linux-ext4.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andreas Dilger <adilger@whamcloud.com>
To: tytso@mit.edu
Cc: linux-ext4@vger.kernel.org, Andreas Dilger <adilger@whamcloud.com>
Subject: [PATCH 4/9] e2fsck: reduce memory usage for many directories
Date: Thu,  6 Feb 2020 18:09:41 -0700	[thread overview]
Message-ID: <1581037786-62789-4-git-send-email-adilger@whamcloud.com> (raw)
In-Reply-To: <1581037786-62789-1-git-send-email-adilger@whamcloud.com>

Pack struct dx_dir_info and dx_dirblock_info properly in memory, to
avoid holes, and fields are not larger than necessary.  This reduces
the memory needed for each hashed dir, according to pahole(1) from:

    struct dx_dir_info {
        /* size: 32, cachelines: 1, members: 6 */
        /* sum members: 26, holes: 1, sum holes: 2 */
        /* padding: 4 */
    };
    struct dx_dirblock_info {
        /* size: 56, cachelines: 1, members: 9 */
        /* sum members: 48, holes: 2, sum holes: 8 */
        /* last cacheline: 56 bytes */
    };

to 8 bytes less for each directory and directory block, and leaves
space for future use if needed (e.g. larger numblocks):

    struct dx_dir_info {
        /* size: 24, cachelines: 1, members: 6 */
        /* sum members: 20, holes: 1, sum holes: 4 */
        /* bit holes: 1, sum bit holes: 7 bits */
    };
    struct dx_dirblock_info {
        /* size: 48, cachelines: 1, members: 9 */
    };

Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Lustre-bug-id: https://jira.whamcloud.com/browse/LU-13197
---
 e2fsck/dx_dirinfo.c |  3 +--
 e2fsck/e2fsck.h     | 14 +++++++-------
 e2fsck/pass2.c      | 12 ++++++------
 3 files changed, 14 insertions(+), 15 deletions(-)

diff --git a/e2fsck/dx_dirinfo.c b/e2fsck/dx_dirinfo.c
index f0f6084..caca3e3 100644
--- a/e2fsck/dx_dirinfo.c
+++ b/e2fsck/dx_dirinfo.c
@@ -73,11 +73,10 @@ void e2fsck_add_dx_dir(e2fsck_t ctx, ext2_ino_t ino, struct ext2_inode *inode,
 	dir->ino = ino;
 	dir->numblocks = num_blocks;
 	dir->hashversion = 0;
-	dir->casefolded_hash = inode->i_flags & EXT4_CASEFOLD_FL;
+	dir->casefolded_hash = !!(inode->i_flags & EXT4_CASEFOLD_FL);
 	dir->dx_block = e2fsck_allocate_memory(ctx, num_blocks
 				       * sizeof (struct dx_dirblock_info),
 				       "dx_block info array");
-
 }
 
 /*
diff --git a/e2fsck/e2fsck.h b/e2fsck/e2fsck.h
index 5e7db42..feb605c 100644
--- a/e2fsck/e2fsck.h
+++ b/e2fsck/e2fsck.h
@@ -104,12 +104,12 @@ struct dir_info {
  * directories which contain a hash tree index.
  */
 struct dx_dir_info {
-	ext2_ino_t		ino; 		/* Inode number */
-	int			numblocks;	/* number of blocks */
-	int			hashversion;
-	short			depth;		/* depth of tree */
-	struct dx_dirblock_info	*dx_block; 	/* Array of size numblocks */
-	int			casefolded_hash;
+	ext2_ino_t		ino;		/* Inode number */
+	short			depth;		/* depth of tree (15 bits) */
+	__u8			hashversion;
+	__u8			casefolded_hash:1;
+	blk_t			numblocks;	/* number of blocks in dir */
+	struct dx_dirblock_info	*dx_block;	/* Array of size numblocks */
 };
 
