X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fextract.c;h=cf8a2c922379b424a3986b6c6871259f7bb6d6b4;hp=68fab3fd2e95f3e30be19bf6e2dbf5514ae600bf;hb=7b587995ba7f2fb4e3a8d23f4be0ad91f1ab066a;hpb=c268a6111e676141c1eb13432cd46c77441a5b77

diff --git a/src/extract.c b/src/extract.c
index 68fab3fd..cf8a2c92 100644
--- a/src/extract.c
+++ b/src/extract.c
@@ -58,12 +58,10 @@
 #include "wimlib/resource.h"
 #include "wimlib/security.h"
 #include "wimlib/unix_data.h"
-#ifdef __WIN32__
-#  include "wimlib/win32.h" /* for realpath() equivalent */
-#endif
-#include "wimlib/xml.h"
 #include "wimlib/wildcard.h"
 #include "wimlib/wim.h"
+#include "wimlib/win32.h" /* for realpath() equivalent */
+#include "wimlib/xml.h"
 
 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x80000000
 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x40000000
@@ -98,7 +96,7 @@ do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 }
 
 static int
-start_file_phase(struct apply_ctx *ctx, uint64_t end_file_count, enum wimlib_progress_msg msg)
+start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
 {
 	ctx->progress.extract.current_file_count = 0;
 	ctx->progress.extract.end_file_count = end_file_count;
@@ -106,13 +104,13 @@ start_file_phase(struct apply_ctx *ctx, uint64_t end_file_count, enum wimlib_pro
 }
 
 int
-start_file_structure_phase(struct apply_ctx *ctx, uint64_t end_file_count)
+start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count)
 {
 	return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
 }
 
 int
-start_file_metadata_phase(struct apply_ctx *ctx, uint64_t end_file_count)
+start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count)
 {
 	return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 }
@@ -136,158 +134,124 @@ end_file_metadata_phase(struct apply_ctx *ctx)
 	return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
 }
 
-#define PWM_ALLOW_WIM_HDR 0x00001
+#define PWM_FOUND_WIM_HDR (-1)
 
-/* Read the header for a blob in a pipable WIM.  */
+/* Read the header for a blob in a pipable WIM.  If @pwm_hdr_ret is not NULL,
+ * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found.  */
 static int
-read_pwm_blob_header(WIMStruct *pwm, struct blob_descriptor *blob,
-		     struct wim_resource_descriptor *rdesc,
-		     int flags, struct wim_header_disk *hdr_ret)
+read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
+		     struct wim_reshdr *reshdr_ret,
+		     struct wim_header_disk *pwm_hdr_ret)
 {
-	union {
-		struct pwm_blob_hdr blob_hdr;
-		struct wim_header_disk pwm_hdr;
-	} buf;
-	struct wim_reshdr reshdr;
 	int ret;
+	struct pwm_blob_hdr blob_hdr;
+	u64 magic;
 
-	ret = full_read(&pwm->in_fd, &buf.blob_hdr, sizeof(buf.blob_hdr));
-	if (ret)
+	ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
+	if (unlikely(ret))
 		goto read_error;
 
-	if ((flags & PWM_ALLOW_WIM_HDR) &&
-	    le64_to_cpu(buf.blob_hdr.magic) == PWM_MAGIC)
-	{
-		BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.blob_hdr));
-		ret = full_read(&pwm->in_fd, &buf.blob_hdr + 1,
-				sizeof(buf.pwm_hdr) - sizeof(buf.blob_hdr));
+	magic = le64_to_cpu(blob_hdr.magic);
 
-		if (ret)
+	if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) {
+		memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr));
+		ret = full_read(&pwm->in_fd,
+				(u8 *)pwm_hdr_ret + sizeof(blob_hdr),
+				sizeof(*pwm_hdr_ret) - sizeof(blob_hdr));
+		if (unlikely(ret))
 			goto read_error;
-		blob->blob_location = BLOB_NONEXISTENT;
-		memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
-		return 0;
+		return PWM_FOUND_WIM_HDR;
 	}
 
