linux-kernel - Re: [PATCH wireless] wifi: wilc1000: unregister wiphy only if it has been registered

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: <87sepur6mo.fsf@kernel.org>
Date: Tue, 07 Jan 2025 17:22:23 +0200
From: Kalle Valo <kvalo@...nel.org>
To: Marek Vasut <marex@...x.de>
Cc: Alexis Lothoré <alexis.lothore@...tlin.com>,  Ajay
 Singh
 <ajay.kathat@...rochip.com>,  Claudiu Beznea <claudiu.beznea@...on.dev>,
  Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
  linux-wireless@...r.kernel.org,  linux-kernel@...r.kernel.org
Subject: Re: [PATCH wireless] wifi: wilc1000: unregister wiphy only if it
 has been registered

Marek Vasut <marex@...x.de> writes:

> On 12/23/24 4:46 PM, Alexis Lothoré wrote:
>
>> There is a specific error path in probe functions in wilc drivers (both
>> sdio and spi) which can lead to kernel panic, as this one for example
>> when using SPI:
>> Unable to handle kernel paging request at virtual address 9f000000
>> when read
>> [9f000000] *pgd=00000000
>> Internal error: Oops: 5 [#1] ARM
>> Modules linked in: wilc1000_spi(+) crc_itu_t crc7 wilc1000 cfg80211 bluetooth ecdh_generic ecc
>> CPU: 0 UID: 0 PID: 106 Comm: modprobe Not tainted 6.13.0-rc3+ #22
>> Hardware name: Atmel SAMA5
>> PC is at wiphy_unregister+0x244/0xc40 [cfg80211]
>> LR is at wiphy_unregister+0x1c0/0xc40 [cfg80211]
>> [...]
>>   wiphy_unregister [cfg80211] from wilc_netdev_cleanup+0x380/0x494 [wilc1000]
>>   wilc_netdev_cleanup [wilc1000] from wilc_bus_probe+0x360/0x834 [wilc1000_spi]
>>   wilc_bus_probe [wilc1000_spi] from spi_probe+0x15c/0x1d4
>>   spi_probe from really_probe+0x270/0xb2c
>>   really_probe from __driver_probe_device+0x1dc/0x4e8
>>   __driver_probe_device from driver_probe_device+0x5c/0x140
>>   driver_probe_device from __driver_attach+0x220/0x540
>>   __driver_attach from bus_for_each_dev+0x13c/0x1a8
>>   bus_for_each_dev from bus_add_driver+0x2a0/0x6a4
>>   bus_add_driver from driver_register+0x27c/0x51c
>>   driver_register from do_one_initcall+0xf8/0x564
>>   do_one_initcall from do_init_module+0x2e4/0x82c
>>   do_init_module from load_module+0x59a0/0x70c4
>>   load_module from init_module_from_file+0x100/0x148
>>   init_module_from_file from sys_finit_module+0x2fc/0x924
>>   sys_finit_module from ret_fast_syscall+0x0/0x1c
>> The issue can easily be reproduced, for example by not wiring
>> correctly
>> a wilc device through SPI (and so, make it unresponsive to early SPI
>> commands). It is due to a recent change decoupling wiphy allocation from
>> wiphy registration, however wilc_netdev_cleanup has not been updated
>> accordingly, letting it possibly call wiphy unregister on a wiphy which
>> has never been registered.
>> Fix this crash by moving wiphy_unregister/wiphy_free out of
>> wilc_netdev_cleanup, and by adjusting error paths in both drivers
>> Fixes: fbdf0c5248dc ("wifi: wilc1000: Register wiphy after reading
>> out chipid")
>> Signed-off-by: Alexis Lothoré <alexis.lothore@...tlin.com>
> Nice find, thank you for fixing it.
>
> Reviewed-by: Marek Vasut <marex@...x.de>

We are quite late in the cycle so I think I'll queue this to
wireless-next for v6.14.

-- 
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches