Mailing List Archive

[PATCH v2] i2c: smbus: Add (LP)DDR5 types to `i2c_register_spd()`
On several systems Linux logs:

i2c i2c-0: Memory type 0x22 not supported yet, not instantiating SPD

1. Supermicro Super Server/X13SAE, BIOS 2.0 10/17/2022
2. Dell Inc. Precision 3660/0PRR48, BIOS 2.9.3 11/22/2023
3. Dell Inc. OptiPlex SFF Plus 7010/0YGWFV, BIOS 1.7.1 08/11/2023
4. Run `git grep 'emory type.*supported yet, not instantiating SPD'` in
the repository of dmesg reports for various computers collected by
Linux users at https://linux-hardware.org. [1]

Add 0x22 and 0x23 for DDR5 according to section 7.18.2 (Memory Device —
Type), table 78 in *System Management BIOS (SMBIOS) Reference
Specification*, version 3.6.0 [2].

The same name *ee1004* as for DDR4 is used.

Successfully tested on Supermicro Super Server/X13SAE, BIOS 2.0 10/17/2022:

[ 5.459383] i801_smbus 0000:00:1f.4: SPD Write Disable is set
[ 5.465180] i801_smbus 0000:00:1f.4: SMBus using PCI interrupt
[ 5.465803] intel_pstate: HWP enabled by BIOS
[ 5.474046] i2c i2c-0: Successfully instantiated SPD at 0x50
[ 5.475379] intel_pstate: Intel P-state driver initializing
[ 5.487129] i2c i2c-0: Successfully instantiated SPD at 0x51
[ 5.487350] intel_pstate: HWP enabled
[ 5.493288] i2c i2c-0: Successfully instantiated SPD at 0x52
[ 5.496481] hid: raw HID events driver (C) Jiri Kosina
[ 5.502695] i2c i2c-0: Successfully instantiated SPD at 0x53

[1]: https://github.com/linuxhw/Dmesg
[2]: https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_3.6.0.pdf

Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
Cc: Jean Delvare <jdelvare@suse.de>
---
1. I have no idea if the name ee1004 is correct.
2. Should I test other things besides looking at the Linux messages?

drivers/i2c/i2c-smbus.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/i2c/i2c-smbus.c b/drivers/i2c/i2c-smbus.c
index 97f338b123b1..0d67a95c0599 100644
--- a/drivers/i2c/i2c-smbus.c
+++ b/drivers/i2c/i2c-smbus.c
@@ -308,7 +308,7 @@ EXPORT_SYMBOL_GPL(i2c_free_slave_host_notify_device);
* target systems are the same.
* Restrictions to automatic SPD instantiation:
* - Only works if all filled slots have the same memory type
- * - Only works for DDR, DDR2, DDR3 and DDR4 for now
+ * - Only works for DDR, DDR2, DDR3, DDR4 and DDR5 for now
* - Only works on systems with 1 to 8 memory slots
*/
#if IS_ENABLED(CONFIG_DMI)
@@ -380,6 +380,8 @@ void i2c_register_spd(struct i2c_adapter *adap)
break;
case 0x1A: /* DDR4 */
case 0x1E: /* LPDDR4 */
+ case 0x22: /* DDR5 */
+ case 0x23: /* LPDDR5 */
name = "ee1004";
break;
default:
--
2.43.0
Re: [PATCH v2] i2c: smbus: Add (LP)DDR5 types to `i2c_register_spd()` [ In reply to ]
Hi Paul,

> 1. I have no idea if the name ee1004 is correct.

I am afraid this won't work. It was tried already in 2023:

https://www.spinics.net/lists/linux-i2c/msg62267.html

Especially note Jean's answer here:

https://www.spinics.net/lists/linux-i2c/msg62420.html

So, DDR5 probably needs a new driver.

But thanks for trying!

Wolfram
Re: [PATCH v2] i2c: smbus: Add (LP)DDR5 types to `i2c_register_spd()` [ In reply to ]
Dear Wolfram,


Am 27.04.24 um 11:14 schrieb Wolfram Sang:

>> 1. I have no idea if the name ee1004 is correct.
>
> I am afraid this won't work. It was tried already in 2023:
>
> https://www.spinics.net/lists/linux-i2c/msg62267.html
>
> Especially note Jean's answer here:
>
> https://www.spinics.net/lists/linux-i2c/msg62420.html
>
> So, DDR5 probably needs a new driver.
>
> But thanks for trying!

Thank you for digging this up. Would it be alright, to adapt the warning
for (LP)DDR5?


Kind regards,

Paul
Re: [PATCH v2] i2c: smbus: Add (LP)DDR5 types to `i2c_register_spd()` [ In reply to ]
> Thank you for digging this up. Would it be alright, to adapt the warning for
> (LP)DDR5?

Sure, good idea. I think it will be helpful to include the standard name
in the message, so people have a starting point, like:

"DDR5 SPD not supported. JEDEC JESD300-5 compatible driver does not exist yet"

Or similar...