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]
Date:   Fri, 4 Feb 2022 10:50:35 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Cc:     Yannick Vignon <yannick.vignon@....nxp.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Giuseppe Cavallaro <peppe.cavallaro@...com>,
        Alexandre Torgue <alexandre.torgue@...s.st.com>,
        Jose Abreu <joabreu@...opsys.com>,
        "David S. Miller" <davem@...emloft.net>,
        Maxime Coquelin <mcoquelin.stm32@...il.com>,
        Antoine Tenart <atenart@...nel.org>,
        Alexander Lobakin <alexandr.lobakin@...el.com>,
        Paolo Abeni <pabeni@...hat.com>, Wei Wang <weiwan@...gle.com>,
        Kumar Kartikeya Dwivedi <memxor@...il.com>,
        Yunsheng Lin <linyunsheng@...wei.com>,
        Arnd Bergmann <arnd@...db.de>, netdev <netdev@...r.kernel.org>,
        Vladimir Oltean <olteanv@...il.com>,
        Xiaoliang Yang <xiaoliang.yang_1@....com>, mingkai.hu@....com,
        Joakim Zhang <qiangqing.zhang@....com>,
        sebastien.laveze@....com, Yannick Vignon <yannick.vignon@....com>
Subject: Re: [PATCH net-next 1/2] net: napi: wake up ksoftirqd if needed
 after scheduling NAPI

On Fri, 4 Feb 2022 19:03:31 +0100 Sebastian Andrzej Siewior wrote:
> On 2022-02-04 09:45:22 [-0800], Jakub Kicinski wrote:
> > Coincidentally, I believe the threaded NAPI wake up is buggy - 
> > we assume the thread is only woken up when NAPI gets scheduled,
> > but IIUC signal delivery and other rare paths may wake up kthreads,
> > randomly.  
> 
> I had to look into NAPI-threads for some reason.
> What I dislike is that after enabling it via sysfs I have to:
> - adjust task priority manual so it is preferred over other threads.
>   This is usually important on RT. But then there is no overload
>   protection.
> 
> - set an affinity-mask for the thread so it does not migrate from one
>   CPU to the other. This is worse for a RT task where the scheduler
>   tries to keep the task running.
> 
> Wouldn't it work to utilize the threaded-IRQ API and use that instead
> the custom thread? Basically the primary handler would what it already
> does (disable the interrupt) and the threaded handler would feed packets
> into the stack. In the overload case one would need to lower the
> thread-priority.

Sounds like an interesting direction if you ask me! That said I have
not been able to make threaded NAPI useful in my experiments / with my
workloads so I'd defer to Wei for confirmation.

To be clear -- are you suggesting that drivers just switch to threaded
NAPI, or a more dynamic approach where echo 1 > /proc/irq/$n/threaded
dynamically engages a thread in a generic fashion?

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