SPI support for intel i82580

[Michael Stickel]

Hi,

we have a added and tested support for the intel i82580 to the nicintel_spi driver.
Is anyone else working on that topic or should I post the patch. This could conflict with the i350 patch http://patchwork.coreboot.org/patch/4394/.
Should our patch be based on a clear 0.9.9 release or on a 0.9.9 + i350 patch.

Background: the intel i82580 quad ethernet controller has two memories. One 16KB or 32KB spi-eeprom for MAC address and other configurations and one 512KB spi-flash for the firmware.
the spi-eeprom can be written with the driver nicintel_eeprom. We made the nessesary modifications in the nicintel_spi driver based on the 0.9.9 release.

best regards
Michael

[Stefan Tauner]

On Wed, 23 Mar 2016 09:01:48 +0100 (CET)
Michael Stickel  wrote:
> Hi,
>
> we have a added and tested support for the intel i82580 to the nicintel_spi driver.
> Is anyone else working on that topic or should I post the patch. This could conflict with the i350 patch http://patchwork.coreboot.org/patch/4394/.
> Should our patch be based on a clear 0.9.9 release or on a 0.9.9 + i350 patch.
>
> Background: the intel i82580 quad ethernet controller has two memories. One 16KB or 32KB spi-eeprom for MAC address and other configurations and one 512KB spi-flash for the firmware.
> the spi-eeprom can be written with the driver nicintel_eeprom. We made the nessesary modifications in the nicintel_spi driver based on the 0.9.9 release.

Hello Michael,

we have no idea if the i350 patch really works and we won't merge it
till someone has tested it. Thus your approach of basing it on the last
commit (i.e. on the 0.9.9 release) is just fine. The conflicts can't be
too bad anyway since the i350 patch is not large at all. Thanks in
advance for the patch!

[Michael Stickel]

Hi Stefan,
> > we have a added and tested support for the intel i82580 to the nicintel_spi driver.
> > Is anyone else working on that topic or should I post the patch. This could conflict with the i350 patch http://patchwork.coreboot.org/patch/4394/.
> > Should our patch be based on a clear 0.9.9 release or on a 0.9.9 + i350 patch.
>
> we have no idea if the i350 patch really works and we won't merge it
> till someone has tested it. Thus your approach of basing it on the last
> commit (i.e. on the 0.9.9 release) is just fine. The conflicts can't be
> too bad anyway since the i350 patch is not large at all. Thanks in
> advance for the patch!

find the patch attached. You will see, that it's impact is as small as with the i350 patch. It's just the id's and the offset to the flash registers.

I have introduced a small helper function called nicintel_spi_get_flash_register_offset. The purpose is to make the code more readable and to move the offset more closely to the definition of the pci-devices.

We have found out some things that might be interesting for others. Is there a place where such things can be put, e.g. a wiki?
- with an unprogrammed eeprom the i82580 detects the eeprom as 16KB, even if the eeprom is 32KB.
- if the eeprom of flash is not programmed (has device-id 1509), the igb driver does not recognize the card and does not enable it. The device must be enable via sysfs prior to programming (e.g. echo 1 > /sys/bus/pci/devices/0000:01:00.0/enable). Only one device needs to be programmed, as they all share the same eeprom.

Any comments to the patch are welcome.

best regards
Michael