X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Fverify.c;h=a83b1d3b0bcbaf03c414c729d9a0462389c951b2;hb=dd95b601226f4c508252ad3cd7c7a2887ee48486;hp=eef45d3ce3ee29eae902c9834302c8f0313964a2;hpb=e1ab245c9f04e34deb2f078b5ab4b98af8cb7b3b;p=wimlib

diff --git a/src/verify.c b/src/verify.c
index eef45d3c..a83b1d3b 100644
--- a/src/verify.c
+++ b/src/verify.c
@@ -1,12 +1,11 @@
 /*
  * verify.c
  *
- * Some functions to verify that stuff in the WIM is valid.  Of course, not
- * *all* the verifications of the input data are in this file.
+ * Verify WIM files.
  */
 
 /*
- * Copyright (C) 2012, 2013 Eric Biggers
+ * Copyright (C) 2012, 2013, 2014 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
@@ -32,207 +31,198 @@
 #include "wimlib/error.h"
 #include "wimlib/lookup_table.h"
 #include "wimlib/metadata.h"
+#include "wimlib/progress.h"
 #include "wimlib/security.h"
 
 static int
-verify_inode(struct wim_inode *inode, const WIMStruct *w)
+lte_fix_refcnt(struct wim_lookup_table_entry *lte, void *ctr)
 {
-	const struct wim_lookup_table *table = w->lookup_table;
-	const struct wim_security_data *sd = wim_const_security_data(w);
-	struct wim_dentry *first_dentry = inode_first_dentry(inode);
-	struct wim_dentry *dentry;
-	int ret = WIMLIB_ERR_INVALID_DENTRY;
-
-	/* Check the security ID.  -1 is valid and means "no security
-	 * descriptor".  Anything else has to be a valid index into the WIM
-	 * image's security descriptors table. */
-	if (inode->i_security_id < -1) {
-		ERROR("Dentry `%"TS"' has an invalid security ID (%d)",
-		      dentry_full_path(first_dentry), inode->i_security_id);
-		goto out;
-	}
-
-	if (inode->i_security_id >= sd->num_entries) {
-		ERROR("Dentry `%"TS"' has an invalid security ID (%d) "
-		      "(there are only %u entries in the security table)",
-		      dentry_full_path(first_dentry), inode->i_security_id,
-		      sd->num_entries);
-		goto out;
-	}
-
-	/* Check that lookup table entries for all the inode's stream exist,
-	 * except if the SHA1 message digest is all 0's, which indicates an
-	 * empty stream.
-	 *
-	 * This check is skipped on split WIMs. */
-	if (w->hdr.total_parts == 1) {
-		for (unsigned i = 0; i <= inode->i_num_ads; i++) {
-			struct wim_lookup_table_entry *lte;
-			const u8 *hash;
-			hash = inode_stream_hash_unresolved(inode, i);
-			lte = __lookup_resource(table, hash);
-			if (!lte && !is_zero_hash(hash)) {
-				ERROR("Could not find lookup table entry for stream "
-				      "%u of dentry `%"TS"'",
-				      i, dentry_full_path(first_dentry));
-				goto out;
-			}
-			if (lte)
-				lte->real_refcnt += inode->i_nlink;
-		}
+	if (lte->refcnt != lte->real_refcnt) {
+		lte->refcnt = lte->real_refcnt;
+		++*(unsigned long *)ctr;
 	}
+	return 0;
+}
 
-	/* Make sure there is only one unnamed data stream. */
-	unsigned num_unnamed_streams = 0;
+static void
+tally_inode_refcnts(const struct wim_inode *inode,
+		    const struct wim_lookup_table *lookup_table)
+{
 	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
-		const u8 *hash;
-		hash = inode_stream_hash_unresolved(inode, i);
-		if (inode_stream_name_nbytes(inode, i) == 0 && !is_zero_hash(hash))
-			num_unnamed_streams++;
-	}
-	if (num_unnamed_streams > 1) {
-		ERROR("Dentry `%"TS"' has multiple (%u) un-named streams",
-		      dentry_full_path(first_dentry), num_unnamed_streams);
-		goto out;
+		struct wim_lookup_table_entry *lte;
+		lte = inode_stream_lte(inode, i, lookup_table);
+		if (lte)
+			lte->real_refcnt += inode->i_nlink;
 	}
+}
 
-	/* Files cannot have multiple DOS names, even if they have multiple
-	 * names in multiple directories (i.e. hard links).
-	 * Source: NTFS-3g authors. */
-	struct wim_dentry *dentry_with_dos_name = NULL;
-	inode_for_each_dentry(dentry, inode) {
-		if (dentry_has_short_name(dentry)) {
-			if (dentry_with_dos_name) {
-				ERROR("Hard-linked file has a DOS name at "
-				      "both `%"TS"' and `%"TS"'",
-				      dentry_full_path(dentry_with_dos_name),
-				      dentry_full_path(dentry));
-				goto out;
-			}
-			dentry_with_dos_name = dentry;
-		}
-	}
 
