staging: lustre: llite: reduce indent in ll_dir_read
authorwang di <di.wang@intel.com>
Fri, 22 Jul 2016 02:44:03 +0000 (22:44 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 15 Aug 2016 16:31:06 +0000 (18:31 +0200)
Instead of making a large chunk of code conditional
based on if a page is valid we reverse the page
validity test and exit the loop if the page is invalid
instead. This allows a section of code to reduce its
indentation one level.

Signed-off-by: wang di <di.wang@intel.com>
Reviewed-on: http://review.whamcloud.com/10761
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-4906
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/staging/lustre/lustre/llite/dir.c

index 5b46dd8..77fadf1 100644 (file)
@@ -508,93 +508,90 @@ int ll_dir_read(struct inode *inode, struct md_op_data *op_data,
        while (rc == 0 && !done) {
                struct lu_dirpage *dp;
                struct lu_dirent  *ent;
+               __u64 hash;
+               __u64 next;
+
+               if (IS_ERR(page)) {
+                       rc = PTR_ERR(page);
+                       break;
+               }
+
+               hash = MDS_DIR_END_OFF;
+               dp = page_address(page);
+               for (ent = lu_dirent_start(dp); ent && !done;
+                    ent = lu_dirent_next(ent)) {
+                       __u16     type;
+                       int         namelen;
+                       struct lu_fid  fid;
+                       __u64     lhash;
+                       __u64     ino;
 
-               if (!IS_ERR(page)) {
                        /*
-                        * If page is empty (end of directory is reached),
-                        * use this value.
+                        * XXX: implement correct swabbing here.
                         */
-                       __u64 hash = MDS_DIR_END_OFF;
-                       __u64 next;
-
-                       dp = page_address(page);
-                       for (ent = lu_dirent_start(dp); ent && !done;
-                            ent = lu_dirent_next(ent)) {
-                               __u16     type;
-                               int         namelen;
-                               struct lu_fid  fid;
-                               __u64     lhash;
-                               __u64     ino;
 
+                       hash = le64_to_cpu(ent->lde_hash);
+                       if (hash < pos)
                                /*
-                                * XXX: implement correct swabbing here.
+                                * Skip until we find target hash
+                                * value.
                                 */
+                               continue;
 
-                               hash = le64_to_cpu(ent->lde_hash);
-                               if (hash < pos)
-                                       /*
-                                        * Skip until we find target hash
-                                        * value.
-                                        */
-                                       continue;
-
-                               namelen = le16_to_cpu(ent->lde_namelen);
-                               if (namelen == 0)
-                                       /*
-                                        * Skip dummy record.
-                                        */
-                                       continue;
-
-                               if (api32 && hash64)
-                                       lhash = hash >> 32;
-                               else
-                                       lhash = hash;
-                               fid_le_to_cpu(&fid, &ent->lde_fid);
-                               ino = cl_fid_build_ino(&fid, api32);
-                               type = ll_dirent_type_get(ent);
-                               ctx->pos = lhash;
-                               /* For 'll_nfs_get_name_filldir()', it will try
-                                * to access the 'ent' through its 'lde_name',
-                                * so the parameter 'name' for 'ctx->actor()'
-                                * must be part of the 'ent'.
+                       namelen = le16_to_cpu(ent->lde_namelen);
+                       if (namelen == 0)
+                               /*
+                                * Skip dummy record.
                                 */
-                               done = !dir_emit(ctx, ent->lde_name,
-                                                namelen, ino, type);
-                       }
-                       next = le64_to_cpu(dp->ldp_hash_end);
-                       if (!done) {
-                               pos = next;
-                               if (pos == MDS_DIR_END_OFF) {
-                                       /*
-                                        * End of directory reached.
-                                        */
-                                       done = 1;
-                                       ll_release_page(page, 0);
-                               } else if (1 /* chain is exhausted*/) {
-                                       /*
-                                        * Normal case: continue to the next
-                                        * page.
-                                        */
-                                       ll_release_page(page,
-                                           le32_to_cpu(dp->ldp_flags) &
-                                                       LDF_COLLIDE);
-                                       next = pos;
-                                       page = ll_get_dir_page(inode, pos,
-                                                              &chain);
-                               } else {
-                                       /*
-                                        * go into overflow page.
-                                        */
-                                       LASSERT(le32_to_cpu(dp->ldp_flags) &
+                               continue;
+
+                       if (api32 && hash64)
+                               lhash = hash >> 32;
+                       else
+                               lhash = hash;
+                       fid_le_to_cpu(&fid, &ent->lde_fid);
+                       ino = cl_fid_build_ino(&fid, api32);
+                       type = ll_dirent_type_get(ent);
+                       ctx->pos = lhash;
+                       /* For 'll_nfs_get_name_filldir()', it will try
+                        * to access the 'ent' through its 'lde_name',
+                        * so the parameter 'name' for 'ctx->actor()'
+                        * must be part of the 'ent'.
+                        */
+                       done = !dir_emit(ctx, ent->lde_name,
+                                        namelen, ino, type);
+               }
+               next = le64_to_cpu(dp->ldp_hash_end);
+               if (!done) {
+                       pos = next;
+                       if (pos == MDS_DIR_END_OFF) {
+                               /*
+                                * End of directory reached.
+                                */
+                               done = 1;
+                               ll_release_page(page, 0);
+                       } else if (1 /* chain is exhausted*/) {
+                               /*
+                                * Normal case: continue to the next
+                                * page.
+                                */
+                               ll_release_page(page,
+                                               le32_to_cpu(dp->ldp_flags) &
                                                LDF_COLLIDE);
-                                       ll_release_page(page, 1);
-                               }
+                               next = pos;
+                               page = ll_get_dir_page(inode, pos,
+                                                      &chain);
                        } else {
-                               pos = hash;
-                               ll_release_page(page, 0);
+                               /*
+                                * go into overflow page.
+                                */
+                               LASSERT(le32_to_cpu(dp->ldp_flags) &
+                                       LDF_COLLIDE);
+                               ll_release_page(page, 1);
                        }
                } else {
-                       rc = PTR_ERR(page);
+                       pos = hash;
+                       ll_release_page(page, 0);
                }
        }