netdev - Re: [PATCH] net: usb: ax88179_178a: fix packet alignment padding

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: <20200616.135535.379478681934951754.davem@davemloft.net>
Date:   Tue, 16 Jun 2020 13:55:35 -0700 (PDT)
From:   David Miller <davem@...emloft.net>
To:     jk@...abs.org
Cc:     netdev@...r.kernel.org, allan@...x.com.tw, freddy@...x.com.tw,
        pfink@...ist-es.de, linux-usb@...r.kernel.org, louis@...x.com.tw
Subject: Re: [PATCH] net: usb: ax88179_178a: fix packet alignment padding

From: Jeremy Kerr <jk@...abs.org>
Date: Tue, 16 Jun 2020 17:08:23 +0800

>> Because that code in this loop makes the same calculations:
>> 
>>                 ax_skb = skb_clone(skb, GFP_ATOMIC);
>>                 if (ax_skb) {
>>                         ax_skb->len = pkt_len;
>>                         ax_skb->data = skb->data + 2;
>>                         skb_set_tail_pointer(ax_skb, pkt_len);
>>                         ax_skb->truesize = pkt_len + sizeof(struct sk_buff);
>>                         ax88179_rx_checksum(ax_skb, pkt_hdr);
>>                         usbnet_skb_return(dev, ax_skb);
>> 
>> So if your change is right, it should be applied to this code block
>> as well.
> 
> Yep, my patch changes that block too (or did I miss something?)

Nope, you didn't miss anything.  I missed that completely.

>> And do we know that it's two extra tail bytes always?  Or some kind
>> of alignment padding the chip performs for every sub-packet?
> 
> I've assumed it's a constant two bytes of prefix padding, as that's all
> I've seen. But it would be great to have more detail from ASIX if
> possible.

I'll wait a bit for the ASIX folks to comment.

It seems logical to me that what the chip does is align up the total
sub-packet length to a multiple of 4 or larger, and then add those two
prefix padding bytes.  Otherwise the prefix padding won't necessarily
and reliably align the IP header after the link level header.

Thanks.