netdev - Re: [net-next PATCH v6 06/10] netdev-genl: Add netlink framework functions for napi

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 for Android: free password hash cracker in your pocket

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

Date: Tue, 24 Oct 2023 19:18:11 -0700
From: "Nambiar, Amritha" <amritha.nambiar@...el.com>
To: Jakub Kicinski <kuba@...nel.org>
CC: <netdev@...r.kernel.org>, <pabeni@...hat.com>,
	<sridhar.samudrala@...el.com>
Subject: Re: [net-next PATCH v6 06/10] netdev-genl: Add netlink framework
 functions for napi

On 10/24/2023 3:39 PM, Jakub Kicinski wrote:
> On Mon, 23 Oct 2023 18:33:50 -0700 Amritha Nambiar wrote:
>> +	rcu_read_lock();
>> +
>> +	napi = napi_by_id(napi_id);
>> +	if (napi)
>> +		err = netdev_nl_napi_fill_one(rsp, napi, info);
>> +	else
>> +		err = -EINVAL;
>> +
>> +	rcu_read_unlock();
> 
> Is rcu_read_lock always going to be sufficient here?
> Reading of the thread, for example, without much locking could
> potentially get problematic.

I think I was inspired by the napi_busy_loop code which works on 
napi->state under rcu_read_lock.

WRT napi_get_doit here (just thinking aloud), so if we have a reader 
thread retrieving napi from the global list for a napi_id, and a writer 
thread (maybe ethtool -L sort which is under rtnl_lock) that changes the 
napi list... Now, the reader may be working on a stale napi instance, or 
if the writer still holds lock until freeing memory containing the napi, 
then the reader may cause a crash... okay, all the more reason for the 
reader to hold rtnl_lock instead of rcu_read_lock.