Patches 1 is a preparation patch.
Patch 2 is the actual change to piix4.

Patch 1 drops the warning about muxed busses.
I didn't feel that the warning only would warrant the additional
complexity it introduces with multiple callers of i2c_register_spd().

If others feel different, maybe a more generic warning/info or a source
code comment would suffice.

On a machine with 32 slots of which 16 are populated only the first 8
slots are addressable, half of which are empty.
Unfortunately I couldn't run a custom kernel for testing.
But manually instantiating ee1004 devices worked as expected,
so the proposed changes should also work.

Tested with spd5118 and two DIMMs.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
---
Changes in v3:
- Apply tags from Guenter
- Squash commits 1-3, and reword the message slightly
- Drop Fixes:
- Link to v2: https://lore.kernel.org/r/20240627-piix4-spd-v2-0-617ce47b8ff4@weissschuh.net

Changes in v2:
- Add new i2c-smbus prep patches
- Only register SPD for port 0
- Link to v1: https://lore.kernel.org/r/20240530-piix4-spd-v1-1-9cbf1abebf41@weissschuh.net

---
Thomas Weißschuh (2):
      i2c: smbus: remove i801 assumptions from SPD probing
      i2c: piix4: Register SPDs

 drivers/i2c/busses/Kconfig     |  1 +
 drivers/i2c/busses/i2c-piix4.c |  4 ++++
 drivers/i2c/i2c-smbus.c        | 15 ++++-----------
 3 files changed, 9 insertions(+), 11 deletions(-)
---
base-commit: 9746c2dd0307c80bd695e4e3065367f3e0154723
change-id: 20240530-piix4-spd-39c156b22959

Best regards,
-- 
Thomas Weißschuh <linux@weissschuh.net>

Hi Thomas,

On Tue, Jul 09, 2024 at 07:35:34PM GMT, Thomas Weißschuh wrote:
> Patches 1 is a preparation patch.
> Patch 2 is the actual change to piix4.
> 
> Patch 1 drops the warning about muxed busses.
> I didn't feel that the warning only would warrant the additional
> complexity it introduces with multiple callers of i2c_register_spd().
> 
> If others feel different, maybe a more generic warning/info or a source
> code comment would suffice.
> 
> On a machine with 32 slots of which 16 are populated only the first 8
> slots are addressable, half of which are empty.
> Unfortunately I couldn't run a custom kernel for testing.
> But manually instantiating ee1004 devices worked as expected,
> so the proposed changes should also work.
> 
> Tested with spd5118 and two DIMMs.
> 
> Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

with the comment added, pushed to

i2c/i2c-host.

Thanks,
Andi

Hi Wolfram,

> Thomas Weißschuh (2):
>       i2c: smbus: remove i801 assumptions from SPD probing

this is yours...

>       i2c: piix4: Register SPDs

... this is mine :-)

Do you want me to pick them both up?

Thanks,
Andi

On Fri, Jul 12, 2024 at 02:02:11AM +0200, Andi Shyti wrote:
> Hi Wolfram,
> 
> > Thomas Weißschuh (2):
> >       i2c: smbus: remove i801 assumptions from SPD probing
> 
> this is yours...
> 
> >       i2c: piix4: Register SPDs
> 
> ... this is mine :-)
> 
> Do you want me to pick them both up?

I'd  really like to have feedback here from Heiner and/or Jean before we
apply this...