-	if (le64_to_cpu(buf.blob_hdr.magic) != PWM_BLOB_MAGIC) {
-		ERROR("Data read on pipe is invalid (expected blob header).");
+	if (unlikely(magic != PWM_BLOB_MAGIC)) {
+		ERROR("Data read on pipe is invalid (expected blob header)");
 		return WIMLIB_ERR_INVALID_PIPABLE_WIM;
 	}
 
-	copy_hash(blob->hash, buf.blob_hdr.hash);
+	copy_hash(hash_ret, blob_hdr.hash);
+
+	reshdr_ret->size_in_wim = 0; /* Not available  */
+	reshdr_ret->flags = le32_to_cpu(blob_hdr.flags);
+	reshdr_ret->offset_in_wim = pwm->in_fd.offset;
+	reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size);
+
+	if (unlikely(reshdr_ret->uncompressed_size == 0)) {
+		ERROR("Data read on pipe is invalid (resource is of 0 size)");
+		return WIMLIB_ERR_INVALID_PIPABLE_WIM;
+	}
 
-	reshdr.size_in_wim = 0;
-	reshdr.flags = le32_to_cpu(buf.blob_hdr.flags);
-	reshdr.offset_in_wim = pwm->in_fd.offset;
-	reshdr.uncompressed_size = le64_to_cpu(buf.blob_hdr.uncompressed_size);
-	wim_res_hdr_to_desc(&reshdr, pwm, rdesc);
-	blob_set_is_located_in_wim_resource(blob, rdesc);
-	blob->flags = rdesc->flags;
-	blob->size = rdesc->uncompressed_size;
-	blob->offset_in_res = 0;
 	return 0;
 
 read_error:
-	ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
+	if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE)
+		ERROR("The pipe ended before all needed data was sent!");
+	else
+		ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
 	return ret;
 }
 
 static int
-read_blobs_from_pipe(struct apply_ctx *ctx,
-		     const struct read_blob_list_callbacks *cbs)
+read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 {
-	struct blob_descriptor *found_blob = NULL;
-	struct wim_resource_descriptor *rdesc = NULL;
-	struct blob_table *blob_table;
 	int ret;
+	u8 hash[SHA1_HASH_SIZE];
+	struct wim_reshdr reshdr;
+	struct wim_header_disk pwm_hdr;
+	struct wim_resource_descriptor rdesc;
+	struct blob_descriptor *blob;
 
-	ret = WIMLIB_ERR_NOMEM;
-	found_blob = new_blob_descriptor();
-	if (!found_blob)
-		goto out;
-
-	rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
-	if (!rdesc)
-		goto out;
-
-	blob_table = ctx->wim->blob_table;
-	memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
+	copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid);
 	ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
 	ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
 	ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
 	if (ret)
-		goto out;
+		return ret;
 
 	while (ctx->num_blobs_remaining) {
-		struct wim_header_disk pwm_hdr;
-		struct blob_descriptor *needed_blob;
 
-		if (found_blob->blob_location != BLOB_NONEXISTENT)
-			blob_unset_is_located_in_wim_resource(found_blob);
-		ret = read_pwm_blob_header(ctx->wim, found_blob, rdesc,
-					   PWM_ALLOW_WIM_HDR, &pwm_hdr);
-		if (ret)
-			goto out;
+		ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
 
-		if ((found_blob->blob_location != BLOB_NONEXISTENT)
-		    && !(found_blob->flags & WIM_RESHDR_FLAG_METADATA)
-		    && (needed_blob = lookup_blob(blob_table, found_blob->hash))
-		    && (needed_blob->out_refcnt))
-		{
-			needed_blob->offset_in_res = found_blob->offset_in_res;
-			needed_blob->flags = found_blob->flags;
-			needed_blob->size = found_blob->size;
+		if (ret == PWM_FOUND_WIM_HDR) {
+			u16 part_number = le16_to_cpu(pwm_hdr.part_number);
+			u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
 
