netdev - RE: [PATCH] net/hyperv: fix erroneous NETDEV_TX

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]

Date:	Wed, 14 Mar 2012 20:00:46 +0000
From:	Haiyang Zhang <haiyangz@...rosoft.com>
To:	Eric Dumazet <eric.dumazet@...il.com>,
	Ben Greear <greearb@...delatech.com>
CC:	David Miller <davem@...emloft.net>,
	netdev <netdev@...r.kernel.org>,
	KY Srinivasan <kys@...rosoft.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: RE: [PATCH] net/hyperv: fix erroneous NETDEV_TX_BUSY use



> -----Original Message-----
> From: Eric Dumazet [mailto:eric.dumazet@...il.com]
> Sent: Wednesday, March 14, 2012 3:17 PM
> To: Ben Greear
> Cc: David Miller; netdev; KY Srinivasan; Haiyang Zhang; Greg Kroah-Hartman
> Subject: Re: [PATCH] net/hyperv: fix erroneous NETDEV_TX_BUSY use
> 
> On Wed, 2012-03-14 at 11:56 -0700, Ben Greear wrote:
> > On 03/14/2012 11:53 AM, Eric Dumazet wrote:
> > > A driver start_xmit() method cannot free skb and return
> NETDEV_TX_BUSY,
> > > since caller is going to reuse freed skb.
> > >
> > > This is mostly a revert of commit bf769375c (staging: hv: fix the
> return
> > > status of netvsc_start_xmit())
> > >
> > > In fact netif_tx_stop_queue() / netif_stop_queue() is needed before
> > > returning NETDEV_TX_BUSY or you can trigger a ksoftirqd fatal loop.
> > >
> > > In case of memory allocation error, only safe way is to drop the
> packet
> > > and return NETDEV_TX_OK
> >
> > Can you not just leave the skb alone, not bump any tx-dropped
> > stats, and return NETDEV_TX_BUSY?
> >
> 
> Nope. Think about OOM.
> 
> If we do that, we requeue packet on qdisc, and schedule TX softirq.
> 
> -> loop
> 
> Really, NETDEV_TX_BUSY is not for this kind of situation.
> 

Actually, I'm working on this bug, and have a patch under our internal 
review now. My fix is actually not freeing the SKB, and return NETDEV_TX_BUSY.

Even if the vmbus ring buffer is busy currently, it can be available during
the next re-try, because the host is taking away data from ring buffer. The 
stop/wake queue mechanism is in the netvsc.c file.

In the out-of-memory case, dropping packet and return NETDEV_TX_OK seems fine.

Thanks,
- Haiyang