X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Funix_capture.c;h=a6c569323757ee14e68ef1e0012570ddb936e996;hb=5a50cccf768a6bf6e9ccfd2337bcedc0cbf1ba6c;hp=f7dd5074e7df7208d4cfc4d8d05a34bfb5e7fd39;hpb=d5a64bb4299105c93b4e9a95e97b579c6f9254af;p=wimlib

diff --git a/src/unix_capture.c b/src/unix_capture.c
index f7dd5074..a6c56932 100644
--- a/src/unix_capture.c
+++ b/src/unix_capture.c
@@ -3,7 +3,7 @@
  */
 
 /*
- * Copyright (C) 2012, 2013, 2014 Eric Biggers
+ * Copyright (C) 2012-2018 Eric Biggers
  *
  * This file is free software; you can redistribute it and/or modify it under
  * the terms of the GNU Lesser General Public License as published by the Free
@@ -19,7 +19,7 @@
  * along with this file; if not, see http://www.gnu.org/licenses/.
  */
 
-#ifndef __WIN32__
+#ifndef _WIN32
 
 #ifdef HAVE_CONFIG_H
 #  include "config.h"
@@ -28,17 +28,21 @@
 #include <dirent.h>
 #include <errno.h>
 #include <fcntl.h>
-#include <limits.h> /* for PATH_MAX */
 #include <sys/stat.h>
+#include <sys/types.h>
+#ifdef HAVE_SYS_XATTR_H
+#  include <sys/xattr.h>
+#endif
 #include <unistd.h>
 
 #include "wimlib/blob_table.h"
-#include "wimlib/capture.h"
 #include "wimlib/dentry.h"
 #include "wimlib/error.h"
 #include "wimlib/reparse.h"
+#include "wimlib/scan.h"
 #include "wimlib/timestamp.h"
 #include "wimlib/unix_data.h"
+#include "wimlib/xattr.h"
 
 #ifdef HAVE_FDOPENDIR
 #  define my_fdopendir(dirfd_p) fdopendir(*(dirfd_p))
@@ -99,14 +103,197 @@ my_fdopendir(int *dirfd_p)
 #  define AT_SYMLINK_NOFOLLOW	0x100
 #endif
 
