Fixes, comments

[wimlib] / src / write.c

diff --git a/src/write.c b/src/write.c
index 9835dba03f67b70e15e5e9a812532774bc702cdd..ccad7f07caa5c3f4b7af309995ab5974d2742aa0 100644 (file)
--- a/src/write.c
+++ b/src/write.c
@@ -187,9 +187,9 @@ get_compress_func(int out_ctype)
  *
  * @chunk:       Uncompressed data of the chunk.
  * @chunk_size:          Size of the chunk (<= WIM_CHUNK_SIZE)
  *
  * @chunk:       Uncompressed data of the chunk.
  * @chunk_size:          Size of the chunk (<= WIM_CHUNK_SIZE)

- * @out_fp:      FILE * to write tho chunk to.
- * @out_ctype:   Compression type to use when writing the chunk (ignored if no
- *                     chunk table provided)
+ * @out_fp:      FILE * to write the chunk to.
+ * @compress:     Compression function to use (NULL if writing uncompressed
+ *                     data).

  * @chunk_tab:   Pointer to chunk table being created.  It is updated with the
  *                     offset of the chunk we write.
  *
  * @chunk_tab:   Pointer to chunk table being created.  It is updated with the
  *                     offset of the chunk we write.
  *


@@ -202,7 +202,7 @@ write_wim_resource_chunk(const void *chunk, unsigned chunk_size,
 {
        const u8 *out_chunk;
        unsigned out_chunk_size;
 {
        const u8 *out_chunk;
        unsigned out_chunk_size;

-       if (chunk_tab) {
+       if (compress) {

                u8 *compressed_chunk = alloca(chunk_size);
 
                out_chunk_size = compress(chunk, chunk_size, compressed_chunk);
                u8 *compressed_chunk = alloca(chunk_size);
 
                out_chunk_size = compress(chunk, chunk_size, compressed_chunk);


@@ -378,12 +378,12 @@ write_wim_resource(struct wim_lookup_table_entry *lte,
        ret = read_resource_prefix(lte, wim_resource_size(lte),
                                   write_resource_cb, &write_ctx, 0);
 
        ret = read_resource_prefix(lte, wim_resource_size(lte),
                                   write_resource_cb, &write_ctx, 0);
 

-       /* Verify SHA1 message digest of the resource,  Or, if the hash we had
-        * before is all 0's, just re-set it to be the new hash. */
+       /* Verify SHA1 message digest of the resource, or set the hash for the
+        * first time. */

        if (write_ctx.doing_sha) {
                u8 md[SHA1_HASH_SIZE];
                sha1_final(md, &write_ctx.sha_ctx);
        if (write_ctx.doing_sha) {
                u8 md[SHA1_HASH_SIZE];
                sha1_final(md, &write_ctx.sha_ctx);

-               if (is_zero_hash(lte->hash)) {
+               if (lte->unhashed) {

                        copy_hash(lte->hash, md);
                } else if (!hashes_equal(md, lte->hash)) {
                        ERROR("WIM resource has incorrect hash!");
                        copy_hash(lte->hash, md);
                } else if (!hashes_equal(md, lte->hash)) {
                        ERROR("WIM resource has incorrect hash!");


@@ -404,7 +404,6 @@ write_wim_resource(struct wim_lookup_table_entry *lte,
                /* Doing a raw write:  The new compressed size is the same as
                 * the compressed size in the other WIM. */
                new_size = lte->resource_entry.size;
                /* Doing a raw write:  The new compressed size is the same as
                 * the compressed size in the other WIM. */
                new_size = lte->resource_entry.size;

-               out_res_entry->flags = lte->resource_entry.flags;

        } else if (out_ctype == WIMLIB_COMPRESSION_TYPE_NONE) {
                /* Using WIMLIB_COMPRESSION_TYPE_NONE:  The new compressed size
                 * is the original size. */
        } else if (out_ctype == WIMLIB_COMPRESSION_TYPE_NONE) {
                /* Using WIMLIB_COMPRESSION_TYPE_NONE:  The new compressed size
                 * is the original size. */


@@ -599,8 +598,35 @@ do_write_streams_progress(union wimlib_progress_info *progress,
        }
 }
 
