Re: Re: nvme-tcp: fix a possible UAF when failing to send request

["zhang.guanghui@cestc.cn" <zhang.guanghui@cestc.cn>]

Hi 
     int fact,  the commit aeacfcefa218f4ed11da478e9b7915a37d1afaff  has been synchronized, but the issue still occurs.

    I think so, when the host failing to send request -13,   maybe only part of the command has been sent or  the controller has receved  a complete command.
but the controller may send nvme_tcp_rsp  and  C2HTermReq consecutively.  Now I am analyzing the controller spdk processing logic. 
when the host uses nvme_tcp_poll,  it receives nvme_tcp_rsp and handles it firstly,    
then the host  have a UAF and crashed.   




zhang.guanghui@cestc.cn



 



From: Maurizio Lombardi



Date: 2025-02-13 00:07



To: Maurizio Lombardi; zhang.guanghui@cestc.cn; sagi; mgurtovoy; kbusch; sashal; chunguang.xu



CC: linux-kernel; linux-nvme; linux-block



Subject: Re: nvme-tcp: fix a possible UAF when failing to send request



On Wed Feb 12, 2025 at 4:33 PM CET, Maurizio Lombardi wrote:



>



> Taking a step back. Let's take a different approach and try to avoid the



> double completion.



>



> The problem here is that apparently we received a nvme_tcp_rsp capsule



> from the target, meaning that the command has been processed (I guess



> the capsule has an error status?)



>



> So maybe only part of the command has been sent?



> Why we receive the rsp capsule at all? Shouldn't this be treated as a fatal



> error by the controller?  

--


 



 



> The NVMe/TCP specification says



 



******



When a controller detects a fatal error, that controller shall:



  1. stop processing any PDUs that arrive on the connection; and



  2. send a C2HTermReq PDU



******



 



And indeed I see in the dmesg this:



 



nvme nvme2: unsupported pdu type (3)



 



This means the controller detected the problem and sent to the host the



C2HTermReq command. Upon receiving this command, the host is supposed to



close the connection.



 



Now I get it.



 



Zhang, do you have commit aeacfcefa218f4ed11da478e9b7915a37d1afaff in



your kernel, I guess you are missing it. Check it please.



 



Maurizio