]> wimlib.net Git - wimlib/blobdiff - src/extract.c
Clean up pipable WIM extraction
[wimlib] / src / extract.c
index 24c3e7518ab1112374bd4a166755443ef8e7f324..0aa63f4bf4c17e15ca70f0f3fcf00048ba7746c6 100644 (file)
@@ -6,22 +6,20 @@
  */
 
 /*
- * Copyright (C) 2012, 2013, 2014 Eric Biggers
+ * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers
  *
- * This file is part of wimlib, a library for working with WIM files.
+ * 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
+ * Software Foundation; either version 3 of the License, or (at your option) any
+ * later version.
  *
- * 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
+ * This file 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 Lesser 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/.
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this file; if not, see http://www.gnu.org/licenses/.
  */
 
 /*
 #  include "config.h"
 #endif
 
+#include <errno.h>
+#include <fcntl.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
 #include "wimlib/apply.h"
+#include "wimlib/assert.h"
+#include "wimlib/blob_table.h"
 #include "wimlib/dentry.h"
 #include "wimlib/encoding.h"
 #include "wimlib/endianness.h"
 #include "wimlib/error.h"
-#include "wimlib/lookup_table.h"
 #include "wimlib/metadata.h"
 #include "wimlib/pathlist.h"
 #include "wimlib/paths.h"
 #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"
 
-#include <errno.h>
-#include <fcntl.h>
-#include <stdlib.h>
-#include <sys/stat.h>
-#include <unistd.h>
-
-#define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
-#define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x40000000
-#define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x20000000
+#define WIMLIB_EXTRACT_FLAG_FROM_PIPE   0x80000000
+#define WIMLIB_EXTRACT_FLAG_IMAGEMODE   0x40000000
 
 /* Keep in sync with wimlib.h  */
 #define WIMLIB_EXTRACT_MASK_PUBLIC                             \
         WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE  |       \
         WIMLIB_EXTRACT_FLAG_WIMBOOT)
 
-/* Check whether the extraction of a dentry should be skipped completely.  */
-static bool
-dentry_is_supported(struct wim_dentry *dentry,
-                   const struct wim_features *supported_features)
+/* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
+ * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA.  */
+int
+do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
 {
-       struct wim_inode *inode = dentry->d_inode;
+       ctx->count_until_file_progress = 500;  /* Arbitrary value to limit calls  */
+       return extract_progress(ctx, msg);
+}
 
-       if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
-               return supported_features->reparse_points ||
-                       (inode_is_symlink(inode) &&
-                        supported_features->symlink_reparse_points);
-       }
-       if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
-               if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
-                       return supported_features->encrypted_directories != 0;
-               else
-                       return supported_features->encrypted_files != 0;
-       }
-       return true;
+static int
+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;
+       return do_file_extract_progress(ctx, msg);
+}
+
+int
+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, u64 end_file_count)
+{
+       return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
+}
 
-#define PWM_ALLOW_WIM_HDR 0x00001
-#define PWM_SILENT_EOF   0x00002
+static int
+end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
+{
+       ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
+       return do_file_extract_progress(ctx, msg);
+}
 
-/* Read the header from a stream in a pipable WIM.  */
+int
+end_file_structure_phase(struct apply_ctx *ctx)
+{
+       return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
+}
+
+int
+end_file_metadata_phase(struct apply_ctx *ctx)
+{
+       return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
+}
+
+#define PWM_FOUND_WIM_HDR (-1)
+
+/* 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_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
-                      struct wim_resource_spec *rspec,
-                      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_stream_hdr stream_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.stream_hdr, sizeof(buf.stream_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) && buf.stream_hdr.magic == PWM_MAGIC) {
-               BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
-               ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
-                               sizeof(buf.pwm_hdr) - sizeof(buf.stream_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;
-               lte->resource_location = RESOURCE_NONEXISTENT;
-               memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
-               return 0;
+               return PWM_FOUND_WIM_HDR;
        }
 
-       if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
-               ERROR("Data read on pipe is invalid (expected stream 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(lte->hash, buf.stream_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.stream_hdr.flags);
-       reshdr.offset_in_wim = pwm->in_fd.offset;
-       reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
-       wim_res_hdr_to_spec(&reshdr, pwm, rspec);
-       lte_bind_wim_resource_spec(lte, rspec);
-       lte->flags = rspec->flags;
-       lte->size = rspec->uncompressed_size;
-       lte->offset_in_res = 0;
        return 0;
 
 read_error:
-       if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
+       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
-load_streams_from_pipe(struct apply_ctx *ctx,
-                      const struct read_stream_list_callbacks *cbs)
+read_blobs_from_pipe(struct apply_ctx *ctx,
+                    const struct read_blob_list_callbacks *cbs)
 {
-       struct wim_lookup_table_entry *found_lte = NULL;
-       struct wim_resource_spec *rspec = NULL;
-       struct wim_lookup_table *lookup_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_lte = new_lookup_table_entry();
-       if (!found_lte)
-               goto out;
-
-       rspec = MALLOC(sizeof(struct wim_resource_spec));
-       if (!rspec)
-               goto out;
-
-       lookup_table = ctx->wim->lookup_table;
        memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
        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) {
 
