X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;ds=sidebyside;f=src%2Fsymlink.c;h=366e54ed36af853517d0d43b5eac0e53619c82c3;hb=ca3f717fa42ec38d85dc020821340e59a15c25df;hp=e7163e584c71b35f810315eed7b285cd8a06151a;hpb=b8dd0c7d735c9e7db3f190a284a6255dab10ef1b;p=wimlib

diff --git a/src/symlink.c b/src/symlink.c
index e7163e58..366e54ed 100644
--- a/src/symlink.c
+++ b/src/symlink.c
@@ -1,33 +1,65 @@
+/*
+ * symlink.c
+ *
+ * Code to read and set symbolic links in WIM files.
+ */
+
+/*
+ * Copyright (C) 2012, 2013 Eric Biggers
+ *
+ * This file is part of wimlib, a library for working with WIM files.
+ *
+ * wimlib 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 3 of the License, or (at your option)
+ * any later version.
+ *
+ * wimlib 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 wimlib; if not, see http://www.gnu.org/licenses/.
+ */
+
 #include "dentry.h"
-#include "io.h"
+#include "buffer_io.h"
 #include "lookup_table.h"
 #include "sha1.h"
 #include <errno.h>
 
+/* UNIX version of getting and setting the data in reparse points */
+#if !defined(__WIN32__)
+
+#include <sys/stat.h>
+
 /*
  * Find the symlink target of a symbolic link or junction point in the WIM.
  *
- * See http://msdn.microsoft.com/en-us/library/cc232006(v=prot.10).aspx
- * Except the first 8 bytes aren't included in the resource (presumably because
- * we already know the reparse tag from the dentry, and we already know the
- * reparse tag len from the lookup table entry resource length).
+ * See http://msdn.microsoft.com/en-us/library/cc232006(v=prot.10).aspx for a
+ * description of the format of the so-called "reparse point data buffers".
+ *
+ * But, in the WIM format, the first 8 bytes of the reparse point data buffer
+ * are omitted, presumably because we already know the reparse tag from the
+ * dentry, and we already know the reparse tag length from the lookup table
+ * entry resource length.
  */
-static ssize_t get_symlink_name(const u8 *resource, size_t resource_len,
-			        char *buf, size_t buf_len,
-			        u32 reparse_tag)
+static ssize_t
+get_symlink_name(const void *resource, size_t resource_len, char *buf,
+		 size_t buf_len, u32 reparse_tag)
 {
-	const u8 *p = resource;
+	const void *p = resource;
 	u16 substitute_name_offset;
 	u16 substitute_name_len;
 	u16 print_name_offset;
 	u16 print_name_len;
 	char *link_target;
+	char *translated_target;
 	size_t link_target_len;
 	ssize_t ret;
 	unsigned header_size;
-	char *translated_target;
-	bool is_absolute;
-	u32 flags;
+	bool translate_slashes;
 
 	if (resource_len < 12)
 		return -EIO;
@@ -35,58 +67,81 @@ static ssize_t get_symlink_name(const u8 *resource, size_t resource_len,
 	p = get_u16(p, &substitute_name_len);
 	p = get_u16(p, &print_name_offset);
 	p = get_u16(p, &print_name_len);
-	get_u32(p, &flags);
 
 	wimlib_assert(reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK ||
 		      reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT);
 
-	/* I think that some junction points incorrectly get marked as symbolic
-	 * links.  So, parse the link buffer as a symlink if the flags seem
-	 * plausible. */
-	if (flags <= 1)
-		reparse_tag = WIM_IO_REPARSE_TAG_SYMLINK;
-
-	if (reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT) {
+	if (reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT)
 		header_size = 8;
-	} else {
-		is_absolute = (flags & 1) ? false : true;
+	else {
 		header_size = 12;
 		p += 4;
 	}
 	if (header_size + substitute_name_offset + substitute_name_len > resource_len)
 		return -EIO;
-	link_target = utf16_to_utf8(p + substitute_name_offset,
-				    substitute_name_len,
-				    &link_target_len);
 
