *) FSYS_BUF + FSYSREISER_CACHE_SIZE))
+/*
+ * The journal cache. For each transaction it contains the number of
+ * blocks followed by the real block numbers of this transaction.
+ *
+ * If the block numbers of some transaction won't fit in this space,
+ * this list is stopped with a 0xffffffff marker and the remaining
+ * uncommitted transactions aren't cached.
+ */
+#define JOURNAL_START ((__u32 *) (INFO + 1))
+#define JOURNAL_END ((__u32 *) (FSYS_BUF + FSYS_BUFLEN))
+
+#ifdef __amd64
+#define BSF "bsfq"
+#else
+#define BSF "bsfl"
+#endif
+static __inline__ unsigned long
+grub_log2 (unsigned long word)
+{
+ __asm__ (BSF " %1,%0"
+ : "=r" (word)
+ : "r" (word));
+ return word;
+}
+#define log2 grub_log2
+
+static __inline__ int
+is_power_of_two (unsigned long word)
+{
+ return (word & -word) == word;
+}
+
+static int
+journal_read (fsi_file_t *ffi, int block, int len, char *buffer)
+{
+ return devread (ffi, (INFO->journal_block + block) << INFO->blocksize_shift,
+ 0, len, buffer);
+}
+
+/* Read a block from ReiserFS file system, taking the journal into
+ * account. If the block nr is in the journal, the block from the
+ * journal taken.
+ */
+static int
+block_read (fsi_file_t *ffi, int blockNr, int start, int len, char *buffer)
+{
+ int transactions = INFO->journal_transactions;
+ int desc_block = INFO->journal_first_desc;
+ int journal_mask = INFO->journal_block_count - 1;
+ int translatedNr = blockNr;
+ __u32 *journal_table = JOURNAL_START;
+ while (transactions-- > 0)
+ {
+ int i = 0;
+ int j_len;
+ if (*journal_table != 0xffffffff)
+ {
+ /* Search for the blockNr in cached journal */
+ j_len = *journal_table++;
+ while (i++ < j_len)
+ {
+ if (*journal_table++ == blockNr)
+ {
+ journal_table += j_len - i;
+ goto found;
+ }
+ }
+ }
+ else
+ {
+ /* This is the end of cached journal marker. The remaining
+ * transactions are still on disk.
+ */
+ struct reiserfs_journal_desc desc;
+ struct reiserfs_journal_commit commit;
+
+ if (! journal_read (ffi, desc_block, sizeof (desc), (char *) &desc))
+ return 0;
+
+ j_len = desc.j_len;
+ while (i < j_len && i < JOURNAL_TRANS_HALF)
+ if (desc.j_realblock[i++] == blockNr)
+ goto found;
+
+ if (j_len >= JOURNAL_TRANS_HALF)
+ {
+ int commit_block = (desc_block + 1 + j_len) & journal_mask;
+ if (! journal_read (ffi, commit_block,
+ sizeof (commit), (char *) &commit))
+ return 0;
+ while (i < j_len)
+ if (commit.j_realblock[i++ - JOURNAL_TRANS_HALF] == blockNr)
+ goto found;
+ }
+ }
+ goto not_found;
+
+ found:
+ translatedNr = INFO->journal_block + ((desc_block + i) & journal_mask);
+#ifdef REISERDEBUG
+ printf ("block_read: block %d is mapped to journal block %d.\n",
+ blockNr, translatedNr - INFO->journal_block);
+#endif
+ /* We must continue the search, as this block may be overwritten
+ * in later transactions.
+ */
+ not_found:
+ desc_block = (desc_block + 2 + j_len) & journal_mask;
+ }
+ return devread (ffi, translatedNr << INFO->blocksize_shift, start, len,
buffer);
+}
+
+/* Init the journal data structure. We try to cache as much as
+ * possible in the JOURNAL_START-JOURNAL_END space, but if it is full
+ * we can still read the rest from the disk on demand.
+ *
+ * The first number of valid transactions and the descriptor block of the
+ * first valid transaction are held in INFO. The transactions are all
+ * adjacent, but we must take care of the journal wrap around.
+ */
+static int
+journal_init (fsi_file_t *ffi)
+{
+ unsigned int block_count = INFO->journal_block_count;
+ unsigned int desc_block;
+ unsigned int commit_block;
+ unsigned int next_trans_id;
+ struct reiserfs_journal_header header;
+ struct reiserfs_journal_desc desc;
+ struct reiserfs_journal_commit commit;
+ __u32 *journal_table = JOURNAL_START;
+
+ journal_read (ffi, block_count, sizeof (header), (char *) &header);
+ desc_block = header.j_first_unflushed_offset;
+ if (desc_block >= block_count)
+ return 0;
+
+ INFO->journal_first_desc = desc_block;
+ next_trans_id = header.j_last_flush_trans_id + 1;
+
+#ifdef REISERDEBUG
+ printf ("journal_init: last flushed %d\n",
+ header.j_last_flush_trans_id);
+#endif
+
+ while (1)
+ {
+ journal_read (ffi, desc_block, sizeof (desc), (char *) &desc);
+ if (substring (JOURNAL_DESC_MAGIC, desc.j_magic)
+ || desc.j_trans_id != next_trans_id
+ || desc.j_mount_id != header.j_mount_id)
+ /* no more valid transactions */
+ break;
+
+ commit_block = (desc_block + desc.j_len + 1) & (block_count - 1);
+ journal_read (ffi, commit_block, sizeof (commit), (char *) &commit);
+ if (desc.j_trans_id != commit.j_trans_id
+ || desc.j_len != commit.j_len)
+ /* no more valid transactions */
+ break;
+
+#ifdef REISERDEBUG
+ printf ("Found valid transaction %d/%d at %d.\n",
+ desc.j_trans_id, desc.j_mount_id, desc_block);
+#endif
+
+ next_trans_id++;
+ if (journal_table < JOURNAL_END)
+ {
+ if ((journal_table + 1 + desc.j_len) >= JOURNAL_END)
+ {
+ /* The table is almost full; mark the end of the cached
+ * journal.*/
+ *journal_table = 0xffffffff;
+ journal_table = JOURNAL_END;
+ }
+ else
+ {
+ int i;
+ /* Cache the length and the realblock numbers in the table.
+ * The block number of descriptor can easily be computed.
+ * and need not to be stored here.