-       while (ctx->num_streams_remaining) {
-               struct wim_header_disk pwm_hdr;
-               struct wim_lookup_table_entry *needed_lte;
+               ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
+
+               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);
+
+                       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))
+                               continue;
+
+                       memcpy(ctx->progress.extract.guid, pwm_hdr.guid, WIM_GUID_LEN);
+                       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;
+
+                       continue;
+               }
 
-               if (found_lte->resource_location != RESOURCE_NONEXISTENT)
-                       lte_unbind_wim_resource_spec(found_lte);
-               ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
-                                            PWM_ALLOW_WIM_HDR, &pwm_hdr);
                if (ret)
-                       goto out;
+                       return ret;
 
-               if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
-                   && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
-                   && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
-                   && (needed_lte->out_refcnt))
+               wim_res_hdr_to_desc(&reshdr, ctx->wim, &rdesc);
+
+               if (!(rdesc.flags & WIM_RESHDR_FLAG_METADATA)
+                   && (blob = lookup_blob(ctx->wim->blob_table, hash))
+                   && (blob->out_refcnt))
                {
-                       needed_lte->offset_in_res = found_lte->offset_in_res;
-                       needed_lte->flags = found_lte->flags;
-                       needed_lte->size = found_lte->size;
+                       blob_set_is_located_in_nonsolid_wim_resource(blob, &rdesc);
 
-                       lte_unbind_wim_resource_spec(found_lte);
-                       lte_bind_wim_resource_spec(needed_lte, rspec);
+                       ret = (*cbs->begin_blob)(blob, cbs->begin_blob_ctx);
 
-                       ret = (*cbs->begin_stream)(needed_lte, 0,
-                                                  cbs->begin_stream_ctx);
-                       if (ret) {
-                               lte_unbind_wim_resource_spec(needed_lte);
-                               goto out;
-                       }
-
-                       ret = extract_stream(needed_lte, needed_lte->size,
-                                            cbs->consume_chunk,
-                                            cbs->consume_chunk_ctx);
+                       if (!ret) {
+                               ret = extract_blob(blob, blob->size,
+                                                  cbs->consume_chunk,
+                                                  cbs->consume_chunk_ctx);
 
-                       ret = (*cbs->end_stream)(needed_lte, ret,
-                                                cbs->end_stream_ctx);
-                       lte_unbind_wim_resource_spec(needed_lte);
-                       if (ret)
-                               goto out;
-                       ctx->num_streams_remaining--;
-               } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
-                       ret = skip_wim_stream(found_lte);
+                               ret = (*cbs->end_blob)(blob, ret,
+                                                      cbs->end_blob_ctx);
+                       }
+                       blob_unset_is_located_in_wim_resource(blob);
                        if (ret)
-                               goto out;
+                               return ret;
+                       ctx->num_blobs_remaining--;
                } else {
-                       u16 part_number = le16_to_cpu(pwm_hdr.part_number);
-                       u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
+                       ret = skip_wim_resource(&rdesc);
+                       if (ret)
+                               return ret;
+               }
+       }
 
-                       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;
-                       }
+       return 0;
+}
+
+/* Creates a temporary file opened for writing.  The open file descriptor is
+ * returned in @fd_ret and its name is returned in @name_ret (dynamically
+ * allocated).  */
+static int
+create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
+{
+       tchar *name;
+       int open_flags;
+       int raw_fd;
+
+retry:
+       name = ttempnam(NULL, T("wimlib"));
+       if (!name) {
+               ERROR_WITH_ERRNO("Failed to create temporary filename");
+               return WIMLIB_ERR_NOMEM;
+       }
+
+       open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
+#ifdef __WIN32__
+       open_flags |= _O_SHORT_LIVED;
+#endif
+       raw_fd = topen(name, open_flags, 0600);
+
+       if (raw_fd < 0) {
+               if (errno == EEXIST) {
+                       FREE(name);
+                       goto retry;
                }
+               ERROR_WITH_ERRNO("Failed to create temporary file "
+                                "\"%"TS"\"", name);
+               FREE(name);
+               return WIMLIB_ERR_OPEN;
        }
-       ret = 0;
-out:
-       if (found_lte->resource_location != RESOURCE_IN_WIM)
-               FREE(rspec);
-       free_lookup_table_entry(found_lte);
-       return ret;
+
+       filedes_init(fd_ret, raw_fd);
+       *name_ret = name;
+       return 0;
 }
 
 static int
-begin_extract_stream_with_progress(struct wim_lookup_table_entry *lte,
-                                  u32 flags, void *_ctx)
+begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
 {
        struct apply_ctx *ctx = _ctx;
 
-       ctx->cur_stream = lte;
+       ctx->cur_blob = blob;
+       ctx->cur_blob_offset = 0;
 
-       return (*ctx->saved_cbs->begin_stream)(lte, flags,
-                                              ctx->saved_cbs->begin_stream_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);
 }
 
 static int
-consume_chunk_with_progress(const void *chunk, size_t size, void *_ctx)
+extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
 {
        struct apply_ctx *ctx = _ctx;
        union wimlib_progress_info *progress = &ctx->progress;
        int ret;
 
