netdev
[Top] [All Lists]

Re: [PATCH][ATM]: [clip] fix race between modifying entry->vccs and clip

To: Mitchell Blank Jr <mitch@xxxxxxxxxx>
Subject: Re: [PATCH][ATM]: [clip] fix race between modifying entry->vccs and clip_start_xmit()
From: chas williams <chas@xxxxxxxxxxxxxxxx>
Date: Tue, 16 Sep 2003 06:59:56 -0400
Cc: "David S. Miller" <davem@xxxxxxxxxx>, romieu@xxxxxxxxxxxxx, netdev@xxxxxxxxxxx
In-reply-to: Message from Mitchell Blank Jr <mitch@xxxxxxxxxx> of "Mon, 15 Sep 2003 15:30:48 PDT." <20030915223048.GD92642@xxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
In message <20030915223048.GD92642@xxxxxxxxxxxxxx>,Mitchell Blank Jr writes:
>That should be the case at least for any VCC on a real interface[*].
>...
>[*] some protocols use psuedo-interfaces for their control connections - in
>    theory I guess they could be different but I don't believe there are
>    any cases where they are.  They wouldn't affect this issue anyways though.

afaict all vcc's are open/closed from user space including any control
vccs (often labelled ATM_VF_META).  this is a bit of a sore point with
some people who want to open atm sockets in the kernel.  if i remember
correctly the atm api explicitly states that you can sleep in atm_dev->close()
so it must be a user context.

<Prev in Thread] Current Thread [Next in Thread>