-	if (!link_target)
-		return -EIO;
+	ret = utf16le_to_tstr((const utf16lechar*)(p + substitute_name_offset),
+			      substitute_name_len,
+			      &link_target, &link_target_len);
+	if (ret)
+		return -errno;
 
 	if (link_target_len + 1 > buf_len) {
 		ret = -ENAMETOOLONG;
 		goto out;
 	}
 
+	DEBUG("Interpeting substitute name \"%s\" (ReparseTag=0x%x)",
+	      link_target, reparse_tag);
+	translate_slashes = true;
 	translated_target = link_target;
-	if (reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT || is_absolute) {
-		if (link_target_len < 7
-		      || memcmp(translated_target, "\\??\\", 4) != 0
-		      || translated_target[4] == '\0'
-		      || translated_target[5] != ':'
-		      || translated_target[6] != '\\') {
-			ret = -EIO;
-			goto out;
-		}
-		translated_target += 4;
-		link_target_len -= 4;
-		/* There's a drive letter, so just leave the backslashes since
-		 * it won't go anyhwere on UNIX anyway... */
+	if (link_target_len >= 7 &&
+	    translated_target[0] == '\\' &&
+	    translated_target[1] == '?' &&
+	    translated_target[2] == '?' &&
+	    translated_target[3] == '\\' &&
+	    translated_target[4] != '\0' &&
+	    translated_target[5] == ':' &&
+	    translated_target[6] == '\\')
+	{
+		/* "Full" symlink or junction (\??\x:\ prefixed path) */
+		translated_target += 6;
+		link_target_len -= 6;
+	} else if (reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT &&
+		   link_target_len >= 12 &&
+		   memcmp(translated_target, "\\\\?\\Volume{", 11) == 0 &&
+		   translated_target[link_target_len - 1] == '\\')
+	{
+		/* Volume junction.  Can't really do anything with it. */
+		translate_slashes = false;
+	} else if (reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK &&
+		   link_target_len >= 3 &&
+		   translated_target[0] != '\0' &&
+		   translated_target[1] == ':' &&
+		   translated_target[2] == '/')
+	{
+		/* "Absolute" symlink, with drive letter */
+		translated_target += 2;
+		link_target_len -= 2;
+	} else if (reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK &&
+		   link_target_len >= 1)
+	{
+		if (translated_target[0] == '/')
+			/* "Absolute" symlink, without drive letter */
+			;
+		else
+			/* "Relative" symlink, without drive letter */
+			;
 	} else {
+		ERROR("Invalid reparse point: \"%s\"", translated_target);
+		ret = -EIO;
+		goto out;
+	}
+
+	if (translate_slashes)
 		for (size_t i = 0; i < link_target_len; i++)
 			if (translated_target[i] == '\\')
 				translated_target[i] = '/';
-	}
-
 	memcpy(buf, translated_target, link_target_len + 1);
 	ret = link_target_len;
 out:
@@ -94,118 +149,116 @@ out:
 	return ret;
 }
 