-			blob_unset_is_located_in_wim_resource(found_blob);
-			blob_set_is_located_in_wim_resource(needed_blob, rdesc);
+			if (part_number == ctx->progress.extract.part_number &&
+			    total_parts == ctx->progress.extract.total_parts &&
+			    guids_equal(pwm_hdr.guid, ctx->progress.extract.guid))
+				continue;
 
-			ret = (*cbs->begin_blob)(needed_blob,
-						 cbs->begin_blob_ctx);
-			if (ret) {
-				blob_unset_is_located_in_wim_resource(needed_blob);
-				goto out;
-			}
+			copy_guid(ctx->progress.extract.guid, pwm_hdr.guid);
+			ctx->progress.extract.part_number = part_number;
+			ctx->progress.extract.total_parts = total_parts;
+			ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
+			if (ret)
+				return ret;
 
-			ret = extract_blob(needed_blob, needed_blob->size,
-					   cbs->consume_chunk,
-					   cbs->consume_chunk_ctx);
+			continue;
+		}
 
-			ret = (*cbs->end_blob)(needed_blob, ret,
-					       cbs->end_blob_ctx);
-			blob_unset_is_located_in_wim_resource(needed_blob);
+		if (ret)
+			return ret;
+
+		if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
+		    && (blob = lookup_blob(ctx->wim->blob_table, hash))
+		    && (blob->out_refcnt))
+		{
+			wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob);
+			ret = read_blob_with_sha1(blob, cbs);
+			blob_unset_is_located_in_wim_resource(blob);
 			if (ret)
-				goto out;
+				return ret;
 			ctx->num_blobs_remaining--;
-		} else if (found_blob->blob_location != BLOB_NONEXISTENT) {
-			ret = skip_wim_resource(found_blob->rdesc);
-			if (ret)
-				goto out;
 		} else {
-			u16 part_number = le16_to_cpu(pwm_hdr.part_number);
-			u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
-
-			if (part_number != ctx->progress.extract.part_number ||
-			    total_parts != ctx->progress.extract.total_parts ||
-			    memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
-				   WIM_GUID_LEN))
-			{
-				ctx->progress.extract.part_number = part_number;
-				ctx->progress.extract.total_parts = total_parts;
-				memcpy(ctx->progress.extract.guid,
-				       pwm_hdr.guid, WIM_GUID_LEN);
-				ret = extract_progress(ctx,
-						       WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
-				if (ret)
-					goto out;
-			}
+			wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
+			ret = skip_wim_resource(&rdesc);
+			if (ret)
+				return ret;
 		}
 	}
-	ret = 0;
-out:
-	if (found_blob && found_blob->blob_location != BLOB_IN_WIM)
-		FREE(rdesc);
-	free_blob_descriptor(found_blob);
-	return ret;
+
+	return 0;
 }
 
 /* Creates a temporary file opened for writing.  The open file descriptor is
@@ -339,8 +303,8 @@ begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 
 	if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
 		return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
-	else
-		return (*ctx->saved_cbs->begin_blob)(blob, ctx->saved_cbs->begin_blob_ctx);
+
+	return call_begin_blob(blob, ctx->saved_cbs);
 }
 
 static int
@@ -415,10 +379,9 @@ extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 					 ctx->tmpfile_name);
 		}
 		return ret;
-	} else {
-		return (*ctx->saved_cbs->consume_chunk)(chunk, size,
-							ctx->saved_cbs->consume_chunk_ctx);
 	}
+
+	return call_consume_chunk(chunk, size, ctx->saved_cbs);
 }
 
 static int
@@ -426,7 +389,7 @@ extract_from_tmpfile(const tchar *tmpfile_name, struct apply_ctx *ctx)
 {
 	struct blob_descriptor tmpfile_blob;
 	struct blob_descriptor *orig_blob = ctx->cur_blob;
-	const struct read_blob_list_callbacks *cbs = ctx->saved_cbs;
+	const struct read_blob_callbacks *cbs = ctx->saved_cbs;
 	int ret;
 	const u32 orig_refcnt = orig_blob->out_refcnt;
 
