Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <6c75a90d-3793-55fc-e525-eca14e562350@gmail.com>
Date:   Tue, 27 Sep 2022 17:22:46 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.13.0
Subject: Re: [dm-devel] [PATCH] kernfs: fix a crash when two processes delete
 the same directory
Content-Language: en-US
To:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Mikulas Patocka <mpatocka@redhat.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        linux-kernel@vger.kernel.org, dm-devel@redhat.com,
        Christoph Lameter <cl@linux.com>
References: <alpine.LRH.2.02.2209260418360.16612@file01.intranet.prod.int.rdu2.redhat.com>
 <YzGYFBv0pdt+DQg+@kroah.com>
From:   Milan Broz <gmazyland@gmail.com>
In-Reply-To: <YzGYFBv0pdt+DQg+@kroah.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 9/26/22 14:16, Greg Kroah-Hartman wrote:
> On Mon, Sep 26, 2022 at 07:04:52AM -0400, Mikulas Patocka wrote:
>> There is a crash when running the cryptsetup testsuite on Fedora Rawhide.
>> It can be reproduced by installing Rawhide with the 6.0-rc6 kernel,
>> downloading and compiling the cryptsetup repository and running this test
>> in a loop for about 15 minuts:
>> 	while ./integrity-compat-test; do :; done
>>
>>   ------------[ cut here ]------------
>>   WARNING: CPU: 0 PID: 50087 at fs/kernfs/dir.c:504 __kernfs_remove.part.0+0x26f/0x2b0
>>   Modules linked in: crc32_generic loop dm_integrity async_xor async_tx tls isofs uinput snd_seq_dummy snd_hrtimer nft_objref nf_conntrack_netbios_ns nf_conntrack_broadcast nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 rfkill ip_set nf_tables nfnetlink qrtr sunrpc snd_hda_codec_generic ledtrig_audio snd_hda_intel iTCO_wdt snd_intel_dspcfg intel_pmc_bxt snd_intel_sdw_acpi iTCO_vendor_support snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device joydev snd_pcm i2c_i801 snd_timer pcspkr i2c_smbus virtio_balloon snd lpc_ich soundcore zram virtio_net net_failover virtio_blk serio_raw failover qxl virtio_console drm_ttm_helper ttm ip6_tables ip_tables fuse qemu_fw_cfg
>>   Unloaded tainted modules: crc32_pclmul():1 pcc_cpufreq():1 pcc_cpufreq():1 acpi_cpufreq():1 edac_mce_amd():1 pcc_cpufreq():1 acpi_cpufreq():1 edac_mce_amd():1 acpi_cpufreq():1 edac_mce_amd():1 pcc_cpufreq():1 acpi_cpufreq():1 pcc_cpufreq():1 edac_mce_amd():1 edac_mce_amd():1 acpi_cpufreq():1 pcc_cpufreq():1 edac_mce_amd():1 acpi_cpufreq():1 pcc_cpufreq():1 edac_mce_amd():1 acpi_cpufreq():1 pcc_cpufreq():1 edac_mce_amd():1 acpi_cpufreq():1 edac_mce_amd():1 pcc_cpufreq():1 edac_mce_amd():1 acpi_cpufreq():1 pcc_cpufreq():1 edac_mce_amd():1 pcc_cpufreq():1 acpi_cpufreq():1 edac_mce_amd():1 pcc_cpufreq():1 acpi_cpufreq():1 acpi_cpufreq():1
>>   CPU: 0 PID: 50087 Comm: integritysetup Not tainted 6.0.0-0.rc6.41.fc38.x86_64 #1
>>   Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.14.0-2 04/01/2014
>>   RIP: 0010:__kernfs_remove.part.0+0x26f/0x2b0

...

> Can you see if 4abc99652812 ("kernfs: fix use-after-free in
> __kernfs_remove") in linux-next fixes this for you or not?  It seems to
> be the same issue, as was also reported at:
> 	https://lore.kernel.org/r/7f489b14-2fdc-3d91-c87e-6a802bd8592d@I-love.SAKURA.ne.jp


I tried it on system where cryptsetup testsuite almost immediately crashed in the integrity test.

With the patch in https://lore.kernel.org/r/7f489b14-2fdc-3d91-c87e-6a802bd8592d@I-love.SAKURA.ne.jp
it now iterates for some time without any problems, so I think it is fixed.

Tested-by: Milan Broz <gmazyland@gmail.com>

Thanks,
Milan