        }
 }
 

+static int
+sha1_chunk(const void *buf, size_t len, void *ctx)
+{
+       sha1_update(ctx, buf, len);
+       return 0;
+}
+
+static int
+sha1_resource(struct wim_lookup_table_entry *lte)
+{
+       int ret;
+       SHA_CTX sha_ctx;
+
+       sha1_init(&sha_ctx);
+       ret = read_resource_prefix(lte, wim_resource_size(lte),
+                                  sha1_chunk, &sha_ctx, 0);
+       if (ret == 0)
+               sha1_final(lte->hash, &sha_ctx);
+       return ret;
+}
+
+enum {
+       STREAMS_MERGED = 0,
+       STREAMS_NOT_MERGED = 1,
+};
+

 static int
 do_write_stream_list(struct list_head *my_resources,
 static int
 do_write_stream_list(struct list_head *my_resources,

+                    struct wim_lookup_table *lookup_table,

                     FILE *out_fp,
                     int out_ctype,
                     wimlib_progress_func_t progress_func,
                     FILE *out_fp,
                     int out_ctype,
                     wimlib_progress_func_t progress_func,


@@ -608,18 +634,61 @@ do_write_stream_list(struct list_head *my_resources,
                     int write_resource_flags)
 {
        int ret;
                     int write_resource_flags)
 {
        int ret;

-       struct wim_lookup_table_entry *lte, *tmp;
+       struct wim_lookup_table_entry *lte;
+
+       while (!list_empty(my_resources)) {
+               lte = container_of(my_resources->next,
+                                  struct wim_lookup_table_entry,
+                                  write_streams_list);
+               list_del(&lte->write_streams_list);
+               if (lte->unhashed && !lte->unique_size) {
+                       struct wim_lookup_table_entry *duplicate_lte;
+                       struct wim_lookup_table_entry **my_ptr;
+
+                       my_ptr = lte->my_ptr;
+                       ret = sha1_resource(lte);
+                       if (ret)
+                               return ret;
+                       duplicate_lte = __lookup_resource(lookup_table, lte->hash);
+                       if (duplicate_lte) {
+                               bool new_stream = (duplicate_lte->out_refcnt == 0);
+                               duplicate_lte->refcnt += lte->refcnt;
+                               duplicate_lte->out_refcnt += lte->refcnt;
+                               *my_ptr = duplicate_lte;
+                               free_lookup_table_entry(lte);
+
+                               if (new_stream) {
+                                       lte = duplicate_lte;
+                                       DEBUG("Stream of length %"PRIu64" is duplicate "
+                                             "with one already in WIM",
+                                             wim_resource_size(duplicate_lte));
+                               } else {
+                                       DEBUG("Discarding duplicate stream of length %"PRIu64,
+                                             wim_resource_size(duplicate_lte));
+                                       goto skip_to_progress;
+                               }
+
+                       } else {
+                               lookup_table_insert(lookup_table, lte);
+                               lte->out_refcnt = lte->refcnt;
+                               lte->unhashed = 0;
+                       }
+               }
+
+               wimlib_assert(lte->out_refcnt != 0);

 
 

-       list_for_each_entry_safe(lte, tmp, my_resources, staging_list) {

                ret = write_wim_resource(lte,
                                         out_fp,
                                         out_ctype,
                                         &lte->output_resource_entry,
                                         write_resource_flags);
                ret = write_wim_resource(lte,
                                         out_fp,
                                         out_ctype,
                                         &lte->output_resource_entry,
                                         write_resource_flags);

-               if (ret != 0)
+               if (ret)

                        return ret;
                        return ret;

-               list_del(&lte->staging_list);
-
+               if (lte->unhashed) {
+                       lookup_table_insert(lookup_table, lte);
+                       lte->unhashed = 0;
+               }
+       skip_to_progress:

                do_write_streams_progress(progress,
                                          progress_func,
                                          wim_resource_size(lte));
                do_write_streams_progress(progress,
                                          progress_func,
                                          wim_resource_size(lte));