+ */
+ *journal_table++ = desc.j_len;
+ for (i = 0; i < desc.j_len && i < JOURNAL_TRANS_HALF; i++)
+ {
+ *journal_table++ = desc.j_realblock[i];
+#ifdef REISERDEBUG
+ printf ("block %d is in journal %d.\n",
+ desc.j_realblock[i], desc_block);
+#endif
+ }
+ for ( ; i < desc.j_len; i++)
+ {
+ *journal_table++ = commit.j_realblock[i-JOURNAL_TRANS_HALF];
+#ifdef REISERDEBUG
+ printf ("block %d is in journal %d.\n",
+ commit.j_realblock[i-JOURNAL_TRANS_HALF],
+ desc_block);
+#endif
+ }
+ }
+ }
+ desc_block = (commit_block + 1) & (block_count - 1);
+ }
+#ifdef REISERDEBUG
+ printf ("Transaction %d/%d at %d isn't valid.\n",
+ desc.j_trans_id, desc.j_mount_id, desc_block);
+#endif
+
+ INFO->journal_transactions
+ = next_trans_id - header.j_last_flush_trans_id - 1;
+ return errnum == 0;
+}
+
+/* check filesystem types and read superblock into memory buffer */
+int
+reiserfs_mount (fsi_file_t *ffi)
+{
+ struct reiserfs_super_block super;
+ int superblock = REISERFS_DISK_OFFSET_IN_BYTES >> SECTOR_BITS;
+
+ if (/*part_length < superblock + (sizeof (super) >> SECTOR_BITS)
+ || */ !devread (ffi, superblock, 0, sizeof (struct
reiserfs_super_block),
+ (char *) &super)
+ || (substring (REISER3FS_SUPER_MAGIC_STRING, super.s_magic) > 0
+ && substring (REISER2FS_SUPER_MAGIC_STRING, super.s_magic) > 0
+ && substring (REISERFS_SUPER_MAGIC_STRING, super.s_magic) > 0)
+ || (/* check that this is not a copy inside the journal log */
+ super.s_journal_block * super.s_blocksize
+ <= REISERFS_DISK_OFFSET_IN_BYTES))
+ {
+ /* Try old super block position */
+ superblock = REISERFS_OLD_DISK_OFFSET_IN_BYTES >> SECTOR_BITS;
+ if (/*part_length < superblock + (sizeof (super) >> SECTOR_BITS)
+ || */ ! devread (ffi, superblock, 0, sizeof (struct
reiserfs_super_block),
+ (char *) &super))
+ return 0;
+
+ if (substring (REISER3FS_SUPER_MAGIC_STRING, super.s_magic) > 0
+ && substring (REISER2FS_SUPER_MAGIC_STRING, super.s_magic) > 0
+ && substring (REISERFS_SUPER_MAGIC_STRING, super.s_magic) > 0)
+ {
+ /* pre journaling super block ? */
+ if (substring (REISERFS_SUPER_MAGIC_STRING,
+ (char*) ((char *) &super + 20)) > 0)
+ return 0;
+
+ super.s_blocksize = REISERFS_OLD_BLOCKSIZE;
+ super.s_journal_block = 0;
+ super.s_version = 0;
+ }
+ }
+
+ /* check the version number. */
+ if (super.s_version > REISERFS_MAX_SUPPORTED_VERSION)
+ return 0;
+
+ INFO->version = super.s_version;
+ INFO->blocksize = super.s_blocksize;
+ INFO->fullblocksize_shift = log2 (super.s_blocksize);
+ INFO->blocksize_shift = INFO->fullblocksize_shift - SECTOR_BITS;
+ INFO->cached_slots =
+ (FSYSREISER_CACHE_SIZE >> INFO->fullblocksize_shift) - 1;
+
+#ifdef REISERDEBUG
+ printf ("reiserfs_mount: version=%d, blocksize=%d\n",
+ INFO->version, INFO->blocksize);
+#endif /* REISERDEBUG */
+
+ /* Clear node cache. */
+ memset (INFO->blocks, 0, sizeof (INFO->blocks));
+
+ if (super.s_blocksize < FSYSREISER_MIN_BLOCKSIZE
+ || super.s_blocksize > FSYSREISER_MAX_BLOCKSIZE
+ || (SECTOR_SIZE << INFO->blocksize_shift) != super.s_blocksize)
+ return 0;
+
+ /* Initialize journal code. If something fails we end with zero
+ * journal_transactions, so we don't access the journal at all.
+ */
+ INFO->journal_transactions = 0;
+ if (super.s_journal_block != 0 && super.s_journal_dev == 0)
+ {
+ INFO->journal_block = super.s_journal_block;
+ INFO->journal_block_count = super.s_journal_size;
+ if (is_power_of_two (INFO->journal_block_count))
+ journal_init (ffi);
+
+ /* Read in super block again, maybe it is in the journal */
+ block_read (ffi, superblock >> INFO->blocksize_shift,
+ 0, sizeof (struct reiserfs_super_block), (char *) &super);
+ }
+
+ if (! block_read (ffi, super.s_root_block, 0, INFO->blocksize, (char*) ROOT))
+ return 0;
+
+ INFO->tree_depth = BLOCKHEAD (ROOT)->blk_level;
+
+#ifdef REISERDEBUG
+ printf ("root read_in: block=%d, depth=%d\n",
+ super.s_root_block, INFO->tree_depth);
+#endif /* REISERDEBUG */
+
+ if (INFO->tree_depth >= MAX_HEIGHT)
+ return 0;
+ if (INFO->tree_depth == DISK_LEAF_NODE_LEVEL)
+ {
+ /* There is only one node in the whole filesystem,
+ * which is simultanously leaf and root */
+ memcpy (LEAF, ROOT, INFO->blocksize);
+ }
+ return 1;
+}
+
+/***************** TREE ACCESSING METHODS *****************************/
+
+/* I assume you are familiar with the ReiserFS tree, if not go to
+ * http://www.namesys.com/content_table.html
+ *
+ * My tree node cache is organized as following
+ * 0 ROOT node
+ * 1 LEAF node (if the ROOT is also a LEAF it is copied here
+ * 2-n other nodes on current path from bottom to top.
+ * if there is not enough space in the cache, the top most are
+ * omitted.
+ *
+ * I have only two methods to find a key in the tree:
+ * search_stat(dir_id, objectid) searches for the stat entry (always
+ * the first entry) of an object.
+ * next_key() gets the next key in tree order.
+ *
+ * This means, that I can only sequential reads of files are
+ * efficient, but this really doesn't hurt for grub.
+ */
+
+/* Read in the node at the current path and depth into the node cache.
+ * You must set INFO->blocks[depth] before.
+ */
+static char *
+read_tree_node (fsi_file_t *ffi, unsigned int blockNr, int depth)
+{
+ char* cache = CACHE(depth);
+ int num_cached = INFO->cached_slots;
+ if (depth < num_cached)
+ {
+ /* This is the cached part of the path. Check if same block is
+ * needed.
