linux-ext4 - [RFC PATCH v1 08/50] fs/ext4/ialloc.c: Replace % with reciprocal

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <202003281643.02SGh9vH007105@sdf.org>
Date:   Mon, 18 Mar 2019 21:32:20 -0400
From:   George Spelvin <lkml@....org>
To:     linux-kernel@...r.kernel.org, lkml@....org
Cc:     "Theodore Ts'o" <tytso@....edu>,
        Andreas Dilger <adilger.kernel@...ger.ca>,
        linux-ext4@...r.kernel.org
Subject: [RFC PATCH v1 08/50] fs/ext4/ialloc.c: Replace % with reciprocal_scale() TO BE VERIFIED

This came about as part of auditing prandom_u32() usage, but
this is a special case: sometimes the starting value comes
from prandom_u32, and sometimes it comes from a hash of a name.

Does the name hash algorithm have to be stable? In that case, this
change would alter it.  But it appears to use s_hash_seed which
is regenerated on "e2fsck -D", so maybe changing it isn't a big deal.

Feedback needed.

Signed-off-by: George Spelvin <lkml@....org>
Cc: "Theodore Ts'o" <tytso@....edu>
Cc: Andreas Dilger <adilger.kernel@...ger.ca>
Cc: linux-ext4@...r.kernel.org
---
 fs/ext4/ialloc.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c
index 7db0c8814f2ec..a4ea89b3ed368 100644
--- a/fs/ext4/ialloc.c
+++ b/fs/ext4/ialloc.c
@@ -457,9 +457,8 @@ static int find_group_orlov(struct super_block *sb, struct inode *parent,
 			grp = hinfo.hash;
 		} else
 			grp = prandom_u32();
-		parent_group = (unsigned)grp % ngroups;
-		for (i = 0; i < ngroups; i++) {
-			g = (parent_group + i) % ngroups;
+		g = parent_group = reciprocal_scale(grp, ngroups);
+		for (i = 0; i < ngroups; i++, ++g == ngroups && (g = 0)) {
 			get_orlov_stats(sb, g, flex_size, &stats);
 			if (!stats.free_inodes)
 				continue;
-- 
2.26.0