@@ -629,6 +698,7 @@ do_write_stream_list(struct list_head *my_resources,
 
 static int
 write_stream_list_serial(struct list_head *stream_list,
 
 static int
 write_stream_list_serial(struct list_head *stream_list,

+                        struct wim_lookup_table *lookup_table,

                         FILE *out_fp,
                         int out_ctype,
                         int write_flags,
                         FILE *out_fp,
                         int out_ctype,
                         int write_flags,


@@ -644,7 +714,9 @@ write_stream_list_serial(struct list_head *stream_list,
        progress->write_streams.num_threads = 1;
        if (progress_func)
                progress_func(WIMLIB_PROGRESS_MSG_WRITE_STREAMS, progress);
        progress->write_streams.num_threads = 1;
        if (progress_func)
                progress_func(WIMLIB_PROGRESS_MSG_WRITE_STREAMS, progress);

-       return do_write_stream_list(stream_list, out_fp,
+       return do_write_stream_list(stream_list,
+                                   lookup_table,
+                                   out_fp,

                                    out_ctype, progress_func,
                                    progress, write_resource_flags);
 }
                                    out_ctype, progress_func,
                                    progress, write_resource_flags);
 }


@@ -1112,6 +1184,7 @@ get_default_num_threads()
 
 static int
 write_stream_list_parallel(struct list_head *stream_list,
 
 static int
 write_stream_list_parallel(struct list_head *stream_list,

+                          struct wim_lookup_table *lookup_table,

                           FILE *out_fp,
                           int out_ctype,
                           int write_flags,
                           FILE *out_fp,
                           int out_ctype,
                           int write_flags,


@@ -1206,6 +1279,7 @@ out_destroy_res_to_compress_queue:
 out_serial:
        WARNING("Falling back to single-threaded compression");
        return write_stream_list_serial(stream_list,
 out_serial:
        WARNING("Falling back to single-threaded compression");
        return write_stream_list_serial(stream_list,

+                                       lookup_table,

                                        out_fp,
                                        out_ctype,
                                        write_flags,
                                        out_fp,
                                        out_ctype,
                                        write_flags,


@@ -1220,18 +1294,19 @@ out_serial:
  * @out_ctype and up to @num_threads compressor threads.
  */
 static int
  * @out_ctype and up to @num_threads compressor threads.
  */
 static int

-write_stream_list(struct list_head *stream_list, FILE *out_fp,
-                 int out_ctype, int write_flags,
-                 unsigned num_threads,
-                 wimlib_progress_func_t progress_func)
+write_stream_list(struct list_head *stream_list,
+                 struct wim_lookup_table *lookup_table,
+                 FILE *out_fp, int out_ctype, int write_flags,
+                 unsigned num_threads, wimlib_progress_func_t progress_func)

 {
        struct wim_lookup_table_entry *lte;
        size_t num_streams = 0;
        u64 total_bytes = 0;
        u64 total_compression_bytes = 0;
        union wimlib_progress_info progress;
 {
        struct wim_lookup_table_entry *lte;
        size_t num_streams = 0;
        u64 total_bytes = 0;
        u64 total_compression_bytes = 0;
        union wimlib_progress_info progress;

+       int ret;

 
 

-       list_for_each_entry(lte, stream_list, staging_list) {
+       list_for_each_entry(lte, stream_list, write_streams_list) {

                num_streams++;
                total_bytes += wim_resource_size(lte);
                if (out_ctype != WIMLIB_COMPRESSION_TYPE_NONE
                num_streams++;
                total_bytes += wim_resource_size(lte);
                if (out_ctype != WIMLIB_COMPRESSION_TYPE_NONE


@@ -1251,21 +1326,24 @@ write_stream_list(struct list_head *stream_list, FILE *out_fp,
 
 #ifdef ENABLE_MULTITHREADED_COMPRESSION
        if (total_compression_bytes >= 1000000 && num_threads != 1)
 
 #ifdef ENABLE_MULTITHREADED_COMPRESSION
        if (total_compression_bytes >= 1000000 && num_threads != 1)

