netdev - RE: question about eth

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] [day] [month] [year] [list]

Date:	Tue, 20 Nov 2012 17:39:30 +0000
From:	"Dmitry Kravkov" <dmitry@...adcom.com>
To:	"Rami Rosen" <roszenrami@...il.com>
cc:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: RE: question about eth_header

> -----Original Message-----
> From: netdev-owner@...r.kernel.org [mailto:netdev-owner@...r.kernel.org]
> On Behalf Of Rami Rosen
> Sent: Tuesday, November 20, 2012 4:51 PM
> To: Dmitry Kravkov
> Cc: netdev@...r.kernel.org
> Subject: Re: question about eth_header
> 
> Hi,
> I think that you should call skb_reset_mac_header()
> before calling eth_hdr() and that skb_reset_mac_header() should not be
> inside eth_hdr(). Following is explanation:
> 
> In the RX path, when we get a packet in the driver, what essentially
> happens is that at this point, we have a pointer to
> the packet buffer in skb->data, and its length in skb->len.

Actually I'm on TX path, and do not intend  to update skb, just received it 
for transmission, I want to pick mac addresess from eth header.
Now bnx2x does it by accessing skb->data (which points to the correct mac)
and I wanted to replace it with eth_hrd instead of doing ugly casting in driver code.
Moreover, in the future, when inner_mac_address will be used for tunneling
acceleration, inner_mac_address will be updated with mac_address offset, and so
may also point to the improper location.   
 
> With ethernet packets, the driver will call eth_type_trans(); please
> look at ethernet drivers code.
> eth_type_trans() indeed sets the MAC header pointer by calling
> skb_reset_mac_header() on the skb.
> 
> Afterwards, the eth_type_trans() will advance skb->data by
> 14, the size of the Ethernet header, and decrease skb->len
> by calling skb_pull_inline(skb, ETH_HLEN).
> 
> In case you are wondering why the skb_reset_mac_header() is not inside
> the eth_hdr(), I think that the reason is that, once
> skb_reset_mac_header() on the skb was called, there are cases when we
> want to make several calls eth_hdr() without each time again resetting
> the mac header, for example, in the bridging code.  There is no reason
> to doing it in terms of performance.

My question is about eth_header() function which is create() callback for
eth_header_ops.

Thanks.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html