On Thu, Jan 14, 2021 at 05:43:27AM +0100, Marek Behún wrote:
> Add SFP_PASSWORD = 0x7b to the diagnostics enumerator. This address is
> described in SFF-8436 and SFF-8636.
However, as I mentioned previously, these are not relevant for SFPs,
but are for QSFPs, and QSFPs have a totally different layout.
The data structure for this enum is described by SFF-8472, and it lists
Vendor Specific Locations [Address A2h, Bytes 120-126]
This is what the SFP_VSL definition covers.
SFF-8436 defines the layout for "QSFP+ 10 Gbs 4X PLUGGABLE TRANSCEIVER":
Table 17 - Lower Memory Map (A0h)
Address Description Type Passive Copper,
Optical
Active Copper,
Module Active Optical
0 Identifier (1 Read-Only R R
Byte)
1-2 Status (2 Bytes) Read-Only See Table 18
3-21 Interrupt Flags Read-Only See Tables 19-21
(19 Bytes)
...
123-126 Password Entry Read/Write O O
Area (optional) 4
Bytes
127 Page Select Byte Read/Write R R
So, SFF-8436 is not relevant - not only is it for a different address
but the whole format is different.
SFF-8636 is "Specification for Management Interface for Cabled
Environments" and defines:
2-Wire Serial Address 1010000x
Lower Page 00h
0 Identifier
1- 2 Status
3- 21 Interrupt Flags
...
123-126 Password Entry Area (Optional)
127 Page Select Byte
So again, SFF-8636 is not relevant for the same reasons as SFF-8436.
We're left with SFF-8472, which mentions no password entry area, but
leaves the area from 120-127 open as "vendor specific", so I don't
think it is appropriate to start adding vendor specific definitions as
if they were generic to sfp.h
I would instead suggest defining this inside mdio-i2c.c as
#define ROLLERBALL_PASSWORD (SFP_VSL + 3)
rather than trying to make this appear generic.
--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!