+#ifdef HAVE_LINUX_XATTR_SUPPORT
+/*
+ * Retrieves the values of the xattrs named by the null-terminated @names of the
+ * file at @path and serializes the xattr names and values into @entries.  If
+ * successful, returns the number of bytes used in @entries.  If unsuccessful,
+ * returns -1 and sets errno (ERANGE if @entries was too small).
+ */
+static ssize_t
+gather_xattr_entries(const char *path, const char *names, size_t names_size,
+		     void *entries, size_t entries_size)
+{
+	const char * const names_end = names + names_size;
+	void * const entries_end = entries + entries_size;
+	const char *name = names;
+	struct wim_xattr_entry *entry = entries;
+
+	do {
+		size_t name_len = strnlen(name, names_end - name);
+		void *value;
+		ssize_t value_len;
+
+		if (name_len == 0 || name_len >= names_end - name) {
+			ERROR("\"%s\": malformed extended attribute names list",
+			      path);
+			errno = EINVAL;
+			return -1;
+		}
+
+		if (name_len > WIM_XATTR_NAME_MAX) {
+			WARNING("\"%s\": name of extended attribute \"%s\" is too long to store",
+				path, name);
+			goto next_name;
+		}
+
+		/*
+		 * Take care to always call lgetxattr() with a nonzero size,
+		 * since zero size means to return the value length only.
+		 */
+		if (entries_end - (void *)entry <=
+		    sizeof(*entry) + name_len + 1) {
+			errno = ERANGE;
+			return -1;
+		}
+
+		entry->name_len = name_len;
+		entry->flags = 0;
+		value = mempcpy(entry->name, name, name_len + 1);
+
+		value_len = lgetxattr(path, name, value, entries_end - value);
+		if (value_len < 0) {
+			if (errno != ERANGE) {
+				ERROR_WITH_ERRNO("\"%s\": unable to read extended attribute \"%s\"",
+						 path, name);
+			}
+			return -1;
+		}
+		if (value_len > WIM_XATTR_SIZE_MAX) {
+			WARNING("\"%s\": value of extended attribute \"%s\" is too large to store",
+				path, name);
+			goto next_name;
+		}
+		entry->value_len = cpu_to_le16(value_len);
+		entry = value + value_len;
+	next_name:
+		name += name_len + 1;
+	} while (name < names_end);
+
+	return (void *)entry - entries;
+}
+
 static int
-unix_scan_regular_file(const char *path, u64 size, struct wim_inode *inode,
+create_xattr_item(const char *path, struct wim_inode *inode,
+		  const char *names, size_t names_size)
+{
+	char _entries[1024];
+	char *entries = _entries;
+	size_t entries_avail = ARRAY_LEN(_entries);
+	ssize_t entries_size;
+	int ret;
+
+retry:
+	/* Serialize the xattrs into @entries */
+	entries_size = gather_xattr_entries(path, names, names_size,
+					    entries, entries_avail);
+	if (entries_size < 0) {
+		ret = WIMLIB_ERR_STAT;
+		if (errno != ERANGE)
+			goto out;
+		/* Not enough space in @entries.  Reallocate it. */
+		if (entries != _entries)
+			FREE(entries);
+		ret = WIMLIB_ERR_NOMEM;
+		entries_avail *= 2;
+		entries = MALLOC(entries_avail);
+		if (!entries)
+			goto out;
+		goto retry;
+	}
+
+	/* Copy @entries into an xattr item associated with @inode */
+	if ((u32)entries_size != entries_size) {
+		ERROR("\"%s\": too much xattr data!", path);
+		ret = WIMLIB_ERR_STAT;
+		goto out;
+	}
+	ret = WIMLIB_ERR_NOMEM;
+	if (!inode_set_xattrs(inode, entries, entries_size))
+		goto out;
+
+	ret = 0;
+out:
+	if (entries != _entries)
+		FREE(entries);
+	return ret;
+}
+
+/*
+ * If the file at @path has Linux-style extended attributes, read them into
+ * memory and add them to @inode as a tagged item.
+ */
+static noinline_for_stack int
+scan_linux_xattrs(const char *path, struct wim_inode *inode)
+{
+	char _names[256];
+	char *names = _names;
+	ssize_t names_size = ARRAY_LEN(_names);
+	int ret = 0;
+
+retry:
+	/* Gather the names of the xattrs of the file at @path */
+	names_size = llistxattr(path, names, names_size);
+	if (names_size == 0) /* No xattrs? */
+		goto out;
+	if (names_size < 0) {
+		/* xattrs unsupported or disabled? */
+		if (errno == ENOTSUP || errno == ENOSYS)
+			goto out;
+		if (errno == ERANGE) {
+			/*
+			 * Not enough space in @names.  Ask for how much space
+			 * we need, then try again.
+			 */
+			names_size = llistxattr(path, NULL, 0);
+			if (names_size == 0)
+				goto out;
+			if (names_size > 0) {
+				if (names != _names)
+					FREE(names);
+				names = MALLOC(names_size);
+				if (!names) {
+					ret = WIMLIB_ERR_NOMEM;
+					goto out;
+				}
+				goto retry;
+			}
+		}
+		/* Some other error occurred. */
+		ERROR_WITH_ERRNO("\"%s\": unable to list extended attributes",
+				 path);
+		ret = WIMLIB_ERR_STAT;
+		goto out;
+	}
+
+	/*
+	 * We have a nonempty list of xattr names.  Gather the xattr values and
+	 * add them as a tagged item.
+	 */
+	ret = create_xattr_item(path, inode, names, names_size);
+out:
+	if (names != _names)
+		FREE(names);
+	return ret;
+}
+#endif /* HAVE_LINUX_XATTR_SUPPORT */
+
+static int
+unix_scan_regular_file(const char *path, u64 blocks, u64 size,
+		       struct wim_inode *inode,
 		       struct list_head *unhashed_blobs)
 {
 	struct blob_descriptor *blob = NULL;
 	struct wim_inode_stream *strm;
 
-	inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
+	/*
+	 * Set FILE_ATTRIBUTE_SPARSE_FILE if the file uses less disk space than
+	 * expected given its size.
+	 */
+	if (blocks < DIV_ROUND_UP(size, 512))
+		inode->i_attributes = FILE_ATTRIBUTE_SPARSE_FILE;
+	else
+		inode->i_attributes = FILE_ATTRIBUTE_NORMAL;
 
 	if (size) {
 		blob = new_blob_descriptor();
@@ -134,31 +321,31 @@ err_nomem:
 
 static int
 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
-				 char *path, size_t path_len,
 				 int dirfd, const char *relpath,
-				 struct capture_params *params);
+				 struct scan_params *params);
 
 static int
 unix_scan_directory(struct wim_dentry *dir_dentry,
-		    char *full_path, size_t full_path_len,
 		    int parent_dirfd, const char *dir_relpath,
-		    struct capture_params *params)
+		    struct scan_params *params)
 {
 
 	int dirfd;
 	DIR *dir;
 	int ret;
 
