From: "Rodríguez Barbarin, José Javier" <JoseJavier.Rodriguez@duagon.com>
To: "gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
"jirislaby@kernel.org" <jirislaby@kernel.org>
Cc: "morbidrsa@gmail.com" <morbidrsa@gmail.com>,
"linux-serial@vger.kernel.org" <linux-serial@vger.kernel.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"jth@kernel.org" <jth@kernel.org>,
"Sanjuán García, Jorge" <Jorge.SanjuanGarcia@duagon.com>,
"Rodríguez Barbarin,
José Javier" <JoseJavier.Rodriguez@duagon.com>
Subject: [PATCH v2 0/1] mcb: Fix crash mcb-core module is removed
Date: Fri, 18 Aug 2023 11:02:54 +0000 [thread overview]
Message-ID: <20230818110230.186700-1-JoseJavier.Rodriguez@duagon.com> (raw)
In-Reply-To: <2023071044-deduct-cycle-a918@gregkh>
When allocating a new mcb_bus the bus_type is added to the mcb_bus itself,
causing an issue when calling mcb_bus_add_devices(). This function is not
only called for each mcb_device under the mcb_bus but for the bus itself.
The crash happens when the mcb_core module is removed, getting
the following error:
[ 286.691693] ------------[ cut here ]------------
[ 286.691695] ida_free called for id=1 which is not allocated.
[ 286.691714] WARNING: CPU: 0 PID: 1719 at lib/idr.c:523 ida_free+0xe0/0x140
[ 286.691715] Modules linked in: snd_hda_codec_hdmi amd64_edac_mod snd_hda_intel edac_mce_amd snd_intel_dspcfg kvm_amd snd_hda_codec amdgpu nls_iso8859_1 ccp snd_hda_core snd_hwdep amd_iommu_v2 kvm snd_pcm gpu_sched crct10dif_pclmul crc32_pclmul snd_seq_midi snd_seq_midi_event ghash_clmulni_intel ttm snd_rawmidi aesni_intel snd_seq binfmt_misc crypto_simd cryptd glue_helper drm_kms_helper snd_seq_device snd_timer drm snd k10temp fb_sys_fops syscopyarea sysfillrect sysimgblt snd_rn_pci_acp3x mcb_pci(-) snd_pci_acp3x soundcore altera_cvp fpga_mgr mcb spi_nor mtd 8250_dw mac_hid sch_fq_codel parport_pc ppdev lp parport ip_tables x_tables autofs4 mmc_block nvme ahci i2c_piix4 libahci i2c_amd_mp2_pci igb nvme_core i2c_algo_bit dca video sdhci_acpi sdhci [last unloaded: 8250_men_mcb]
[ 286.691752] CPU: 0 PID: 1719 Comm: modprobe Not tainted 5.4.702+ #11
[ 286.691753] Hardware name: MEN F027/n/a, BIOS 1.03 04/20/2021
[ 286.691756] RIP: 0010:ida_free+0xe0/0x140
[ 286.691759] Code: a8 31 f6 e8 12 f7 00 00 eb 4b 4c 0f a3 28 72 21 48 8b 7d a8 4c 89 f6 e8 8e ad 02 00 89 de 48 c7 c7 e8 02 83 b5 e8 b0 7a 5d ff <0f> 0b e9 67 ff ff ff 4c 0f b3 28 48 8d 7d a8 31 f6 e8 da e0 00 00
[ 286.691761] RSP: 0018:ffff9a56c38f7bd8 EFLAGS: 00010282
[ 286.691763] RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000006
[ 286.691764] RDX: 0000000000000007 RSI: 0000000000000096 RDI: ffff8d881fa1c8c0
[ 286.691765] RBP: ffff9a56c38f7c30 R08: 0000000000000487 R09: 0000000000000004
[ 286.691766] R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000001
[ 286.691767] R13: 0000000000000001 R14: 0000000000000202 R15: 0000000000000001
[ 286.691769] FS: 00007fb78e303540(0000) GS:ffff8d881fa00000(0000) knlGS:0000000000000000
[ 286.691770] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 286.691771] CR2: 00007ffe92b2ce98 CR3: 000000079fd9c000 CR4: 00000000003406f0
[ 286.691772] Call Trace:
[ 286.691781] mcb_free_bus+0x2b/0x40 [mcb]
[ 286.691785] device_release+0x2c/0x80
[ 286.691787] kobject_put+0xb9/0x1d0
[ 286.691790] put_device+0x13/0x20
As mcb_bus_add_devices() is called for the mcb_bus itself, the function
tries to cast the incorrectly passed struct mcb_bus to mcb_device. Both
structs have the same layout:
struct mcb_bus {
struct device dev;
struct device *carrier;
int bus_nr;
...
};
struct mcb_device {
struct device dev;
struct mcb_bus *bus;
bool is_added;
...
};
This incorrect casting is causing a wrong behaviour in
mcb_bus_add_devices() where the member bus_nr is casted to is_added,
meaning that when bus_nr is "0", the function continues and sets bus_nr
to "1" (is_added = true)
If we have 2 buses (one for each F215 board), the function ida_alloc()
will give the value "0" and "1" to each bus respectively, but as both
buses are included themselves in the devices' lists, after the call to
mcb_bus_add_devices(), the buses will have the value "1" and "1". For
this reason, when the mcb-core module is removed, the error raises as
the ida resource with value "1" is being released twice, leaking
the ida resource with value "0".
changes for V2:
* create a dedicated bus_type for mcb_bus and mcb_device structs
instead of removing bus_type for mcb_bus.
This patch is based on linux-next (next-20230817)
Jose Javier Rodriguez Barbarin (1):
mcb: create dedicated bus_type for mcb_bus and mcb_device
drivers/mcb/mcb-core.c | 43 ++++++++++++++++++++++++++++++++++++------
1 file changed, 37 insertions(+), 6 deletions(-)
--
2.34.1
next parent reply other threads:[~2023-08-18 11:04 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <2023071044-deduct-cycle-a918@gregkh>
2023-08-18 11:02 ` Rodríguez Barbarin, José Javier [this message]
2023-08-18 11:03 ` [PATCH v2 1/1] mcb: create dedicated bus_type for mcb_bus and mcb_device Rodríguez Barbarin, José Javier
2023-08-19 9:56 ` gregkh
2023-08-23 8:07 ` Sanjuán García, Jorge
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=20230818110230.186700-1-JoseJavier.Rodriguez@duagon.com \
--to=josejavier.rodriguez@duagon.com \
--cc=Jorge.SanjuanGarcia@duagon.com \
--cc=gregkh@linuxfoundation.org \
--cc=jirislaby@kernel.org \
--cc=jth@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-serial@vger.kernel.org \
--cc=morbidrsa@gmail.com \
/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