-void *make_symlink_reparse_data_buf(const char *symlink_target, size_t *len_ret)
+static int
+make_symlink_reparse_data_buf(const char *symlink_target,
+			      size_t *len_ret, void **buf_ret)
 {
-	size_t utf8_len = strlen(symlink_target);
-	size_t utf16_len;
-	char *name_utf16 = utf8_to_utf16(symlink_target, utf8_len, &utf16_len);
-	if (!name_utf16)
-		return NULL;
-	/*DEBUG("utf16_len = %zu", utf16_len);*/
-	for (size_t i = 0; i < utf16_len / 2; i++)
-		if (((u16*)name_utf16)[i] == to_le16('/'))
-			((u16*)name_utf16)[i] = to_le16('\\');
-	size_t len = 12 + utf16_len * 2;
-	void *buf = MALLOC(len);
-	if (!buf)
-		goto out;
-	/* XXX Fix absolute paths */
-
-	u8 *p = buf;
-	p = put_u16(p, 0); /* Substitute name offset */
-	p = put_u16(p, utf16_len); /* Substitute name length */
-	p = put_u16(p, utf16_len); /* Print name offset */
-	p = put_u16(p, utf16_len); /* Print name length */
-	p = put_u32(p, 1);
-	p = put_bytes(p, utf16_len, name_utf16);
-	p = put_bytes(p, utf16_len, name_utf16);
-	/*DEBUG("utf16_len = %zu, len = %zu", utf16_len, len);*/
-	*len_ret = len;
-out:
-	FREE(name_utf16);
-	return buf;
-}
+	utf16lechar *name_utf16le;
+	size_t name_utf16le_nbytes;
+	int ret;
 
-/* Get the symlink target from a dentry that's already checked to be either a
- * "real" symlink or a junction point. */
-ssize_t dentry_readlink(const struct dentry *dentry, char *buf, size_t buf_len,
-			const WIMStruct *w)
-{
-	struct resource_entry *res_entry;
-	struct lookup_table_entry *lte;
-	u16 i = 0;
-	const u8 *hash = dentry->hash;
-
-	wimlib_assert(dentry_is_symlink(dentry));
-
-	while (1) {
-		if ((lte = __lookup_resource(w->lookup_table, hash)))
-			break;
-		if (i == dentry->num_ads)
-			return -EIO;
-		hash = dentry->ads_entries[i].hash;
-		i++;
-	}
-	res_entry = &lte->resource_entry;
-	if (res_entry->original_size > 10000)
-		return -EIO;
+	ret = tstr_to_utf16le(symlink_target, strlen(symlink_target),
+			      &name_utf16le, &name_utf16le_nbytes);
+	if (ret)
+		return ret;
+
+	for (size_t i = 0; i < name_utf16le_nbytes / 2; i++)
+		if (name_utf16le[i] == cpu_to_le16('/'))
+			name_utf16le[i] = cpu_to_le16('\\');
 
-	char __res_buf[res_entry->original_size];
-	const char *res_buf;
-	if (lte->is_symlink && lte->symlink_buf) {
-		res_buf = lte->symlink_buf;
+	size_t len = 12 + (name_utf16le_nbytes + 2) * 2;
+	void *buf = MALLOC(len);
+	if (buf) {
+		void *p = buf;
+		p = put_u16(p, 0); /* Substitute name offset */
+		p = put_u16(p, name_utf16le_nbytes); /* Substitute name length */
+		p = put_u16(p, name_utf16le_nbytes + 2); /* Print name offset */
+		p = put_u16(p, name_utf16le_nbytes); /* Print name length */
+		p = put_u32(p, 1); /* flags: 0 if relative link, otherwise 1 */
+		p = put_bytes(p, name_utf16le_nbytes, name_utf16le);
+		p = put_u16(p, 0);
+		p = put_bytes(p, name_utf16le_nbytes, name_utf16le);
+		p = put_u16(p, 0);
+		*len_ret = len;
+		*buf_ret = buf;
+		ret = 0;
 	} else {
-		res_buf = __res_buf;
-		if (read_full_resource(w->fp, res_entry->size, 
-				       res_entry->original_size,
-				       res_entry->offset,
-				       wim_resource_compression_type(w, res_entry),
-				       __res_buf) != 0)
-			return -EIO;
-		res_buf = __res_buf;
+		ret = WIMLIB_ERR_NOMEM;
 	}
-	return get_symlink_name(res_buf, res_entry->original_size, buf,
-				buf_len, dentry->reparse_tag);
+	FREE(name_utf16le);
+	return ret;
 }
 
