RE: [PATCH 1/1] Drivers: hv: Implement the file copy service

[KY Srinivasan]

> -----Original Message-----
> From: Dan Carpenter [mailto:dan.carpenter@xxxxxxxxxx]
> Sent: Thursday, January 09, 2014 12:09 PM
> To: KY Srinivasan
> Cc: olaf@xxxxxxxxx; gregkh@xxxxxxxxxxxxxxxxxxx; jasowang@xxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; apw@xxxxxxxxxxxxx; devel@xxxxxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH 1/1] Drivers: hv: Implement the file copy service
> 
> We've had this discussion before where you urge me to trust the host...

I am just implementing the protocol specification given by the host. If I cannot trust the
specified protocol, I am not sure what else can be done here. 

> 
>  Problem: This code is racy.
> Solution: The host will only send one message at a time.

The code is not racy given the protocol that is specified. While I could have 
blindly trusted the host, this driver code actually reads only one packet at a time.
When we get a transaction from the host, we do not process any more transactions until
the current transaction is fully processed. So what is the race condition here?

> 
> Now I have to audit the user space code on the host and I don't feel
> like doing that so you win.

I don't think you need to audit any code.

> 
> I wish we had a better way to do IPC.  If kdbus were ready, that might
> have worked for this, and it's a better solution because both sender and
> reciever code will be written in a less trusting way.

I am not sure how kdbus would help you here. We are talking about communicating
between the host and the guest here. 

K. Y
> 
> regards,
> dan carpenter

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/