You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@subversion.apache.org by st...@apache.org on 2012/11/18 22:45:55 UTC
svn commit: r1410995 [2/2] - in /subversion/trunk: build.conf
tools/server-side/fsfs-stats.c
Copied: subversion/trunk/tools/server-side/fsfs-stats.c (from r1410722, subversion/trunk/tools/server-side/fsfs-reorg.c)
URL: http://svn.apache.org/viewvc/subversion/trunk/tools/server-side/fsfs-stats.c?p2=subversion/trunk/tools/server-side/fsfs-stats.c&p1=subversion/trunk/tools/server-side/fsfs-reorg.c&r1=1410722&r2=1410995&rev=1410995&view=diff
==============================================================================
--- subversion/trunk/tools/server-side/fsfs-reorg.c (original)
+++ subversion/trunk/tools/server-side/fsfs-stats.c Sun Nov 18 21:45:53 2012
@@ -22,6 +22,7 @@
#include <assert.h>
+#include <sys/stat.h>
#include <apr.h>
#include <apr_general.h>
@@ -40,6 +41,7 @@
#include "private/svn_string_private.h"
#include "private/svn_subr_private.h"
#include "private/svn_dep_compat.h"
+#include <../test/testutil.h>
#ifndef _
#define _(x) x
@@ -47,293 +49,88 @@
#define ERROR_TAG "diff: "
-/* forward declarations */
-typedef struct noderev_t noderev_t;
-typedef struct revision_info_t revision_info_t;
-
-/* A FSFS rev file is sequence of fragments and unused space (the latter
- * only being inserted by this tool and not during ordinary SVN operation).
- *
- * This type defines the type of any fragment.
- *
- * Please note that the classification as "property", "dir" or "file"
- * fragments is only to be used while determining the future placement
- * of a representation. If the rep is shared, the same rep may be used
- * as *any* of the 3 kinds.
- */
-enum fragment_kind_t
+typedef enum rep_kind_t
{
- /* the 2 number line containing changes and root node offsets */
- header_fragment,
+ unused_rep,
- /* list of all changes in a revision */
- changes_fragment,
+ dir_property_rep,
- /* (the textual representation of) a noderev */
- noderev_fragment,
-
- /* a property rep (including PLAIN / DELTA header) */
- property_fragment,
+ file_property_rep,
/* a directory rep (including PLAIN / DELTA header) */
- dir_fragment,
+ dir_rep,
/* a file rep (including PLAIN / DELTA header) */
- file_fragment
-};
-
-/* A fragment. This is used to represent the final ordering, i.e. there
- * will be an array containing elements of this type that basically put
- * a fragment at some location in the target file.
- */
-typedef struct fragment_t
-{
- /* position in the target file */
- apr_size_t position;
-
- /* kind of fragment */
- enum fragment_kind_t kind;
-
- /* pointer to the fragment struct; type depends on KIND */
- void *data;
-} fragment_t;
-
-/* Location info for a single revision.
- */
-typedef struct revision_location_t
-{
- /* pack file offset (manifest value), 0 for non-packed files */
- apr_size_t offset;
-
- /* offset of the changes list relative to OFFSET */
- apr_size_t changes;
+ file_rep
+} rep_kind_t;
- /* length of the changes list on bytes */
- apr_size_t changes_len;
-
- /* first offset behind the revision data in the pack file (file length
- * for non-packed revs) */
- apr_size_t end;
-} revision_location_t;
-
-/* Absolute position and size of some item.
+/* A representation fragment.
*/
-typedef struct location_t
+typedef struct representation_t
{
/* absolute offset in the file */
apr_size_t offset;
/* item length in bytes */
apr_size_t size;
-} location_t;
-
-/* A parsed directory entry. Note that instances of this struct may be
- * shared between different DIRECTORY_T containers.
- */
-typedef struct direntry_t
-{
- /* (local) entry / path name */
- const char *name;
-
- /* strlen (name) */
- apr_size_t name_len;
-
- /* node rev providing ID and representation(s) */
- noderev_t *node;
-} direntry_t;
-
-/* Representation of a parsed directory content.
- */
-typedef struct directory_t
-{
- /* array of pointers to DIRENTRY_T */
- apr_array_header_t *entries;
-
- /* MD5 of the textual representation. Will be set lazily as a side-effect
- * of determining the length of this dir's textual representation. */
- unsigned char target_md5[16];
-
- /* (expanded) length of the textual representation.
- * Determined lazily during the write process. */
- apr_size_t size;
-} directory_t;
-
-/* A representation fragment.
- */
-typedef struct representation_t
-{
- /* location in the source file */
- location_t original;
- /* location in the reordered target file */
- location_t target;
-
- /* length of the PLAIN / DELTA line in the source file in bytes */
- apr_size_t header_size;
+ apr_size_t expanded_size;
/* deltification base, or NULL if there is none */
struct representation_t *delta_base;
/* revision that contains this representation
* (may be referenced by other revisions, though) */
- revision_info_t *revision;
-
- /* representation content parsed as a directory. This will be NULL, if
- * *no* directory noderev uses this representation. */
- directory_t *dir;
-
- /* the source content has a PLAIN header, so we may simply copy the
- * source content into the target */
- svn_boolean_t is_plain;
-
- /* coloring flag used in the reordering algorithm to keep track of
- * representations that still need to be placed. */
- svn_boolean_t covered;
-} representation_t;
-
-/* A node rev.
- */
-struct noderev_t
-{
- /* location within the source file */
- location_t original;
-
- /* location within the reorganized target file. */
- location_t target;
-
- /* predecessor node, or NULL if there is none */
- noderev_t *predecessor;
-
- /* content representation; may be NULL if there is none */
- representation_t *text;
+
+ apr_uint32_t revision;
+ apr_uint32_t ref_count;
- /* properties representation; may be NULL if there is none */
- representation_t *props;
+ struct
+ {
+ /* length of the PLAIN / DELTA line in the source file in bytes */
+ apr_size_t header_size : 12;
- /* revision that this noderev belongs to */
- revision_info_t *revision;
+ rep_kind_t kind : 3;
- /* coloring flag used in the reordering algorithm to keep track of
- * representations that still need to be placed. */
- svn_boolean_t covered;
-};
+ /* the source content has a PLAIN header, so we may simply copy the
+ * source content into the target */
+ svn_boolean_t is_plain : 1;
+ };
+
+} representation_t;
/* Represents a single revision.
* There will be only one instance per revision. */
-struct revision_info_t
+typedef struct revision_info_t
{
/* number of this revision */
svn_revnum_t revision;
- /* position in the source file */
- revision_location_t original;
+ /* pack file offset (manifest value), 0 for non-packed files */
+ apr_size_t offset;
+
+ /* offset of the changes list relative to OFFSET */
+ apr_size_t changes;
- /* position in the reorganized target file */
- revision_location_t target;
+ /* length of the changes list on bytes */
+ apr_size_t changes_len;
- /* noderev of the root directory */
- noderev_t *root_noderev;
+ /* offset of the changes list relative to OFFSET */
+ apr_size_t change_count;
- /* all noderevs_t of this revision (in no particular order),
- * i.e. those that point back to this struct */
- apr_array_header_t *node_revs;
+ /* first offset behind the revision data in the pack file (file length
+ * for non-packed revs) */
+ apr_size_t end;
+
+ apr_size_t dir_noderev_count;
+ apr_size_t file_noderev_count;
+ apr_size_t dir_noderev_size;
+ apr_size_t file_noderev_size;
/* all representation_t of this revision (in no particular order),
* i.e. those that point back to this struct */
apr_array_header_t *representations;
-};
-
-/* Represents a packed revision file.
- */
-typedef struct revision_pack_t
-{
- /* first revision in the pack file */
- svn_revnum_t base;
-
- /* revision_info_t* of all revisions in the pack file; in revision order. */
- apr_array_header_t *info;
-
- /* list of fragments to place in the target pack file; in target order. */
- apr_array_header_t *fragments;
-
- /* source pack file length */
- apr_size_t filesize;
-
- /* temporary value. Equal to the number of bytes in the target pack file
- * already allocated to fragments. */
- apr_size_t target_offset;
-} revision_pack_t;
-
-/* Cache for revision source content. All content is stored in DATA and
- * the HASH maps revision number to an svn_string_t instance whose data
- * member points into DATA.
- *
- * Once TOTAL_SIZE exceeds LIMIT, all content will be discarded. Similarly,
- * the hash gets cleared every 10000 insertions to keep the HASH_POOL
- * memory usage in check.
- */
-typedef struct content_cache_t
-{
- /* pool used for HASH */
- apr_pool_t *hash_pool;
-
- /* svn_revnum_t -> svn_string_t.
- * The strings become (potentially) invalid when adding new cache entries. */
- apr_hash_t *hash;
-
- /* data buffer. the first TOTAL_SIZE bytes are actually being used. */
- char *data;
-
- /* DATA capacity */
- apr_size_t limit;
-
- /* number of bytes used in DATA */
- apr_size_t total_size;
-
- /* number of insertions since the last hash cleanup */
- apr_size_t insert_count;
-} content_cache_t;
-
-/* A cached directory. In contrast to directory_t, this stored the data as
- * the plain hash that the normal FSFS will use to serialize & diff dirs.
- */
-typedef struct dir_cache_entry_t
-{
- /* revision containing the representation */
- svn_revnum_t revision;
-
- /* offset of the representation within that revision */
- apr_size_t offset;
-
- /* key-value representation of the directory entries */
- apr_hash_t *hash;
-} dir_cache_entry_t;
-
-/* Directory cache. (revision, offset) will be mapped directly into the
- * ENTRIES array of ENTRY_COUNT buckets (many entries will be NULL).
- * Two alternating pools will be used to allocate dir content.
- *
- * If the INSERT_COUNT exceeds a given limit, the pools get exchanged and
- * the older of the two will be cleared. This is to keep dir objects valid
- * for at least one insertion.
- */
-typedef struct dir_cache_t
-{
- /* fixed-size array of ENTRY_COUNT elements */
- dir_cache_entry_t *entries;
-
- /* currently used for entry allocations */
- apr_pool_t *pool1;
-
- /* previously used for entry allocations */
- apr_pool_t *pool2;
-
- /* size of ENTRIES in elements */
- apr_size_t entry_count;
-
- /* number of directory elements added. I.e. usually >> #cached dirs */
- apr_size_t insert_count;
-} dir_cache_t;
+} revision_info_t;
/* A cached, undeltified txdelta window.
*/
@@ -398,19 +195,10 @@ typedef struct fs_fs_t
/* all revisions */
apr_array_header_t *revisions;
- /* all packed files */
- apr_array_header_t *packs;
-
/* empty representation.
* Used as a dummy base for DELTA reps without base. */
representation_t *null_base;
- /* revision content cache */
- content_cache_t *cache;
-
- /* directory hash cache */
- dir_cache_t *dir_cache;
-
/* undeltified txdelta window cache */
window_cache_t *window_cache;
} fs_fs_t;
@@ -458,238 +246,46 @@ open_rev_or_pack_file(apr_file_t **file,
* in *CONTENT.