@@ -450,7 +413,7 @@ extract_from_tmpfile(const tchar *tmpfile_name, struct apply_ctx *ctx)
 		 * blob descriptor to callbacks provided by the extraction
 		 * backend as opposed to the tmpfile blob descriptor, since they
 		 * shouldn't actually read data from the blob other than through
-		 * the read_blob_prefix() call below.  But for
+		 * the read_blob_with_cbs() call below.  But for
 		 * WIMLIB_EXTRACT_FLAG_WIMBOOT mode on Windows it does matter
 		 * because it needs access to the original WIM resource
 		 * descriptor in order to create the external backing reference.
@@ -459,17 +422,17 @@ extract_from_tmpfile(const tchar *tmpfile_name, struct apply_ctx *ctx)
 		orig_blob->out_refcnt = 1;
 		orig_blob->inline_blob_extraction_targets[0] = targets[i];
 
-		ret = (*cbs->begin_blob)(orig_blob, cbs->begin_blob_ctx);
+		ret = call_begin_blob(orig_blob, cbs);
 		if (ret)
 			break;
 
-		/* Extra SHA-1 isn't necessary here, but it shouldn't hurt as
-		 * this case is very rare anyway.  */
-		ret = extract_blob(&tmpfile_blob, tmpfile_blob.size,
-				   cbs->consume_chunk,
-				   cbs->consume_chunk_ctx);
+		struct read_blob_callbacks wrapper_cbs = {
+			.consume_chunk	= cbs->consume_chunk,
+			.ctx		= cbs->ctx,
+		};
+		ret = read_blob_with_cbs(&tmpfile_blob, &wrapper_cbs);
 
-		ret = (*cbs->end_blob)(orig_blob, ret, cbs->end_blob_ctx);
+		ret = call_end_blob(orig_blob, ret, cbs);
 		if (ret)
 			break;
 	}
@@ -491,10 +454,9 @@ end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
 		tunlink(ctx->tmpfile_name);
 		FREE(ctx->tmpfile_name);
 		return status;
-	} else {
-		return (*ctx->saved_cbs->end_blob)(blob, status,
-						   ctx->saved_cbs->end_blob_ctx);
 	}
+
+	return call_end_blob(blob, status, ctx->saved_cbs);
 }
 
 /*
@@ -505,27 +467,23 @@ end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
  *
  * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
  *
- * This also works if the WIM is being read from a pipe, whereas attempting to
- * read blobs directly (e.g. with read_full_blob_into_buf()) will not.
+ * This also works if the WIM is being read from a pipe.
  *
  * This also will split up blobs that will need to be extracted to more than
  * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
  * Therefore, the apply_operations implementation need not worry about running
  * out of file descriptors, unless it might open more than one file descriptor
- * per nominal destination (e.g. Win32 currently might because the destination
- * file system might not support hard links).
+ * per 'blob_extraction_target' (e.g. Win32 currently might because the
+ * destination file system might not support hard links).
  */
 int
-extract_blob_list(struct apply_ctx *ctx,
-		  const struct read_blob_list_callbacks *cbs)
+extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
 {
-	struct read_blob_list_callbacks wrapper_cbs = {
-		.begin_blob        = begin_extract_blob_wrapper,
-		.begin_blob_ctx    = ctx,
-		.consume_chunk     = extract_chunk_wrapper,
-		.consume_chunk_ctx = ctx,
-		.end_blob          = end_extract_blob_wrapper,
-		.end_blob_ctx      = ctx,
+	struct read_blob_callbacks wrapper_cbs = {
+		.begin_blob	= begin_extract_blob_wrapper,
+		.consume_chunk	= extract_chunk_wrapper,
+		.end_blob	= end_extract_blob_wrapper,
+		.ctx		= ctx,
 	};
 	ctx->saved_cbs = cbs;
 	if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
@@ -569,7 +527,7 @@ extract_dentry_to_stdout(struct wim_dentry *dentry,
 	}
 
