linux-kernel - Re: [PATCH] netpoll: WARN_ONCE for start_xmit returns with interrupts enabled

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]

Message-Id: <1250765040.17891.91.camel@dengdd-desktop>
Date:	Thu, 20 Aug 2009 18:44:00 +0800
From:	DDD <Dongdong.deng@...driver.com>
To:	Matt Mackall <mpm@...enic.com>
Cc:	linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...e.hu>,
	David Miller <davem@...emloft.net>, netdev@...r.kernel.org
Subject: Re: [PATCH] netpoll: WARN_ONCE for start_xmit returns with
	interrupts enabled

On Wed, 2009-08-19 at 05:43 -0500, Matt Mackall wrote: 
> On Wed, 2009-08-19 at 18:43 +0800, DDD wrote:
> > The NETPOLL API requires that interrupts remain disabled in
> > netpoll_send_skb().
> > 
> > Add a WARN_ONCE when ndo_start_xmit returns with interrupts enabled
> > in netpoll_send_skb().
> > 
> > Signed-off-by: Dongdong Deng <dongdong.deng@...driver.com>
> > ---
> >  net/core/netpoll.c |    6 ++++++
> >  1 files changed, 6 insertions(+), 0 deletions(-)
> > 
> > diff --git a/net/core/netpoll.c b/net/core/netpoll.c
> > index df30feb..d38103f 100644
> > --- a/net/core/netpoll.c
> > +++ b/net/core/netpoll.c
> > @@ -319,6 +319,12 @@ static void netpoll_send_skb(struct netpoll *np, struct sk_buff *skb)
> >  
> >  			udelay(USEC_PER_POLL);
> >  		}
> > +
> > +		WARN_ONCE(!irqs_disabled(),
> > +			KERN_WARNING "netpoll_send_skb(): [net driver %s]"
> > +			"ndo_start_xmit() shouldn't return with interrupts"
> > +			"enabled!", dev->name);
> 
> Hmm, perhaps:
> 
> "netpoll_send_skb(): %s enabled interrupts in poll (%pF)\n",
> dev->name, ops->ndo_start_xmit
> 
> That'll give us a real function name and maybe even a module name.

Thanks, I will modify it as your suggestions.

When I test the WRAN_ONCE, I find that some net drivers still have this problem,
such as using local_irq_disable/local_irq_enable in ndo_start_xmit() or other cases.


I will do an new patch and hope it could fix this problem completely in all the net drivers that support NETPOLL.


I check the drivers that support NETPOLL one by one again(almost 100 drivers),
and following drivers's fix will be included by my patch:

1: ixp2000/ixpdev.c : 66 local_irq_disable

2: fec_mpc52xx.c : 319 spin_unlock_irq

3: macb.c:602 : 638 spin_unlock_irq

4: mlx4/en_netdev.c : 448 
mlx4_en_xmit() -> mlx4_en_xmit_poll() -> spin_trylock_irq

5: smc91x.c 
It seems that there is a lots work to let smc91x work well with netpoll.


Thanks,
Dongdong

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