*/
static svn_error_t *
-read_rev_or_pack_file(svn_stringbuf_t **content,
+rev_or_pack_file_size(apr_off_t *file_size,
fs_fs_t *fs,
svn_revnum_t rev,
apr_pool_t *pool)
{
- return svn_stringbuf_from_file2(content,
- rev_or_pack_file_name(fs, rev, pool),
- pool);
-}
-
-/* Return a new content cache with the given size LIMIT. Use POOL for
- * all cache-related allocations.
- */
-static content_cache_t *
-create_content_cache(apr_pool_t *pool,
- apr_size_t limit)
-{
- content_cache_t *result = apr_pcalloc(pool, sizeof(*result));
-
- result->hash_pool = svn_pool_create(pool);
- result->hash = svn_hash__make(result->hash_pool);
- result->limit = limit;
- result->total_size = 0;
- result->insert_count = 0;
- result->data = apr_palloc(pool, limit);
-
- return result;
-}
-
-/* Return the content of revision REVISION from CACHE. Return NULL upon a
- * cache miss. This is a cache-internal function.
- */
-static svn_string_t *
-get_cached_content(content_cache_t *cache,
- svn_revnum_t revision)
-{
- return apr_hash_get(cache->hash, &revision, sizeof(revision));
-}
-
-/* Take the content in DATA and store it under REVISION in CACHE.
- * This is a cache-internal function.
- */
-static void
-set_cached_content(content_cache_t *cache,
- svn_revnum_t revision,
- svn_string_t *data)
-{
- svn_string_t *content;
- svn_revnum_t *key;
-
- /* double insertion? -> broken cache logic */
- assert(get_cached_content(cache, revision) == NULL);
-
- /* purge the cache upon overflow */
- if (cache->total_size + data->len > cache->limit)
- {
- /* the hash pool grows slowly over time; clear it once in a while */
- if (cache->insert_count > 10000)
- {
- svn_pool_clear(cache->hash_pool);
- cache->hash = svn_hash__make(cache->hash_pool);
- cache->insert_count = 0;
- }
- else
- cache->hash = svn_hash__make(cache->hash_pool);
-
- cache->total_size = 0;
-
- /* buffer overflow / revision too large */
- if (data->len > cache->limit)
- SVN_ERR_MALFUNCTION_NO_RETURN();
- }
-
- /* copy data to cache and update he index (hash) */
- content = apr_palloc(cache->hash_pool, sizeof(*content));
- content->data = cache->data + cache->total_size;
- content->len = data->len;
+ apr_file_t *file;
+ apr_finfo_t finfo;
- memcpy(cache->data + cache->total_size, data->data, data->len);
- cache->total_size += data->len;
-
- key = apr_palloc(cache->hash_pool, sizeof(*key));
- *key = revision;
+ SVN_ERR(open_rev_or_pack_file(&file, fs, rev, pool));
+ SVN_ERR(svn_io_file_info_get(&finfo, APR_FINFO_SIZE, file, pool));
+ SVN_ERR(svn_io_file_close(file, pool));
- apr_hash_set(cache->hash, key, sizeof(*key), content);
- ++cache->insert_count;
+ *file_size = finfo.size;
+ return SVN_NO_ERROR;
}
/* Get the file content of revision REVISION in FS and return it in *DATA.
* Use SCRATCH_POOL for temporary allocations.
*/
static svn_error_t *
-get_content(svn_string_t **data,
+get_content(svn_stringbuf_t **content,
fs_fs_t *fs,
svn_revnum_t revision,
- apr_pool_t *scratch_pool)
+ apr_off_t offset,
+ apr_size_t len,
+ apr_pool_t *pool)
{
apr_file_t *file;
- revision_info_t *revision_info;
- svn_stringbuf_t *temp;
- apr_off_t temp_offset;
-
- /* try getting the data from our cache */
- svn_string_t *result = get_cached_content(fs->cache, revision);
- if (result)
- {
- *data = result;
- return SVN_NO_ERROR;
- }
-
- /* not in cache. Is the revision valid at all? */
- if (revision - fs->start_revision > fs->revisions->nelts)
- return svn_error_createf(SVN_ERR_BAD_VERSION_FILE_FORMAT, NULL,
- _("Unknown revision %ld"), revision);
- revision_info = APR_ARRAY_IDX(fs->revisions,
- revision - fs->start_revision,
- revision_info_t*);
-
- /* read the revision content. Assume that the file has *not* been
- * reorg'ed, yet, i.e. all data is in one place. */
- temp = svn_stringbuf_create_ensure( revision_info->original.end
- - revision_info->original.offset,
- scratch_pool);
- temp->len = revision_info->original.end - revision_info->original.offset;
- SVN_ERR(open_rev_or_pack_file(&file, fs, revision, scratch_pool));
-
- temp_offset = revision_info->original.offset;
- SVN_ERR(svn_io_file_seek(file, APR_SET, &temp_offset,
- scratch_pool));
- SVN_ERR_ASSERT(temp_offset < APR_SIZE_MAX);
- revision_info->original.offset = (apr_size_t)temp_offset;
- SVN_ERR(svn_io_file_read(file, temp->data, &temp->len, scratch_pool));
-
- /* cache the result and return it */
- set_cached_content(fs->cache, revision,
- svn_stringbuf__morph_into_string(temp));
- *data = get_cached_content(fs->cache, revision);
-
- return SVN_NO_ERROR;
-}
-
-/* Return a new directory cache with ENTRY_COUNT buckets in its index.
- * Use POOL for all cache-related allocations.
- */
-static dir_cache_t *
-create_dir_cache(apr_pool_t *pool,
- apr_size_t entry_count)
-{
- dir_cache_t *result = apr_pcalloc(pool, sizeof(*result));
-
- result->pool1 = svn_pool_create(pool);
- result->pool2 = svn_pool_create(pool);
- result->entry_count = entry_count;
- result->insert_count = 0;
- result->entries = apr_pcalloc(pool, sizeof(*result->entries) * entry_count);
-
- return result;
-}
-
-/* Return the position within FS' dir cache ENTRIES index for the given
- * (REVISION, OFFSET) pair. This is a cache-internal function.
- */
-static apr_size_t
-get_dir_cache_index(fs_fs_t *fs,
- svn_revnum_t revision,
- apr_size_t offset)
-{
- return (revision + offset * 0xd1f3da69) % fs->dir_cache->entry_count;
-}
-
-/* Return the currently active pool of FS' dir cache. Note that it may be
- * cleared after *2* insertions.
- */
-static apr_pool_t *
-get_cached_dir_pool(fs_fs_t *fs)
-{
- return fs->dir_cache->pool1;
-}
-
-/* Return the cached directory content stored in REPRESENTAION within FS.
- * If that has not been found in cache, return NULL.
- */
-static apr_hash_t *
-get_cached_dir(fs_fs_t *fs,
- representation_t *representation)
-{
- svn_revnum_t revision = representation->revision->revision;
- apr_size_t offset = representation->original.offset;
-
- apr_size_t i = get_dir_cache_index(fs, revision, offset);
- dir_cache_entry_t *entry = &fs->dir_cache->entries[i];
-
- return entry->offset == offset && entry->revision == revision
- ? entry->hash
- : NULL;
-}
-
-/* Cache the directory HASH for REPRESENTAION within FS.
- */
-static void
-set_cached_dir(fs_fs_t *fs,
- representation_t *representation,
- apr_hash_t *hash)
-{
- /* select the entry to use */
- svn_revnum_t revision = representation->revision->revision;
- apr_size_t offset = representation->original.offset;
+ apr_pool_t * file_pool = svn_pool_create(pool);
- apr_size_t i = get_dir_cache_index(fs, revision, offset);
- dir_cache_entry_t *entry = &fs->dir_cache->entries[i];
+ SVN_ERR(open_rev_or_pack_file(&file, fs, revision, file_pool));
+ SVN_ERR(svn_io_file_seek(file, APR_SET, &offset, pool));
- /* clean the cache and rotate pools at regular intervals */
- fs->dir_cache->insert_count += apr_hash_count(hash);
- if (fs->dir_cache->insert_count >= fs->dir_cache->entry_count * 100)
- {
- apr_pool_t *pool;
-
- svn_pool_clear(fs->dir_cache->pool2);
- memset(fs->dir_cache->entries,
- 0,
- sizeof(*fs->dir_cache->entries) * fs->dir_cache->entry_count);
- fs->dir_cache->insert_count = 0;
-
- pool = fs->dir_cache->pool2;
- fs->dir_cache->pool2 = fs->dir_cache->pool1;
- fs->dir_cache->pool1 = pool;
- }
+ *content = svn_stringbuf_create_ensure(len, pool);
+ (*content)->len = len;
+ SVN_ERR(svn_io_file_read_full2(file, (*content)->data, len,
+ NULL, NULL, pool));
+ svn_pool_destroy(file_pool);
- /* write data to cache */
- entry->hash = hash;
- entry->offset = offset;
- entry->revision = revision;
+ return SVN_NO_ERROR;
}
/* Return a new txdelta window cache with ENTRY_COUNT buckets in its index
@@ -731,8 +327,8 @@ get_cached_window(fs_fs_t *fs,
representation_t *representation,
apr_pool_t *pool)
{
- svn_revnum_t revision = representation->revision->revision;
- apr_size_t offset = representation->original.offset;
+ svn_revnum_t revision = representation->revision;
+ apr_size_t offset = representation->offset;
apr_size_t i = get_window_cache_index(fs, revision, offset);
window_cache_entry_t *entry = &fs->window_cache->entries[i];
@@ -750,8 +346,8 @@ set_cached_window(fs_fs_t *fs,
svn_stringbuf_t *window)
{
/* select entry */
- svn_revnum_t revision = representation->revision->revision;
- apr_size_t offset = representation->original.offset;
+ svn_revnum_t revision = representation->revision;
+ apr_size_t offset = representation->offset;
apr_size_t i = get_window_cache_index(fs, revision, offset);
window_cache_entry_t *entry = &fs->window_cache->entries[i];
@@ -822,8 +418,6 @@ read_revision_header(apr_size_t *changes
apr_size_t *changes_len,
apr_size_t *root_noderev,
svn_stringbuf_t *file_content,
- apr_size_t start,
- apr_size_t end,
apr_pool_t *pool)
{
char buf[64];
@@ -834,10 +428,10 @@ read_revision_header(apr_size_t *changes
/* Read in this last block, from which we will identify the last line. */
len = sizeof(buf);
- if (start + len > end)
- len = end - start;
+ if (len > file_content->len)
+ len = file_content->len;
- memcpy(buf, file_content->data + end - len, len);
+ memcpy(buf, file_content->data + file_content->len - len, len);
/* The last byte should be a newline. */
if (buf[(apr_ssize_t)len - 1] != '\n')
@@ -863,7 +457,7 @@ read_revision_header(apr_size_t *changes
*root_noderev = (apr_size_t)val;
