netdev - Re: [PATCH v3 iproute] ip: Add support for netdev events to monitor

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: <20170409.144721.342305587683511172.davem@davemloft.net>
Date:   Sun, 09 Apr 2017 14:47:21 -0700 (PDT)
From:   David Miller <davem@...emloft.net>
To:     dsa@...ulusnetworks.com
Cc:     vyasevic@...hat.com, netdev@...r.kernel.org,
        roopa@...ulusnetworks.com
Subject: Re: [PATCH v3 iproute] ip: Add support for netdev events to monitor

From: David Ahern <dsa@...ulusnetworks.com>
Date: Sat, 8 Apr 2017 22:54:07 -0400

> On 4/8/17 10:33 PM, David Miller wrote:
>> From: David Ahern <dsa@...ulusnetworks.com>
>> Date: Sat, 8 Apr 2017 18:24:06 -0400
>> 
>>> per comments on the email thread about reducing notifications, the
>>> kernel patch for this should be reverted (and hence this iproute2 patch
>>> is not needed) in favor of using a bitmask. Right now there are too many
>>> redundant notifications to userspace.
>> 
>> I must have missed something in all the discussion, which patch needs
>> to be reverted from my tree exactly?
>> 
> 
> Here's the thread:
> https://www.spinics.net/lists/netdev/msg429154.html
> 
> The comment is that def12888c161 is adding a uapi that leads to way too
> many notifications (e.g., on a setlink).
> 
> It would be more efficient (read less notifications) to have do_setlink
> emit a single message with the IFLA_EVENT (or something else
> appropriately named) that indicates what attributes changed. Right now,
> a change MTU leads to 3 notifications causing unnecessary churn in
> userspace to track what the state of the link is.

Ok, I'll queue up the revert.  I guess this means your rtnetlink
patch set is going to need changes or a respin, therefore.