+       ctx->cur_blob_offset += size;
+
        if (likely(ctx->supported_features.hard_links)) {
                progress->extract.completed_bytes +=
-                       (u64)size * ctx->cur_stream->out_refcnt;
+                       (u64)size * ctx->cur_blob->out_refcnt;
+               if (ctx->cur_blob_offset == ctx->cur_blob->size)
+                       progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
        } else {
-               const struct stream_owner *owners = stream_owners(ctx->cur_stream);
-               for (u32 i = 0; i < ctx->cur_stream->out_refcnt; i++) {
-                       const struct wim_inode *inode = owners[i].inode;
+               const struct blob_extraction_target *targets =
+                       blob_extraction_targets(ctx->cur_blob);
+               for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
+                       const struct wim_inode *inode = targets[i].inode;
                        const struct wim_dentry *dentry;
 
                        list_for_each_entry(dentry,
@@ -300,6 +345,8 @@ consume_chunk_with_progress(const void *chunk, size_t size, void *_ctx)
                                            d_extraction_alias_node)
                        {
                                progress->extract.completed_bytes += size;
+                               if (ctx->cur_blob_offset == ctx->cur_blob->size)
+                                       progress->extract.completed_streams++;
                        }
                }
        }
@@ -314,49 +361,156 @@ consume_chunk_with_progress(const void *chunk, size_t size, void *_ctx)
                {
                        ctx->next_progress = UINT64_MAX;
                } else {
-                       ctx->next_progress += progress->extract.total_bytes / 128;
-                       if (ctx->next_progress > progress->extract.total_bytes)
+                       /* Send new message as soon as another 1/128 of the
+                        * total has been extracted.  (Arbitrary number.)  */
+                       ctx->next_progress =
+                               progress->extract.completed_bytes +
+                                       progress->extract.total_bytes / 128;
+
+                       /* ... Unless that would be more than 5000000 bytes, in
+                        * which case send the next after the next 5000000
+                        * bytes.  (Another arbitrary number.)  */
+                       if (progress->extract.completed_bytes + 5000000 <
+                           ctx->next_progress)
+                               ctx->next_progress =
+                                       progress->extract.completed_bytes + 5000000;
+
+                       /* ... But always send a message as soon as we're
+                        * completely done.  */
+                       if (progress->extract.total_bytes < ctx->next_progress)
                                ctx->next_progress = progress->extract.total_bytes;
                }
        }
-       return (*ctx->saved_cbs->consume_chunk)(chunk, size,
-                                               ctx->saved_cbs->consume_chunk_ctx);
+
+       if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
+               /* Just extracting to temporary file for now.  */
+               ret = full_write(&ctx->tmpfile_fd, chunk, size);
+               if (ret) {
+                       ERROR_WITH_ERRNO("Error writing data to "
+                                        "temporary file \"%"TS"\"",
+                                        ctx->tmpfile_name);
+               }
+               return ret;
+       } else {
+               return (*ctx->saved_cbs->consume_chunk)(chunk, size,
+                                                       ctx->saved_cbs->consume_chunk_ctx);
+       }
+}
+
+static int
+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;
+       int ret;
+       const u32 orig_refcnt = orig_blob->out_refcnt;
+
+       BUILD_BUG_ON(MAX_OPEN_FILES <
+                    ARRAY_LEN(orig_blob->inline_blob_extraction_targets));
+
+       struct blob_extraction_target *targets = orig_blob->blob_extraction_targets;
+
+       /* Copy the blob's data from the temporary file to each of its targets.
+        *
+        * This is executed only in the very uncommon case that a blob is being
+        * extracted to more than MAX_OPEN_FILES targets!  */
+
+       memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
+       tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
+       tmpfile_blob.file_on_disk = ctx->tmpfile_name;
+       ret = 0;
+       for (u32 i = 0; i < orig_refcnt; i++) {
+
+               /* Note: it usually doesn't matter whether we pass the original
+                * 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
+                * 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.
+                */
+
+               orig_blob->out_refcnt = 1;
+               orig_blob->inline_blob_extraction_targets[0] = targets[i];
+
+               ret = (*cbs->begin_blob)(orig_blob, cbs->begin_blob_ctx);
+               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);
+
+               ret = (*cbs->end_blob)(orig_blob, ret, cbs->end_blob_ctx);
+               if (ret)
+                       break;
+       }
+       FREE(targets);
+       orig_blob->out_refcnt = 0;
+       return ret;
+}
+
+static int
+end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
+{
+       struct apply_ctx *ctx = _ctx;
+
+       if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
+               filedes_close(&ctx->tmpfile_fd);
+               if (!status)
+                       status = extract_from_tmpfile(ctx->tmpfile_name, ctx);
+               filedes_invalidate(&ctx->tmpfile_fd);
+               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);
+       }
 }
 
