netdev - Re: [net-next.git 3/8 (V2)] stmmac: add the initial tx coalesce schema

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: <1347577875.13258.70.camel@deadeye.wl.decadent.org.uk>
Date:	Fri, 14 Sep 2012 00:11:15 +0100
From:	Ben Hutchings <bhutchings@...arflare.com>
To:	David Miller <davem@...emloft.net>
CC:	<peppe.cavallaro@...com>, <netdev@...r.kernel.org>
Subject: Re: [net-next.git 3/8 (V2)] stmmac: add the initial tx coalesce
 schema

On Thu, 2012-09-13 at 17:37 -0400, David Miller wrote:
> From: Ben Hutchings <bhutchings@...arflare.com>
> Date: Thu, 13 Sep 2012 22:10:50 +0100
> 
> > On Thu, 2012-09-13 at 16:46 -0400, David Miller wrote:
> >> From: Ben Hutchings <bhutchings@...arflare.com>
> >> Date: Thu, 13 Sep 2012 21:42:51 +0100
> >> 
> >> Well written NAPI drivers never need to disable hardware interrupts
> >> in their ->poll() method and it's callers, neither should you.
> > 
> > Perhaps you should get round to reviewing netpoll, because it does
> > exactly this.
> 
> Then I don't understand the point you're trying to make.
> 
> Hardware interrupt disabling has absolutely no place in the
> NAPI polling fast paths.
> 
> If NAPI drivers can't be implemented without hardware interrupt
> toggling in ->poll(), we've failed.

Right.

The problem being that NAPI poll functions *are* sometimes called in
hardware interrupt context.  Thus, any spinlock that may be taken by a
NAPI handler, may well need to be taken with spinlock_irq or
spinlock_irqsave elsewhere.  (This is horrible and I think it's well
past time that we ripped the NAPI polling out of netpoll.)

I think you're right that stmmac_tx() (completion handler?) doesn't need
to disable hardware interrupts, but sadly stmmac_xmit() does right now
unless Giuseppe can work out how to make their interaction lockless.

Ben.

-- 
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

--
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