-               return write_stream_list_parallel(stream_list,
-                                                 out_fp,
-                                                 out_ctype,
-                                                 write_flags,
-                                                 num_threads,
-                                                 progress_func,
-                                                 &progress);
+               ret = write_stream_list_parallel(stream_list,
+                                                lookup_table,
+                                                out_fp,
+                                                out_ctype,
+                                                write_flags,
+                                                num_threads,
+                                                progress_func,
+                                                &progress);

        else
 #endif
        else
 #endif

-               return write_stream_list_serial(stream_list,
-                                               out_fp,
-                                               out_ctype,
-                                               write_flags,
-                                               progress_func,
-                                               &progress);
+               ret = write_stream_list_serial(stream_list,
+                                              lookup_table,
+                                              out_fp,
+                                              out_ctype,
+                                              write_flags,
+                                              progress_func,
+                                              &progress);
+       return ret;

 }
 
 struct lte_overwrite_prepare_args {
 }
 
 struct lte_overwrite_prepare_args {


@@ -1311,7 +1389,7 @@ wim_prepare_streams(WIMStruct *wim, off_t end_offset,
        int ret;
 
        for (int i = 0; i < wim->hdr.image_count; i++) {
        int ret;
 
        for (int i = 0; i < wim->hdr.image_count; i++) {

-               ret = lte_overwrite_prepare(wim->image_metadata[i].metadata_lte,
+               ret = lte_overwrite_prepare(wim->image_metadata[i]->metadata_lte,

                                            &args);
                if (ret)
                        return ret;
                                            &args);
                if (ret)
                        return ret;


@@ -1320,17 +1398,73 @@ wim_prepare_streams(WIMStruct *wim, off_t end_offset,
                                      lte_overwrite_prepare, &args);
 }
 
                                      lte_overwrite_prepare, &args);
 }
 

+struct stream_size_table {
+       struct hlist_head *array;
+       size_t num_entries;
+       size_t capacity;
+};
+
+static int
+init_stream_size_table(struct stream_size_table *tab, size_t capacity)
+{
+       tab->array = CALLOC(capacity, sizeof(tab->array[0]));
+       if (!tab->array)
+               return WIMLIB_ERR_NOMEM;
+       tab->num_entries = 0;
+       tab->capacity = capacity;
+       return 0;
+}
+
+static void
+destroy_stream_size_table(struct stream_size_table *tab)
+{
+       FREE(tab->array);
+}
+
+static int
+stream_size_table_insert(struct wim_lookup_table_entry *lte, void *_tab)
+{
+       struct stream_size_table *tab = _tab;
+       size_t pos;
+       struct wim_lookup_table_entry *hashed_lte;
+       struct hlist_node *tmp;
+
+       pos = hash_u64(wim_resource_size(lte)) % tab->capacity;
+       lte->unique_size = 1;
+       hlist_for_each_entry(hashed_lte, tmp, &tab->array[pos], hash_list_2) {
+               if (wim_resource_size(hashed_lte) == wim_resource_size(lte)) {
+                       lte->unique_size = 0;
+                       hashed_lte->unique_size = 0;
+                       break;
+               }
+       }
+
+       hlist_add_head(&lte->hash_list_2, &tab->array[pos]);
+       tab->num_entries++;
+       return 0;
+}
+
+
+struct find_streams_ctx {
+       struct list_head stream_list;
+       struct stream_size_table stream_size_tab;
+};
+

 static int
 inode_find_streams_to_write(struct wim_inode *inode,
                            struct wim_lookup_table *table,
 static int
 inode_find_streams_to_write(struct wim_inode *inode,
                            struct wim_lookup_table *table,

-                           struct list_head *stream_list)
+                           struct list_head *stream_list,
+                           struct stream_size_table *tab)

 {
        struct wim_lookup_table_entry *lte;
        for (unsigned i = 0; i <= inode->i_num_ads; i++) {
                lte = inode_stream_lte(inode, i, table);
                if (lte) {
 {
        struct wim_lookup_table_entry *lte;
        for (unsigned i = 0; i <= inode->i_num_ads; i++) {
                lte = inode_stream_lte(inode, i, table);
                if (lte) {

-                       if (lte->out_refcnt == 0)
-                               list_add_tail(&lte->staging_list, stream_list);
+                       if (lte->out_refcnt == 0) {
+                               if (lte->unhashed)
+                                       stream_size_table_insert(lte, tab);
+                               list_add_tail(&lte->write_streams_list, stream_list);
+                       }