 /*
- * Read the list of single-instance streams to extract and feed their data into
- * the specified callback functions.
+ * Read the list of blobs to extract and feed their data into the specified
+ * callback functions.
  *
- * This handles checksumming each stream.
+ * This handles checksumming each blob.
  *
  * 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 streams directly (e.g. with read_full_stream_into_buf()) will not.
+ * read blobs directly (e.g. with read_full_blob_into_buf()) will not.
+ *
+ * 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 'blob_extraction_target' (e.g. Win32 currently might because the
+ * destination file system might not support hard links).
  */
 int
-extract_stream_list(struct apply_ctx *ctx,
-                   const struct read_stream_list_callbacks *cbs)
+extract_blob_list(struct apply_ctx *ctx,
+                 const struct read_blob_list_callbacks *cbs)
 {
-       struct read_stream_list_callbacks wrapper_cbs = {
-               .begin_stream      = begin_extract_stream_with_progress,
-               .begin_stream_ctx  = ctx,
-               .consume_chunk     = consume_chunk_with_progress,
+       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_stream        = cbs->end_stream,
-               .end_stream_ctx    = cbs->end_stream_ctx,
+               .end_blob          = end_extract_blob_wrapper,
+               .end_blob_ctx      = ctx,
        };
-       if (ctx->progfunc) {
-               ctx->saved_cbs = cbs;
-               cbs = &wrapper_cbs;
-       }
+       ctx->saved_cbs = cbs;
        if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
-               return load_streams_from_pipe(ctx, cbs);
+               return read_blobs_from_pipe(ctx, &wrapper_cbs);
        } else {
-               return read_stream_list(&ctx->stream_list,
-                                       offsetof(struct wim_lookup_table_entry,
-                                                extraction_list),
-                                       cbs, VERIFY_STREAM_HASHES);
+               return read_blob_list(&ctx->blob_list,
+                                     offsetof(struct blob_descriptor,
+                                              extraction_list),
+                                     &wrapper_cbs, VERIFY_BLOB_HASHES);
        }
 }
 
@@ -367,38 +521,39 @@ extract_stream_list(struct apply_ctx *ctx,
  * unnamed data stream only.  */
 static int
 extract_dentry_to_stdout(struct wim_dentry *dentry,
-                        const struct wim_lookup_table *lookup_table)
+                        const struct blob_table *blob_table)
 {
        struct wim_inode *inode = dentry->d_inode;
-       struct wim_lookup_table_entry *lte;
+       struct blob_descriptor *blob;
        struct filedes _stdout;
 
        if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
-                                  FILE_ATTRIBUTE_DIRECTORY))
+                                  FILE_ATTRIBUTE_DIRECTORY |
+                                  FILE_ATTRIBUTE_ENCRYPTED))
        {
                ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
                      "extracted to standard output", dentry_full_path(dentry));
                return WIMLIB_ERR_NOT_A_REGULAR_FILE;
        }
 
-       lte = inode_unnamed_lte(inode, lookup_table);
-       if (!lte) {
-               const u8 *hash = inode_unnamed_stream_hash(inode);
+       blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
+       if (!blob) {
+               const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
                if (!is_zero_hash(hash))
-                       return stream_not_found_error(inode, hash);
+                       return blob_not_found_error(inode, hash);
                return 0;
        }
 
        filedes_init(&_stdout, STDOUT_FILENO);
-       return extract_full_stream_to_fd(lte, &_stdout);
+       return extract_full_blob_to_fd(blob, &_stdout);
 }
 
 static int
 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
-                          const struct wim_lookup_table *lookup_table)
+                          const struct blob_table *blob_table)
 {
        for (size_t i = 0; i < num_dentries; i++) {
-               int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
+               int ret = extract_dentry_to_stdout(dentries[i], blob_table);
                if (ret)
                        return ret;
        }
@@ -537,6 +692,7 @@ destroy_dentry_list(struct list_head *dentry_list)
                inode = dentry->d_inode;
                dentry_reset_extraction_list_node(dentry);
                inode->i_visited = 0;
+               inode->i_can_externally_back = 0;
                if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
                        FREE(dentry->d_extraction_name);
                dentry->d_extraction_name = NULL;
@@ -545,13 +701,13 @@ destroy_dentry_list(struct list_head *dentry_list)
 }
 
 static void
-destroy_stream_list(struct list_head *stream_list)
+destroy_blob_list(struct list_head *blob_list)
 {
-       struct wim_lookup_table_entry *lte;
+       struct blob_descriptor *blob;
 
-       list_for_each_entry(lte, stream_list, extraction_list)
-               if (lte->out_refcnt > ARRAY_LEN(lte->inline_stream_owners))
-                       FREE(lte->stream_owners);
+       list_for_each_entry(blob, blob_list, extraction_list)
+               if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
+                       FREE(blob->blob_extraction_targets);
 }
 
 #ifdef __WIN32__
@@ -607,18 +763,17 @@ dentry_calculate_extraction_name(struct wim_dentry *dentry,
 {
        int ret;
 
-       if (!dentry_is_supported(dentry, &ctx->supported_features))
-               goto skip_dentry;
-
        if (dentry_is_root(dentry))
                return 0;
 
+#ifdef WITH_NTFS_3G
        if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
                dentry->d_extraction_name = dentry->file_name;
                dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
                                                   sizeof(utf16lechar);
                return 0;
        }
+#endif
 
        if (!ctx->supported_features.case_sensitive_filenames) {
                struct wim_dentry *other;
@@ -698,8 +853,7 @@ out_replace:
 
                utf16le_put_tstr(tchar_name);
 