-	/* Directories with multiple links have not been tested. XXX */
-	if (inode->i_nlink > 1 && inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
-		ERROR("Hard-linked directory `%"TS"' is unsupported",
-		      dentry_full_path(first_dentry));
-		goto out;
-	}
+static int
+tally_image_refcnts(WIMStruct *wim)
+{
+	const struct wim_image_metadata *imd;
+	const struct wim_inode *inode;
 
-	inode->i_verified = 1;
-	ret = 0;
-out:
-	return ret;
+	imd = wim_get_current_image_metadata(wim);
+	image_for_each_inode(inode, imd)
+		tally_inode_refcnts(inode, wim->lookup_table);
+	return 0;
 }
 
-/* Run some miscellaneous verifications on a WIM dentry */
+
+/* Ideally this would be unnecessary... however, the WIMs for Windows 8 are
+ * screwed up because some lookup table entries are referenced more times than
+ * their stated reference counts.  So theoretically, if we delete all the
+ * references to a stream and then remove it, it might still be referenced
+ * somewhere else, making a file be missing from the WIM... So, work around this
+ * problem by looking at ALL the images to re-calculate the reference count of
+ * EVERY lookup table entry.  This only absolutely has to be done before an image
+ * is deleted or before an image is mounted read-write. */
 int
-verify_dentry(struct wim_dentry *dentry, void *wim)
+wim_recalculate_refcnts(WIMStruct *wim)
 {
+	unsigned long num_ltes_with_bogus_refcnt = 0;
 	int ret;
-	WIMStruct *w = wim;
-
-	/* Verify the associated inode, but only one time no matter how many
-	 * dentries it has (unless we are doing a full verification of the WIM,
-	 * in which case we need to force the inode to be verified again.) */
-	if (!dentry->d_inode->i_verified) {
-		ret = verify_inode(dentry->d_inode, w);
-		if (ret != 0)
-			return ret;
-	}
 
-	/* Make sure root dentry is unnamed, while every other dentry has at
-	 * least a long name.
-	 *
-	 * I am assuming that dentries having only a DOS name is illegal; i.e.,
-	 * Windows will always combine the Win32 name and DOS name for a file
-	 * into a single WIM dentry, even if they are stored separately on NTFS.
-	 * (This seems to be the case...) */
-	if (dentry_is_root(dentry)) {
-		if (dentry_has_long_name(dentry) || dentry_has_short_name(dentry)) {
-			ERROR("The root dentry has a nonempty name!");
-			return WIMLIB_ERR_INVALID_DENTRY;
-		}
-	} else {
-		if (!dentry_has_long_name(dentry)) {
-			ERROR("Dentry `%"TS"' has no long name!",
-			      dentry_full_path(dentry));
-			return WIMLIB_ERR_INVALID_DENTRY;
-		}
-	}
-
-#if 0
-	/* Check timestamps */
-	if (inode->i_last_access_time < inode->i_creation_time ||
-	    inode->i_last_write_time < inode->i_creation_time) {
-		WARNING("Dentry `%"TS"' was created after it was last accessed or "
-			"written to", dentry->full_path);
+	for_lookup_table_entry(wim->lookup_table, lte_zero_real_refcnt, NULL);
+	ret = for_image(wim, WIMLIB_ALL_IMAGES, tally_image_refcnts);
+	if (ret)
+		return ret;
+	num_ltes_with_bogus_refcnt = 0;
+	for_lookup_table_entry(wim->lookup_table, lte_fix_refcnt,
+			       &num_ltes_with_bogus_refcnt);
+	if (num_ltes_with_bogus_refcnt != 0) {
+		WARNING("%lu stream(s) had incorrect reference count.",
+			num_ltes_with_bogus_refcnt);
 	}
-#endif
+	wim->refcnts_ok = 1;
+	return 0;
+}
 
+static int
+append_lte_to_list(struct wim_lookup_table_entry *lte, void *_list)
+{
+	list_add(&lte->extraction_list, (struct list_head *)_list);
 	return 0;
 }
 
+struct verify_stream_list_ctx {
+	wimlib_progress_func_t progfunc;
+	void *progctx;
+	union wimlib_progress_info *progress;
+};
+
 static int
-image_run_full_verifications(WIMStruct *w)
+end_verify_stream(struct wim_lookup_table_entry *lte, int status, void *_ctx)
 {
-	struct wim_image_metadata *imd;
-	struct wim_inode *inode;
+	struct verify_stream_list_ctx *ctx = _ctx;
 