                        lte->out_refcnt += inode->i_nlink;
                }
        }
                        lte->out_refcnt += inode->i_nlink;
                }
        }


@@ -1340,31 +1474,85 @@ inode_find_streams_to_write(struct wim_inode *inode,
 static int
 image_find_streams_to_write(WIMStruct *w)
 {
 static int
 image_find_streams_to_write(WIMStruct *w)
 {

+       struct wim_image_metadata *imd;
+       struct find_streams_ctx *ctx;

        struct wim_inode *inode;
        struct wim_inode *inode;

-       struct hlist_node *cur;
-       struct hlist_head *inode_list;
+       struct wim_lookup_table_entry *lte;
+
+       ctx = w->private;
+       imd = wim_get_current_image_metadata(w);
+
+       image_for_each_unhashed_stream(lte, imd) {
+               lte->out_refcnt = 0;
+               wimlib_assert(lte->unhashed);
+               wimlib_assert(lte->my_ptr != NULL);
+       }

 
 

-       inode_list = &wim_get_current_image_metadata(w)->inode_list;
-       hlist_for_each_entry(inode, cur, inode_list, i_hlist) {
+       /* Go through this image's inodes to find any streams that have not been
+        * found yet. */
+       image_for_each_inode(inode, imd) {

                inode_find_streams_to_write(inode, w->lookup_table,
                inode_find_streams_to_write(inode, w->lookup_table,

-                                           (struct list_head*)w->private);
+                                           &ctx->stream_list,
+                                           &ctx->stream_size_tab);

        }
        return 0;
 }
 
        }
        return 0;
 }
 

+/* Given a WIM that from which one or all of the images is being written, build
+ * the list of unique streams ('struct wim_lookup_table_entry's) that must be
+ * written, plus any unhashed streams that need to be written but may be
+ * identical to other hashed or unhashed streams being written.  These unhashed
+ * streams are checksummed while the streams are being written.  To aid this
+ * process, the member @unique_size is set to 1 on streams that have a unique
+ * size and therefore must be written.
+ *
+ * The out_refcnt member of each 'struct wim_lookup_table_entry' is set to
+ * indicate the number of times the stream is referenced in only the streams
+ * that are being written; this may still be adjusted later when unhashed
+ * streams are being resolved.
+ */
+static int
+prepare_stream_list(WIMStruct *wim, int image, struct list_head *stream_list)
+{
+       int ret;
+       struct find_streams_ctx ctx;
+
+       for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
+       ret = init_stream_size_table(&ctx.stream_size_tab, 9001);
+       if (ret)
+               return ret;
+       for_lookup_table_entry(wim->lookup_table, stream_size_table_insert,
+                              &ctx.stream_size_tab);
+       INIT_LIST_HEAD(&ctx.stream_list);
+       wim->private = &ctx;
+       for_image(wim, image, image_find_streams_to_write);
+       destroy_stream_size_table(&ctx.stream_size_tab);
+
+       INIT_LIST_HEAD(stream_list);
+       list_splice(&ctx.stream_list, stream_list);
+       return 0;
+}
+
+/* Writes the streams for the specified @image in @wim to @wim->out_fp.
+ */

 static int
 static int

-write_wim_streams(WIMStruct *w, int image, int write_flags,
-                            unsigned num_threads,
-                            wimlib_progress_func_t progress_func)
+write_wim_streams(WIMStruct *wim, int image, int write_flags,
+                 unsigned num_threads,
+                 wimlib_progress_func_t progress_func)

 {
 {

+       int ret;
+       struct list_head stream_list;

 
 

-       for_lookup_table_entry(w->lookup_table, lte_zero_out_refcnt, NULL);
-       LIST_HEAD(stream_list);
-       w->private = &stream_list;
-       for_image(w, image, image_find_streams_to_write);
-       return write_stream_list(&stream_list, w->out_fp,
-                                wimlib_get_compression_type(w), write_flags,
-                                num_threads, progress_func);
+       ret = prepare_stream_list(wim, image, &stream_list);
+       if (ret)
+               return ret;
+       return write_stream_list(&stream_list,
+                                wim->lookup_table,
+                                wim->out_fp,
+                                wimlib_get_compression_type(wim),
+                                write_flags,
+                                num_threads,
+                                progress_func);