-               dentry->d_extraction_name = memdup(fixed_name,
-                                                  2 * fixed_name_num_chars + 2);
+               dentry->d_extraction_name = TSTRDUP(fixed_name);
                if (!dentry->d_extraction_name)
                        return WIMLIB_ERR_NOMEM;
                dentry->d_extraction_name_nchars = fixed_name_num_chars;
@@ -756,34 +910,35 @@ dentry_list_calculate_extraction_names(struct list_head *dentry_list,
 
 static int
 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
-                      struct wim_lookup_table *lookup_table)
+                      struct blob_table *blob_table)
 {
        struct wim_inode *inode = dentry->d_inode;
-       struct wim_lookup_table_entry *lte;
+       struct blob_descriptor *blob;
        int ret;
        bool force = false;
 
-       /* Special case:  when extracting from a pipe, the WIM lookup table is
+       /* Special case:  when extracting from a pipe, the WIM blob table is
         * initially empty, so "resolving" an inode's streams is initially not
-        * possible.  However, we still need to keep track of which streams,
-        * identified by SHA1 message digests, need to be extracted, so we
-        * "resolve" the inode's streams anyway by allocating new entries.  */
+        * possible.  However, we still need to keep track of which blobs,
+        * identified by SHA-1 message digests, need to be extracted, so we
+        * "resolve" the inode's streams anyway by allocating a 'struct
+        * blob_descriptor' for each one.  */
        if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
                force = true;
-       ret = inode_resolve_streams(inode, lookup_table, force);
+       ret = inode_resolve_streams(inode, blob_table, force);
        if (ret)
                return ret;
-       for (u32 i = 0; i <= inode->i_num_ads; i++) {
-               lte = inode_stream_lte_resolved(inode, i);
-               if (lte)
-                       lte->out_refcnt = 0;
+       for (unsigned i = 0; i < inode->i_num_streams; i++) {
+               blob = stream_blob_resolved(&inode->i_streams[i]);
+               if (blob)
+                       blob->out_refcnt = 0;
        }
        return 0;
 }
 
 /*
  * For each dentry to be extracted, resolve all streams in the corresponding
- * inode and set 'out_refcnt' in each to 0.
+ * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
  *
  * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
  */
@@ -797,7 +952,7 @@ dentry_list_resolve_streams(struct list_head *dentry_list,
        list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
                ret = dentry_resolve_streams(dentry,
                                             ctx->extract_flags,
-                                            ctx->wim->lookup_table);
+                                            ctx->wim->blob_table);
                if (ret)
                        return ret;
        }
@@ -805,73 +960,129 @@ dentry_list_resolve_streams(struct list_head *dentry_list,
 }
 
 static int
-ref_stream(struct wim_lookup_table_entry *lte, u32 stream_idx,
-          struct wim_dentry *dentry, struct apply_ctx *ctx)
+ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
+          struct apply_ctx *ctx)
 {
        struct wim_inode *inode = dentry->d_inode;
-       struct stream_owner *stream_owners;
+       struct blob_descriptor *blob = stream_blob_resolved(strm);
+       struct blob_extraction_target *targets;
 
-       if (!lte)
+       if (!blob)
                return 0;
 
-       /* Tally the size only for each extraction of the stream (not hard
-        * links).  */
+       /* Tally the size only for each actual extraction of the stream (not
+        * additional hard links to the inode).  */
        if (inode->i_visited && ctx->supported_features.hard_links)
                return 0;
 
-       ctx->progress.extract.total_bytes += lte->size;
-       ctx->progress.extract.num_streams++;
+       ctx->progress.extract.total_bytes += blob->size;
+       ctx->progress.extract.total_streams++;
 
        if (inode->i_visited)
                return 0;
 
-       /* Add stream to the dentry_list only one time, even if it's going
-        * to be extracted to multiple inodes.  */
-       if (lte->out_refcnt == 0) {
-               list_add_tail(&lte->extraction_list, &ctx->stream_list);
-               ctx->num_streams_remaining++;
+       /* Add each blob to 'ctx->blob_list' only one time, regardless of how
+        * many extraction targets it will have.  */
+       if (blob->out_refcnt == 0) {
+               list_add_tail(&blob->extraction_list, &ctx->blob_list);
+               ctx->num_blobs_remaining++;
        }
 
-       /* If inode not yet been visited, append it to the stream_owners array.  */
-       if (lte->out_refcnt < ARRAY_LEN(lte->inline_stream_owners)) {
-               stream_owners = lte->inline_stream_owners;
+       /* Set this stream as an extraction target of 'blob'.  */
+
+       if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
+               targets = blob->inline_blob_extraction_targets;
        } else {
-               struct stream_owner *prev_stream_owners;
-               size_t alloc_stream_owners;
+               struct blob_extraction_target *prev_targets;
+               size_t alloc_blob_extraction_targets;
 