-	dirfd = my_openat(full_path, parent_dirfd, dir_relpath, O_RDONLY);
+	dirfd = my_openat(params->cur_path, parent_dirfd, dir_relpath, O_RDONLY);
 	if (dirfd < 0) {
-		ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
+		ERROR_WITH_ERRNO("\"%s\": Can't open directory",
+				 params->cur_path);
 		return WIMLIB_ERR_OPENDIR;
 	}
 
 	dir_dentry->d_inode->i_attributes = FILE_ATTRIBUTE_DIRECTORY;
 	dir = my_fdopendir(&dirfd);
 	if (!dir) {
-		ERROR_WITH_ERRNO("\"%s\": Can't open directory", full_path);
+		ERROR_WITH_ERRNO("\"%s\": Can't open directory",
+				 params->cur_path);
 		close(dirfd);
 		return WIMLIB_ERR_OPENDIR;
 	}
@@ -168,6 +355,7 @@ unix_scan_directory(struct wim_dentry *dir_dentry,
 		struct dirent *entry;
 		struct wim_dentry *child;
 		size_t name_len;
+		size_t orig_path_len;
 
 		errno = 0;
 		entry = readdir(dir);
@@ -175,30 +363,26 @@ unix_scan_directory(struct wim_dentry *dir_dentry,
 			if (errno) {
 				ret = WIMLIB_ERR_READ;
 				ERROR_WITH_ERRNO("\"%s\": Error reading directory",
-						 full_path);
+						 params->cur_path);
 			}
 			break;
 		}
 
-		if (entry->d_name[0] == '.' &&
-		    (entry->d_name[1] == '\0' ||
-		     (entry->d_name[1] == '.' && entry->d_name[2] == '\0')))
+		name_len = strlen(entry->d_name);
+
+		if (should_ignore_filename(entry->d_name, name_len))
 			continue;
 
-		full_path[full_path_len] = '/';
-		name_len = strlen(entry->d_name);
-		memcpy(&full_path[full_path_len + 1], entry->d_name, name_len + 1);
-		ret = unix_build_dentry_tree_recursive(&child,
-						       full_path,
-						       full_path_len + 1 + name_len,
-						       dirfd,
-						       &full_path[full_path_len + 1],
-						       params);
-		full_path[full_path_len] = '\0';
+		ret = WIMLIB_ERR_NOMEM;
+		if (!pathbuf_append_name(params, entry->d_name, name_len,
+					 &orig_path_len))
+			break;
+		ret = unix_build_dentry_tree_recursive(&child, dirfd,
+						       entry->d_name, params);
+		pathbuf_truncate(params, orig_path_len);
 		if (ret)
 			break;
-		if (child)
-			dentry_add_child(dir_dentry, child);
+		attach_scanned_tree(dir_dentry, child, params->blob_table);
 	}
 	closedir(dir);
 	return ret;
@@ -278,23 +462,24 @@ unix_relativize_link_target(char *target, u64 ino, u64 dev)
 }
 
 static noinline_for_stack int
-unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
-		  struct wim_inode *inode, struct capture_params *params)
+unix_scan_symlink(int dirfd, const char *relpath,
+		  struct wim_inode *inode, struct scan_params *params)
 {
 	char orig_target[REPARSE_POINT_MAX_SIZE];
 	char *target = orig_target;
 	int ret;
 
 	/* Read the UNIX symbolic link target.  */
-	ret = my_readlinkat(full_path, dirfd, relpath, target,
+	ret = my_readlinkat(params->cur_path, dirfd, relpath, target,
 			    sizeof(orig_target));
 	if (unlikely(ret < 0)) {
 		ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
-				 full_path);
+				 params->cur_path);
 		return WIMLIB_ERR_READLINK;
 	}
 	if (unlikely(ret >= sizeof(orig_target))) {
-		ERROR("\"%s\": target of symbolic link is too long", full_path);
+		ERROR("\"%s\": target of symbolic link is too long",
+		      params->cur_path);
 		return WIMLIB_ERR_READLINK;
 	}
 	target[ret] = '\0';
@@ -304,7 +489,6 @@ unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 	if (target[0] == '/' && (params->add_flags & WIMLIB_ADD_FLAG_RPFIX)) {
 		int status = WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK;
 