 	filedes_init(&_stdout, STDOUT_FILENO);
-	return extract_full_blob_to_fd(blob, &_stdout);
+	return extract_blob_to_fd(blob, &_stdout);
 }
 
 static int
@@ -1199,7 +1157,7 @@ inode_tally_features(const struct wim_inode *inode,
 		else
 			features->other_reparse_points++;
 	}
-	if (inode->i_security_id != -1)
+	if (inode_has_security_descriptor(inode))
 		features->security_descriptors++;
 	if (inode_has_unix_data(inode))
 		features->unix_data++;
@@ -1920,21 +1878,19 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
 	 * write_pipable_wim() for more details about the format of pipable
 	 * WIMs.)  */
 	{
-		struct blob_descriptor xml_blob;
-		struct wim_resource_descriptor xml_rdesc;
-		ret = read_pwm_blob_header(pwm, &xml_blob, &xml_rdesc, 0, NULL);
+		u8 hash[SHA1_HASH_SIZE];
+
+		ret = read_pwm_blob_header(pwm, hash,
+					   &pwm->hdr.xml_data_reshdr, NULL);
 		if (ret)
 			goto out_wimlib_free;
 
-		if (!(xml_blob.flags & WIM_RESHDR_FLAG_METADATA))
-		{
+		if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
 			ERROR("Expected XML data, but found non-metadata resource.");
 			ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
 			goto out_wimlib_free;
 		}
 
-		wim_res_desc_to_hdr(&xml_rdesc, &pwm->hdr.xml_data_reshdr);
-
 		ret = read_wim_xml_data(pwm);
 		if (ret)
 			goto out_wimlib_free;
@@ -1972,37 +1928,38 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
 
 	/* Load the needed metadata resource.  */
 	for (i = 1; i <= pwm->hdr.image_count; i++) {
-		struct blob_descriptor *metadata_blob;
 		struct wim_image_metadata *imd;
+		struct wim_reshdr reshdr;
 		struct wim_resource_descriptor *metadata_rdesc;
 
-		metadata_blob = new_blob_descriptor();
-		if (metadata_blob == NULL) {
-			ret = WIMLIB_ERR_NOMEM;
-			goto out_wimlib_free;
-		}
-		metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
-		if (metadata_rdesc == NULL) {
-			ret = WIMLIB_ERR_NOMEM;
-			free_blob_descriptor(metadata_blob);
+		imd = pwm->image_metadata[i - 1];
+
+		ret = WIMLIB_ERR_NOMEM;
+		imd->metadata_blob = new_blob_descriptor();
+		if (!imd->metadata_blob)
 			goto out_wimlib_free;
-		}
 
-		ret = read_pwm_blob_header(pwm, metadata_blob, metadata_rdesc, 0, NULL);
-		imd = pwm->image_metadata[i - 1];
-		imd->metadata_blob = metadata_blob;
-		if (ret) {
-			FREE(metadata_rdesc);
+		imd->metadata_blob->is_metadata = 1;
+
+		ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
+					   &reshdr, NULL);
+		if (ret)
 			goto out_wimlib_free;
-		}
 
-		if (!(metadata_blob->flags & WIM_RESHDR_FLAG_METADATA)) {
+		if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
 			ERROR("Expected metadata resource, but found "
-			      "non-metadata resource.");
+			      "non-metadata resource");
 			ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
 			goto out_wimlib_free;
 		}
 
+		ret = WIMLIB_ERR_NOMEM;
+		metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
+		if (!metadata_rdesc)
+			goto out_wimlib_free;
+		wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
+					    imd->metadata_blob);
+
 		if (i == image) {
 			/* Metadata resource is for the image being extracted.
 			 * Parse it and save the metadata in memory.  */