SVN_ERR(svn_cstring_strtoui64(&val, space+1, 0, APR_SIZE_MAX, 10));
*changes = (apr_size_t)val;
- *changes_len = end - *changes - start - (buf + len - line) + 1;
+ *changes_len = file_content->len - *changes - (buf + len - line) + 1;
return SVN_NO_ERROR;
}
@@ -996,90 +590,9 @@ key_matches(svn_string_t *string, const
}
static int
-compare_noderev_offsets(const void *data, const void *key)
-{
- apr_ssize_t diff = (*(const noderev_t **)data)->original.offset
- - *(const apr_size_t *)key;
-
- /* sizeof(int) may be < sizeof(ssize_t) */
- if (diff < 0)
- return -1;
- return diff > 0 ? 1 : 0;
-}
-
-static svn_error_t *
-parse_revnode_pos(revision_info_t **revision_info,
- apr_size_t *offset,
- fs_fs_t *fs,
- svn_string_t *id)
-{
- int revision;
- apr_uint64_t temp;
-
- const char *revision_pos = strrchr(id->data, 'r');
- char *offset_pos = (char *)strchr(id->data, '/');
-
- if (revision_pos == NULL || offset_pos == NULL)
- return svn_error_createf(SVN_ERR_BAD_VERSION_FILE_FORMAT, NULL,
- _("Invalid node id '%s'"), id->data);
-
- *offset_pos = 0;
- SVN_ERR(svn_cstring_atoi(&revision, revision_pos + 1));
- SVN_ERR(svn_cstring_strtoui64(&temp, offset_pos + 1, 0, APR_SIZE_MAX, 10));
- *offset = (apr_size_t)temp;
- *offset_pos = '/';
-
- if (revision - fs->start_revision > fs->revisions->nelts)
- return svn_error_createf(SVN_ERR_BAD_VERSION_FILE_FORMAT, NULL,
- _("Unknown revision %d"), revision);
-
- *revision_info = APR_ARRAY_IDX(fs->revisions,
- revision - fs->start_revision,
- revision_info_t*);
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-find_noderev(noderev_t **result,
- revision_info_t *revision_info,
- apr_size_t offset)
-{
- int idx = svn_sort__bsearch_lower_bound(&offset,
- revision_info->node_revs,
- compare_noderev_offsets);
- if ((idx < 0) || (idx >= revision_info->node_revs->nelts))
- return svn_error_createf(SVN_ERR_BAD_VERSION_FILE_FORMAT, NULL,
- _("No noderev found at offset %" APR_OFF_T_FMT),
- offset);
-
- *result = APR_ARRAY_IDX(revision_info->node_revs, idx, noderev_t *);
- if ((*result)->original.offset != offset)
- return svn_error_createf(SVN_ERR_BAD_VERSION_FILE_FORMAT, NULL,
- _("No noderev found at offset %" APR_OFF_T_FMT),
- offset);
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-parse_pred(noderev_t **result,
- fs_fs_t *fs,
- svn_string_t *id)
-{
- apr_size_t offset;
- revision_info_t *revision_info;
-
- SVN_ERR(parse_revnode_pos(&revision_info, &offset, fs, id));
- SVN_ERR(find_noderev(result, revision_info, offset));
-
- return SVN_NO_ERROR;
-}
-
-static int
compare_representation_offsets(const void *data, const void *key)
{
- apr_ssize_t diff = (*(const representation_t **)data)->original.offset
+ apr_ssize_t diff = (*(const representation_t **)data)->offset
- *(const apr_size_t *)key;
/* sizeof(int) may be < sizeof(ssize_t) */
@@ -1118,7 +631,7 @@ find_representation(int *idx,
{
representation_t *result
= APR_ARRAY_IDX(info->representations, *idx, representation_t *);
- if (result->original.offset == offset)
+ if (result->offset == offset)
return result;
}
@@ -1187,25 +700,33 @@ parse_representation(representation_t **
apr_uint64_t offset;
apr_uint64_t size;
+ apr_uint64_t expanded_size;
int idx;
char *c = (char *)value->data;
SVN_ERR(svn_cstring_atoi(&revision, svn_cstring_tokenize(" ", &c)));
SVN_ERR(svn_cstring_strtoui64(&offset, svn_cstring_tokenize(" ", &c), 0, APR_SIZE_MAX, 10));
SVN_ERR(svn_cstring_strtoui64(&size, svn_cstring_tokenize(" ", &c), 0, APR_SIZE_MAX, 10));
+ SVN_ERR(svn_cstring_strtoui64(&expanded_size, svn_cstring_tokenize(" ", &c), 0, APR_SIZE_MAX, 10));
result = find_representation(&idx, fs, &revision_info, revision, (apr_size_t)offset);
if (!result)
{
+ apr_size_t header_size;
+ svn_boolean_t is_plain;
+
result = apr_pcalloc(pool, sizeof(*result));
- result->revision = revision_info;
- result->original.offset = (apr_size_t)offset;
- result->original.size = (apr_size_t)size;
- SVN_ERR(read_rep_base(&result->delta_base, &result->header_size,
- &result->is_plain, fs, file_content,
- (apr_size_t)offset + revision_info->original.offset,
+ result->revision = revision;
+ result->expanded_size = (apr_size_t)(expanded_size ? expanded_size : size);
+ result->offset = (apr_size_t)offset;
+ result->size = (apr_size_t)size;
+ SVN_ERR(read_rep_base(&result->delta_base, &header_size,
+ &is_plain, fs, file_content,
+ (apr_size_t)offset,
pool, scratch_pool));
+ result->header_size = header_size;
+ result->is_plain = is_plain;
svn_sort__array_insert(&result, revision_info->representations, idx);
}
@@ -1214,29 +735,68 @@ parse_representation(representation_t **
return SVN_NO_ERROR;
}
+/* Get the file content of revision REVISION in FS and return it in *DATA.
+ * Use SCRATCH_POOL for temporary allocations.
+ */
+static svn_error_t *
+get_rep_content(svn_stringbuf_t **content,
+ fs_fs_t *fs,
+ representation_t *representation,
+ svn_stringbuf_t *file_content,
+ apr_pool_t *pool)
+{
+ apr_off_t offset;
+ svn_revnum_t revision = representation->revision;
+ revision_info_t *revision_info = APR_ARRAY_IDX(fs->revisions,
+ revision - fs->start_revision,
+ revision_info_t*);
+
+ /* not in cache. Is the revision valid at all? */
+ if (revision - fs->start_revision > fs->revisions->nelts)
+ return svn_error_createf(SVN_ERR_FS_CORRUPT, NULL,
+ _("Unknown revision %ld"), revision);
+
+ if (file_content)
+ {
+ offset = representation->offset
+ + representation->header_size;
+ *content = svn_stringbuf_ncreate(file_content->data + offset,
+ representation->size, pool);
+ }
+ else
+ {
+ offset = revision_info->offset
+ + representation->offset
+ + representation->header_size;
+ SVN_ERR(get_content(content, fs, revision, offset,
+ representation->size, pool));
+ }
+
+ return SVN_NO_ERROR;
+}
+
+
/* Skip forwards to THIS_CHUNK in REP_STATE and then read the next delta
window into *NWIN. */
static svn_error_t *
read_windows(apr_array_header_t **windows,
fs_fs_t *fs,
representation_t *representation,
+ svn_stringbuf_t *file_content,
apr_pool_t *pool)
{
- svn_string_t *content;
- svn_string_t data;
+ svn_stringbuf_t *content;
svn_stream_t *stream;
- apr_size_t offset = representation->original.offset
- + representation->header_size;
char version;
apr_size_t len = sizeof(version);
*windows = apr_array_make(pool, 0, sizeof(svn_txdelta_window_t *));
- SVN_ERR(get_content(&content, fs, representation->revision->revision, pool));
+ SVN_ERR(get_rep_content(&content, fs, representation, file_content, pool));
- data.data = content->data + offset + 3;
- data.len = representation->original.size - 3;
- stream = svn_stream_from_string(&data, pool);
+ content->data += 3;
+ content->len -= 3;
+ stream = svn_stream_from_stringbuf(content, pool);
SVN_ERR(svn_stream_read(stream, &version, &len));
while (TRUE)
@@ -1259,25 +819,6 @@ read_windows(apr_array_header_t **window
return SVN_NO_ERROR;
}
-static svn_error_t *
-read_plain(svn_stringbuf_t **content,
- fs_fs_t *fs,
- representation_t *representation,
- apr_pool_t *pool)
-{
- svn_string_t *data;
- apr_size_t offset = representation->original.offset
- + representation->header_size;
-
- SVN_ERR(get_content(&data, fs, representation->revision->revision, pool));
-
- *content = svn_stringbuf_ncreate(data->data + offset,
- representation->original.size,
- pool);
-
- return SVN_NO_ERROR;
-}
-
/* Get the undeltified window that is a result of combining all deltas
from the current desired representation identified in *RB with its
base representation. Store the window in *RESULT. */
@@ -1285,6 +826,7 @@ static svn_error_t *
get_combined_window(svn_stringbuf_t **content,
fs_fs_t *fs,
representation_t *representation,
+ svn_stringbuf_t *file_content,
apr_pool_t *pool)
{
int i;
@@ -1295,16 +837,16 @@ get_combined_window(svn_stringbuf_t **co
apr_pool_t *iter_pool = svn_pool_create(pool);
if (representation->is_plain)
- return read_plain(content, fs, representation, pool);
+ return get_rep_content(content, fs, representation, file_content, pool);
*content = get_cached_window(fs, representation, pool);
if (*content)
return SVN_NO_ERROR;
- SVN_ERR(read_windows(&windows, fs, representation, sub_pool));
+ SVN_ERR(read_windows(&windows, fs, representation, file_content, sub_pool));
if (representation->delta_base && representation->delta_base->revision)
SVN_ERR(get_combined_window(&base_content, fs,
- representation->delta_base, sub_pool));
+ representation->delta_base, NULL, sub_pool));
else
base_content = svn_stringbuf_create_empty(sub_pool);
@@ -1337,8 +879,7 @@ get_combined_window(svn_stringbuf_t **co
}
static svn_error_t *
-read_noderev(noderev_t **noderev,
- fs_fs_t *fs,
+read_noderev(fs_fs_t *fs,
svn_stringbuf_t *file_content,
apr_size_t offset,
revision_info_t *revision_info,
@@ -1346,157 +887,93 @@ read_noderev(noderev_t **noderev,
apr_pool_t *scratch_pool);
static svn_error_t *
-get_noderev(noderev_t **noderev,
- fs_fs_t *fs,
- svn_stringbuf_t *file_content,
- apr_size_t offset,
- revision_info_t *revision_info,
- apr_pool_t *pool,
- apr_pool_t *scratch_pool)
-{
- int idx = svn_sort__bsearch_lower_bound(&offset,
- revision_info->node_revs,
- compare_noderev_offsets);
- if ((idx < 0) || (idx >= revision_info->node_revs->nelts))
- SVN_ERR(read_noderev(noderev, fs, file_content, offset, revision_info,
- pool, scratch_pool));
- else
- {