 #define DX_DIRBLOCK_ROOT	1
@@ -120,8 +120,8 @@ struct dx_dir_info {
 
 struct dx_dirblock_info {
 	int		type;
-	blk64_t		phys;
 	int		flags;
+	blk64_t		phys;
 	blk64_t		parent;
 	blk64_t		previous;
 	ext2_dirhash_t	min_hash;
diff --git a/e2fsck/pass2.c b/e2fsck/pass2.c
index 5c3f7b8..0fa6233 100644
--- a/e2fsck/pass2.c
+++ b/e2fsck/pass2.c
@@ -71,8 +71,8 @@ static int allocate_dir_block(e2fsck_t ctx,
 			      struct ext2_db_entry2 *dir_blocks_info,
 			      char *buf, struct problem_context *pctx);
 static void clear_htree(e2fsck_t ctx, ext2_ino_t ino);
-static int htree_depth(struct dx_dir_info *dx_dir,
-		       struct dx_dirblock_info *dx_db);
+static short htree_depth(struct dx_dir_info *dx_dir,
+			 struct dx_dirblock_info *dx_db);
 static EXT2_QSORT_TYPE special_dir_block_cmp(const void *a, const void *b);
 
 struct check_dir_struct {
@@ -132,7 +132,7 @@ void e2fsck_pass2(e2fsck_t ctx)
 	struct dx_dirblock_info	*dx_db;
 	int			b;
 	ext2_ino_t		i;
-	int			depth;
+	short			depth;
 	problem_t		code;
 	int			bad_dir;
 	int (*check_dir_func)(ext2_filsys fs,
@@ -311,10 +311,10 @@ cleanup:
 }
 
 #define MAX_DEPTH 32000
-static int htree_depth(struct dx_dir_info *dx_dir,
-		       struct dx_dirblock_info *dx_db)
+static short htree_depth(struct dx_dir_info *dx_dir,
+			 struct dx_dirblock_info *dx_db)
 {
-	int	depth = 0;
+	short depth = 0;
 
 	while (dx_db->type != DX_DIRBLOCK_ROOT && depth < MAX_DEPTH) {
 		dx_db = &dx_dir->dx_block[dx_db->parent];
-- 
1.8.0


  parent reply	other threads:[~2020-02-07  1:17 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-07  1:09 [PATCH 1/9] e2fsck: fix e2fsck_allocate_memory() overflow Andreas Dilger
2020-02-07  1:09 ` [PATCH 2/9] e2fsck: use proper types for variables Andreas Dilger
2020-02-29 23:27   ` Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 3/9] e2fsck: avoid mallinfo() if over 2GB allocated Andreas Dilger
2020-02-29 23:28   ` Theodore Y. Ts'o
2020-02-07  1:09 ` Andreas Dilger [this message]
2020-02-29 23:29   ` [PATCH 4/9] e2fsck: reduce memory usage for many directories Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 5/9] debugfs: allow comment lines in command file Andreas Dilger
2020-02-29 23:32   ` Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 6/9] debugfs: print inode numbers as unsigned Andreas Dilger
2020-02-29 23:34   ` Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 7/9] e2fsck: fix overflow if more than 4B inodes Andreas Dilger
2020-02-29 23:35   ` Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 8/9] e2fsck: consistently use ext2fs_get_mem() Andreas Dilger
2020-02-29 23:36   ` Theodore Y. Ts'o
2020-03-04 23:23     ` Theodore Y. Ts'o
2020-02-07  1:09 ` [PATCH 9/9] misc: handle very large files with filefrag Andreas Dilger
2020-03-04 23:27   ` Theodore Y. Ts'o
2020-02-12  0:58 ` [PATCH] " Andreas Dilger
2020-02-12  1:09   ` Andreas Dilger
2020-02-12  1:07 ` [PATCH] e2fsck: avoid overflow with very large dirs Andreas Dilger
2020-03-04 23:39   ` Theodore Y. Ts'o
2020-02-29 23:25 ` [PATCH 1/9] e2fsck: fix e2fsck_allocate_memory() overflow Theodore Y. Ts'o

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1581037786-62789-4-git-send-email-adilger@whamcloud.com \
    --to=adilger@whamcloud.com \
    --cc=linux-ext4@vger.kernel.org \
    --cc=tytso@mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).