X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Funix_capture.c;h=d869f258836f6a40bbe78f025ceef872ee2fc14b;hb=088dff37aa334c218e1cac96cc847f5dd14f7124;hp=a61d91f25f61fdc5f2cf6b924f0db2a9e60c37da;hpb=3de1ec66f778edda19865482d685bc6f4e17faf7;p=wimlib

diff --git a/src/unix_capture.c b/src/unix_capture.c
index a61d91f2..d869f258 100644
--- a/src/unix_capture.c
+++ b/src/unix_capture.c
@@ -3,7 +3,7 @@
  */
 
 /*
- * Copyright (C) 2012, 2013, 2014 Eric Biggers
+ * Copyright (C) 2012-2016 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
@@ -30,15 +30,20 @@
 #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))
@@ -99,52 +104,241 @@ my_fdopendir(int *dirfd_p)
 #  define AT_SYMLINK_NOFOLLOW	0x100
 #endif
 
+#ifdef HAVE_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 wimlib_xattr_entry *entry = entries;
+
+	wimlib_assert((uintptr_t)entries % 4 == 0 &&
+		      entries_size % 4 == 0 && names_size != 0);
+	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 ||
+		    (u16)name_len != name_len) {
+			ERROR("\"%s\": malformed extended attribute names list",
+			      path);
+			errno = EINVAL;
+			return -1;
+		}
+
+		/*
+		 * Note: we 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) {
+			errno = ERANGE;
+			return -1;
+		}
+
+		entry->name_len = cpu_to_le16(name_len);
+		entry->reserved = 0;
+		value = mempcpy(entry->name, name, name_len);
+
+		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 ((u32)value_len != value_len) {
+			ERROR("\"%s\": value of extended attribute \"%s\" is too large",
+			      path, name);
+			errno = EINVAL;
+			return -1;
+		}
+		entry->value_len = cpu_to_le32(value_len);
+
+		/*
+		 * Zero-pad the entry to the next 4-byte boundary.
+		 * Note: because we've guaranteed that @entries_size is a
+		 * multiple of 4, this cannot overflow the @entries buffer.
+		 */
+		value += value_len;
+		while ((uintptr_t)value & 3) {
+			*(u8 *)value = 0;
+			value++;
+		}
+
+		entry = value;
+		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] _aligned_attribute(4);
+	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_linux_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_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;
+	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) {
-		char *file_on_disk = STRDUP(path);
-		if (!file_on_disk)
-			return WIMLIB_ERR_NOMEM;
 		blob = new_blob_descriptor();
-		if (!blob) {
-			FREE(file_on_disk);
-			return WIMLIB_ERR_NOMEM;
-		}
-		blob->file_on_disk = file_on_disk;
-		blob->file_inode = inode;
+		if (unlikely(!blob))
+			goto err_nomem;
+		blob->file_on_disk = STRDUP(path);
+		if (unlikely(!blob->file_on_disk))
+			goto err_nomem;
 		blob->blob_location = BLOB_IN_FILE_ON_DISK;
 		blob->size = size;
-	} else {
-		blob = NULL;
+		blob->file_inode = inode;
 	}
 
 	strm = inode_add_stream(inode, STREAM_TYPE_DATA, NO_STREAM_NAME, blob);
-	if (!strm) {
-		free_blob_descriptor(blob);
-		return WIMLIB_ERR_NOMEM;
-	}
+	if (unlikely(!strm))
+		goto err_nomem;
+
 	prepare_unhashed_blob(blob, inode, strm->stream_id, unhashed_blobs);
 	return 0;
+
+err_nomem:
+	free_blob_descriptor(blob);
+	return WIMLIB_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;
@@ -182,13 +376,12 @@ unix_scan_directory(struct wim_dentry *dir_dentry,
 			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,
@@ -199,39 +392,66 @@ unix_scan_directory(struct wim_dentry *dir_dentry,
 		full_path[full_path_len] = '\0';
 		if (ret)
 			break;
-		if (child)
-			dentry_add_child(dir_dentry, child);
+		attach_scanned_tree(dir_dentry, child, params->blob_table);
 	}
 	closedir(dir);
 	return ret;
 }
 