- *noderev = APR_ARRAY_IDX(revision_info->node_revs, idx, noderev_t *);
- if ((*noderev)->original.offset != offset)
- SVN_ERR(read_noderev(noderev, fs, file_content, offset, revision_info,
- pool, scratch_pool));
- }
+parse_dir(fs_fs_t *fs,
+ svn_stringbuf_t *file_content,
+ representation_t *representation,
+ revision_info_t *revision_info,
+ apr_pool_t *pool,
+ apr_pool_t *scratch_pool)
+{
+ svn_stringbuf_t *text;
+ apr_pool_t *iter_pool;
+ apr_pool_t *text_pool;
+ const char *current;
+ const char *revision_key;
+ apr_size_t key_len;
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-read_dir(apr_hash_t **hash,
- fs_fs_t *fs,
- representation_t *representation,
- apr_pool_t *scratch_pool)
-{
- svn_stringbuf_t *text;
- apr_pool_t *text_pool;
- svn_stream_t *stream;
- apr_pool_t *pool;
-
- *hash = get_cached_dir(fs, representation);
- if (*hash)
+ if (representation == NULL)
return SVN_NO_ERROR;
- pool = get_cached_dir_pool(fs);
- *hash = svn_hash__make(pool);
- if (representation != NULL)
- {
- text_pool = svn_pool_create(scratch_pool);
- SVN_ERR(get_combined_window(&text, fs, representation, text_pool));
- stream = svn_stream_from_stringbuf(text, text_pool);
- SVN_ERR(svn_hash_read2(*hash, stream, SVN_HASH_TERMINATOR, pool));
- svn_pool_destroy(text_pool);
- }
+ iter_pool = svn_pool_create(scratch_pool);
+ text_pool = svn_pool_create(scratch_pool);
- set_cached_dir(fs, representation, *hash);
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-parse_dir(fs_fs_t *fs,
- svn_stringbuf_t *file_content,
- representation_t *representation,
- apr_pool_t *pool,
- apr_pool_t *scratch_pool)
-{
- apr_hash_t *hash;
- apr_hash_index_t *hi;
- apr_pool_t *iter_pool = svn_pool_create(scratch_pool);
- apr_hash_t *base_dir = svn_hash__make(scratch_pool);
-
- if (representation == NULL)
- return SVN_NO_ERROR;
+ SVN_ERR(get_combined_window(&text, fs, representation, file_content,
+ text_pool));
+ current = text->data;
- if (representation->delta_base && representation->delta_base->dir)
+ revision_key = apr_psprintf(text_pool, "r%d/", representation->revision);
+ key_len = strlen(revision_key);
+
+ /* Translate the string dir entries into real entries. */
+ while (*current != 'E')
{
- apr_array_header_t *dir = representation->delta_base->dir->entries;
- int i;
+ char *next;
- for (i = 0; i < dir->nelts; ++i)
+ current = strchr(current, '\n');
+ if (current)
+ current = strchr(current+1, '\n');
+ if (current)
+ current = strchr(current+1, '\n');
+ next = current ? strchr(++current, '\n') : NULL;
+ if (next == NULL)
+ return svn_error_createf(SVN_ERR_FS_CORRUPT, NULL,
+ _("Corrupt directory representation in rev %d at offset %ld"),
+ representation->revision,
+ representation->offset);
+
+ *next = 0;
+ current = strstr(current, revision_key);
+ if (current)
{
- direntry_t *entry = APR_ARRAY_IDX(dir, i, direntry_t *);
- apr_hash_set(base_dir, entry->name, entry->name_len, entry);
- }
- }
+ apr_uint64_t offset;
- SVN_ERR(read_dir(&hash, fs, representation, scratch_pool));
+ SVN_ERR(svn_cstring_strtoui64(&offset, current + key_len, 0,
+ APR_SIZE_MAX, 10));
+ SVN_ERR(read_noderev(fs, file_content, (apr_size_t)offset,
+ revision_info, pool, iter_pool));
- representation->dir = apr_pcalloc(pool, sizeof(*representation->dir));
- representation->dir->entries
- = apr_array_make(pool, apr_hash_count(hash), sizeof(direntry_t *));
-
- /* Translate the string dir entries into real entries. */
- for (hi = apr_hash_first(pool, hash); hi; hi = apr_hash_next(hi))
- {
- const char *name = svn__apr_hash_index_key(hi);
- svn_string_t *str_val = svn__apr_hash_index_val(hi);
- apr_size_t offset;
- revision_info_t *revision_info;
- apr_size_t name_len = strlen(name);
- direntry_t *entry = base_dir
- ? apr_hash_get(base_dir, name, name_len)
- : NULL;
-
- SVN_ERR(parse_revnode_pos(&revision_info, &offset, fs, str_val));
-
- if ( !entry
- || !entry->node->text
- || entry->node->text->revision != revision_info
- || entry->node->original.offset != offset)
- {
- direntry_t *new_entry = apr_pcalloc(pool, sizeof(*entry));
- new_entry->name_len = name_len;
- if (entry)
- new_entry->name = entry->name;
- else
- new_entry->name = apr_pstrdup(pool, name);
-
- entry = new_entry;
- SVN_ERR(get_noderev(&entry->node, fs, file_content, offset,
- revision_info, pool, iter_pool));
+ svn_pool_clear(iter_pool);
}
-
- APR_ARRAY_PUSH(representation->dir->entries, direntry_t *) = entry;
- svn_pool_clear(iter_pool);
+ current = next+1;
}
svn_pool_destroy(iter_pool);
+ svn_pool_destroy(text_pool);
return SVN_NO_ERROR;
}
static svn_error_t *
-read_noderev(noderev_t **noderev,
- fs_fs_t *fs,
+read_noderev(fs_fs_t *fs,
svn_stringbuf_t *file_content,
apr_size_t offset,
revision_info_t *revision_info,
apr_pool_t *pool,
apr_pool_t *scratch_pool)
{
- noderev_t *result = apr_pcalloc(pool, sizeof(*result));
svn_string_t *line;
+ representation_t *text = NULL;
+ representation_t *props = NULL;
+ apr_size_t start_offset = offset;
svn_boolean_t is_dir = FALSE;
scratch_pool = svn_pool_create(scratch_pool);
- result->original.offset = offset;
while (1)
{
svn_string_t key;
svn_string_t value;
char *sep;
- const char *start = file_content->data + offset
- + revision_info->original.offset;
+ const char *start = file_content->data + offset;
const char *end = strchr(start, '\n');
line = svn_string_ncreate(start, end - start, scratch_pool);
@@ -1520,37 +997,57 @@ read_noderev(noderev_t **noderev,
if (key_matches(&key, "type"))
is_dir = strcmp(value.data, "dir") == 0;
- else if (key_matches(&key, "pred"))
- SVN_ERR(parse_pred(&result->predecessor, fs, &value));
else if (key_matches(&key, "text"))
- SVN_ERR(parse_representation(&result->text, fs, file_content,
- &value, revision_info,
- pool, scratch_pool));
+ {
+ SVN_ERR(parse_representation(&text, fs, file_content,
+ &value, revision_info,
+ pool, scratch_pool));
+ if (++text->ref_count == 1)
+ text->kind = is_dir ? dir_rep : file_rep;
+ }
else if (key_matches(&key, "props"))
- SVN_ERR(parse_representation(&result->props, fs, file_content,
- &value, revision_info,
- pool, scratch_pool));
+ {
+ SVN_ERR(parse_representation(&props, fs, file_content,
+ &value, revision_info,
+ pool, scratch_pool));
+ if (++props->ref_count == 1)
+ props->kind = is_dir ? dir_property_rep : file_property_rep;
+ }
}
- result->revision = revision_info;
- result->original.size = offset - result->original.offset;
-
- svn_sort__array_insert(&result,
- revision_info->node_revs,
- svn_sort__bsearch_lower_bound(&offset,
- revision_info->node_revs,
- compare_noderev_offsets));
-
- if (is_dir)
- SVN_ERR(parse_dir(fs, file_content, result->text,
+ if (is_dir && text && text->ref_count == 1)
+ SVN_ERR(parse_dir(fs, file_content, text, revision_info,
pool, scratch_pool));
+ if (is_dir)
+ {
+ revision_info->dir_noderev_size += offset - start_offset;
+ revision_info->dir_noderev_count++;
+ }
+ else
+ {
+ revision_info->file_noderev_size += offset - start_offset;
+ revision_info->file_noderev_count++;
+ }
svn_pool_destroy(scratch_pool);
- *noderev = result;
return SVN_NO_ERROR;
}
+static apr_size_t
+get_change_count(const char *changes,
+ apr_size_t len)
+{
+ apr_size_t lines = 0;
+ const char *end = changes + len;
+
+ for (; changes < end; ++changes)
+ if (*changes == '\n')
+ ++lines;
+
+ return lines / 2;
+}
+
static void print_progress(svn_revnum_t revision)
{
printf("%8ld", revision);
@@ -1566,58 +1063,48 @@ read_pack_file(fs_fs_t *fs,
apr_pool_t *local_pool = svn_pool_create(pool);
apr_pool_t *iter_pool = svn_pool_create(local_pool);
int i;
- svn_stringbuf_t *file_content;
- revision_pack_t *revisions;
+ apr_off_t file_size = 0;
const char *pack_folder = get_pack_folder(fs, base, local_pool);
- SVN_ERR(read_rev_or_pack_file(&file_content, fs, base, local_pool));
-
- revisions = apr_pcalloc(pool, sizeof(*revisions));
- revisions->base = base;
- revisions->fragments = NULL;
- revisions->info = apr_array_make(pool,
- fs->max_files_per_dir,
- sizeof(revision_info_t*));
- revisions->filesize = file_content->len;
- APR_ARRAY_PUSH(fs->packs, revision_pack_t*) = revisions;
SVN_ERR(read_manifest(&manifest, fs, pack_folder, local_pool));
if (manifest->nelts != fs->max_files_per_dir)
return svn_error_create(SVN_ERR_FS_CORRUPT, NULL, NULL);
+ SVN_ERR(rev_or_pack_file_size(&file_size, fs, base, pool));
+
for (i = 0; i < manifest->nelts; ++i)
{
apr_size_t root_node_offset;
- svn_string_t rev_content;
+ svn_stringbuf_t *rev_content;
revision_info_t *info = apr_pcalloc(pool, sizeof(*info));
- info->node_revs = apr_array_make(iter_pool, 4, sizeof(noderev_t*));
info->representations = apr_array_make(iter_pool, 4, sizeof(representation_t*));
info->revision = base + i;
- info->original.offset = APR_ARRAY_IDX(manifest, i, apr_size_t);
- info->original.end = i+1 < manifest->nelts
+ info->offset = APR_ARRAY_IDX(manifest, i, apr_size_t);
+ info->end = i+1 < manifest->nelts
? APR_ARRAY_IDX(manifest, i+1 , apr_size_t)
- : file_content->len;
- SVN_ERR(read_revision_header(&info->original.changes,
- &info->original.changes_len,
+ : file_size;
+
+ SVN_ERR(get_content(&rev_content, fs, info->revision,
+ info->offset,
+ info->end - info->offset,
+ iter_pool));
+
+ SVN_ERR(read_revision_header(&info->changes,
+ &info->changes_len,
&root_node_offset,
- file_content,
- APR_ARRAY_IDX(manifest, i , apr_size_t),
- info->original.end,
