You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Daniel Shahaf <da...@elego.de> on 2011/09/07 16:19:19 UTC
Re: svn commit: r1166184 -
/subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c
This will assert with
% cat $REPOS/db/format
6
layout linear
%
stsp@apache.org wrote on Wed, Sep 07, 2011 at 13:53:59 -0000:
> Author: stsp
> Date: Wed Sep 7 13:53:59 2011
> New Revision: 1166184
>
> URL: http://svn.apache.org/viewvc?rev=1166184&view=rev
> Log:
> On the fs-successor-ids branch, shard files within db/successors/ into
> directories, such that no more than MAX_FILES_PER_DIR files are created
> in a directory. This follows the layout specified in the 'format' file,
> just like the sibling directories of db/successors/ do.
>
> * subversion/libsvn_fs_fs/fs_fs.c
> (path_successor_ids_shard, path_successor_revisions_shard,
> path_successor_node_revs_shard): New. Return the appropriate shard for
> a given revision.
> (path_successor_ids, path_successor_revisions,
> path_successor_node_revs): Use above new helper function to compute paths.
> (make_successor_ids_dirs): Create shards named '0'.
> (update_successor_map): Create new shards as necessary.
>
> Modified:
> subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c
>
> Modified: subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c
> URL: http://svn.apache.org/viewvc/subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c?rev=1166184&r1=1166183&r2=1166184&view=diff
> ==============================================================================
> --- subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c (original)
> +++ subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c Wed Sep 7 13:53:59 2011
> @@ -253,26 +253,66 @@ path_rev(svn_fs_t *fs, svn_revnum_t rev,
> }
>
> static const char *
> -path_successor_ids(svn_fs_t *fs, svn_revnum_t rev, apr_pool_t *pool)
> +path_successor_ids_shard(svn_fs_t *fs, svn_revnum_t rev, apr_pool_t *pool)
> {
> - long filenum = rev / FSFS_SUCCESSORS_MAX_REVS_PER_FILE;
> + fs_fs_data_t *ffd = fs->fsap_data;
>
> + assert(ffd->max_files_per_dir);
> return svn_dirent_join_many(pool, fs->path, PATH_SUCCESSORS_TOP_DIR,
> PATH_SUCCESSORS_IDS_DIR,
> - apr_psprintf(pool, "%ld", filenum), NULL);
> + apr_psprintf(pool, "%ld",
> + rev / ffd->max_files_per_dir),
> + NULL);
> }
>
> static const char *
> -path_successor_revisions(svn_fs_t *fs, svn_revnum_t rev, apr_pool_t *pool)
> +path_successor_ids(svn_fs_t *fs, svn_revnum_t rev, apr_pool_t *pool)
> {
> long filenum = rev / FSFS_SUCCESSORS_MAX_REVS_PER_FILE;
>
> + return svn_dirent_join_many(pool, path_successor_ids_shard(fs, rev, pool),
> + apr_psprintf(pool, "%ld", filenum), NULL);
> +}
> +
> +static const char *
> +path_successor_revisions_shard(svn_fs_t *fs, svn_revnum_t rev,
> + apr_pool_t *pool)
> +{
> + fs_fs_data_t *ffd = fs->fsap_data;
> +
> + assert(ffd->max_files_per_dir);
> return svn_dirent_join_many(pool, fs->path, PATH_SUCCESSORS_TOP_DIR,
> PATH_SUCCESSORS_REVISIONS_DIR,
> + apr_psprintf(pool, "%ld",
> + rev / ffd->max_files_per_dir),
> + NULL);
> +}
> +
> +static const char *
> +path_successor_revisions(svn_fs_t *fs, svn_revnum_t rev, apr_pool_t *pool)
> +{
> + long filenum = rev / FSFS_SUCCESSORS_MAX_REVS_PER_FILE;
> +
> + return svn_dirent_join_many(pool,
> + path_successor_revisions_shard(fs, rev, pool),
> apr_psprintf(pool, "%ld", filenum), NULL);
> }
>
> static const char *
> +path_successor_node_revs_shard(svn_fs_t *fs, svn_revnum_t rev,
> + apr_pool_t *pool)
> +{
> + fs_fs_data_t *ffd = fs->fsap_data;
> +
> + assert(ffd->max_files_per_dir);
> + return svn_dirent_join_many(pool, fs->path, PATH_SUCCESSORS_TOP_DIR,
> + PATH_SUCCESSORS_NODE_REVS_DIR,
> + apr_psprintf(pool, "%ld",
> + rev / ffd->max_files_per_dir),
> + NULL);
> +}
> +
> +static const char *
> path_successor_node_revs(svn_fs_t *fs, const char *node_rev_id,
> apr_pool_t *pool)
> {
> @@ -285,8 +325,8 @@ path_successor_node_revs(svn_fs_t *fs, c
> rev = svn_fs_fs__id_rev(id);
> filenum = rev / FSFS_SUCCESSORS_MAX_REVS_PER_FILE;
>
> - return svn_dirent_join_many(pool, fs->path, PATH_SUCCESSORS_TOP_DIR,