-	imd = wim_get_current_image_metadata(w);
-	image_for_each_inode(inode, imd)
-		inode->i_verified = 0;
-	return for_dentry_in_tree(imd->root_dentry, verify_dentry, w);
+	if (status)
+		return status;
+
+	ctx->progress->verify_streams.completed_streams++;
+	ctx->progress->verify_streams.completed_bytes += lte->size;
+
+	return call_progress(ctx->progfunc, WIMLIB_PROGRESS_MSG_VERIFY_STREAMS,
+			     ctx->progress, ctx->progctx);
 }
 
 static int
-lte_fix_refcnt(struct wim_lookup_table_entry *lte, void *ctr)
+verify_image_streams_present(struct wim_image_metadata *imd,
+			     struct wim_lookup_table *lookup_table)
 {
-	if (lte->refcnt != lte->real_refcnt) {
-	#ifdef ENABLE_ERROR_MESSAGES
-		WARNING("The following lookup table entry has a reference "
-			"count of %u, but", lte->refcnt);
-		WARNING("We found %u references to it",
-			lte->real_refcnt);
-		print_lookup_table_entry(lte, stderr);
-	#endif
-		lte->refcnt = lte->real_refcnt;
-		++*(unsigned long *)ctr;
+	struct wim_inode *inode;
+	int ret;
+
+	image_for_each_inode(inode, imd) {
+		ret = inode_resolve_streams(inode, lookup_table, false);
+		if (ret)
+			return ret;
 	}
 	return 0;
 }
 
-/* Ideally this would be unnecessary... however, the WIMs for Windows 8 are
- * screwed up because some lookup table entries are referenced more times than
- * their stated reference counts.  So theoretically, if we delete all the
- * references to a stream and then remove it, it might still be referenced
- * somewhere else, making a file be missing from the WIM... So, work around this
- * problem by looking at ALL the images to re-calculate the reference count of
- * EVERY lookup table entry.  This only absolutely has to be done before an image
- * is deleted or before an image is mounted read-write. */
-int
-wim_run_full_verifications(WIMStruct *w)
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_verify_wim(WIMStruct *wim, int verify_flags)
 {
 	int ret;
+	LIST_HEAD(stream_list);
+	union wimlib_progress_info progress;
+	struct verify_stream_list_ctx ctx;
+	struct wim_lookup_table_entry *lte;
+	struct read_stream_list_callbacks cbs = {
+		.end_stream = end_verify_stream,
+		.end_stream_ctx = &ctx,
+	};
+
+	/* Check parameters  */
+
+	if (!wim)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	if (verify_flags)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	/* Verify the images  */
+
+	if (wim_has_metadata(wim)) {
+
+		memset(&progress, 0, sizeof(progress));
+		progress.verify_image.wimfile = wim->filename;
+		progress.verify_image.total_images = wim->hdr.image_count;
+
+		for (int i = 1; i <= wim->hdr.image_count; i++) {
+
+			progress.verify_image.current_image = i;
+
+			ret = call_progress(wim->progfunc, WIMLIB_PROGRESS_MSG_BEGIN_VERIFY_IMAGE,
+					    &progress, wim->progctx);
+			if (ret)
+				return ret;
 
-	for_lookup_table_entry(w->lookup_table, lte_zero_real_refcnt, NULL);
-
-	w->all_images_verified = 1; /* Set *before* image_run_full_verifications,
-				       because of check in read_metadata_resource() */
-	ret = for_image(w, WIMLIB_ALL_IMAGES, image_run_full_verifications);
-	if (ret == 0) {
-		unsigned long num_ltes_with_bogus_refcnt = 0;
-		for_lookup_table_entry(w->lookup_table, lte_fix_refcnt,
-				       &num_ltes_with_bogus_refcnt);
-		if (num_ltes_with_bogus_refcnt != 0) {
-			WARNING("A total of %lu entries in the WIM's stream "
-				"lookup table had to have\n"
-				"          their reference counts fixed.",
-				num_ltes_with_bogus_refcnt);
+			ret = select_wim_image(wim, i);
+			if (ret)
+				return ret;
+
+			ret = verify_image_streams_present(wim_get_current_image_metadata(wim),
+							   wim->lookup_table);
+			if (ret)
+				return ret;
+
+			ret = call_progress(wim->progfunc, WIMLIB_PROGRESS_MSG_END_VERIFY_IMAGE,
+					    &progress, wim->progctx);
+			if (ret)
+				return ret;
 		}
 	} else {
-		w->all_images_verified = 0;
+		WARNING("\"%"TS"\" does not contain image metadata.  Skipping image verification.",
+			wim->filename);
+	}
+
+	/* Verify the streams  */
+
+	for_lookup_table_entry(wim->lookup_table, append_lte_to_list, &stream_list);
+
+	memset(&progress, 0, sizeof(progress));
+
+	progress.verify_streams.wimfile = wim->filename;
+	list_for_each_entry(lte, &stream_list, extraction_list) {
+		progress.verify_streams.total_streams++;
+		progress.verify_streams.total_bytes += lte->size;
 	}
-	return ret;
+
+	ctx.progfunc = wim->progfunc;
+	ctx.progctx = wim->progctx;
+	ctx.progress = &progress;
+
+	ret = call_progress(ctx.progfunc, WIMLIB_PROGRESS_MSG_VERIFY_STREAMS,
+			    ctx.progress, ctx.progctx);
+	if (ret)
+		return ret;
+
+	return read_stream_list(&stream_list,
+				offsetof(struct wim_lookup_table_entry,
+					 extraction_list),
+				&cbs, VERIFY_STREAM_HASHES);
 }