+ rev_content,
iter_pool));
-
- APR_ARRAY_PUSH(revisions->info, revision_info_t*) = info;
- APR_ARRAY_PUSH(fs->revisions, revision_info_t*) = info;
- rev_content.data = file_content->data + info->original.offset;
- rev_content.len = info->original.end - info->original.offset;
- set_cached_content(fs->cache, info->revision, &rev_content);
-
- SVN_ERR(read_noderev(&info->root_noderev, fs, file_content,
+ info->change_count
+ = get_change_count(rev_content->data + info->changes,
+ info->changes_len);
+ SVN_ERR(read_noderev(fs, rev_content,
root_node_offset, info, pool, iter_pool));
- info->node_revs = apr_array_copy(pool, info->node_revs);
info->representations = apr_array_copy(pool, info->representations);
+ APR_ARRAY_PUSH(fs->revisions, revision_info_t*) = info;
svn_pool_clear(iter_pool);
}
@@ -1635,44 +1122,35 @@ read_revision_file(fs_fs_t *fs,
{
apr_size_t root_node_offset;
apr_pool_t *local_pool = svn_pool_create(pool);
- svn_stringbuf_t *file_content;
- svn_string_t rev_content;
- revision_pack_t *revisions = apr_pcalloc(pool, sizeof(*revisions));
+ svn_stringbuf_t *rev_content;
revision_info_t *info = apr_pcalloc(pool, sizeof(*info));
+ apr_off_t file_size = 0;
- SVN_ERR(read_rev_or_pack_file(&file_content, fs, revision, local_pool));
+ SVN_ERR(rev_or_pack_file_size(&file_size, fs, revision, pool));
- info->node_revs = apr_array_make(pool, 4, sizeof(noderev_t*));
info->representations = apr_array_make(pool, 4, sizeof(representation_t*));
info->revision = revision;
- info->original.offset = 0;
- info->original.end = file_content->len;
- SVN_ERR(read_revision_header(&info->original.changes,
- &info->original.changes_len,
+ info->offset = 0;
+ info->end = file_size;
+
+ SVN_ERR(get_content(&rev_content, fs, revision, 0, file_size, local_pool));
+
+ SVN_ERR(read_revision_header(&info->changes,
+ &info->changes_len,
&root_node_offset,
- file_content,
- 0,
- info->original.end,
+ rev_content,
local_pool));
APR_ARRAY_PUSH(fs->revisions, revision_info_t*) = info;
- revisions->base = revision;
- revisions->fragments = NULL;
- revisions->info = apr_array_make(pool, 1, sizeof(revision_info_t*));
- revisions->filesize = file_content->len;
- APR_ARRAY_PUSH(revisions->info, revision_info_t*) = info;
- APR_ARRAY_PUSH(fs->packs, revision_pack_t*) = revisions;
-
- rev_content.data = file_content->data + info->original.offset;
- rev_content.len = info->original.end - info->original.offset;
- set_cached_content(fs->cache, info->revision, &rev_content);
+ info->change_count
+ = get_change_count(rev_content->data + info->changes,
+ info->changes_len);
- SVN_ERR(read_noderev(&info->root_noderev, fs, file_content,
+ SVN_ERR(read_noderev(fs, rev_content,
root_node_offset, info,
pool, local_pool));
- APR_ARRAY_PUSH(info->node_revs, noderev_t*) = info->root_noderev;
if (revision % fs->max_files_per_dir == 0)
print_progress(revision);
@@ -1690,18 +1168,14 @@ read_revisions(fs_fs_t **fs,
apr_pool_t *pool)
{
svn_revnum_t revision;
- apr_size_t content_cache_size;
apr_size_t window_cache_size;
- apr_size_t dir_cache_size;
/* determine cache sizes */
if (memsize < 100)
memsize = 100;
- content_cache_size = memsize * 7 / 10 > 4000 ? 4000 : memsize * 7 / 10;
- window_cache_size = memsize * 2 / 10 * 1024 * 1024;
- dir_cache_size = (memsize / 10) * 16000;
+ window_cache_size = memsize * 1024 * 1024;
SVN_ERR(fs_open(fs, path, pool));
@@ -1710,19 +1184,7 @@ read_revisions(fs_fs_t **fs,
(*fs)->revisions = apr_array_make(pool,
(*fs)->max_revision + 1 - (*fs)->start_revision,
sizeof(revision_info_t *));
- (*fs)->packs = apr_array_make(pool,
- ((*fs)->min_unpacked_rev - (*fs)->start_revision)
- / (*fs)->max_files_per_dir,
- sizeof(revision_pack_t *));
(*fs)->null_base = apr_pcalloc(pool, sizeof(*(*fs)->null_base));
- (*fs)->cache = create_content_cache
- (apr_allocator_owner_get
- (svn_pool_create_allocator(FALSE)),
- content_cache_size * 1024 * 1024);
- (*fs)->dir_cache = create_dir_cache
- (apr_allocator_owner_get
- (svn_pool_create_allocator(FALSE)),
- dir_cache_size);
(*fs)->window_cache = create_window_cache
(apr_allocator_owner_get
(svn_pool_create_allocator(FALSE)),
@@ -1739,966 +1201,194 @@ read_revisions(fs_fs_t **fs,
return SVN_NO_ERROR;
}
-static apr_size_t
-get_max_offset_len(const revision_pack_t *pack)
+typedef struct rep_pack_stats_t
{
- apr_size_t max_future_size = pack->filesize * 2 + 10000;
- apr_size_t result = 0;
-
- while (max_future_size > 0)
- {
- ++result;
- max_future_size /= 10;
- }
+ apr_int64_t count;
+ apr_int64_t packed_size;
+ apr_int64_t expanded_size;
+ apr_int64_t overhead_size;
+} rep_pack_stats_t;
- return result;
-}
-
-static svn_error_t *
-add_revisions_pack_heads(revision_pack_t *pack,
- apr_pool_t *pool)
+typedef struct representation_stats_t
{
- int i;
- revision_info_t *info;
- apr_size_t offset_len = get_max_offset_len(pack);
- fragment_t fragment;
-
- /* allocate fragment arrays */
-
- int fragment_count = 1;
- for (i = 0; i < pack->info->nelts; ++i)
- {
- info = APR_ARRAY_IDX(pack->info, i, revision_info_t*);
- fragment_count += info->node_revs->nelts
- + info->representations->nelts
- + 2;
- }
-
- pack->target_offset = pack->info->nelts > 1 ? 64 : 0;
- pack->fragments = apr_array_make(pool,
- fragment_count,
- sizeof(fragment_t));
-
- /* put revision headers first */
-
- for (i = 0; i < pack->info->nelts - 1; ++i)
- {
- info = APR_ARRAY_IDX(pack->info, i, revision_info_t*);
- info->target.offset = pack->target_offset;
-
- fragment.data = info;
- fragment.kind = header_fragment;
- fragment.position = pack->target_offset;
- APR_ARRAY_PUSH(pack->fragments, fragment_t) = fragment;
-
- pack->target_offset += 2 * offset_len + 3;
- }
-
- info = APR_ARRAY_IDX(pack->info, pack->info->nelts - 1, revision_info_t*);
- info->target.offset = pack->target_offset;
+ rep_pack_stats_t total;
+ rep_pack_stats_t uniques;
+ rep_pack_stats_t shared;
- /* followed by the changes list */
-
- for (i = 0; i < pack->info->nelts; ++i)
- {
- info = APR_ARRAY_IDX(pack->info, i, revision_info_t*);
-
- info->target.changes = pack->target_offset - info->target.offset;
- info->target.changes_len = info->original.changes_len;
-
- fragment.data = info;
- fragment.kind = changes_fragment;
- fragment.position = pack->target_offset;
- APR_ARRAY_PUSH(pack->fragments, fragment_t) = fragment;
-
- pack->target_offset += info->original.changes_len;
- }
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-get_target_offset(apr_size_t **current_pos,
- apr_array_header_t **fragments,
- fs_fs_t *fs,
- revision_info_t *info)
-{
- int i;
- revision_pack_t *pack;
- svn_revnum_t revision = info->revision;
-
- if (fs->min_unpacked_rev > revision)
- {
- i = (revision - fs->start_revision) / fs->max_files_per_dir;
- }
- else
- {
- i = (fs->min_unpacked_rev - fs->start_revision) / fs->max_files_per_dir;
- i += revision - fs->min_unpacked_rev;
- }
-
- pack = APR_ARRAY_IDX(fs->packs, i, revision_pack_t*);
- *current_pos = &pack->target_offset;
- *fragments = pack->fragments;
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-add_noderev_recursively(fs_fs_t *fs,
- noderev_t *node,
- apr_pool_t *pool);
-
-static svn_error_t *
-add_representation_recursively(fs_fs_t *fs,
- representation_t *representation,
- enum fragment_kind_t kind,
- apr_pool_t *pool)
-{
- apr_size_t *current_pos;
- apr_array_header_t *fragments;
- fragment_t fragment;
-
- if ( representation == NULL
- || representation->covered
- || (representation->dir && kind != dir_fragment)
- || representation == fs->null_base)
- return SVN_NO_ERROR;
-
- SVN_ERR(get_target_offset(¤t_pos, &fragments,
- fs, representation->revision));
- representation->target.offset = *current_pos;
- representation->covered = TRUE;
-
- fragment.data = representation;
- fragment.kind = kind;
- fragment.position = *current_pos;
- APR_ARRAY_PUSH(fragments, fragment_t) = fragment;
-
- if ( kind != dir_fragment
- && representation->delta_base && representation->delta_base->dir)
- {
- apr_pool_t *text_pool = svn_pool_create(pool);
- svn_stringbuf_t *content;
-
- get_combined_window(&content, fs, representation, text_pool);
- representation->target.size = content->len;
- *current_pos += representation->target.size + 13;
-
- svn_pool_destroy(text_pool);
- }
- else
- if ( kind == dir_fragment
- || (representation->delta_base && representation->delta_base->dir))
- {
- if (representation->original.size < 50)
- *current_pos += 300;
- else
- *current_pos += representation->original.size * 3 + 150;
- }
- else
- {
- representation->target.size = representation->original.size;
-
- if (representation->delta_base &&
- (representation->delta_base != fs->null_base))
- *current_pos += representation->original.size + 50;
- else
- *current_pos += representation->original.size + 13;
- }
-
- if (representation->delta_base)
- SVN_ERR(add_representation_recursively(fs,
- representation->delta_base,
- kind,
- pool));
-
- if (representation->dir)
- {
- int i;
- apr_array_header_t *entries = representation->dir->entries;
-
- for (i = 0; i < entries->nelts; ++i)
- {
- direntry_t *entry = APR_ARRAY_IDX(entries, i, direntry_t *);
- if (entry->node)
- SVN_ERR(add_noderev_recursively(fs, entry->node, pool));
- }
- }
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-add_noderev_recursively(fs_fs_t *fs,
- noderev_t *node,
- apr_pool_t *pool)