> - PATH_SUCCESSORS_NODE_REVS_DIR,
> + return svn_dirent_join_many(pool,
> + path_successor_node_revs_shard(fs, rev, pool),
> apr_psprintf(pool, "%ld", filenum), NULL);
> }
>
> @@ -1282,19 +1322,21 @@ make_successor_ids_dirs(svn_fs_t *fs, ap
> {
> const char *top_dir = svn_dirent_join(fs->path, PATH_SUCCESSORS_TOP_DIR,
> pool);
> - const char *node_revs_dir = svn_dirent_join(top_dir,
> - PATH_SUCCESSORS_NODE_REVS_DIR,
> - pool);
> - const char *revs_dir = svn_dirent_join(top_dir,
> - PATH_SUCCESSORS_REVISIONS_DIR, pool);
> - const char *data_dir = svn_dirent_join(top_dir, PATH_SUCCESSORS_IDS_DIR,
> - pool);
> + const char *node_revs_dir = svn_dirent_join_many(
> + pool, top_dir, PATH_SUCCESSORS_NODE_REVS_DIR,
> + "0", NULL, pool);
> + const char *revisions_dir = svn_dirent_join_many(
> + pool, top_dir, PATH_SUCCESSORS_REVISIONS_DIR,
> + "0", NULL, pool);
> + const char *ids_dir = svn_dirent_join_many(pool, top_dir,
> + PATH_SUCCESSORS_IDS_DIR,
> + "0", NULL, pool);
> apr_file_t *file;
> apr_uint64_t n;
>
> SVN_ERR(svn_io_make_dir_recursively(node_revs_dir, pool));
> - SVN_ERR(svn_io_make_dir_recursively(revs_dir, pool));
> - SVN_ERR(svn_io_make_dir_recursively(data_dir, pool));
> + SVN_ERR(svn_io_make_dir_recursively(revisions_dir, pool));
> + SVN_ERR(svn_io_make_dir_recursively(ids_dir, pool));
>
> /* Create empty successor-ID data for revision zero.
> * No successors were created in this revision. */
> @@ -6227,6 +6269,47 @@ update_successor_map(svn_fs_t *fs,
> SVN_ERR(update_successor_node_revs_files(&node_revs_tempfiles, fs,
> new_rev, successor_ids, pool));
>
> + if (new_rev % FSFS_SUCCESSORS_MAX_REVS_PER_FILE == 0)
> + {
> + /* Create the shards for new successor files. We don't care if this
> + * fails because the shards already existed for some reason. */
> + svn_error_t *err;
> + const char *new_dir = path_successor_ids_shard(fs, new_rev, pool);
> + err = svn_io_dir_make(new_dir, APR_OS_DEFAULT, pool);
> + if (err && !APR_STATUS_IS_EEXIST(err->apr_err))
> + SVN_ERR(err);
> + svn_error_clear(err);
> + SVN_ERR(svn_fs_fs__dup_perms(new_dir,
> + svn_dirent_join(fs->path,
> + PATH_SUCCESSORS_IDS_DIR,
> + pool),
> + pool));
> +
> + new_dir = path_successor_revisions_shard(fs, new_rev, pool);
> + err = svn_io_dir_make(new_dir, APR_OS_DEFAULT, pool);
> + if (err && !APR_STATUS_IS_EEXIST(err->apr_err))
> + SVN_ERR(err);
> + svn_error_clear(err);
> + SVN_ERR(svn_fs_fs__dup_perms(new_dir,
> + svn_dirent_join(
> + fs->path,
> + PATH_SUCCESSORS_REVISIONS_DIR,
> + pool),
> + pool));
> +
> + new_dir = path_successor_node_revs_shard(fs, new_rev, pool);
> + err = svn_io_dir_make(new_dir, APR_OS_DEFAULT, pool);
> + if (err && !APR_STATUS_IS_EEXIST(err->apr_err))
> + SVN_ERR(err);
> + svn_error_clear(err);
> + SVN_ERR(svn_fs_fs__dup_perms(new_dir,
> + svn_dirent_join(
> + fs->path,
> + PATH_SUCCESSORS_NODE_REVS_DIR,
> + pool),
> + pool));
> + }
> +
> /* Move temporary files into place. */
> if (new_rev > 1 && new_rev % FSFS_SUCCESSORS_MAX_REVS_PER_FILE != 0)
> perms_reference = successor_ids_abspath;
>
>
Re: svn commit: r1166184 -
/subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c
Posted by Daniel Shahaf <da...@elego.de>.
Stefan Sperling wrote on Wed, Sep 07, 2011 at 16:46:21 +0200:
> On Wed, Sep 07, 2011 at 05:19:19PM +0300, Daniel Shahaf wrote:
> > This will assert with
> >
> > % cat $REPOS/db/format
> > 6
> > layout linear
> > %
>
> Good catch. The assert() is from code I copy-pasted. Will look into it.
You need to decide at what level to check the format number. For
example, in make_successor_dirs() you check the format number in the
function itself; the other option is to assert() the format number in
the function and check it at the caller, and I believe that's what I did
last time I had to make that call.
Re: svn commit: r1166184 -
/subversion/branches/fs-successor-ids/subversion/libsvn_fs_fs/fs_fs.c
Posted by Stefan Sperling <st...@elego.de>.
On Wed, Sep 07, 2011 at 05:19:19PM +0300, Daniel Shahaf wrote:
> This will assert with
>
> % cat $REPOS/db/format
> 6
> layout linear
> %
Good catch. The assert() is from code I copy-pasted. Will look into it.