linux-kernel - Re: [PATCH 1/2] stmmac: introduce flag to dynamically disable TX offload for rockchip devices

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: <c863f88a-65e2-4093-7cc1-f37e4443b871@arm.com>
Date:   Wed, 3 Apr 2019 17:12:03 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     "Leonidas P. Papadakos" <papadakospan@...il.com>,
        Jose Abreu <jose.abreu@...opsys.com>
Cc:     Philipp Tomsich <philipp.tomsich@...obroma-systems.com>,
        Heiko Stübner <heiko@...ech.de>,
        Christoph Müllner 
        <christoph.muellner@...obroma-systems.com>,
        Maxime Coquelin <mcoquelin.stm32@...il.com>,
        Alexandre Torgue <alexandre.torgue@...com>,
        netdev@...r.kernel.org, LKML <linux-kernel@...r.kernel.org>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Klaus Goger <klaus.goger@...obroma-systems.com>
Subject: Re: [PATCH 1/2] stmmac: introduce flag to dynamically disable TX
 offload for rockchip devices

On 03/04/2019 16:55, Leonidas P. Papadakos wrote:
> 
>> If snps,no-pbl-x8 does indeed have a performace hit, should the 
>> workaround be to
>> a) turn tx checksumming off or
>> b) snps,no-pbl-x8 set?
>>
>> I'll test to see the difference
> 
> Update: It really seems like snps,no-pbl-x8 is the better option.
> So I say, replace snps,force_thresh_dma_mode with it.

Yes, I would expect software checksumming to have a much more noticeable 
impact (in fact I've already been trying to get round to benchmarking 
some arm64 checksum optimisations on my RK3328 precisely because of this 
issue).

If I'm interpreting the register descriptions in the Rockchip TRMs 
correctly, it seems like no-pbl-x8 is a relatively big hammer and there 
should still be room to tune things a bit closer to the maximum limits - 
I'll have another play this evening to see if I've understood things right.

Robin.