-{
- apr_size_t *current_pos;
- apr_array_header_t *fragments;
- fragment_t fragment;
-
- if (node->covered)
- return SVN_NO_ERROR;
-
- SVN_ERR(get_target_offset(¤t_pos, &fragments, fs, node->revision));
- node->covered = TRUE;
- node->target.offset = *current_pos;
-
- fragment.data = node;
- fragment.kind = noderev_fragment;
- fragment.position = *current_pos;
- APR_ARRAY_PUSH(fragments, fragment_t) = fragment;
-
- *current_pos += node->original.size + 40;
-
- if (node->text && node->text->dir)
- SVN_ERR(add_representation_recursively(fs, node->text, dir_fragment, pool));
- else
- SVN_ERR(add_representation_recursively(fs, node->text, file_fragment, pool));
-
- SVN_ERR(add_representation_recursively(fs, node->props, property_fragment, pool));
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-add_revisions_pack_tail(revision_pack_t *pack,
- apr_pool_t *pool)
-{
- int i;
- revision_info_t *info;
- apr_size_t offset_len = get_max_offset_len(pack);
- fragment_t fragment;
-
- /* put final revision header last and fix up revision lengths */
-
- info = APR_ARRAY_IDX(pack->info, pack->info->nelts-1, revision_info_t*);
-
- fragment.data = info;
- fragment.kind = header_fragment;
- fragment.position = pack->target_offset;
- APR_ARRAY_PUSH(pack->fragments, fragment_t) = fragment;
-
- pack->target_offset += 2 * offset_len + 3;
-
- for (i = 0; i < pack->info->nelts; ++i)
- {
- info = APR_ARRAY_IDX(pack->info, i, revision_info_t*);
- info->target.end = pack->target_offset;
- }
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-reorder_revisions(fs_fs_t *fs,
- apr_pool_t *pool)
-{
- int i, k;
-
- /* headers and changes */
-
- for (i = 0; i < fs->packs->nelts; ++i)
- {
- revision_pack_t *pack = APR_ARRAY_IDX(fs->packs, i, revision_pack_t*);
- SVN_ERR(add_revisions_pack_heads(pack, pool));
- }
-
- /* representations & nodes */
-
- for (i = fs->revisions->nelts-1; i >= 0; --i)
- {
- revision_info_t *info = APR_ARRAY_IDX(fs->revisions, i, revision_info_t*);
- for (k = info->node_revs->nelts - 1; k >= 0; --k)
- {
- noderev_t *node = APR_ARRAY_IDX(info->node_revs, k, noderev_t*);
- SVN_ERR(add_noderev_recursively(fs, node, pool));
- }
-
- if (info->revision % fs->max_files_per_dir == 0)
- print_progress(info->revision);
- }
-
- /* pack file tails */
-
- for (i = 0; i < fs->packs->nelts; ++i)
- {
- revision_pack_t *pack = APR_ARRAY_IDX(fs->packs, i, revision_pack_t*);
- SVN_ERR(add_revisions_pack_tail(pack, pool));
- }
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-get_fragment_content(svn_string_t **content,
- fs_fs_t *fs,
- fragment_t *fragment,
- apr_pool_t *pool);
-
-static svn_error_t *
-update_noderevs(fs_fs_t *fs,
- revision_pack_t *pack,
- apr_pool_t *pool)
-{
- int i;
- apr_pool_t *itempool = svn_pool_create(pool);
-
- for (i = 0; i < pack->fragments->nelts; ++i)
- {
- fragment_t *fragment = &APR_ARRAY_IDX(pack->fragments, i, fragment_t);
- if (fragment->kind == dir_fragment)
- {
- svn_string_t *content;
-
- SVN_ERR(get_fragment_content(&content, fs, fragment, itempool));
- svn_pool_clear(itempool);
- }
- }
-
- svn_pool_destroy(itempool);
+ apr_int64_t references;
+ apr_int64_t expanded_size;
+} representation_stats_t;
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-get_content_length(apr_size_t *length,
- fs_fs_t *fs,
- fragment_t *fragment,
- svn_boolean_t add_padding,
- apr_pool_t *pool)
+typedef struct node_stats_t
{
- svn_string_t *content;
-
- SVN_ERR(get_fragment_content(&content, fs, fragment, pool));
- if (add_padding)
- switch (fragment->kind)
- {
- case dir_fragment:
- *length = content->len + 16;
- break;
- case noderev_fragment:
- *length = content->len + 3;
- break;
- default:
- *length = content->len;
- break;
- }
- else
- *length = content->len;
-
- return SVN_NO_ERROR;
-}
+ apr_int64_t count;
+ apr_int64_t size;
+} node_stats_t;
static void
-move_fragment(fragment_t *fragment,
- apr_size_t new_position)
+add_rep_pack_stats(rep_pack_stats_t *stats,
+ representation_t *rep)
{
- revision_info_t *info;
- representation_t *representation;
- noderev_t *node;
+ stats->count++;
- fragment->position = new_position;
-
- switch (fragment->kind)
- {
- case header_fragment:
- info = fragment->data;
- info->target.offset = new_position;
- break;
-
- case changes_fragment:
- info = fragment->data;
- info->target.changes = new_position - info->target.offset;
- break;
-
- case property_fragment:
- case file_fragment:
- case dir_fragment:
- representation = fragment->data;
- representation->target.offset = new_position;
- break;
-
- case noderev_fragment:
- node = fragment->data;
- node->target.offset = new_position;
- break;
- }
+ stats->packed_size += rep->size;
+ stats->expanded_size += rep->expanded_size;
+ stats->overhead_size += rep->header_size + 7;
}
-static svn_error_t *
-pack_revisions(fs_fs_t *fs,
- revision_pack_t *pack,
- apr_pool_t *pool)
+static void
+add_rep_stats(representation_stats_t *stats,
+ representation_t *rep)
{
- int i;
- fragment_t *fragment, *next;
- svn_boolean_t needed_to_expand;
- revision_info_t *info;
- apr_size_t current_pos, len, old_len;
-
- apr_pool_t *itempool = svn_pool_create(pool);
-
- SVN_ERR(update_noderevs(fs, pack, pool));
-
- current_pos = pack->info->nelts > 1 ? 64 : 0;
- for (i = 0; i + 1 < pack->fragments->nelts; ++i)
- {
- fragment = &APR_ARRAY_IDX(pack->fragments, i, fragment_t);
- SVN_ERR(get_content_length(&len, fs, fragment, TRUE, itempool));
- move_fragment(fragment, current_pos);
- current_pos += len;
-
- svn_pool_clear(itempool);
- }
-
- fragment = &APR_ARRAY_IDX(pack->fragments, pack->fragments->nelts-1, fragment_t);
- fragment->position = current_pos;
-
- do
- {
- needed_to_expand = FALSE;
- current_pos = pack->info->nelts > 1 ? 64 : 0;
-
- for (i = 0; i + 1 < pack->fragments->nelts; ++i)
- {
- fragment = &APR_ARRAY_IDX(pack->fragments, i, fragment_t);
- next = &APR_ARRAY_IDX(pack->fragments, i + 1, fragment_t);
- old_len = next->position - fragment->position;
-
- SVN_ERR(get_content_length(&len, fs, fragment, FALSE, itempool));
-
- if (len > old_len)
- {
- len = (apr_size_t)(len * 1.1) + 10;
- needed_to_expand = TRUE;
- }
- else
- len = old_len;
-
- if (i == pack->info->nelts - 1)
- {
- info = APR_ARRAY_IDX(pack->info, pack->info->nelts - 1, revision_info_t*);
- info->target.offset = current_pos;
- }
-
- move_fragment(fragment, current_pos);
- current_pos += len;
-
- svn_pool_clear(itempool);
- }
-
- fragment = &APR_ARRAY_IDX(pack->fragments, pack->fragments->nelts-1, fragment_t);
- fragment->position = current_pos;
-
- SVN_ERR(get_content_length(&len, fs, fragment, FALSE, itempool));
- current_pos += len;
-
- for (i = 0; i < pack->info->nelts; ++i)
- {
- info = APR_ARRAY_IDX(pack->info, i, revision_info_t*);
- info->target.end = current_pos;
- }
- }
- while (needed_to_expand);
-
- svn_pool_destroy(itempool);
+ add_rep_pack_stats(&stats->total, rep);
+ if (rep->ref_count == 1)
+ add_rep_pack_stats(&stats->uniques, rep);
+ else
+ add_rep_pack_stats(&stats->shared, rep);
- return SVN_NO_ERROR;
+ stats->references += rep->ref_count;
+ stats->expanded_size += rep->ref_count * rep->expanded_size;
}
-static svn_error_t *
-write_revisions(fs_fs_t *fs,
- revision_pack_t *pack,
+static void
+print_rep_stats(representation_stats_t *stats,
apr_pool_t *pool)
{
- int i;
- fragment_t *fragment = NULL;
- svn_string_t *content;
-
- apr_pool_t *itempool = svn_pool_create(pool);
- apr_pool_t *iterpool = svn_pool_create(pool);
-
- apr_file_t *file;
- apr_size_t current_pos = 0;
- svn_stringbuf_t *null_buffer = svn_stringbuf_create_empty(iterpool);
-
- const char *dir = apr_psprintf(iterpool, "%s/new/%ld%s",
- fs->path, pack->base / fs->max_files_per_dir,
- pack->info->nelts > 1 ? ".pack" : "");
- SVN_ERR(svn_io_make_dir_recursively(dir, pool));
- SVN_ERR(svn_io_file_open(&file,
- pack->info->nelts > 1
- ? apr_psprintf(iterpool, "%s/pack", dir)
- : apr_psprintf(iterpool, "%s/%ld", dir, pack->base),
- APR_WRITE | APR_CREATE | APR_BUFFERED,
- APR_OS_DEFAULT,
- iterpool));
-
- for (i = 0; i < pack->fragments->nelts; ++i)
- {
- apr_size_t padding;
- fragment = &APR_ARRAY_IDX(pack->fragments, i, fragment_t);
- SVN_ERR(get_fragment_content(&content, fs, fragment, itempool));
-
- SVN_ERR_ASSERT(fragment->position >= current_pos);
- if ( fragment->kind == header_fragment
- && i+1 < pack->fragments->nelts)
- padding = APR_ARRAY_IDX(pack->fragments, i+1, fragment_t).position -
- content->len - current_pos;
- else
- padding = fragment->position - current_pos;
-
- if (padding)
- {
- while (null_buffer->len < padding)
- svn_stringbuf_appendbyte(null_buffer, 0);
-
- SVN_ERR(svn_io_file_write_full(file,
- null_buffer->data,
- padding,
- NULL,
- itempool));
- current_pos += padding;
- }
-
- SVN_ERR(svn_io_file_write_full(file,
- content->data,
- content->len,
- NULL,
- itempool));
- current_pos += content->len;
-
- svn_pool_clear(itempool);
- }
-
- apr_file_close(file);
-
- if (pack->info->nelts > 1)
- {
- svn_stream_t *stream;
- SVN_ERR(svn_io_file_open(&file,
- apr_psprintf(iterpool, "%s/manifest", dir),
- APR_WRITE | APR_CREATE | APR_BUFFERED,
- APR_OS_DEFAULT,
- iterpool));
- stream = svn_stream_from_aprfile2(file, FALSE, iterpool);
-
- for (i = 0; i < pack->info->nelts; ++i)
- {
- revision_info_t *info = APR_ARRAY_IDX(pack->info, i,
- revision_info_t *);