+ */
+ if (blockNr == INFO->blocks[depth])
+ return cache;
+ }
+ else
+ cache = CACHE(num_cached);
+
+#ifdef REISERDEBUG
+ printf (" next read_in: block=%d (depth=%d)\n",
+ blockNr, depth);
+#endif /* REISERDEBUG */
+ if (! block_read (ffi, blockNr, 0, INFO->blocksize, cache))
+ return 0;
+ /* Make sure it has the right node level */
+ if (BLOCKHEAD (cache)->blk_level != depth)
+ {
+ errnum = ERR_FSYS_CORRUPT;
+ return 0;
+ }
+
+ INFO->blocks[depth] = blockNr;
+ return cache;
+}
+
+/* Get the next key, i.e. the key following the last retrieved key in
+ * tree order. INFO->current_ih and
+ * INFO->current_info are adapted accordingly. */
+static int
+next_key (fsi_file_t *ffi)
+{
+ int depth;
+ struct item_head *ih = INFO->current_ih + 1;
+ char *cache;
+
+#ifdef REISERDEBUG
+ printf ("next_key:\n old ih: key %d:%d:%d:%d version:%d\n",
+ INFO->current_ih->ih_key.k_dir_id,
+ INFO->current_ih->ih_key.k_objectid,
+ INFO->current_ih->ih_key.u.v1.k_offset,
+ INFO->current_ih->ih_key.u.v1.k_uniqueness,
+ INFO->current_ih->ih_version);
+#endif /* REISERDEBUG */
+
+ if (ih == &ITEMHEAD[BLOCKHEAD (LEAF)->blk_nr_item])
+ {
+ depth = DISK_LEAF_NODE_LEVEL;
+ /* The last item, was the last in the leaf node.
+ * Read in the next block
+ */
+ do
+ {
+ if (depth == INFO->tree_depth)
+ {
+ /* There are no more keys at all.
+ * Return a dummy item with MAX_KEY */
+ ih = (struct item_head *) &BLOCKHEAD (LEAF)->blk_right_delim_key;
+ goto found;
+ }
+ depth++;
+#ifdef REISERDEBUG
+ printf (" depth=%d, i=%d\n", depth, INFO->next_key_nr[depth]);
+#endif /* REISERDEBUG */
+ }
+ while (INFO->next_key_nr[depth] == 0);
+
+ if (depth == INFO->tree_depth)
+ cache = ROOT;
+ else if (depth <= INFO->cached_slots)
+ cache = CACHE (depth);
+ else
+ {
+ cache = read_tree_node (ffi, INFO->blocks[depth], depth);
+ if (! cache)
+ return 0;
+ }
+
+ do
+ {
+ int nr_item = BLOCKHEAD (cache)->blk_nr_item;
+ int key_nr = INFO->next_key_nr[depth]++;
+#ifdef REISERDEBUG
+ printf (" depth=%d, i=%d/%d\n", depth, key_nr, nr_item);
+#endif /* REISERDEBUG */
+ if (key_nr == nr_item)
+ /* This is the last item in this block, set the next_key_nr to 0 */
+ INFO->next_key_nr[depth] = 0;
+
+ cache = read_tree_node (ffi, DC (cache)[key_nr].dc_block_number,
--depth);
+ if (! cache)
+ return 0;
+ }
+ while (depth > DISK_LEAF_NODE_LEVEL);
+
+ ih = ITEMHEAD;
+ }
+ found:
+ INFO->current_ih = ih;
+ INFO->current_item = &LEAF[ih->ih_item_location];
+#ifdef REISERDEBUG
+ printf (" new ih: key %d:%d:%d:%d version:%d\n",
+ INFO->current_ih->ih_key.k_dir_id,
+ INFO->current_ih->ih_key.k_objectid,
+ INFO->current_ih->ih_key.u.v1.k_offset,
+ INFO->current_ih->ih_key.u.v1.k_uniqueness,
+ INFO->current_ih->ih_version);
+#endif /* REISERDEBUG */
+ return 1;
+}
+
+/* preconditions: reiserfs_mount already executed, therefore
+ * INFO block is valid
+ * returns: 0 if error (errnum is set),
+ * nonzero iff we were able to find the key successfully.
+ * postconditions: on a nonzero return, the current_ih and
+ * current_item fields describe the key that equals the
+ * searched key. INFO->next_key contains the next key after
+ * the searched key.
+ * side effects: messes around with the cache.
+ */
+static int
+search_stat (fsi_file_t *ffi, __u32 dir_id, __u32 objectid)
+{
+ char *cache;
+ int depth;
+ int nr_item;
+ int i;
+ struct item_head *ih;
+#ifdef REISERDEBUG
+ printf ("search_stat:\n key %d:%d:0:0\n", dir_id, objectid);
+#endif /* REISERDEBUG */
+
+ depth = INFO->tree_depth;
+ cache = ROOT;
+
+ while (depth > DISK_LEAF_NODE_LEVEL)
+ {
+ struct key *key;
+ nr_item = BLOCKHEAD (cache)->blk_nr_item;
+
+ key = KEY (cache);
+
+ for (i = 0; i < nr_item; i++)
+ {
+ if (key->k_dir_id > dir_id
+ || (key->k_dir_id == dir_id
+ && (key->k_objectid > objectid
+ || (key->k_objectid == objectid
+ && (key->u.v1.k_offset
+ | key->u.v1.k_uniqueness) > 0))))
+ break;
+ key++;
+ }
+
+#ifdef REISERDEBUG
+ printf (" depth=%d, i=%d/%d\n", depth, i, nr_item);
+#endif /* REISERDEBUG */
+ INFO->next_key_nr[depth] = (i == nr_item) ? 0 : i+1;
+ cache = read_tree_node (ffi, DC (cache)[i].dc_block_number, --depth);
+ if (! cache)
+ return 0;
+ }
+
+ /* cache == LEAF */
+ nr_item = BLOCKHEAD (LEAF)->blk_nr_item;
+ ih = ITEMHEAD;
+ for (i = 0; i < nr_item; i++)
+ {
+ if (ih->ih_key.k_dir_id == dir_id
+ && ih->ih_key.k_objectid == objectid
+ && ih->ih_key.u.v1.k_offset == 0
+ && ih->ih_key.u.v1.k_uniqueness == 0)
+ {
+#ifdef REISERDEBUG
+ printf (" depth=%d, i=%d/%d\n", depth, i, nr_item);
