linux-kernel - Re: Synopsys Ethernet QoS Driver

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <937252db-9538-2cf6-c8fa-82b558531c51@st.com>
Date:   Mon, 21 Nov 2016 15:36:58 +0100
From:   Giuseppe CAVALLARO <peppe.cavallaro@...com>
To:     Joao Pinto <Joao.Pinto@...opsys.com>,
        Rayagond Kokatanur <rayagond@...avyalabs.com>,
        Rabin Vincent <rabin@....in>
CC:     mued dib <kreptor@...il.com>, David Miller <davem@...emloft.net>,
        Jeff Kirsher <jeffrey.t.kirsher@...el.com>,
        <jiri@...lanox.com>, <saeedm@...lanox.com>, <idosch@...lanox.com>,
        netdev <netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <CARLOS.PALMINHA@...opsys.com>, <andreas.irestal@...s.com>,
        <alexandre.torgue@...com>, <lars.persson@...s.com>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>
Subject: Re: Synopsys Ethernet QoS Driver

Hello Joao

On 11/21/2016 2:48 PM, Joao Pinto wrote:
> Synopsys QoS IP is a separated hardware component, so it should be reusable by
> all implementations using it and so have its own "core driver" and platform +
> pci glue drivers. This is necessary for example in hardware validation, where
> you prototype an IP and instantiate its drivers and test it.
>
> Was there a strong reason to integrate QoS features directly in stmmac and not
> in synopsys/dwc_eth_qos.*?

We decided to enhance the stmmac on supporting the QoS for several
reasons; for example the common APIs that the driver already exposed and
actually suitable for other SYNP chips. Then, PTP, EEE,
S/RGMII, MMC could be shared among different chips with a minimal
effort.  This meant a lot of code already ready.

For sure, the net-core, Ethtool, mdio parts were reused. Same for the
glue logic files.
For the latter, this helped to easily bring-up new platforms also
because the stmmac uses the HW cap register to auto-configure many
parts of the MAC core, DMA and modules. This helped many users, AFAIK.

For validation purpose, this is my experience, the stmmac helped
a lot because people used the same code to validate different HW
and it was easy to switch to a platform to another one in order to
verify / check if the support was ok or if a regression was introduced.
This is important for complex supports like PTP or EEE.

Hoping this can help.

Do not hesitate to contact me for further details

peppe