netdev - Re: r8169: is the work queue is initialized at wrong place?

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: <2387149.B9WVEQm2aY@al>
Date:	Fri, 19 Jul 2013 00:41:57 +0200
From:	Peter Wu <lekensteyn@...il.com>
To:	Francois Romieu <romieu@...zoreil.com>
Cc:	netdev@...r.kernel.org,
	Realtek linux nic maintainers <nic_swsd@...ltek.com>
Subject: Re: r8169: is the work queue is initialized at wrong place?

On Thursday 18 July 2013 23:53:43 Francois Romieu wrote:
> Peter Wu <lekensteyn@...il.com> :
> [...]
> 
> > However, this queue is only initialized in rtl_open:
> > 
> > 6673         INIT_WORK(&tp->wk.work, rtl_task);
> > 6674
> > 6675         smp_mb();
> > 
> > Shouldn't this INIT_WORK be done in rtl_init_one ?
> 
> (or cancel_work_sync in rtl8169_close to reduce the scope)
Will this also work with multiple adapters? I currently have an on-board chip 
using the r8169 driver and a separate PCI card.

> I do not see how the current code could hurt but it's really ugly.
When I googled for the warning, I found some hint about something not being 
initialized. From that I guessed that rtl_open is never called when the 
network interface is not brought up (which seems to match the documentation of 
struct net_device_ops[1]).

> > I do not know what this smp_mb is used for in this context, so I leave
> > an appropriate patch up to you.
> 
> It does not cost much and it enforces an initialized state. IRQ - real
> or shared ones - and napi poll could happen really soon after this point.
> There is probably some implicit memory barrier the driver could rely on.
> It would be a bit hackish though.
>
> I'll check the history and cook a patch.

Thanks.

Regards,
Peter

 [1]: http://lxr.free-electrons.com/source/include/linux/netdevice.h#L716
--
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