+#endif /* REISERDEBUG */
+ INFO->current_ih = ih;
+ INFO->current_item = &LEAF[ih->ih_item_location];
+ return 1;
+ }
+ ih++;
+ }
+ errnum = ERR_FSYS_CORRUPT;
+ return 0;
+}
+
+int
+reiserfs_read (fsi_file_t *ffi, char *buf, int len)
+{
+ unsigned int blocksize;
+ unsigned int offset;
+ unsigned int to_read;
+ char *prev_buf = buf;
+
+#ifdef REISERDEBUG
+ printf ("reiserfs_read: filepos=%d len=%d, offset=%x:%x\n",
+ filepos, len, (__u64) IH_KEY_OFFSET (INFO->current_ih) - 1);
+#endif /* REISERDEBUG */
+
+ if (INFO->current_ih->ih_key.k_objectid != INFO->fileinfo.k_objectid
+ || IH_KEY_OFFSET (INFO->current_ih) > filepos + 1)
+ {
+ search_stat (ffi, INFO->fileinfo.k_dir_id, INFO->fileinfo.k_objectid);
+ goto get_next_key;
+ }
+
+ while (! errnum)
+ {
+ if (INFO->current_ih->ih_key.k_objectid != INFO->fileinfo.k_objectid)
+ break;
+
+ offset = filepos - IH_KEY_OFFSET (INFO->current_ih) + 1;
+ blocksize = INFO->current_ih->ih_item_len;
+
+#ifdef REISERDEBUG
+ printf (" loop: filepos=%d len=%d, offset=%d blocksize=%d\n",
+ filepos, len, offset, blocksize);
+#endif /* REISERDEBUG */
+
+ if (IH_KEY_ISTYPE(INFO->current_ih, TYPE_DIRECT)
+ && offset < blocksize)
+ {
+#ifdef REISERDEBUG
+ printf ("direct_read: offset=%d, blocksize=%d\n",
+ offset, blocksize);
+#endif /* REISERDEBUG */
+ to_read = blocksize - offset;
+ if (to_read > len)
+ to_read = len;
+
+ if (disk_read_hook != NULL)
+ {
+ disk_read_func = disk_read_hook;
+
+ block_read (ffi, INFO->blocks[DISK_LEAF_NODE_LEVEL],
+ (INFO->current_item - LEAF + offset), to_read, buf);
+
+ disk_read_func = NULL;
+ }
+ else
+ memcpy (buf, INFO->current_item + offset, to_read);
+ goto update_buf_len;
+ }
+ else if (IH_KEY_ISTYPE(INFO->current_ih, TYPE_INDIRECT))
+ {
+ blocksize = (blocksize >> 2) << INFO->fullblocksize_shift;
+#ifdef REISERDEBUG
+ printf ("indirect_read: offset=%d, blocksize=%d\n",
+ offset, blocksize);
+#endif /* REISERDEBUG */
+
+ while (offset < blocksize)
+ {
+ __u32 blocknr = ((__u32 *) INFO->current_item)
+ [offset >> INFO->fullblocksize_shift];
+ int blk_offset = offset & (INFO->blocksize-1);
+
+ to_read = INFO->blocksize - blk_offset;
+ if (to_read > len)
+ to_read = len;
+
+ disk_read_func = disk_read_hook;
+
+ /* Journal is only for meta data. Data blocks can be read
+ * directly without using block_read
+ */
+ devread (ffi, blocknr << INFO->blocksize_shift,
+ blk_offset, to_read, buf);
+
+ disk_read_func = NULL;
+ update_buf_len:
+ len -= to_read;
+ buf += to_read;
+ offset += to_read;
+ filepos += to_read;
+ if (len == 0)
+ goto done;
+ }
+ }
+ get_next_key:
+ next_key (ffi);
+ }
+ done:
+ return errnum ? 0 : buf - prev_buf;
+}
+
+
+/* preconditions: reiserfs_mount already executed, therefore
+ * INFO block is valid
+ * returns: 0 if error, nonzero iff we were able to find the file successfully
+ * postconditions: on a nonzero return, INFO->fileinfo contains the info
+ * of the file we were trying to look up, filepos is 0 and filemax is
+ * the size of the file.
+ */
+int
+reiserfs_dir (fsi_file_t *ffi, char *dirname)
+{
+ struct reiserfs_de_head *de_head;
+ char *rest, ch;
+ __u32 dir_id, objectid, parent_dir_id = 0, parent_objectid = 0;
+#ifndef STAGE1_5
+ int do_possibilities = 0;
+#endif /* ! STAGE1_5 */
+ char linkbuf[PATH_MAX]; /* buffer for following symbolic links */
+ int link_count = 0;
+ int mode;
+
+ dir_id = REISERFS_ROOT_PARENT_OBJECTID;
+ objectid = REISERFS_ROOT_OBJECTID;
+
+ while (1)
+ {
+#ifdef REISERDEBUG
+ printf ("dirname=%s\n", dirname);
+#endif /* REISERDEBUG */
+
+ /* Search for the stat info first. */
+ if (! search_stat (ffi, dir_id, objectid))
+ return 0;
+
+#ifdef REISERDEBUG
+ printf ("sd_mode=%x sd_size=%d\n",
+ ((struct stat_data *) INFO->current_item)->sd_mode,
+ ((struct stat_data *) INFO->current_item)->sd_size);
+#endif /* REISERDEBUG */
+
+ mode = ((struct stat_data *) INFO->current_item)->sd_mode;
+
+ /* If we've got a symbolic link, then chase it. */
+ if (S_ISLNK (mode))
+ {
+ int len;
+ if (++link_count > MAX_LINK_COUNT)
+ {
+ errnum = ERR_SYMLINK_LOOP;
+ return 0;
+ }
+
+ /* Get the symlink size. */
+ filemax = ((struct stat_data *) INFO->current_item)->sd_size;
+
+ /* Find out how long our remaining name is. */
+ len = 0;
+ while (dirname[len] && !isspace (dirname[len]))
+ len++;
+
+ if (filemax + len > sizeof (linkbuf) - 1)
+ {
+ errnum = ERR_FILELENGTH;
+ return 0;
+ }
+
+ /* Copy the remaining name to the end of the symlink data.