-               if (lte->out_refcnt == ARRAY_LEN(lte->inline_stream_owners)) {
-                       prev_stream_owners = NULL;
-                       alloc_stream_owners = ARRAY_LEN(lte->inline_stream_owners);
+               if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
+                       prev_targets = NULL;
+                       alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
                } else {
-                       prev_stream_owners = lte->stream_owners;
-                       alloc_stream_owners = lte->alloc_stream_owners;
+                       prev_targets = blob->blob_extraction_targets;
+                       alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
                }
 
-               if (lte->out_refcnt == alloc_stream_owners) {
-                       alloc_stream_owners *= 2;
-                       stream_owners = REALLOC(prev_stream_owners,
-                                              alloc_stream_owners *
-                                               sizeof(stream_owners[0]));
-                       if (!stream_owners)
+               if (blob->out_refcnt == alloc_blob_extraction_targets) {
+                       alloc_blob_extraction_targets *= 2;
+                       targets = REALLOC(prev_targets,
+                                         alloc_blob_extraction_targets *
+                                         sizeof(targets[0]));
+                       if (!targets)
                                return WIMLIB_ERR_NOMEM;
-                       if (!prev_stream_owners) {
-                               memcpy(stream_owners,
-                                      lte->inline_stream_owners,
-                                      sizeof(lte->inline_stream_owners));
+                       if (!prev_targets) {
+                               memcpy(targets,
+                                      blob->inline_blob_extraction_targets,
+                                      sizeof(blob->inline_blob_extraction_targets));
                        }
-                       lte->stream_owners = stream_owners;
-                       lte->alloc_stream_owners = alloc_stream_owners;
+                       blob->blob_extraction_targets = targets;
+                       blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
                }
-               stream_owners = lte->stream_owners;
+               targets = blob->blob_extraction_targets;
        }
-       stream_owners[lte->out_refcnt].inode = inode;
-       if (stream_idx == 0) {
-               stream_owners[lte->out_refcnt].stream_name = NULL;
-       } else {
-               stream_owners[lte->out_refcnt].stream_name =
-                       inode->i_ads_entries[stream_idx - 1].stream_name;
+       targets[blob->out_refcnt].inode = inode;
+       targets[blob->out_refcnt].stream = strm;
+       blob->out_refcnt++;
+       return 0;
+}
+
+static int
+ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
+                    struct wim_inode_stream *strm, struct apply_ctx *ctx)
+{
+       bool need_stream = false;
+       switch (strm->stream_type) {
+       case STREAM_TYPE_DATA:
+               if (stream_is_named(strm)) {
+                       /* Named data stream  */
+                       if (ctx->supported_features.named_data_streams)
+                               need_stream = true;
+               } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
+                                                   FILE_ATTRIBUTE_ENCRYPTED))
+                          && !(inode_is_symlink(inode)
+                               && !ctx->supported_features.reparse_points
+                               && ctx->supported_features.symlink_reparse_points))
+               {
+                       /*
+                        * Unnamed data stream.  Skip if any of the following is true:
+                        *
+                        * - file is a directory
+                        * - file is encrypted
+                        * - backend needs to create the file as UNIX symlink
+                        * - backend will extract the stream as externally backed
+                        */
+                       if (ctx->apply_ops->will_externally_back) {
+                               int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx);
+                               if (ret > 0) /* Error?  */
+                                       return ret;
+                               if (ret < 0) /* Won't externally back?  */
+                                       need_stream = true;
+                       } else {
+                               need_stream = true;
+                       }
+               }
+               break;
+       case STREAM_TYPE_REPARSE_POINT:
+               wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
+               if (ctx->supported_features.reparse_points ||
+                   (inode_is_symlink(inode) &&
+                    ctx->supported_features.symlink_reparse_points))
+                       need_stream = true;
+               break;
+       case STREAM_TYPE_EFSRPC_RAW_DATA:
+               wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
+               if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
+                       if (ctx->supported_features.encrypted_directories)
+                               need_stream = true;
+               } else {
+                       if (ctx->supported_features.encrypted_files)
+                               need_stream = true;
+               }
+               break;
        }
-       lte->out_refcnt++;
+       if (need_stream)
+               return ref_stream(strm, dentry, ctx);
        return 0;
 }
 
@@ -879,51 +1090,25 @@ static int
 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
 {
        struct wim_inode *inode = dentry->d_inode;
-       int ret;
-
-       /* The unnamed data stream will always be extracted, except in an
-        * unlikely case.  */
-       if (!inode_is_encrypted_directory(inode)) {
-               u16 stream_idx;
-               struct wim_lookup_table_entry *stream;
-
-               stream = inode_unnamed_stream_resolved(inode, &stream_idx);
-               ret = ref_stream(stream, stream_idx, dentry, ctx);
+       for (unsigned i = 0; i < inode->i_num_streams; i++) {
+               int ret = ref_stream_if_needed(dentry, inode,
+                                              &inode->i_streams[i], ctx);
                if (ret)
                        return ret;
        }
-
-       /* Named data streams will be extracted only if supported in the current
-        * extraction mode and volume, and to avoid complications, if not doing
-        * a linked extraction.  */
-       if (ctx->supported_features.named_data_streams) {
-               for (u16 i = 0; i < inode->i_num_ads; i++) {
-                       if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
-                               continue;
-                       ret = ref_stream(inode->i_ads_entries[i].lte, i + 1,
-                                        dentry, ctx);
-                       if (ret)
-                               return ret;
-               }
-       }
        inode->i_visited = 1;
        return 0;
 }
 