-static int dentry_set_symlink_buf(struct dentry *dentry,
-				  const u8 symlink_buf_hash[])
+/* Get the symlink target from a WIM inode.
+ *
+ * The inode may be either a "real" symlink (reparse tag
+ * WIM_IO_REPARSE_TAG_SYMLINK), or it may be a junction point (reparse tag
+ * WIM_IO_REPARSE_TAG_MOUNT_POINT).
+ */
+ssize_t
+inode_readlink(const struct wim_inode *inode, char *buf, size_t buf_len,
+	       const WIMStruct *w, bool threadsafe)
 {
-	struct ads_entry *ads_entries;
-
-	ads_entries = CALLOC(2, sizeof(struct ads_entry));
-	if (!ads_entries)
-		return WIMLIB_ERR_NOMEM;
-	memcpy(ads_entries[1].hash, symlink_buf_hash, WIM_HASH_SIZE);
-	dentry_free_ads_entries(dentry);
-	dentry->num_ads = 2;
-	dentry->ads_entries = ads_entries;
-	return 0;
+	const struct wim_lookup_table_entry *lte;
+	int ret;
+
+	wimlib_assert(inode_is_symlink(inode));
+
+	lte = inode_unnamed_lte(inode, w->lookup_table);
+	if (!lte)
+		return -EIO;
+
+	if (wim_resource_size(lte) > REPARSE_POINT_MAX_SIZE)
+		return -EIO;
+
+	u8 res_buf[wim_resource_size(lte)];
+	ret = read_full_resource_into_buf(lte, res_buf, threadsafe);
+	if (ret)
+		return -EIO;
+	return get_symlink_name(res_buf, wim_resource_size(lte), buf,
+				buf_len, inode->i_reparse_tag);
 }
 
-int dentry_set_symlink(struct dentry *dentry, const char *target,
-		       struct lookup_table *lookup_table)
+/*
+ * Sets @inode to be a symbolic link pointing to @target.
+ *
+ * A lookup table entry for the symbolic link data buffer is created and
+ * inserted into @lookup_table, unless there is an existing lookup table entry
+ * for the exact same data, in which its reference count is incremented.
+ *
+ * The lookup table entry is returned in @lte_ret.
+ *
+ * On failure @dentry and @lookup_table are not modified.
+ */
+int
+inode_set_symlink(struct wim_inode *inode,
+		  const char *target,
+		  struct wim_lookup_table *lookup_table,
+		  struct wim_lookup_table_entry **lte_ret)
 
 {
 	int ret;
 	size_t symlink_buf_len;
-	struct lookup_table_entry *lte = NULL, *existing_lte;
-	u8 symlink_buf_hash[WIM_HASH_SIZE];
+	struct wim_lookup_table_entry *lte = NULL, *existing_lte;
+	u8 symlink_buf_hash[SHA1_HASH_SIZE];
 	void *symlink_buf;
-	
-	symlink_buf = make_symlink_reparse_data_buf(target, &symlink_buf_len);
-	if (!symlink_buf)
-		return WIMLIB_ERR_NOMEM;
+
+	ret = make_symlink_reparse_data_buf(target, &symlink_buf_len,
+					    &symlink_buf);
+	if (ret)
+		return ret;
 
 	DEBUG("Made symlink reparse data buf (len = %zu, name len = %zu)",
-			symlink_buf_len, ret);
-	
+			symlink_buf_len, symlink_buf_len);
+
 	sha1_buffer(symlink_buf, symlink_buf_len, symlink_buf_hash);
 
 	existing_lte = __lookup_resource(lookup_table, symlink_buf_hash);
 
