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-prev] [day] [month] [year] [list]
Message-Id: <20230127132202.6083-1-konishi.ryusuke@gmail.com>
Date:   Fri, 27 Jan 2023 22:22:02 +0900
From:   Ryusuke Konishi <konishi.ryusuke@...il.com>
To:     Andrew Morton <akpm@...ux-foundation.org>
Cc:     linux-nilfs <linux-nilfs@...r.kernel.org>,
        syzbot <syzbot+cbff7a52b6f99059e67f@...kaller.appspotmail.com>,
        syzkaller-bugs@...glegroups.com,
        LKML <linux-kernel@...r.kernel.org>
Subject: [PATCH] nilfs2: prevent WARNING in nilfs_dat_commit_end()

If nilfs2 reads a corrupted disk image and its DAT metadata file contains
invalid lifetime data for a virtual block number, a kernel warning can be
generated by the WARN_ON check in nilfs_dat_commit_end() and can panic
if the kernel is booted with panic_on_warn.

This patch avoids the issue with a sanity check that treats it as an
error.

Since error return is not allowed in the execution phase of
nilfs_dat_commit_end(), this inserts that sanity check in
nilfs_dat_prepare_end(), which prepares for nilfs_dat_commit_end().

As the error code, -EINVAL is returned to notify bmap layer of the
metadata corruption.  When the bmap layer sees this code, it handles the
abnormal situation and replaces the return code with -EIO as it should.

Link: https://lkml.kernel.org/r/000000000000154d2c05e9ec7df6@google.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@...il.com>
Reported-by: syzbot+cbff7a52b6f99059e67f@...kaller.appspotmail.com
Tested-by: Ryusuke Konishi <konishi.ryusuke@...il.com>
---
Andrew, please add this patch to the queue.  This fixes another
WARN_ON hit in fs/nilfs2/dat.c for a corrupted disk image pattern.

Thanks,
Ryusuke Konishi

fs/nilfs2/dat.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/fs/nilfs2/dat.c b/fs/nilfs2/dat.c
index 1e7f653c1df7..9cf6ba58f585 100644
--- a/fs/nilfs2/dat.c
+++ b/fs/nilfs2/dat.c
@@ -158,6 +158,7 @@ void nilfs_dat_commit_start(struct inode *dat, struct nilfs_palloc_req *req,
 int nilfs_dat_prepare_end(struct inode *dat, struct nilfs_palloc_req *req)
 {
 	struct nilfs_dat_entry *entry;
+	__u64 start;
 	sector_t blocknr;
 	void *kaddr;
 	int ret;
@@ -169,6 +170,7 @@ int nilfs_dat_prepare_end(struct inode *dat, struct nilfs_palloc_req *req)
 	kaddr = kmap_atomic(req->pr_entry_bh->b_page);
 	entry = nilfs_palloc_block_get_entry(dat, req->pr_entry_nr,
 					     req->pr_entry_bh, kaddr);
+	start = le64_to_cpu(entry->de_start);
 	blocknr = le64_to_cpu(entry->de_blocknr);
 	kunmap_atomic(kaddr);
 
@@ -179,6 +181,15 @@ int nilfs_dat_prepare_end(struct inode *dat, struct nilfs_palloc_req *req)
 			return ret;
 		}
 	}
+	if (unlikely(start > nilfs_mdt_cno(dat))) {
+		nilfs_err(dat->i_sb,
+			  "vblocknr = %llu has abnormal lifetime: start cno (= %llu) > current cno (= %llu)",
+			  (unsigned long long)req->pr_entry_nr,
+			  (unsigned long long)start,
+			  (unsigned long long)nilfs_mdt_cno(dat));
+		nilfs_dat_abort_entry(dat, req);
+		return -EINVAL;
+	}
 
 	return 0;
 }
-- 
2.34.1

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