+ Note that DIRNAME and LINKBUF may overlap! */
+ grub_memmove (linkbuf + filemax, dirname, len+1);
+
+ INFO->fileinfo.k_dir_id = dir_id;
+ INFO->fileinfo.k_objectid = objectid;
+ filepos = 0;
+ if (! next_key (ffi)
+ || reiserfs_read (ffi, linkbuf, filemax) != filemax)
+ {
+ if (! errnum)
+ errnum = ERR_FSYS_CORRUPT;
+ return 0;
+ }
+
+#ifdef REISERDEBUG
+ printf ("symlink=%s\n", linkbuf);
+#endif /* REISERDEBUG */
+
+ dirname = linkbuf;
+ if (*dirname == '/')
+ {
+ /* It's an absolute link, so look it up in root. */
+ dir_id = REISERFS_ROOT_PARENT_OBJECTID;
+ objectid = REISERFS_ROOT_OBJECTID;
+ }
+ else
+ {
+ /* Relative, so look it up in our parent directory. */
+ dir_id = parent_dir_id;
+ objectid = parent_objectid;
+ }
+
+ /* Now lookup the new name. */
+ continue;
+ }
+
+ /* if we have a real file (and we're not just printing possibilities),
+ then this is where we want to exit */
+
+ if (! *dirname || isspace (*dirname))
+ {
+ if (! S_ISREG (mode))
+ {
+ errnum = ERR_BAD_FILETYPE;
+ return 0;
+ }
+
+ filepos = 0;
+ filemax = ((struct stat_data *) INFO->current_item)->sd_size;
+
+ /* If this is a new stat data and size is > 4GB set filemax to
+ * maximum
+ */
+ if (INFO->current_ih->ih_version == ITEM_VERSION_2
+ && ((struct stat_data *) INFO->current_item)->sd_size_hi > 0)
+ filemax = 0xffffffff;
+
+ INFO->fileinfo.k_dir_id = dir_id;
+ INFO->fileinfo.k_objectid = objectid;
+ return next_key (ffi);
+ }
+
+ /* continue with the file/directory name interpretation */
+ while (*dirname == '/')
+ dirname++;
+ if (! S_ISDIR (mode))
+ {
+ errnum = ERR_BAD_FILETYPE;
+ return 0;
+ }
+ for (rest = dirname; (ch = *rest) && ! isspace (ch) && ch != '/';
rest++);
+ *rest = 0;
+
+# ifndef STAGE1_5
+ if (print_possibilities && ch != '/')
+ do_possibilities = 1;
+# endif /* ! STAGE1_5 */
+
+ while (1)
+ {
+ char *name_end;
+ int num_entries;
+
+ if (! next_key (ffi))
+ return 0;
+#ifdef REISERDEBUG
+ printf ("ih: key %d:%d:%d:%d version:%d\n",
+ INFO->current_ih->ih_key.k_dir_id,
+ INFO->current_ih->ih_key.k_objectid,
+ INFO->current_ih->ih_key.u.v1.k_offset,
+ INFO->current_ih->ih_key.u.v1.k_uniqueness,
+ INFO->current_ih->ih_version);
+#endif /* REISERDEBUG */
+
+ if (INFO->current_ih->ih_key.k_objectid != objectid)
+ break;
+
+ name_end = INFO->current_item + INFO->current_ih->ih_item_len;
+ de_head = (struct reiserfs_de_head *) INFO->current_item;
+ num_entries = INFO->current_ih->u.ih_entry_count;
+ while (num_entries > 0)
+ {
+ char *filename = INFO->current_item + de_head->deh_location;
+ char tmp = *name_end;
+ if ((de_head->deh_state & DEH_Visible))
+ {
+ int cmp;
+ /* Directory names in ReiserFS are not null
+ * terminated. We write a temporary 0 behind it.
+ * NOTE: that this may overwrite the first block in
+ * the tree cache. That doesn't hurt as long as we
+ * don't call next_key () in between.
+ */
+ *name_end = 0;
+ cmp = substring (dirname, filename);
+ *name_end = tmp;
+# ifndef STAGE1_5
+ if (do_possibilities)
+ {
+ if (cmp <= 0)
+ {
+ if (print_possibilities > 0)
+ print_possibilities = -print_possibilities;
+ *name_end = 0;
+ print_a_completion (filename);
+ *name_end = tmp;
+ }
+ }
+ else
+# endif /* ! STAGE1_5 */
+ if (cmp == 0)
+ goto found;
+ }
+ /* The beginning of this name marks the end of the next name.
+ */
+ name_end = filename;
+ de_head++;
+ num_entries--;
+ }
+ }
+
+# ifndef STAGE1_5
+ if (print_possibilities < 0)
+ return 1;
+# endif /* ! STAGE1_5 */
+
+ errnum = ERR_FILE_NOT_FOUND;
+ *rest = ch;
+ return 0;
+
+ found:
+
+ *rest = ch;
+ dirname = rest;
+
+ parent_dir_id = dir_id;
+ parent_objectid = objectid;
+ dir_id = de_head->deh_dir_id;
+ objectid = de_head->deh_objectid;
+ }
+}
+
+int
+reiserfs_embed (fsi_file_t *ffi, int *start_sector, int needed_sectors)
+{
+ struct reiserfs_super_block super;
+ int num_sectors;
+
+ if (! devread (ffi, REISERFS_DISK_OFFSET_IN_BYTES >> SECTOR_BITS, 0,
+ sizeof (struct reiserfs_super_block), (char *) &super))
+ return 0;
+
+ *start_sector = 1; /* reserve first sector for stage1 */
+ if ((substring (REISERFS_SUPER_MAGIC_STRING, super.s_magic) <= 0
+ || substring (REISER2FS_SUPER_MAGIC_STRING, super.s_magic) <= 0
+ || substring (REISER3FS_SUPER_MAGIC_STRING, super.s_magic) <= 0)
+ && (/* check that this is not a super block copy inside
+ * the journal log */
+ super.s_journal_block * super.s_blocksize
+ > REISERFS_DISK_OFFSET_IN_BYTES))
+ num_sectors = (REISERFS_DISK_OFFSET_IN_BYTES >> SECTOR_BITS) - 1;
+ else
+ num_sectors = (REISERFS_OLD_DISK_OFFSET_IN_BYTES >> SECTOR_BITS) - 1;
+
+ return (needed_sectors <= num_sectors);
+}
+
+fsi_plugin_ops_t *
+fsi_init_plugin(int version, fsi_plugin_t *fp, const char **name)
+{
+ static fsig_plugin_ops_t ops = {
+ FSIMAGE_PLUGIN_VERSION,
+ .fpo_mount = reiserfs_mount,
+ .fpo_dir = reiserfs_dir,
+ .fpo_read = reiserfs_read
+ };
+
+ *name = "reiserfs";
+ return (fsig_init(fp, &ops));
+}
diff --git a/tools/libfsimage/ufs/Makefile b/tools/libfsimage/ufs/Makefile
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/ufs/Makefile
@@ -0,0 +1,13 @@
+XEN_ROOT = ../../..
+
+LIB_SRCS-y = fsys_ufs.c
+
+FS = ufs
+
+.PHONY: all
+all: fs-all
+
+.PHONY: install
+install: fs-install
+
+include $(XEN_ROOT)/tools/libfsimage/Rules.mk
diff --git a/tools/libfsimage/ufs/fsys_ufs.c b/tools/libfsimage/ufs/fsys_ufs.c
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/ufs/fsys_ufs.c
@@ -0,0 +1,276 @@
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 2006 Free Software Foundation, Inc.
+ *
+ * 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 of the License, 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.
