netdev - Re: NOHZ: local_softirq

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: <1256308311.12174.38.camel@johannes.local>
Date:	Fri, 23 Oct 2009 16:31:51 +0200
From:	Johannes Berg <johannes@...solutions.net>
To:	Tilman Schmidt <tilman@...p.cc>
Cc:	Jarek Poplawski <jarkao2@...il.com>,
	David Miller <davem@...emloft.net>, hidave.darkstar@...il.com,
	linux-kernel@...r.kernel.org, tglx@...utronix.de,
	linux-wireless@...r.kernel.org, linux-ppp@...r.kernel.org,
	netdev@...r.kernel.org, paulus@...ba.org,
	isdn4linux <isdn4linux@...tserv.isdn4linux.de>,
	i4ldeveloper <i4ldeveloper@...tserv.isdn4linux.de>,
	Karsten Keil <isdn@...ux-pingi.de>
Subject: Re: NOHZ: local_softirq_pending 08

On Fri, 2009-10-23 at 16:27 +0200, Tilman Schmidt wrote:

> >> --- a/drivers/isdn/i4l/isdn_ppp.c
> >> +++ b/drivers/isdn/i4l/isdn_ppp.c
> >> @@ -1174,7 +1174,10 @@ isdn_ppp_push_higher(isdn_net_dev * net_dev, isdn_net_local * lp, struct sk_buff
> >>  #endif /* CONFIG_IPPP_FILTER */
> >>  	skb->dev = dev;
> >>  	skb_reset_mac_header(skb);
> >> -	netif_rx(skb);
> >> +	if (in_interrupt())
> >> +		netif_rx(skb);
> >> +	else
> >> +		netif_rx_ni(skb);
> > 
> > So you've verified that the entire i4l stack can cope with being called
> > twice on the same CPU, from different contexts?
> 
> What makes you think so?

I thought I'd explained this in my other email. *sigh*

You're squelching a warning, which comes from the fact that you're
calling something that calls into netif_rx() with softirqs enabled. That
would seem to imply that potentially a softirq could at the same time
call into that code too.

Basically, what happens now is this:

disable softirqs
call into i4l/ppp
i4l/ppp code
call netif_rx()
more code
enable softirqs


You're changing it to

call into i4l/ppp
i4l/ppp code
call netif_rx_ni()
more code

so the entire chunks "i4l/ppp code" and "more code" are now no longer
protected against being interrupted by a softirq that runs the same
code, maybe for a different device, on the same CPU.

johannes

Download attachment "signature.asc" of type "application/pgp-signature" (802 bytes)