-		params->progress.scan.cur_path = full_path;
 		params->progress.scan.symlink_target = target;
 
 		target = unix_relativize_link_target(target,
@@ -315,22 +499,28 @@ unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 			inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
 			status = WIMLIB_SCAN_DENTRY_FIXED_SYMLINK;
 		}
-		ret = do_capture_progress(params, status, NULL);
+		ret = do_scan_progress(params, status, NULL);
 		if (ret)
 			return ret;
 	}
 
 	/* Translate the UNIX symlink target into a Windows reparse point.  */
 	ret = wim_inode_set_symlink(inode, target, params->blob_table);
-	if (ret)
+	if (unlikely(ret)) {
+		if (ret == WIMLIB_ERR_INVALID_UTF8_STRING) {
+			ERROR("\"%s\": target of symbolic link is not valid "
+			      "UTF-8.  This is not supported.",
+			      params->cur_path);
+		}
 		return ret;
+	}
 
 	/* On Windows, a reparse point can be set on both directory and
 	 * non-directory files.  Usually, a link that is intended to point to a
 	 * (non-)directory is stored as a reparse point on a (non-)directory
 	 * file.  Replicate this behavior by examining the target file.  */
 	struct stat stbuf;
-	if (my_fstatat(full_path, dirfd, relpath, &stbuf, 0) == 0 &&
+	if (my_fstatat(params->cur_path, dirfd, relpath, &stbuf, 0) == 0 &&
 	    S_ISDIR(stbuf.st_mode))
 		inode->i_attributes |= FILE_ATTRIBUTE_DIRECTORY;
 	return 0;
@@ -338,9 +528,8 @@ unix_scan_symlink(const char *full_path, int dirfd, const char *relpath,
 
 static int
 unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
-				 char *full_path, size_t full_path_len,
 				 int dirfd, const char *relpath,
-				 struct capture_params *params)
+				 struct scan_params *params)
 {
 	struct wim_dentry *tree = NULL;
 	struct wim_inode *inode = NULL;
@@ -348,7 +537,7 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 	struct stat stbuf;
 	int stat_flags;
 
-	ret = try_exclude(full_path, params);
+	ret = try_exclude(params);
 	if (unlikely(ret < 0)) /* Excluded? */
 		goto out_progress;
 	if (unlikely(ret > 0)) /* Error? */
@@ -360,10 +549,11 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 	else
 		stat_flags = AT_SYMLINK_NOFOLLOW;
 
-	ret = my_fstatat(full_path, dirfd, relpath, &stbuf, stat_flags);
+	ret = my_fstatat(params->cur_path, dirfd, relpath, &stbuf, stat_flags);
 
 	if (ret) {
-		ERROR_WITH_ERRNO("\"%s\": Can't read metadata", full_path);
+		ERROR_WITH_ERRNO("\"%s\": Can't read metadata",
+				 params->cur_path);
 		ret = WIMLIB_ERR_STAT;
 		goto out;
 	}
@@ -377,23 +567,26 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 			    WIMLIB_ADD_FLAG_NO_UNSUPPORTED_EXCLUDE)
 			{
 				ERROR("\"%s\": File type is unsupported",
-				      full_path);
+				      params->cur_path);
 				ret = WIMLIB_ERR_UNSUPPORTED_FILE;
 				goto out;
 			}
-			params->progress.scan.cur_path = full_path;
-			ret = do_capture_progress(params,
-						  WIMLIB_SCAN_DENTRY_UNSUPPORTED,
-						  NULL);
+			ret = do_scan_progress(params,
+					       WIMLIB_SCAN_DENTRY_UNSUPPORTED,
+					       NULL);
 			goto out;
 		}
 	}
 
 	ret = inode_table_new_dentry(params->inode_table, relpath,
-				     stbuf.st_ino, stbuf.st_dev,
-				     S_ISDIR(stbuf.st_mode), &tree);
-	if (ret)
+				     stbuf.st_ino, stbuf.st_dev, false, &tree);
+	if (unlikely(ret)) {
+		if (ret == WIMLIB_ERR_INVALID_UTF8_STRING) {
+			ERROR("\"%s\": filename is not valid UTF-8.  "
+			      "This is not supported.", params->cur_path);
+		}
 		goto out;
+	}
 
 	inode = tree->d_inode;
 