+ */
+/*
+ * Copyright 2006 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+/* From Solaris usr/src/stand/lib/fs/ufs/ufsops.c */
+
+#include <fsimage_grub.h>
+
+#include "ufs.h"
+
+/* These are the pools of buffers, etc. */
+
+#define SUPERBLOCK ((struct fs *)(FSYS_BUF + 0x2000))
+#define INODE ((struct icommon *)(FSYS_BUF + 0x1000))
+#define DIRENT (FSYS_BUF + 0x4000)
+#define INDIRBLK1 ((grub_daddr32_t *)(FSYS_BUF + 0x4000)) /* 2+ indir blk */
+#define INDIRBLK0 ((grub_daddr32_t *)(FSYS_BUF+ 0x6000)) /* 1st
indirect blk */
+
+#define indirblk0 (*fsig_int1(ffi))
+#define indirblk1 (*fsig_int2(ffi))
+
+static int openi(fsi_file_t *, grub_ino_t);
+static grub_ino_t dlook(fsi_file_t *, grub_ino_t, char *);
+static grub_daddr32_t sbmap(fsi_file_t *, grub_daddr32_t);
+
+/* read superblock and check fs magic */
+int
+ufs_mount(fsi_file_t *ffi)
+{
+ if (/*! IS_PC_SLICE_TYPE_SOLARIS(current_slice) || */
+ !devread(ffi, UFS_SBLOCK, 0, UFS_SBSIZE, (char *)SUPERBLOCK) ||
+ SUPERBLOCK->fs_magic != UFS_MAGIC)
+ return 0;
+
+ return 1;
+}
+
+
+/*
+ * searching for a file, if successful, inode will be loaded in INODE
+ * The entry point should really be named ufs_open(char *pathname).
+ * For now, keep it consistent with the rest of fsys modules.
+ */
+int
+ufs_dir(fsi_file_t *ffi, char *dirname)
+{
+ grub_ino_t inode = ROOTINO; /* start from root */
+ char *fname, ch;
+
+ indirblk0 = indirblk1 = 0;
+
+ /* skip leading slashes */
+ while (*dirname == '/')
+ dirname++;
+
+ while (inode && *dirname && !isspace(*dirname)) {
+ if (!openi(ffi, inode))
+ return 0;
+
+ /* parse for next path component */
+ fname = dirname;
+ while (*dirname && !isspace(*dirname) && *dirname != '/')
+ dirname++;
+ ch = *dirname;
+ *dirname = 0; /* ensure null termination */
+
+ inode = dlook(ffi, inode, fname);
+ *dirname = ch;
+ while (*dirname == '/')
+ dirname++;
+ }
+
+ /* return 1 only if inode exists and is a regular file */
+ if (! openi(ffi, inode))
+ return (0);
+ filepos = 0;
+ filemax = INODE->ic_sizelo;
+ return (inode && ((INODE->ic_smode & IFMT) == IFREG));
+}
+
+/*
+ * This is the high-level read function.
+ */
+int
+ufs_read(fsi_file_t *ffi, char *buf, int len)
+{
+ int off, size, ret = 0, ok;
+ grub_daddr32_t lblk, dblk;
+
+ while (len) {
+ off = blkoff(SUPERBLOCK, filepos);
+ lblk = lblkno(SUPERBLOCK, filepos);
+ size = SUPERBLOCK->fs_bsize;
+ size -= off;
+ if (size > len)
+ size = len;
+
+ if ((dblk = sbmap(ffi, lblk)) <= 0) {
+ /* we are in a file hole, just zero the buf */
+ grub_memset(buf, 0, size);
+ } else {
+ disk_read_func = disk_read_hook;
+ ok = devread(ffi, fsbtodb(SUPERBLOCK, dblk),
+ off, size, buf);
+ disk_read_func = 0;
+ if (!ok)
+ return 0;
+ }
+ buf += size;
+ len -= size;
+ filepos += size;
+ ret += size;
+ }
+
+ return (ret);
+}
+
+int
+ufs_embed (int *start_sector, int needed_sectors)
+{
+ if (needed_sectors > 14)
+ return 0;
+
+ *start_sector = 2;
+ return 1;
+}
+
+/* read inode and place content in INODE */
+static int
+openi(fsi_file_t *ffi, grub_ino_t inode)
+{
+ grub_daddr32_t dblk;
+ int off;
+
+ /* get block and byte offset into the block */
+ dblk = fsbtodb(SUPERBLOCK, itod(SUPERBLOCK, inode));
+ off = itoo(SUPERBLOCK, inode) * sizeof (struct icommon);
+
+ return (devread(ffi, dblk, off, sizeof (struct icommon), (char
*)INODE));
+}
+
+/*
+ * Performs fileblock mapping. Convert file block no. to disk block no.
+ * Returns 0 when block doesn't exist and <0 when block isn't initialized
+ * (i.e belongs to a hole in the file).
+ */
+grub_daddr32_t
+sbmap(fsi_file_t *ffi, grub_daddr32_t bn)
+{
+ int level, bound, i, index;
+ grub_daddr32_t nb, blkno;
+ grub_daddr32_t *db = INODE->ic_db;
+
+ /* blocks 0..UFS_NDADDR are direct blocks */
+ if (bn < UFS_NDADDR) {
+ return db[bn];
+ }
+
+ /* determine how many levels of indirection. */
+ level = 0;
+ bn -= UFS_NDADDR;
+ bound = UFS_NINDIR(SUPERBLOCK);
+ while (bn >= bound) {
+ level++;
+ bn -= bound;
+ bound *= UFS_NINDIR(SUPERBLOCK);
+ }
+ if (level >= UFS_NIADDR) /* bn too big */
+ return ((grub_daddr32_t)0);
+
+ /* fetch the first indirect block */
+ nb = INODE->ic_ib[level];
+ if (nb == 0) {
+ return ((grub_daddr32_t)0);
+ }
+ if (indirblk0 != nb) {
+ indirblk0 = 0;
+ blkno = fsbtodb(SUPERBLOCK, nb);
+ if (!devread(ffi, blkno, 0, SUPERBLOCK->fs_bsize,
+ (char *)INDIRBLK0))
+ return (0);
+ indirblk0 = nb;
+ }
+ bound /= UFS_NINDIR(SUPERBLOCK);
+ index = (bn / bound) % UFS_NINDIR(SUPERBLOCK);
+ nb = INDIRBLK0[index];
+
+ /* fetch through the indirect blocks */
+ for (i = 1; i <= level; i++) {
+ if (indirblk1 != nb) {
+ blkno = fsbtodb(SUPERBLOCK, nb);
+ if (!devread(ffi, blkno, 0, SUPERBLOCK->fs_bsize,
+ (char *)INDIRBLK1))
+ return (0);
+ indirblk1 = nb;
+ }
+ bound /= UFS_NINDIR(SUPERBLOCK);
+ index = (bn / bound) % UFS_NINDIR(SUPERBLOCK);
+ nb = INDIRBLK1[index];
+ if (nb == 0)
+ return ((grub_daddr32_t)0);
+ }
+
+ return (nb);
+}
+
+/* search directory content for name, return inode number */
+static grub_ino_t
+dlook(fsi_file_t *ffi, grub_ino_t dir_ino, char *name)
+{
+ int loc, off;
+ grub_daddr32_t lbn, dbn, dblk;
+ struct direct *dp;
+
+ if ((INODE->ic_smode & IFMT) != IFDIR)
+ return 0;
+
+ loc = 0;
+ while (loc < INODE->ic_sizelo) {
+ /* offset into block */
+ off = blkoff(SUPERBLOCK, loc);
+ if (off == 0) { /* need to read in a new block */
+ /* get logical block number */