- SVN_ERR(svn_stream_printf(stream, itempool,
- "%" APR_UINT64_T_FMT "\n",
- info->target.offset));
- svn_pool_clear(itempool);
- }
- }
-
- svn_pool_destroy(itempool);
- svn_pool_destroy(iterpool);
-
- return SVN_NO_ERROR;
+ printf(_("%20s bytes in %12s reps\n"
+ "%20s bytes in %12s shared reps\n"
+ "%20s bytes expanded size\n"
+ "%20s bytes expanded shared size\n"
+ "%20s bytes with rep-sharing off\n"
+ "%20s shared references\n"),
+ svn__i64toa_sep(stats->total.packed_size, ',', pool),
+ svn__i64toa_sep(stats->total.count, ',', pool),
+ svn__i64toa_sep(stats->shared.packed_size, ',', pool),
+ svn__i64toa_sep(stats->shared.count, ',', pool),
+ svn__i64toa_sep(stats->total.expanded_size, ',', pool),
+ svn__i64toa_sep(stats->shared.expanded_size, ',', pool),
+ svn__i64toa_sep(stats->expanded_size, ',', pool),
+ svn__i64toa_sep(stats->references - stats->total.count, ',', pool));
}
-static svn_error_t *
-pack_and_write_revisions(fs_fs_t *fs,
- apr_pool_t *pool)
-{
- int i;
-
- SVN_ERR(svn_io_make_dir_recursively(apr_psprintf(pool, "%s/new",
- fs->path),
- pool));
-
- for (i = 0; i < fs->packs->nelts; ++i)
- {
- revision_pack_t *pack = APR_ARRAY_IDX(fs->packs, i, revision_pack_t*);
- if (pack->base % fs->max_files_per_dir == 0)
- print_progress(pack->base);
-
- SVN_ERR(pack_revisions(fs, pack, pool));
- SVN_ERR(write_revisions(fs, pack, pool));
- }
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-get_updated_dir(svn_string_t **content,
- fs_fs_t *fs,
- representation_t *representation,
- apr_pool_t *pool,
- apr_pool_t *scratch_pool)
+static void
+print_stats(fs_fs_t *fs,
+ apr_pool_t *pool)
{
- apr_hash_t *hash;
- apr_pool_t *hash_pool = svn_pool_create(scratch_pool);
- apr_array_header_t *dir = representation->dir->entries;
- int i;
- svn_stream_t *stream;
- svn_stringbuf_t *result;
+ int i, k;
- SVN_ERR(read_dir(&hash, fs, representation, scratch_pool));
- hash = apr_hash_copy(hash_pool, hash);
- for (i = 0; i < dir->nelts; ++i)
- {
- char buffer[256];
- svn_string_t *new_val;
- apr_size_t pos;
- direntry_t *entry = APR_ARRAY_IDX(dir, i, direntry_t *);
- svn_string_t *str_val = apr_hash_get(hash, entry->name, entry->name_len);
- if (str_val == NULL)
- return svn_error_createf(SVN_ERR_FS_CORRUPT, NULL,
- _("Dir entry '%s' not found"), entry->name);
-
- SVN_ERR_ASSERT(str_val->len < sizeof(buffer));
-
- memcpy(buffer, str_val->data, str_val->len+1);
- pos = strchr(buffer, '/') - buffer + 1;
- pos += svn__ui64toa(buffer + pos, entry->node->target.offset - entry->node->revision->target.offset);
- new_val = svn_string_ncreate(buffer, pos, hash_pool);
-
- apr_hash_set(hash, entry->name, entry->name_len, new_val);
- }
-
- result = svn_stringbuf_create_ensure(representation->target.size, pool);
- stream = svn_stream_from_stringbuf(result, hash_pool);
- SVN_ERR(svn_hash_write2(hash, stream, SVN_HASH_TERMINATOR, hash_pool));
- svn_pool_destroy(hash_pool);
-
- *content = svn_stringbuf__morph_into_string(result);
+ representation_stats_t file_rep_stats = { { 0 } };
+ representation_stats_t dir_rep_stats = { { 0 } };
+ representation_stats_t file_prop_rep_stats = { { 0 } };
+ representation_stats_t dir_prop_rep_stats = { { 0 } };
+ representation_stats_t total_rep_stats = { { 0 } };
+
+ node_stats_t dir_node_stats = { 0 };
+ node_stats_t file_node_stats = { 0 };
+ node_stats_t total_node_stats = { 0 };
+
+ apr_int64_t total_size = 0;
+ apr_int64_t change_count = 0;
+ apr_int64_t change_len = 0;
- return SVN_NO_ERROR;
-}
-
-struct diff_write_baton_t
-{
- svn_stream_t *stream;
- apr_size_t size;
-};
-
-static svn_error_t *
-diff_write_handler(void *baton,
- const char *data,
- apr_size_t *len)
-{
- struct diff_write_baton_t *whb = baton;
-
- SVN_ERR(svn_stream_write(whb->stream, data, len));
- whb->size += *len;
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-diff_stringbufs(svn_stringbuf_t *diff,
- apr_size_t *inflated_size,
- svn_string_t *base,
- svn_string_t *content,
- apr_pool_t *pool)
-{
- svn_txdelta_window_handler_t diff_wh;
- void *diff_whb;
- struct diff_write_baton_t whb;
-
- svn_stream_t *stream;
- svn_stream_t *source = svn_stream_from_string(base, pool);
- svn_stream_t *target = svn_stream_from_stringbuf(diff, pool);
-
- /* Prepare to write the svndiff data. */
- svn_txdelta_to_svndiff3(&diff_wh,
- &diff_whb,
- target,
- 1,
- SVN_DELTA_COMPRESSION_LEVEL_DEFAULT,
- pool);
-
- whb.stream = svn_txdelta_target_push(diff_wh, diff_whb, source, pool);
- whb.size = 0;
-
- stream = svn_stream_create(&whb, pool);
- svn_stream_set_write(stream, diff_write_handler);
-
- SVN_ERR(svn_stream_write(stream, content->data, &content->len));
- SVN_ERR(svn_stream_close(whb.stream));
- SVN_ERR(svn_stream_close(stream));
-
- *inflated_size = whb.size;
- return SVN_NO_ERROR;
-}
-
-static void
-update_id(svn_stringbuf_t *node_rev,
- const char *key,
- noderev_t *node)
-{
- char *newline_pos = 0;
- char *pos;
-
- pos = strstr(node_rev->data, key);
- if (pos)
- pos = strchr(pos, '/');
- if (pos)
- newline_pos = strchr(++pos, '\n');
-
- if (pos && newline_pos)
- {
- char temp[SVN_INT64_BUFFER_SIZE];
- apr_size_t len = svn__i64toa(temp, node->target.offset - node->revision->target.offset);
- svn_stringbuf_replace(node_rev,
- pos - node_rev->data, newline_pos - pos,
- temp, len);
- }
-}
-
-static void
-update_text(svn_stringbuf_t *node_rev,
- const char *key,
- representation_t *representation,
- apr_pool_t *scratch_pool)
-{
- apr_size_t key_len = strlen(key);
- char *pos = strstr(node_rev->data, key);
- char *val_pos;
-
- if (!pos)
- return;
-
- val_pos = pos + key_len;
- if (representation->dir)
- {
- char *newline_pos = strchr(val_pos, '\n');
- svn_checksum_t checksum = {representation->dir->target_md5,
- svn_checksum_md5};
- const char* temp = apr_psprintf(scratch_pool, "%ld %" APR_UINT64_T_FMT " %"
- APR_UINT64_T_FMT" %" APR_SIZE_T_FMT " %s",
- representation->revision->revision,
- representation->target.offset - representation->revision->target.offset,
- representation->target.size,
- representation->dir->size,
- svn_checksum_to_cstring(&checksum,
- scratch_pool));
-
- svn_stringbuf_replace(node_rev,
- val_pos - node_rev->data, newline_pos - val_pos,
- temp, strlen(temp));
- }
- else
+ for (i = 0; i < fs->revisions->nelts; ++i)
{
- const char* temp;
- char *end_pos = strchr(val_pos, ' ');
+ revision_info_t *revision = APR_ARRAY_IDX(fs->revisions, i,
+ revision_info_t *);
+ change_count += revision->change_count;
+ change_len += revision->changes_len;
+ total_size += revision->end - revision->offset;
+
+ dir_node_stats.count += revision->dir_noderev_count;
+ dir_node_stats.size += revision->dir_noderev_size;
+ file_node_stats.count += revision->file_noderev_count;
+ file_node_stats.size += revision->file_noderev_size;
+ total_node_stats.count += revision->dir_noderev_count
+ + revision->file_noderev_count;
+ total_node_stats.size += revision->dir_noderev_size
+ + revision->file_noderev_size;
- val_pos = end_pos + 1;
- end_pos = strchr(strchr(val_pos, ' ') + 1, ' ');
- temp = apr_psprintf(scratch_pool, "%" APR_UINT64_T_FMT " %" APR_UINT64_T_FMT,
- representation->target.offset - representation->revision->target.offset,
- representation->target.size);
-
- svn_stringbuf_replace(node_rev,
- val_pos - node_rev->data, end_pos - val_pos,
- temp, strlen(temp));
- }
-}
-
-static svn_error_t *
-get_fragment_content(svn_string_t **content,
- fs_fs_t *fs,
- fragment_t *fragment,
- apr_pool_t *pool)
-{
- revision_info_t *info;
- representation_t *representation;
- noderev_t *node;
- svn_string_t *revision_content, *base_content;
- svn_stringbuf_t *header, *node_rev, *text;
- apr_size_t header_size;
- svn_checksum_t *checksum = NULL;
-
- switch (fragment->kind)
- {
- case header_fragment:
- info = fragment->data;
- *content = svn_string_createf(pool,
- "\n%" APR_UINT64_T_FMT " %" APR_UINT64_T_FMT "\n",
- info->root_noderev->target.offset - info->target.offset,
- info->target.changes);
- return SVN_NO_ERROR;
-
- case changes_fragment:
- info = fragment->data;
- SVN_ERR(get_content(&revision_content, fs, info->revision, pool));
-
- *content = svn_string_create_empty(pool);
- (*content)->data = revision_content->data + info->original.changes;
- (*content)->len = info->target.changes_len;
- return SVN_NO_ERROR;
-
- case property_fragment:
- case file_fragment:
- representation = fragment->data;
- SVN_ERR(get_content(&revision_content, fs,
- representation->revision->revision, pool));
-
- if (representation->delta_base)
- if (representation->delta_base->dir)
+ for (k = 0; k < revision->representations->nelts; ++k)
+ {
+ representation_t *rep = APR_ARRAY_IDX(revision->representations,
+ k, representation_t *);
+ switch(rep->kind)
{
- SVN_ERR(get_combined_window(&text, fs, representation, pool));
- representation->target.size = text->len;
-
- svn_stringbuf_insert(text, 0, "PLAIN\n", 6);
- svn_stringbuf_appendcstr(text, "ENDREP\n");
- *content = svn_stringbuf__morph_into_string(text);
-
- return SVN_NO_ERROR;
+ case file_rep:
+ add_rep_stats(&file_rep_stats, rep);
+ break;
+ case dir_rep:
+ add_rep_stats(&dir_rep_stats, rep);
+ break;
+ case file_property_rep:
+ add_rep_stats(&file_prop_rep_stats, rep);
+ break;
+ case dir_property_rep:
+ add_rep_stats(&dir_prop_rep_stats, rep);
+ break;
+ default:
+ break;
}
- else
- if (representation->delta_base == fs->null_base)