 }
 
 /*
 }
 
 /*


@@ -1497,7 +1685,7 @@ finish_write(WIMStruct *w, int image, int write_flags,
        } else {
                memcpy(&hdr.boot_metadata_res_entry,
                       &w->image_metadata[
        } else {
                memcpy(&hdr.boot_metadata_res_entry,
                       &w->image_metadata[

-                         hdr.boot_idx - 1].metadata_lte->output_resource_entry,
+                         hdr.boot_idx - 1]->metadata_lte->output_resource_entry,

                       sizeof(struct resource_entry));
        }
 
                       sizeof(struct resource_entry));
        }
 


@@ -1509,7 +1697,7 @@ finish_write(WIMStruct *w, int image, int write_flags,
        }
 
        ret = write_header(&hdr, out);
        }
 
        ret = write_header(&hdr, out);

-       if (ret != 0)
+       if (ret)

                goto out;
 
        if (write_flags & WIMLIB_WRITE_FLAG_FSYNC) {
                goto out;
 
        if (write_flags & WIMLIB_WRITE_FLAG_FSYNC) {


@@ -1627,19 +1815,19 @@ wimlib_write(WIMStruct *w, const tchar *path,
        }
 
        ret = begin_write(w, path, write_flags);
        }
 
        ret = begin_write(w, path, write_flags);

-       if (ret != 0)
+       if (ret)

                goto out;
 
        ret = write_wim_streams(w, image, write_flags, num_threads,
                                progress_func);
                goto out;
 
        ret = write_wim_streams(w, image, write_flags, num_threads,
                                progress_func);

-       if (ret != 0)
+       if (ret)

                goto out;
 
        if (progress_func)
                progress_func(WIMLIB_PROGRESS_MSG_WRITE_METADATA_BEGIN, NULL);
 
        ret = for_image(w, image, write_metadata_resource);
                goto out;
 
        if (progress_func)
                progress_func(WIMLIB_PROGRESS_MSG_WRITE_METADATA_BEGIN, NULL);
 
        ret = for_image(w, image, write_metadata_resource);

-       if (ret != 0)
+       if (ret)

                goto out;
 
        if (progress_func)
                goto out;
 
        if (progress_func)


@@ -1656,7 +1844,7 @@ static bool
 any_images_modified(WIMStruct *w)
 {
        for (int i = 0; i < w->hdr.image_count; i++)
 any_images_modified(WIMStruct *w)
 {
        for (int i = 0; i < w->hdr.image_count; i++)

-               if (w->image_metadata[i].modified)
+               if (w->image_metadata[i]->modified)

                        return true;
        return false;
 }
                        return true;
        return false;
 }


@@ -1786,7 +1974,9 @@ overwrite_wim_inplace(WIMStruct *w, int write_flags,
        if (!list_empty(&stream_list)) {
                DEBUG("Writing newly added streams (offset = %"PRIu64")",
                      old_wim_end);
        if (!list_empty(&stream_list)) {
                DEBUG("Writing newly added streams (offset = %"PRIu64")",
                      old_wim_end);

-               ret = write_stream_list(&stream_list, w->out_fp,
+               ret = write_stream_list(&stream_list,
+                                       w->lookup_table,
+                                       w->out_fp,

                                        wimlib_get_compression_type(w),
                                        write_flags, num_threads,
                                        progress_func);
                                        wimlib_get_compression_type(w),
                                        write_flags, num_threads,
                                        progress_func);


@@ -1797,10 +1987,10 @@ overwrite_wim_inplace(WIMStruct *w, int write_flags,
        }
 
        for (int i = 0; i < w->hdr.image_count; i++) {
        }
 
        for (int i = 0; i < w->hdr.image_count; i++) {

-               if (w->image_metadata[i].modified) {
+               if (w->image_metadata[i]->modified) {

                        select_wim_image(w, i + 1);
                        ret = write_metadata_resource(w);
                        select_wim_image(w, i + 1);
                        ret = write_metadata_resource(w);

-                       if (ret != 0)
+                       if (ret)

                                goto out_ftruncate;
                }
        }
                                goto out_ftruncate;
                }
        }