+ lbn = lblkno(SUPERBLOCK, loc);
+ /* resolve indrect blocks */
+ dbn = sbmap(ffi, lbn);
+ if (dbn == 0)
+ return (0);
+
+ dblk = fsbtodb(SUPERBLOCK, dbn);
+ if (!devread(ffi, dblk, 0, SUPERBLOCK->fs_bsize,
+ (char *)DIRENT)) {
+ return 0;
+ }
+ }
+
+ dp = (struct direct *)(DIRENT + off);
+ if (dp->d_ino && substring(name, dp->d_name) == 0)
+ return (dp->d_ino);
+ loc += dp->d_reclen;
+ }
+ return (0);
+}
+
+fsi_plugin_ops_t *
+fsi_init_plugin(int version, fsi_plugin_t *fp, const char **name)
+{
+ static fsig_plugin_ops_t ops = {
+ FSIMAGE_PLUGIN_VERSION,
+ .fpo_mount = ufs_mount,
+ .fpo_dir = ufs_dir,
+ .fpo_read = ufs_read
+ };
+
+ *name = "ufs";
+ return (fsig_init(fp, &ops));
+}
diff --git a/tools/libfsimage/ufs/ufs.h b/tools/libfsimage/ufs/ufs.h
new file mode 100644
--- /dev/null
+++ b/tools/libfsimage/ufs/ufs.h
@@ -0,0 +1,228 @@
+/*
+ * Copyright 2005 Sun Microsystems, Inc. All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _GRUB_UFS_H
+#define _GRUB_UFS_H_
+
+/* ufs specific constants */
+#define UFS_SBLOCK 16
+#define UFS_SBSIZE 8192
+#define UFS_MAGIC 0x011954
+#define ROOTINO 2 /* i number of all roots */
+#define UFS_NDADDR 12 /* direct blocks */
+#define UFS_NIADDR 3 /* indirect blocks */
+#define MAXMNTLEN 512
+#define MAXCSBUFS 32
+#define MAXNAMELEN 256
+
+/* file types */
+#define IFMT 0xf000
+#define IFREG 0x8000
+#define IFDIR 0x4000
+
+typedef unsigned char grub_uchar_t;
+typedef unsigned short grub_ushort_t;
+typedef unsigned short grub_o_mode_t;
+typedef unsigned short grub_o_uid_t;
+typedef unsigned short grub_o_gid_t;
+typedef uint32_t grub_ino_t;
+typedef int32_t grub_int32_t;
+typedef int32_t grub_uid_t;
+typedef int32_t grub_gid_t;
+typedef uint32_t grub_uint32_t;
+typedef uint32_t grub_daddr32_t;
+typedef uint32_t grub_time32_t;
+typedef struct { int val[2]; } grub_quad_t;
+
+struct timeval32 {
+ grub_time32_t tv_sec;
+ grub_int32_t tv_usec;
+};
+
+/*
+ * Per cylinder group information; summarized in blocks allocated
+ * from first cylinder group data blocks. These blocks have to be
+ * read in from fs_csaddr (size fs_cssize) in addition to the
+ * super block.
+ *
+ * N.B. sizeof (struct csum) must be a power of two in order for
+ * the ``fs_cs'' macro to work (see below).
+ */
+struct csum {
+ grub_int32_t cs_ndir; /* number of directories */
+ grub_int32_t cs_nbfree; /* number of free blocks */
+ grub_int32_t cs_nifree; /* number of free inodes */
+ grub_int32_t cs_nffree; /* number of free frags */
+};
+
+/* Ufs super block */
+struct fs {
+ grub_uint32_t fs_link; /* linked list of file systems */
+ grub_uint32_t fs_rolled; /* logging only: fs fully rolled */
+ grub_daddr32_t fs_sblkno; /* addr of super-block in filesys */
+ grub_daddr32_t fs_cblkno; /* offset of cyl-block in filesys */
+ grub_daddr32_t fs_iblkno; /* offset of inode-blocks in filesys */
+ grub_daddr32_t fs_dblkno; /* offset of first data after cg */
+ grub_int32_t fs_cgoffset; /* cylinder group offset in cylinder */
+ grub_int32_t fs_cgmask; /* used to calc mod fs_ntrak */
+ grub_time32_t fs_time; /* last time written */
+ grub_int32_t fs_size; /* number of blocks in fs */
+ grub_int32_t fs_dsize; /* number of data blocks in fs */
+ grub_int32_t fs_ncg; /* number of cylinder groups */
+ grub_int32_t fs_bsize; /* size of basic blocks in fs */
+ grub_int32_t fs_fsize; /* size of frag blocks in fs */
+ grub_int32_t fs_frag; /* number of frags in a block in fs */
+ /* these are configuration parameters */
+ grub_int32_t fs_minfree; /* minimum percentage of free blocks */
+ grub_int32_t fs_rotdelay; /* num of ms for optimal next block */
+ grub_int32_t fs_rps; /* disk revolutions per second */
+ /* these fields can be computed from the others */
+ grub_int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */
+ grub_int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */
+ grub_int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */
+ grub_int32_t fs_fshift; /* ``numfrags'' calc number of frags */
+ /* these are configuration parameters */
+ grub_int32_t fs_maxcontig; /* max number of contiguous blks */
+ grub_int32_t fs_maxbpg; /* max number of blks per cyl group */
+ /* these fields can be computed from the others */
+ grub_int32_t fs_fragshift; /* block to frag shift */
+ grub_int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */
+ grub_int32_t fs_sbsize; /* actual size of super block */
+ grub_int32_t fs_csmask; /* csum block offset */
+ grub_int32_t fs_csshift; /* csum block number */
+ grub_int32_t fs_nindir; /* value of NINDIR */
+ grub_int32_t fs_inopb; /* value of INOPB */
+ grub_int32_t fs_nspf; /* value of NSPF */
+ /* yet another configuration parameter */
+ grub_int32_t fs_optim; /* optimization preference, see below */
+ /* these fields are derived from the hardware */
+ /* USL SVR4 compatibility */
+ /*
+ * * USL SVR4 compatibility
+ *
+ * There was a significant divergence here between Solaris and
+ * SVR4 for x86. By swapping these two members in the superblock,
+ * we get read-only compatibility of SVR4 filesystems. Otherwise
+ * there would be no compatibility. This change was introduced
+ * during bootstrapping of Solaris on x86. By making this ifdef'ed
+ * on byte order, we provide ongoing compatibility across all
+ * platforms with the same byte order, the highest compatibility
+ * that can be achieved.