 /*
- * For each dentry to be extracted, iterate through the data streams of the
- * corresponding inode.  For each such stream that is not to be ignored due to
- * the supported features or extraction flags, add it to the list of streams to
- * be extracted (ctx->stream_list) if not already done so.
- *
- * Also builds a mapping from each stream to the inodes referencing it.
+ * Given a list of dentries to be extracted, build the list of blobs that need
+ * to be extracted, and for each blob determine the streams to which that blob
+ * will be extracted.
  *
- * This also initializes the extract progress info with byte and stream
+ * This also initializes the extract progress info with byte and blob
  * information.
  *
  * ctx->supported_features must be filled in.
- *
- * Possible error codes: WIMLIB_ERR_NOMEM.
  */
 static int
 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
@@ -981,7 +1166,7 @@ inode_tally_features(const struct wim_inode *inode,
                features->not_context_indexed_files++;
        if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
                features->sparse_files++;
-       if (inode_has_named_stream(inode))
+       if (inode_has_named_data_stream(inode))
                features->named_data_streams++;
        if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
                features->reparse_points++;
@@ -1032,6 +1217,18 @@ do_feature_check(const struct wim_features *required_features,
                 const struct wim_features *supported_features,
                 int extract_flags)
 {
+       /* Encrypted files.  */
+       if (required_features->encrypted_files &&
+           !supported_features->encrypted_files)
+               WARNING("Ignoring EFS-encrypted data of %lu files",
+                       required_features->encrypted_files);
+
+       /* Named data streams.  */
+       if (required_features->named_data_streams &&
+           !supported_features->named_data_streams)
+               WARNING("Ignoring named data streams of %lu files",
+                       required_features->named_data_streams);
+
        /* File attributes.  */
        if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
                /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
@@ -1069,18 +1266,6 @@ do_feature_check(const struct wim_features *required_features,
                                required_features->encrypted_directories);
        }
 
-       /* Encrypted files.  */
-       if (required_features->encrypted_files &&
-           !supported_features->encrypted_files)
-               WARNING("Ignoring %lu encrypted files",
-                       required_features->encrypted_files);
-
-       /* Named data streams.  */
-       if (required_features->named_data_streams &&
-           (!supported_features->named_data_streams))
-               WARNING("Ignoring named data streams of %lu files",
-                       required_features->named_data_streams);
-
        /* Hard links.  */
        if (required_features->hard_links && !supported_features->hard_links)
                WARNING("Extracting %lu hard links as independent files",
@@ -1100,12 +1285,11 @@ do_feature_check(const struct wim_features *required_features,
        {
                if (supported_features->symlink_reparse_points) {
                        if (required_features->other_reparse_points) {
-                               WARNING("Ignoring %lu non-symlink/junction "
-                                       "reparse point files",
+                               WARNING("Ignoring reparse data of %lu non-symlink/junction files",
                                        required_features->other_reparse_points);
                        }
                } else {
-                       WARNING("Ignoring %lu reparse point files",
+                       WARNING("Ignoring reparse data of %lu files",
                                required_features->reparse_points);
                }
        }
@@ -1189,7 +1373,7 @@ extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
 
        if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
                ret = extract_dentries_to_stdout(trees, num_trees,
-                                                wim->lookup_table);
+                                                wim->blob_table);
                goto out;
        }
 
@@ -1227,7 +1411,9 @@ extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
                                                                         wim->current_image);
                ctx->progress.extract.target = target;
        }
-       INIT_LIST_HEAD(&ctx->stream_list);
+       INIT_LIST_HEAD(&ctx->blob_list);
+       filedes_invalidate(&ctx->tmpfile_fd);
+       ctx->apply_ops = ops;
 
        ret = (*ops->get_supported_features)(target, &ctx->supported_features);
        if (ret)
@@ -1248,21 +1434,26 @@ extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
        if (ret)
                goto out_cleanup;
 
+       if (unlikely(list_empty(&dentry_list))) {
+               WARNING("There is nothing to extract!");
+               goto out_cleanup;
+       }
+
        ret = dentry_list_resolve_streams(&dentry_list, ctx);
        if (ret)
                goto out_cleanup;
 
+       dentry_list_build_inode_alias_lists(&dentry_list);
+
        ret = dentry_list_ref_streams(&dentry_list, ctx);
        if (ret)
                goto out_cleanup;
 
