[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <129a670d-14b5-4f9e-b01b-0a88a0017741@gmail.com>
Date: Sat, 7 Jun 2025 18:24:53 +0200
From: Heiner Kallweit <hkallweit1@...il.com>
To: Claudiu Beznea <claudiu.beznea@...on.dev>,
Abin Joseph <abin.joseph@....com>, nicolas.ferre@...rochip.com,
andrew+netdev@...n.ch, davem@...emloft.net, edumazet@...gle.com,
kuba@...nel.org, pabeni@...hat.com
Cc: git@....com, netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH net-next v2] net: macb: Add shutdown operation support
On 07.06.2025 14:28, Claudiu Beznea wrote:
> Hi, Abin,
>
> On 07.06.2025 10:57, Abin Joseph wrote:
>> Implement the shutdown hook to ensure clean and complete deactivation of
>> MACB controller. The shutdown sequence is protected with 'rtnl_lock()'
>> to serialize access and prevent race conditions while detaching and
>> closing the network device. This ensure a safe transition when the Kexec
>> utility calls the shutdown hook, facilitating seamless loading and
>> booting of a new kernel from the currently running one.
>>
>> Signed-off-by: Abin Joseph <abin.joseph@....com>
>> ---
>>
>> Changes in v2:
>> Update the commit description
>> Update the code to call the close only when admin is up
>>
>> ---
>> drivers/net/ethernet/cadence/macb_main.c | 14 ++++++++++++++
>> 1 file changed, 14 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/cadence/macb_main.c b/drivers/net/ethernet/cadence/macb_main.c
>> index e1e8bd2ec155..5bb08f518d54 100644
>> --- a/drivers/net/ethernet/cadence/macb_main.c
>> +++ b/drivers/net/ethernet/cadence/macb_main.c
>> @@ -5650,6 +5650,19 @@ static int __maybe_unused macb_runtime_resume(struct device *dev)
>> return 0;
>> }
>>
>> +static void macb_shutdown(struct platform_device *pdev)
>> +{
>> + struct net_device *netdev = platform_get_drvdata(pdev);
>> +
>> + rtnl_lock();
>> + netif_device_detach(netdev);
>> +
>> + if (netif_running(netdev))
>> + dev_close(netdev);
>
> Apart from closing the interface should we also invoke the remove?
>
As part of closing the interface dev_close() sends notifiers.
I don't think that's something we want during shutdown.
Typically the shutdown() callback just has to quiesce the device.
> Consider the new loaded kernel will not use this interface and some
> hardware resources may remain on while not used by the newly loaded kernel.
>
> E.g., there is this code in macb_remove():
>
> if (!pm_runtime_suspended(&pdev->dev)) {
> macb_clks_disable(bp->pclk, bp->hclk, bp->tx_clk,
> bp->rx_clk, bp->tsu_clk);
> pm_runtime_set_suspended(&pdev->dev);
> }
>
>
> Thank you,
> Claudiu
>
>> +
>> + rtnl_unlock();
>> +}
>> +
>> static const struct dev_pm_ops macb_pm_ops = {
>> SET_SYSTEM_SLEEP_PM_OPS(macb_suspend, macb_resume)
>> SET_RUNTIME_PM_OPS(macb_runtime_suspend, macb_runtime_resume, NULL)
>> @@ -5663,6 +5676,7 @@ static struct platform_driver macb_driver = {
>> .of_match_table = of_match_ptr(macb_dt_ids),
>> .pm = &macb_pm_ops,
>> },
>> + .shutdown = macb_shutdown,
>> };
>>
>> module_platform_driver(macb_driver);
>
>
Powered by blists - more mailing lists