-/* Given an absolute symbolic link target @dest (UNIX-style, beginning
- * with '/'), determine whether it points into the directory specified by
- * @ino and @dev.  If so, return the target modified to be "absolute"
- * relative to this directory.  Otherwise, return NULL.  */
+/*
+ * Given an absolute symbolic link target (UNIX-style, beginning with '/'),
+ * determine whether it points into the directory identified by @ino and @dev.
+ * If yes, return the suffix of @target which is relative to this directory, but
+ * retaining leading slashes.  If no, return @target.
+ *
+ * Here are some examples, assuming that the @ino/@dev directory is "/home/e":
+ *
+ *	Original target		New target
+ *	---------------		----------
+ *	/home/e/test		/test
+ *	/home/e/test/		/test/
+ *	//home//e//test//	//test//
+ *	/home/e						(empty string)
+ *	/home/e/		/
+ *	/usr/lib		/usr/lib		(external link)
+ *
+ * Because of the possibility of other links into the @ino/@dev directory and/or
+ * multiple path separators, we can't simply do a string comparison; instead we
+ * need to stat() each ancestor directory.
+ *
+ * If the link points directly to the @ino/@dev directory with no trailing
+ * slashes, then the new target will be an empty string.  This is not a valid
+ * UNIX symlink target, but we store this in the archive anyway since the target
+ * is intended to be de-relativized when the link is extracted.
+ */
 static char *
-unix_fixup_abslink(char *dest, u64 ino, u64 dev)
+unix_relativize_link_target(char *target, u64 ino, u64 dev)
 {
-	char *p = dest;
+	char *p = target;
 
 	do {
 		char save;
 		struct stat stbuf;
 		int ret;
 
-		/* Skip non-slashes.  */
-		while (*p && *p != '/')
+		/* Skip slashes (guaranteed to be at least one here)  */
+		do {
 			p++;
+		} while (*p == '/');
+
+		/* End of string?  */
+		if (!*p)
+			break;
 
-		/* Skip slashes.  */
-		while (*p && *p == '/')
+		/* Skip non-slashes (guaranteed to be at least one here)  */
+		do {
 			p++;
+		} while (*p && *p != '/');
 
-		/* Get inode and device for this prefix.  */
+		/* Get the inode and device numbers for this prefix.  */
 		save = *p;
 		*p = '\0';
-		ret = stat(dest, &stbuf);
+		ret = stat(target, &stbuf);
 		*p = save;
 
 		if (ret) {
@@ -243,83 +463,71 @@ unix_fixup_abslink(char *dest, u64 ino, u64 dev)
 		if (stbuf.st_ino == ino && stbuf.st_dev == dev) {
 			/* Link points inside directory tree being captured.
 			 * Return abbreviated path.  */
-			*--p = '/';
-			while (p > dest && *(p - 1) == '/')
-				p--;
 			return p;
 		}
 	} while (*p);
 
 	/* Link does not point inside directory tree being captured.  */
-	return NULL;
+	return target;
 }
 
-static int
+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)
+		  struct wim_inode *inode, struct scan_params *params)
 {
-	char deref_name_buf[4096];
-	ssize_t deref_name_len;
-	char *dest;
+	char orig_target[REPARSE_POINT_MAX_SIZE];
+	char *target = orig_target;
 	int ret;
 
-	inode->i_attributes = FILE_ATTRIBUTE_REPARSE_POINT;
-	inode->i_reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
-
-	/* The idea here is to call readlink() to get the UNIX target of the
-	 * symbolic link, then turn the target into a reparse point data buffer
-	 * that contains a relative or absolute symbolic link. */
-	deref_name_len = my_readlinkat(full_path, dirfd, relpath,
-				       deref_name_buf, sizeof(deref_name_buf) - 1);
-	if (deref_name_len < 0) {
+	/* Read the UNIX symbolic link target.  */
+	ret = my_readlinkat(full_path, dirfd, relpath, target,
+			    sizeof(orig_target));
+	if (unlikely(ret < 0)) {
 		ERROR_WITH_ERRNO("\"%s\": Can't read target of symbolic link",
 				 full_path);
 		return WIMLIB_ERR_READLINK;
 	}
+	if (unlikely(ret >= sizeof(orig_target))) {
+		ERROR("\"%s\": target of symbolic link is too long", full_path);
+		return WIMLIB_ERR_READLINK;
+	}
+	target[ret] = '\0';
 
