netdev - Re: [RFC 1/3] mdio-bitbang: add SMI0 mode support

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, 10 May 2019 14:36:31 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     Michael Grzeschik <mgr@...gutronix.de>
Cc:     Tristram.Ha@...rochip.com, kernel@...gutronix.de,
        UNGLinuxDriver@...rochip.com, netdev@...r.kernel.org
Subject: Re: [RFC 1/3] mdio-bitbang: add SMI0 mode support

> > > +/* Serial Management Interface (SMI) uses the following frame format:
> > > + *
> > > + *       preamble|start|Read/Write|  PHY   |  REG  |TA|   Data bits      | Idle
> > > + *               |frame| OP code  |address |address|  |                  |
> > > + * read | 32x1´s | 01  |    00    | 1xRRR  | RRRRR |Z0| 00000000DDDDDDDD |  Z
> > > + * write| 32x1´s | 01  |    00    | 0xRRR  | RRRRR |10| xxxxxxxxDDDDDDDD |  Z
> > > + *
> > > + * The register number is encoded with the 5 least significant bits in REG
> > > + * and the 3 most significant bits in PHY
> > > + */
> > > +#define MII_ADDR_SMI0 (1<<31)
> > > +
> > 
> > Michael
> > 
> > This is a Micrel Proprietary protocol. So we should reflect this in
> > the name. MII_ADDR_MICREL_SMI? Why the 0? Are there different
> > versions? Maybe replace all SMI0 with MICREL_SMI in mdio-bitbang.c
> 
> There are two variants of the SMI interface.

Hi Michael

O.K, that explains the 0.

> 
> The KSZ8863/73/93 Products use the above Variant described as "SMI0".
> 
> The KSZ8864/95 Products use another layout:
> 
>       preamble|start|Read/Write|  PHY   |  REG  |TA|   Data bits      | Idle
>               |frame| OP code  |address |address|  |                  |
> read | 32x1´s | 01  |    10    | RR11R  | RRRRR |Z0| 00000000DDDDDDDD |  Z
> write| 32x1´s | 01  |    01    | RR11R  | RRRRR |10| xxxxxxxxDDDDDDDD |  Z
> 
> So they describe their write/read operation in the OP code rather then
> the PHY address.

At a first look, i think a standard MDIO bus controller can do this?
If so, we don't need a second define, just some code in the switch
driver which shuffles bits around.

> 
> We could change the SMI index to SMI_KSZ88X3 for the current SMI0 to
> give it a more descriptive name.

That seems sensible. In the mv88e6xxx driver, we name things based on
the first device to introduce the feature.

    Andrew