- header = svn_stringbuf_create("DELTA\n", pool);
- else
- header = svn_stringbuf_createf(pool,
- "DELTA %ld %" APR_UINT64_T_FMT " %" APR_UINT64_T_FMT "\n",
- representation->delta_base->revision->revision,
- representation->delta_base->target.offset
- - representation->delta_base->revision->target.offset,
- representation->delta_base->target.size);
- else
- header = svn_stringbuf_create("PLAIN\n", pool);
-
- header_size = strchr(revision_content->data +
- representation->original.offset, '\n') -
- revision_content->data -
- representation->original.offset + 1;
- svn_stringbuf_appendbytes(header,
- revision_content->data +
- representation->original.offset +
- header_size,
- representation->original.size);
- svn_stringbuf_appendcstr(header, "ENDREP\n");
- *content = svn_stringbuf__morph_into_string(header);
- return SVN_NO_ERROR;
-
- case dir_fragment:
- representation = fragment->data;
- SVN_ERR(get_updated_dir(&revision_content, fs, representation,
- pool, pool));
- SVN_ERR(svn_checksum(&checksum, svn_checksum_md5,
- revision_content->data, revision_content->len,
- pool));
- memcpy(representation->dir->target_md5,
- checksum->digest,
- sizeof(representation->dir->target_md5));
-
- if (representation->delta_base)
- {
- if (representation->delta_base->dir == NULL)
- {
- header = svn_stringbuf_create("DELTA\n", pool);
- base_content = svn_string_create_empty(pool);
- }
- else
- {
- representation_t *base_rep = representation->delta_base;
- header = svn_stringbuf_createf(pool,
- "DELTA %ld %" APR_UINT64_T_FMT " %" APR_UINT64_T_FMT "\n",
- base_rep->revision->revision,
- base_rep->target.offset - base_rep->revision->target.offset,
- base_rep->target.size);
- SVN_ERR(get_updated_dir(&base_content, fs, base_rep,
- pool, pool));
- }
-
- header_size = header->len;
- SVN_ERR(diff_stringbufs(header, &representation->dir->size,
- base_content,
- revision_content, pool));
- representation->target.size = header->len - header_size;
- svn_stringbuf_appendcstr(header, "ENDREP\n");
- *content = svn_stringbuf__morph_into_string(header);
- }
- else
- {
- representation->target.size = revision_content->len;
- representation->dir->size = revision_content->len;
- *content = svn_string_createf(pool, "PLAIN\n%sENDREP\n",
- revision_content->data);
- }
-
- return SVN_NO_ERROR;
-
- case noderev_fragment:
- node = fragment->data;
- SVN_ERR(get_content(&revision_content, fs,
- node->revision->revision, pool));
- node_rev = svn_stringbuf_ncreate(revision_content->data +
- node->original.offset,
- node->original.size,
- pool);
-
- update_id(node_rev, "id: ", node);
- update_id(node_rev, "pred: ", node->predecessor);
- update_text(node_rev, "text: ", node->text, pool);
- update_text(node_rev, "props: ", node->props, pool);
-
- *content = svn_stringbuf__morph_into_string(node_rev);
- return SVN_NO_ERROR;
- }
-
- SVN_ERR_ASSERT(0);
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-prepare_repo(const char *path, apr_pool_t *pool)
-{
- svn_node_kind_t kind;
-
- const char *old_path = svn_dirent_join(path, "db/old", pool);
- const char *new_path = svn_dirent_join(path, "new", pool);
- const char *revs_path = svn_dirent_join(path, "db/revs", pool);
- const char *old_rep_cache_path = svn_dirent_join(path, "db/rep-cache.db.old", pool);
- const char *rep_cache_path = svn_dirent_join(path, "db/rep-cache.db", pool);
-
- SVN_ERR(svn_io_check_path(old_path, &kind, pool));
- if (kind == svn_node_dir)
- {
- SVN_ERR(svn_io_remove_dir2(new_path, TRUE, NULL, NULL, pool));
- SVN_ERR(svn_io_file_move(revs_path, new_path, pool));
- SVN_ERR(svn_io_file_move(old_path, revs_path, pool));
- SVN_ERR(svn_io_remove_dir2(new_path, TRUE, NULL, NULL, pool));
- }
- SVN_ERR(svn_io_check_path(old_rep_cache_path, &kind, pool));
- if (kind == svn_node_file)
- SVN_ERR(svn_io_file_move(old_rep_cache_path, rep_cache_path, pool));
-
- return SVN_NO_ERROR;
-}
-
-static svn_error_t *
-activate_new_revs(const char *path, apr_pool_t *pool)
-{
- svn_node_kind_t kind;
-
- const char *old_path = svn_dirent_join(path, "db/old", pool);
- const char *new_path = svn_dirent_join(path, "new", pool);
- const char *revs_path = svn_dirent_join(path, "db/revs", pool);
- const char *old_rep_cache_path = svn_dirent_join(path, "db/rep-cache.db.old", pool);
- const char *rep_cache_path = svn_dirent_join(path, "db/rep-cache.db", pool);
-
- SVN_ERR(svn_io_check_path(old_path, &kind, pool));
- if (kind == svn_node_none)
- {
- SVN_ERR(svn_io_file_move(revs_path, old_path, pool));
- SVN_ERR(svn_io_file_move(new_path, revs_path, pool));
+ add_rep_stats(&total_rep_stats, rep);
+ }
}
- SVN_ERR(svn_io_check_path(old_rep_cache_path, &kind, pool));
- if (kind == svn_node_none)
- SVN_ERR(svn_io_file_move(rep_cache_path, old_rep_cache_path, pool));
-
- return SVN_NO_ERROR;
+ printf("\nGlobal statistics:\n");
+ printf(_("%20s bytes in %12s revisions\n"
+ "%20s bytes in %12s changes\n"
+ "%20s bytes in %12s node revision records\n"
+ "%20s bytes in %12s representations\n"
+ "%20s bytes expanded representation size\n"
+ "%20s bytes with rep-sharing off\n"),
+ svn__i64toa_sep(total_size, ',', pool),
+ svn__i64toa_sep(fs->revisions->nelts, ',', pool),
+ svn__i64toa_sep(change_len, ',', pool),
+ svn__i64toa_sep(change_count, ',', pool),
+ svn__i64toa_sep(total_node_stats.size, ',', pool),
+ svn__i64toa_sep(total_node_stats.count, ',', pool),
+ svn__i64toa_sep(total_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(total_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(total_rep_stats.total.expanded_size, ',', pool),
+ svn__i64toa_sep(total_rep_stats.expanded_size, ',', pool));
+
+ printf("\nNoderev statistics:\n");
+ printf(_("%20s bytes in %12s nodes total\n"
+ "%20s bytes in %12s directory noderevs\n"
+ "%20s bytes in %12s file noderevs\n"),
+ svn__i64toa_sep(total_node_stats.size, ',', pool),
+ svn__i64toa_sep(total_node_stats.count, ',', pool),
+ svn__i64toa_sep(dir_node_stats.size, ',', pool),
+ svn__i64toa_sep(dir_node_stats.count, ',', pool),
+ svn__i64toa_sep(file_node_stats.size, ',', pool),
+ svn__i64toa_sep(file_node_stats.count, ',', pool));
+
+ printf("\nRepresentation statistics:\n");
+ printf(_("%20s bytes in %12s representations total\n"
+ "%20s bytes in %12s directory representations\n"
+ "%20s bytes in %12s file representations\n"
+ "%20s bytes in %12s directory property representations\n"
+ "%20s bytes in %12s file property representations\n"
+ "%20s bytes in header & footer overhead\n"),
+ svn__i64toa_sep(total_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(total_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(dir_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(dir_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(file_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(file_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(dir_prop_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(dir_prop_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(file_prop_rep_stats.total.packed_size, ',', pool),
+ svn__i64toa_sep(file_prop_rep_stats.total.count, ',', pool),
+ svn__i64toa_sep(total_rep_stats.total.overhead_size, ',', pool));
+
+ printf("\nDirectory representation statistics:\n");
+ print_rep_stats(&dir_rep_stats, pool);
+ printf("\nFile representation statistics:\n");
+ print_rep_stats(&file_rep_stats, pool);
+ printf("\nDirectory property representation statistics:\n");
+ print_rep_stats(&dir_prop_rep_stats, pool);
+ printf("\nFile property representation statistics:\n");
+ print_rep_stats(&file_prop_rep_stats, pool);
}
static void
@@ -2707,16 +1397,13 @@ print_usage(svn_stream_t *ostream, const
{
svn_error_clear(svn_stream_printf(ostream, pool,
"\n"
- "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! WARNING !!!!!!!!!!!!!!!!!!!!!!!!!!!!!\n"
- "!!! This is an experimental tool. Don't use it on production data !!!\n"
- "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!\n"
- "\n"
"Usage: %s <repo> <cachesize>\n"
"\n"
- "Optimize the repository at local path <repo> staring from revision 0.\n"
+ "Read the repository at local path <repo> starting at revision 0,\n"
+ "count statistical information and write that data to stdout.\n"
"Use up to <cachesize> MB of memory for caching. This does not include\n"
"temporary representation of the repository structure, i.e. the actual\n"
- "memory will be higher and <cachesize> be the lower limit.\n",
+ "memory may be considerably higher.\n",
progname));
}
@@ -2761,33 +1448,12 @@ int main(int argc, const char *argv[])
repo_path = argv[1];
start_revision = 0;
- printf("\nPreparing repository\n");
- svn_err = prepare_repo(repo_path, pool);
+ printf("Reading revisions\n");
+ svn_err = read_revisions(&fs, repo_path, start_revision, memsize, pool);
+ printf("\n");
- if (!svn_err)
- {
- printf("Reading revisions\n");
- svn_err = read_revisions(&fs, repo_path, start_revision, memsize, pool);
- }
+ print_stats(fs, pool);
- if (!svn_err)
- {
- printf("\nReordering revision content\n");
- svn_err = reorder_revisions(fs, pool);
- }
-
- if (!svn_err)
- {
- printf("\nPacking and writing revisions\n");
- svn_err = pack_and_write_revisions(fs, pool);
- }
-
- if (!svn_err)
- {
- printf("\nSwitch to new revs\n");
- svn_err = activate_new_revs(repo_path, pool);
- }
-
if (svn_err)
{
[... 2 lines stripped ...]