-	dest = deref_name_buf;
-
-	dest[deref_name_len] = '\0';
-
-	if ((params->add_flags & WIMLIB_ADD_FLAG_RPFIX) &&
-	     dest[0] == '/')
-	{
-		char *fixed_dest;
+	/* If the link is absolute and reparse point fixups are enabled, then
+	 * change it to be "absolute" relative to the tree being captured.  */
+	if (target[0] == '/' && (params->add_flags & WIMLIB_ADD_FLAG_RPFIX)) {
+		int status = WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK;
 
-		/* RPFIX (reparse point fixup) mode:  Change target of absolute
-		 * symbolic link to be "absolute" relative to the tree being
-		 * captured.  */
-		fixed_dest = unix_fixup_abslink(dest,
-						params->capture_root_ino,
-						params->capture_root_dev);
 		params->progress.scan.cur_path = full_path;
-		params->progress.scan.symlink_target = deref_name_buf;
-		if (fixed_dest) {
-			/* Link points inside the tree being captured, so it was
-			 * fixed.  */
-			inode->i_not_rpfixed = 0;
-			dest = fixed_dest;
-			ret = do_capture_progress(params,
-						  WIMLIB_SCAN_DENTRY_FIXED_SYMLINK,
-						  NULL);
-		} else {
-			/* Link points outside the tree being captured, so it
-			 * was not fixed.  */
-			ret = do_capture_progress(params,
-						  WIMLIB_SCAN_DENTRY_NOT_FIXED_SYMLINK,
-						  NULL);
+		params->progress.scan.symlink_target = target;
+
+		target = unix_relativize_link_target(target,
+						     params->capture_root_ino,
+						     params->capture_root_dev);
+		if (target != orig_target) {
+			/* Link target was fixed.  */
+			inode->i_rp_flags &= ~WIM_RP_FLAG_NOT_FIXED;
+			status = WIMLIB_SCAN_DENTRY_FIXED_SYMLINK;
 		}
+		ret = do_scan_progress(params, status, NULL);
 		if (ret)
 			return ret;
 	}
-	ret = wim_inode_set_symlink(inode, dest, params->blob_table);
-	if (ret)
+
+	/* Translate the UNIX symlink target into a Windows reparse point.  */
+	ret = wim_inode_set_symlink(inode, target, params->blob_table);
+	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.", full_path);
+		}
 		return ret;
+	}
 
-	/* Unfortunately, Windows seems to have the concept of "file" symbolic
-	 * links as being different from "directory" symbolic links...  so
-	 * FILE_ATTRIBUTE_DIRECTORY needs to be set on the symbolic link if the
-	 * *target* of the symbolic link is a directory.  */
+	/* 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 &&
 	    S_ISDIR(stbuf.st_mode))
@@ -331,7 +539,7 @@ 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;
@@ -339,10 +547,10 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 	struct stat stbuf;
 	int stat_flags;
 
-	ret = try_exclude(full_path, full_path_len, params);
-	if (ret < 0) /* Excluded? */
+	ret = try_exclude(full_path, params);
+	if (unlikely(ret < 0)) /* Excluded? */
 		goto out_progress;
-	if (ret > 0) /* Error? */
+	if (unlikely(ret > 0)) /* Error? */
 		goto out;
 
 	if (params->add_flags & (WIMLIB_ADD_FLAG_DEREFERENCE |
@@ -373,18 +581,22 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 				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.", full_path);
+		}
 		goto out;
+	}
 
 	inode = tree->d_inode;
 
@@ -412,6 +624,11 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 			ret = WIMLIB_ERR_NOMEM;
 			goto out;
 		}
+#ifdef HAVE_XATTR_SUPPORT
+		ret = scan_linux_xattrs(full_path, inode);
+		if (ret)
+			goto out;
+#endif
 	}
 
 	if (params->add_flags & WIMLIB_ADD_FLAG_ROOT) {
@@ -421,8 +638,9 @@ 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(full_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);
@@ -437,14 +655,14 @@ unix_build_dentry_tree_recursive(struct wim_dentry **tree_ret,
 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, full_path);
 	}
 	*tree_ret = tree;
 	return ret;
@@ -455,13 +673,12 @@ 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
@@ -472,8 +689,7 @@ 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;