@@ -421,6 +614,11 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 			ret = WIMLIB_ERR_NOMEM;
 			goto out;
 		}
+#ifdef HAVE_LINUX_XATTR_SUPPORT
+		ret = scan_linux_xattrs(params->cur_path, inode);
+		if (ret)
+			goto out;
+#endif
 	}
 
 	if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
@@ -430,30 +628,28 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 	}
 
 	if (S_ISREG(stbuf.st_mode)) {
-		ret = unix_scan_regular_file(full_path, stbuf.st_size,
-					     inode, params->unhashed_blobs);
+		ret = unix_scan_regular_file(params->cur_path, stbuf.st_blocks,
+					     stbuf.st_size, inode,
+					     params->unhashed_blobs);
 	} else if (S_ISDIR(stbuf.st_mode)) {
-		ret = unix_scan_directory(tree, full_path, full_path_len,
-					  dirfd, relpath, params);
+		ret = unix_scan_directory(tree, dirfd, relpath, params);
 	} else if (S_ISLNK(stbuf.st_mode)) {
-		ret = unix_scan_symlink(full_path, dirfd, relpath,
-					inode, params);
+		ret = unix_scan_symlink(dirfd, relpath, inode, params);
 	}
 
 	if (ret)
 		goto out;
 
 out_progress:
-	params->progress.scan.cur_path = full_path;
 	if (likely(tree))
-		ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
+		ret = do_scan_progress(params, WIMLIB_SCAN_DENTRY_OK, inode);
 	else
-		ret = do_capture_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
+		ret = do_scan_progress(params, WIMLIB_SCAN_DENTRY_EXCLUDED, NULL);
 out:
 	if (unlikely(ret)) {
 		free_dentry_tree(tree, params->blob_table);
 		tree = NULL;
-		ret = report_capture_error(params, ret, full_path);
+		ret = report_scan_error(params, ret);
 	}
 	*tree_ret = tree;
 	return ret;
@@ -464,13 +660,12 @@ out:
  *	Builds a tree of WIM dentries from an on-disk directory tree (UNIX
  *	version; no NTFS-specific data is captured).
  *
- * @root_ret:   Place to return a pointer to the root of the dentry tree.  Only
- *		modified if successful.  Set to NULL if the file or directory was
- *		excluded from capture.
+ * @root_ret:   Place to return a pointer to the root of the dentry tree.  Set
+ *		to NULL if the file or directory was excluded from capture.
  *
  * @root_disk_path:  The path to the root of the directory tree on disk.
  *
- * @params:     See doc for `struct capture_params'.
+ * @params:     See doc for `struct scan_params'.
  *
  * @return:	0 on success, nonzero on failure.  It is a failure if any of
  *		the files cannot be `stat'ed, or if any of the needed
@@ -481,31 +676,16 @@ out:
  */
 int
 unix_build_dentry_tree(struct wim_dentry **root_ret,
-		       const char *root_disk_path,
-		       struct capture_params *params)
+		       const char *root_disk_path, struct scan_params *params)
 {
-	size_t path_len;
-	size_t path_bufsz;
-	char *path_buf;
 	int ret;
 
-	path_len = strlen(root_disk_path);
-	path_bufsz = min(32790, PATH_MAX + 1);
-
-	if (path_len >= path_bufsz)
-		return WIMLIB_ERR_INVALID_PARAM;
-
-	path_buf = MALLOC(path_bufsz);
-	if (!path_buf)
-		return WIMLIB_ERR_NOMEM;
-	memcpy(path_buf, root_disk_path, path_len + 1);
-
-	params->capture_root_nchars = path_len;
+	ret = pathbuf_init(params, root_disk_path);
+	if (ret)
+		return ret;
 
-	ret = unix_build_dentry_tree_recursive(root_ret, path_buf, path_len,
-					       AT_FDCWD, path_buf, params);
-	FREE(path_buf);
-	return ret;
+	return unix_build_dentry_tree_recursive(root_ret, AT_FDCWD,
+						root_disk_path, params);
 }
 
-#endif /* !__WIN32__ */
+#endif /* !_WIN32 */