netdev - Re: [PATCH net] igb: fix deadlock caused by taking RTNL in RPM resume path

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <5675a5ef-5aa0-3f05-1c44-a91ce90d5f38@gmail.com>
Date:   Tue, 30 Nov 2021 07:33:55 +0100
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Stephen Hemminger <stephen@...workplumber.org>
Cc:     Jakub Kicinski <kuba@...nel.org>,
        David Miller <davem@...emloft.net>,
        Jesse Brandeburg <jesse.brandeburg@...el.com>,
        Tony Nguyen <anthony.l.nguyen@...el.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        intel-wired-lan <intel-wired-lan@...ts.osuosl.org>
Subject: Re: [PATCH net] igb: fix deadlock caused by taking RTNL in RPM resume
 path

On 30.11.2021 00:09, Stephen Hemminger wrote:
> On Mon, 29 Nov 2021 22:14:06 +0100
> Heiner Kallweit <hkallweit1@...il.com> wrote:
> 
>> diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
>> index dd208930f..8073cce73 100644
>> --- a/drivers/net/ethernet/intel/igb/igb_main.c
>> +++ b/drivers/net/ethernet/intel/igb/igb_main.c
>> @@ -9254,7 +9254,7 @@ static int __maybe_unused igb_suspend(struct device *dev)
>>  	return __igb_shutdown(to_pci_dev(dev), NULL, 0);
>>  }
>>  
>> -static int __maybe_unused igb_resume(struct device *dev)
>> +static int __maybe_unused __igb_resume(struct device *dev, bool rpm)
>>  {
>>  	struct pci_dev *pdev = to_pci_dev(dev);
>>  	struct net_device *netdev = pci_get_drvdata(pdev);
>> @@ -9297,17 +9297,24 @@ static int __maybe_unused igb_resume(struct device *dev)
>>  
>>  	wr32(E1000_WUS, ~0);
>>  
>> -	rtnl_lock();
>> +	if (!rpm)
>> +		rtnl_lock();
>>  	if (!err && netif_running(netdev))
>>  		err = __igb_open(netdev, true);
>>  
>>  	if (!err)
>>  		netif_device_attach(netdev);
>> -	rtnl_unlock();
>> +	if (!rpm)
>> +		rtnl_unlock();
>>  
>>  	return err;
>>  }
>>  
>> +static int __maybe_unused igb_resume(struct device *dev)
>> +{
>> +	return __igb_resume(dev, false);
>> +}
>> +
>>  static int __maybe_unused igb_runtime_idle(struct device *dev)
>>  {
>>  	struct net_device *netdev = dev_get_drvdata(dev);
>> @@ -9326,7 +9333,7 @@ static int __maybe_unused igb_runtime_suspend(struct device *dev)
>>  
>>  static int __maybe_unused igb_runtime_resume(struct device *dev)
>>  {
>> -	return igb_resume(dev);
>> +	return __igb_resume(dev, true);
>>  }
> 
> Rather than conditional locking which is one of the seven deadly sins of SMP,
> why not just have __igb_resume() be the locked version where lock is held by caller?
> 
In this case we'd have to duplicate quite some code from igb_resume().
Even more simple alternative would be to remove RTNL from igb_resume().
Then we'd remove RTNL from RPM and system resume path. Should be ok as well.
I just didn't want to change two paths at once.