 	if (existing_lte) {
-		existing_lte->refcnt++;
+		lte = existing_lte;
+		FREE(symlink_buf);
+		symlink_buf = NULL;
 	} else {
 		DEBUG("Creating new lookup table entry for symlink buf");
 		lte = new_lookup_table_entry();
@@ -213,26 +266,109 @@ int dentry_set_symlink(struct dentry *dentry, const char *target,
 			ret = WIMLIB_ERR_NOMEM;
 			goto out_free_symlink_buf;
 		}
-		lte->is_symlink = true;
-		lte->symlink_buf = symlink_buf;
+		lte->resource_location            = RESOURCE_IN_ATTACHED_BUFFER;
+		lte->attached_buffer              = symlink_buf;
 		lte->resource_entry.original_size = symlink_buf_len;
-		lte->resource_entry.size = symlink_buf_len;
-		memcpy(lte->hash, symlink_buf_hash, WIM_HASH_SIZE);
+		copy_hash(lte->hash, symlink_buf_hash);
 	}
 
-	ret = dentry_set_symlink_buf(dentry, symlink_buf_hash);
-
-	if (ret != 0)
-		goto out_free_lte;
+	inode->i_lte = lte;
+	inode->i_resolved = 1;
 
 	DEBUG("Loaded symlink buf");
 
-	if (!existing_lte)
+	if (existing_lte)
+		lte->refcnt++;
+	else
 		lookup_table_insert(lookup_table, lte);
+	if (lte_ret)
+		*lte_ret = lte;
 	return 0;
-out_free_lte:
-	FREE(lte);
 out_free_symlink_buf:
 	FREE(symlink_buf);
 	return ret;
 }
+
+static int
+unix_get_ino_and_dev(const char *path, u64 *ino_ret, u64 *dev_ret)
+{
+	struct stat stbuf;
+	if (stat(path, &stbuf)) {
+		WARNING_WITH_ERRNO("Failed to stat \"%s\"", path);
+		/* Treat as a link pointing outside the capture root (it
+		 * most likely is). */
+		return WIMLIB_ERR_STAT;
+	} else {
+		*ino_ret = stbuf.st_ino;
+		*dev_ret = stbuf.st_dev;
+		return 0;
+	}
+}
+
+#endif /* !defined(__WIN32__) */
+
+#ifdef __WIN32__
+#  include "win32.h"
+#  define RP_PATH_SEPARATOR L'\\'
+#  define os_get_ino_and_dev win32_get_file_and_vol_ids
+#else
+#  define RP_PATH_SEPARATOR '/'
+#  define os_get_ino_and_dev unix_get_ino_and_dev
+#endif
+
+/* Fix up reparse points--- mostly shared between UNIX and Windows */
+tchar *
+fixup_symlink(tchar *dest, u64 capture_root_ino, u64 capture_root_dev)
+{
+	tchar *p = dest;
+
+#ifdef __WIN32__
+	/* Skip over drive letter */
+	if (*p != RP_PATH_SEPARATOR)
+		p += 2;
+#endif
+
+	DEBUG("Fixing symlink or junction \"%"TS"\"", dest);
+	for (;;) {
+		tchar save;
+		int ret;
+		u64 ino;
+		u64 dev;
+
+		while (*p == RP_PATH_SEPARATOR)
+			p++;
+
+		save = *p;
+		*p = T('\0');
+		ret = os_get_ino_and_dev(dest, &ino, &dev);
+		*p = save;
+
+		if (ino == capture_root_ino && dev == capture_root_dev) {
+			/* Link points inside capture root.  Return abbreviated
+			 * path. */
+			if (*p == T('\0'))
+				*(p - 1) = RP_PATH_SEPARATOR;
+			while (p - 1 >= dest && *(p - 1) == RP_PATH_SEPARATOR)
+				p--;
+		#ifdef __WIN32__
+			/* Add back drive letter */
+			if (*dest != RP_PATH_SEPARATOR) {
+				*--p = *(dest + 1);
+				*--p = *dest;
+			}
+		#endif
+			wimlib_assert(p >= dest);
+			return p;
+		}
+
+		if (*p == T('\0')) {
+			/* Link points outside capture root. */
+			return NULL;
+		}
+
+		do {
+			p++;
+		} while (*p != RP_PATH_SEPARATOR && *p != T('\0'));
+	}
+}
+