Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.21\))
Subject: Re: bug with large_dir in 5.12.17
From:   =?utf-8?B?0JHQu9Cw0LPQvtC00LDRgNC10L3QutC+INCQ0YDRgtGR0Lw=?= 
        <artem.blagodarenko@gmail.com>
In-Reply-To: <YPl/boTCfc3rlJLU@fisica.ufpr.br>
Date:   Thu, 29 Jul 2021 22:23:35 +0300
Cc:     linux-ext4@vger.kernel.org, Theodore Tso <tytso@google.com>,
        Andreas Dilger <adilger@dilger.ca>
Content-Transfer-Encoding: quoted-printable
Message-Id: <5FE9762B-6C6B-4A44-AC99-22192B76C060@gmail.com>
References: <YPl/boTCfc3rlJLU@fisica.ufpr.br>
To:     Carlos Carvalho <carlos@fisica.ufpr.br>
Precedence: bulk

Hello,

It looks like the fix b5776e7524afbd4569978ff790864755c438bba7 "ext4: =
fix potential htree index checksum corruption=E2=80=9D introduced this =
regression.
I reverted it and my test from previous message passed the dangerous =
level of 1570000 names count.
Now test is still in progress. 2520000 names are already created.

I am searching the way to fix this.

Best regards,
Artem Blagodarenko.

> On 22 Jul 2021, at 17:23, Carlos Carvalho <carlos@fisica.ufpr.br> =
wrote:
>=20
> There is a bug when enabling large_dir in 5.12.17. I got this during a =
backup:
>=20
> index full, reach max htree level :2
> Large directory feature is not enabled on this filesystem
>=20
> So I unmounted, ran tune2fs -O large_dir /dev/device and mounted =
again. However
> this error appeared:
>=20
> dx_probe:864: inode #576594294: block 144245: comm rsync: directory =
leaf block found instead of index block
>=20
> I unmounted, ran fsck and it "salvaged" a bunch of directories. =
However at the
> next backup run the same errors appeared again.
>=20
> This is with vanilla 5.2.17.