[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <95288bb1-8931-4d18-b8f6-25a4f6148afe@intel.com>
Date: Mon, 27 Jan 2025 09:11:41 +0100
From: Przemek Kitszel <przemyslaw.kitszel@...el.com>
To: "Singh, Krishneil K" <krishneil.k.singh@...el.com>,
"intel-wired-lan@...ts.osuosl.org" <intel-wired-lan@...ts.osuosl.org>, "Emil
Tantilov" <emil.s.tantilov@...el.com>, Jakub Kicinski <kuba@...nel.org>
CC: "netdev@...r.kernel.org" <netdev@...r.kernel.org>, Ahmed Zaki
<ahmed.zaki@...el.com>
Subject: Re: [PATCH iwl-net v1 0/4] fix locking issue
On 1/27/25 07:06, Singh, Krishneil K wrote:
>
>> -----Original Message-----
>> From: Tarun K Singh <tarun.k.singh@...el.com>
>> Sent: Tuesday, November 5, 2024 10:49 AM
>> To: intel-wired-lan@...ts.osuosl.org
>> Cc: netdev@...r.kernel.org
>> Subject: [PATCH iwl-net v1 0/4] fix locking issue
>>
>> This series fix deadlock issues in the driver. The first patch changes
>> argument of function 'idpf_vport_ctrl_lock' to adapter. The second patch
>> renames 'vport_ctrl_lock' to 'vport_cfg_lock'. The first 2 patches make the
>> third patch easier to review. The third patch fixes the locking issue,
>> and the fourth patch prevents lockdep from complaining.
>>
>> Ahmed Zaki (1):
>> idpf: add lock class key
>>
>> Tarun K Singh (3):
>> idpf: Change function argument
>> idpf: rename vport_ctrl_lock
>> idpf: Add init, reinit, and deinit control lock
>>
>> drivers/net/ethernet/intel/idpf/idpf.h | 49 ++++++++----
>> .../net/ethernet/intel/idpf/idpf_ethtool.c | 59 +++++++-------
>> drivers/net/ethernet/intel/idpf/idpf_lib.c | 80 +++++++++++++------
>> drivers/net/ethernet/intel/idpf/idpf_main.c | 39 ++++++---
>> 4 files changed, 149 insertions(+), 78 deletions(-)
>>
>> --
>> 2.46.0
>>
>
> After additional testing on this [series|patch] we discovered an issue after doing ifdown while device is in a reset, which can result in a <null pointer dereference(?)>.
Thank you for the report,
Could you please check if this issue could be reproduced without the
series?
We have yet another patch by Emil, that aims to fix such issues by
detaching the device prior to the reset.
@netdev
I would like to consider adding "in reset" state for the netdev,
to prevent such behavior in more civilized way though.
Powered by blists - more mailing lists