+ */
+ grub_int32_t fs_state; /* file system state time stamp */
+ grub_int32_t fs_si; /* summary info state - lufs only */
+ grub_int32_t fs_trackskew; /* sector 0 skew, per track */
+ /* unique id for this filesystem (currently unused and unmaintained) */
+ /* In 4.3 Tahoe this space is used by fs_headswitch and fs_trkseek */
+ /* Neither of those fields is used in the Tahoe code right now but */
+ /* there could be problems if they are. */
+ grub_int32_t fs_id[2]; /* file system id */
+ /* sizes determined by number of cylinder groups and their sizes */
+ grub_daddr32_t fs_csaddr; /* blk addr of cyl grp summary area */
+ grub_int32_t fs_cssize; /* size of cyl grp summary area */
+ grub_int32_t fs_cgsize; /* cylinder group size */
+ /* these fields are derived from the hardware */
+ grub_int32_t fs_ntrak; /* tracks per cylinder */
+ grub_int32_t fs_nsect; /* sectors per track */
+ grub_int32_t fs_spc; /* sectors per cylinder */
+ /* this comes from the disk driver partitioning */
+ grub_int32_t fs_ncyl; /* cylinders in file system */
+ /* these fields can be computed from the others */
+ grub_int32_t fs_cpg; /* cylinders per group */
+ grub_int32_t fs_ipg; /* inodes per group */
+ grub_int32_t fs_fpg; /* blocks per group * fs_frag */
+ /* this data must be re-computed after crashes */
+ struct csum fs_cstotal; /* cylinder summary information */
+ /* these fields are cleared at mount time */
+ char fs_fmod; /* super block modified flag */
+ char fs_clean; /* file system state flag */
+ char fs_ronly; /* mounted read-only flag */
+ char fs_flags; /* largefiles flag, etc. */
+ char fs_fsmnt[MAXMNTLEN]; /* name mounted on */
+ /* these fields retain the current block allocation info */
+ grub_int32_t fs_cgrotor; /* last cg searched */
+ /*
+ * The following used to be fs_csp[MAXCSBUFS]. It was not
+ * used anywhere except in old utilities. We removed this
+ * in 5.6 and expect fs_u.fs_csp to be used instead.
+ * We no longer limit fs_cssize based on MAXCSBUFS.
+ */
+ union { /* fs_cs (csum) info */
+ grub_uint32_t fs_csp_pad[MAXCSBUFS];
+ struct csum *fs_csp;
+ } fs_u;
+ grub_int32_t fs_cpc; /* cyl per cycle in postbl */
+ short fs_opostbl[16][8]; /* old rotation block list head */
+ grub_int32_t fs_sparecon[51]; /* reserved for future constants */
+ grub_int32_t fs_version; /* minor version of MTB ufs */
+ grub_int32_t fs_logbno; /* block # of embedded log */
+ grub_int32_t fs_reclaim; /* reclaim open, deleted files */
+ grub_int32_t fs_sparecon2; /* reserved for future constant */
+ /* USL SVR4 compatibility */
+ grub_int32_t fs_npsect; /* # sectors/track including spares */
+ grub_quad_t fs_qbmask; /* ~fs_bmask - for use with quad size */
+ grub_quad_t fs_qfmask; /* ~fs_fmask - for use with quad size */
+ grub_int32_t fs_postblformat; /* fmt of positional layout tables */
+ grub_int32_t fs_nrpos; /* number of rotaional positions */
+ grub_int32_t fs_postbloff; /* (short) rotation block list head */
+ grub_int32_t fs_rotbloff; /* (grub_uchar_t) blocks for each */
+ /* rotation */
+ grub_int32_t fs_magic; /* magic number */
+ grub_uchar_t fs_space[1]; /* list of blocks for each rotation */
+ /* actually longer */
+};
+
+struct icommon {
+ grub_o_mode_t ic_smode; /* 0: mode and type of file */
+ short ic_nlink; /* 2: number of links to file */
+ grub_o_uid_t ic_suid; /* 4: owner's user id */
+ grub_o_gid_t ic_sgid; /* 6: owner's group id */
+ grub_uint32_t ic_sizelo; /* 8: number of bytes in file */
+ grub_uint32_t ic_sizehi; /* 12: number of bytes in file */
+ struct timeval32 ic_atime; /* 16: time last accessed */
+ struct timeval32 ic_mtime; /* 24: time last modified */
+ struct timeval32 ic_ctime; /* 32: last time inode changed */
+ grub_daddr32_t ic_db[UFS_NDADDR]; /* 40: disk block addresses */
+ grub_daddr32_t ic_ib[UFS_NIADDR]; /* 88: indirect blocks */
+ grub_int32_t ic_flags; /* 100: cflags */
+ grub_int32_t ic_blocks; /* 104: 512 byte blocks actually held */
+ grub_int32_t ic_gen; /* 108: generation number */
+ grub_int32_t ic_shadow; /* 112: shadow inode */
+ grub_uid_t ic_uid; /* 116: long EFT version of uid */
+ grub_gid_t ic_gid; /* 120: long EFT version of gid */
+ grub_uint32_t ic_oeftflag; /* 124: extended attr directory ino, */
+ /* 0 = none */
+};
+
+struct direct {
+ grub_ino_t d_ino;
+ grub_ushort_t d_reclen;
+ grub_ushort_t d_namelen;
+ char d_name[MAXNAMELEN + 1];
+};
+
+/* inode macros */
+#define INOPB(fs) ((fs)->fs_inopb)
+#define itoo(fs, x) ((x) % (grub_uint32_t)INOPB(fs))
+#define itog(fs, x) ((x) / (grub_uint32_t)(fs)->fs_ipg)
+#define itod(fs, x) ((grub_daddr32_t)(cgimin(fs, itog(fs, x)) + \
+ (blkstofrags((fs), \
+ ((x) % (grub_uint32_t)(fs)->fs_ipg / (grub_uint32_t)INOPB(fs))))))
+
+/* block conversion macros */
+#define UFS_NINDIR(fs) ((fs)->fs_nindir) /* # of indirects */
+#define blkoff(fs, loc) ((int)((loc & ~(fs)->fs_bmask)))
+#define lblkno(fs, loc) ((grub_int32_t)((loc) >> (fs)->fs_bshift))
+/* frag to blk */
+#define fsbtodb(fs, b) (((grub_daddr32_t)(b)) << (fs)->fs_fsbtodb)
+#define blkstofrags(fs, b) ((b) << (fs)->fs_fragshift)
+
+/* cynlinder group macros */
+#define cgbase(fs, c) ((grub_daddr32_t)((fs)->fs_fpg * (c)))
+#define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode
block */
+#define cgstart(fs, c) \
+ (cgbase(fs, c) + (fs)->fs_cgoffset * ((c) & ~((fs)->fs_cgmask)))
+
+#endif /* !_GRUB_UFS_H */
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home