-       dentry_list_build_inode_alias_lists(&dentry_list);
-
        if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
                /* When extracting from a pipe, the number of bytes of data to
                 * extract can't be determined in the normal way (examining the
-                * lookup table), since at this point all we have is a set of
-                * SHA1 message digests of streams that need to be extracted.
+                * blob table), since at this point all we have is a set of
+                * SHA-1 message digests of blobs that need to be extracted.
                 * However, we can get a reasonably accurate estimate by taking
                 * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
                 * data.  This does assume that a full image is being extracted,
@@ -1305,7 +1496,7 @@ extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
                                       WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
                                       WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
 out_cleanup:
-       destroy_stream_list(&ctx->stream_list);
+       destroy_blob_list(&ctx->blob_list);
        destroy_dentry_list(&dentry_list);
        FREE(ctx);
 out:
@@ -1315,23 +1506,20 @@ out:
 static int
 mkdir_if_needed(const tchar *target)
 {
-       struct stat stbuf;
-       if (tstat(target, &stbuf)) {
-               if (errno == ENOENT) {
-                       if (tmkdir(target, 0755)) {
-                               ERROR_WITH_ERRNO("Failed to create directory "
-                                                "\"%"TS"\"", target);
-                               return WIMLIB_ERR_MKDIR;
-                       }
-               } else {
-                       ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
-                       return WIMLIB_ERR_STAT;
-               }
-       } else if (!S_ISDIR(stbuf.st_mode)) {
-               ERROR("\"%"TS"\" is not a directory", target);
-               return WIMLIB_ERR_NOTDIR;
-       }
-       return 0;
+       if (!tmkdir(target, 0755))
+               return 0;
+
+       if (errno == EEXIST)
+               return 0;
+
+#ifdef __WIN32__
+       /* _wmkdir() fails with EACCES if called on a drive root directory.  */
+       if (errno == EACCES)
+               return 0;
+#endif
+
+       ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
+       return WIMLIB_ERR_MKDIR;
 }
 
 /* Make sure the extraction flags make sense, and update them if needed.  */
@@ -1362,12 +1550,16 @@ check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
        }
 #endif
 
-#ifndef __WIN32__
        if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
+#ifdef __WIN32__
+               if (!wim->filename)
+                       return WIMLIB_ERR_NO_FILENAME;
+#else
                ERROR("WIMBoot extraction is only supported on Windows!");
                return WIMLIB_ERR_UNSUPPORTED;
-       }
 #endif
+       }
+
 
        if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
                              WIMLIB_EXTRACT_FLAG_NORPFIX |
@@ -1449,7 +1641,7 @@ do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
        if (ret)
                return ret;
 
-       ret = wim_checksum_unhashed_streams(wim);
+       ret = wim_checksum_unhashed_blobs(wim);
        if (ret)
                return ret;
 
@@ -1570,8 +1762,6 @@ extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
        int image;
        const tchar *image_name;
 
-       extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
-
        if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
                ERROR("Cannot extract multiple images in NTFS extraction mode.");
                return WIMLIB_ERR_INVALID_PARAM;
@@ -1673,8 +1863,8 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
 
        /* Read the WIM header from the pipe and get a WIMStruct to represent
         * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
-        * wimlib_open_wim(), getting a WIMStruct in this way will result in
-        * an empty lookup table, no XML data read, and no filename set.  */
+        * wimlib_open_wim(), getting a WIMStruct in this way will result in an
+        * empty blob table, no XML data read, and no filename set.  */
        ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
                                    progfunc, progctx);
        if (ret)
@@ -1706,22 +1896,19 @@ wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
         * write_pipable_wim() for more details about the format of pipable
         * WIMs.)  */
        {
-               struct wim_lookup_table_entry xml_lte;
-               struct wim_resource_spec xml_rspec;
-               ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 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_lte.flags & WIM_RESHDR_FLAG_METADATA))
-               {
-                       ERROR("Expected XML data, but found non-metadata "
-                             "stream.");
+               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_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
-
                ret = read_wim_xml_data(pwm);
                if (ret)
                        goto out_wimlib_free;
@@ -1759,48 +1946,50 @@ 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 wim_lookup_table_entry *metadata_lte;
                struct wim_image_metadata *imd;
-               struct wim_resource_spec *metadata_rspec;
+               struct wim_reshdr reshdr;
+               struct wim_resource_descriptor *metadata_rdesc;
 
-               metadata_lte = new_lookup_table_entry();
-               if (metadata_lte == NULL) {
-                       ret = WIMLIB_ERR_NOMEM;
-                       goto out_wimlib_free;
-               }
-               metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
-               if (metadata_rspec == NULL) {
-                       ret = WIMLIB_ERR_NOMEM;
-                       free_lookup_table_entry(metadata_lte);
+               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_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
-               imd = pwm->image_metadata[i - 1];
-               imd->metadata_lte = metadata_lte;
-               if (ret) {
-                       FREE(metadata_rspec);
+               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_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
+               if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
                        ERROR("Expected metadata resource, but found "
-                             "non-metadata stream.");
+                             "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_res_hdr_to_desc(&reshdr, pwm, metadata_rdesc);
+               blob_set_is_located_in_nonsolid_wim_resource(imd->metadata_blob,
+                                                            metadata_rdesc);
+
                if (i == image) {
                        /* Metadata resource is for the image being extracted.
                         * Parse it and save the metadata in memory.  */
-                       ret = read_metadata_resource(pwm, imd);
+                       ret = read_metadata_resource(imd);
                        if (ret)
                                goto out_wimlib_free;
                        imd->modified = 1;
                } else {
                        /* Metadata resource is not for the image being
                         * extracted.  Skip over it.  */
-                       ret = skip_wim_stream(metadata_lte);
+                       ret = skip_wim_resource(metadata_rdesc);
                        if (ret)
                                goto out_wimlib_free;
                }