mctp NVMe smart-log Connection timed out (NVMe drive Micron 7450)

[sooraj-emb]

Hi dev,
I am using mctp protocol to check smart log of nvme micron device but ending up in an error,

Identify Controller failed, no quirks applied
smart log: Connection timed out

out of 20+ trial,
1 time got successful log,

nvme cli : version i am using -> nvme-cli-2.x
libnvme version i am using ->Release v1.16.1
mctp from code contruct till commit : 890f81df67966213698f2b40fe2dca6df25d19a7 ( 15 march 2026 )

root@machine: nvme smart-log mctp:1,9 LIBNVME_MI_PROBE_ENABLED=true
Smart Log for NVME device: mctp:1,9 namespace-id:ffffffff
critical_warning : 0
temperature : 38 °C (311 K, 100 °F)
available_spare : 100%
available_spare_threshold : 10%
percentage_used : 0%
endurance group critical warning summary: 0
Data Units Read : 9232 (4.73 GB)
Data Units Written : 1939416 (992.98 GB)
host_read_commands : 247675
host_write_commands : 1203119
controller_busy_time : 11
power_cycles : 2157
power_on_hours : 1427
unsafe_shutdowns : 2101
media_errors : 0
num_err_log_entries : 0
Warning Temperature Time : 0
Critical Composite Temperature Time : 0
Temperature Sensor 1 : 43 °C (316 K, 109 °F)
Temperature Sensor 2 : 40 °C (313 K, 104 °F)
Temperature Sensor 3 : 38 °C (311 K, 100 °F)
Thermal Management T1 Trans Count : 0
Thermal Management T2 Trans Count : 0
Thermal Management T1 Total Time : 0
Thermal Management T2 Total Time : 0
root@machine:~# nvme fw-log mctp:1,9 LIBNVME_MI_PROBE_ENABLED=true
Firmware Log for device:mctp:1,9
afi : 0x1
frs1 : 0x20303032554d3245 (E2MU200.)
frs2 : 0x20303032554d3245 (E2MU200.)
frs3 : 0x20303032554d3245 (E2MU200.)

anyone help to get this fix,

mctp_nvme_cli_test.txt

Thanks

[igaw]

@jk-ozlabs

[jk-ozlabs]

I suspect your device may not be reliably responding to MCTP messages; I would suggest capturing MCTP traffic with:

tcpdump -npi mctpi2c43 -w mctp.pcap

then reproducing the failure.

Once the failure occurs, stop the tcpdump, and attach / link the resulting mctp.pcap here.

[jk-ozlabs]

Or, even better, trace at the i2c layer:

# echo 1 > /sys/kernel/debug/tracing/events/i2c/enable
# echo 1 > /sys/kernel/debug/tracing/events/i2c_slave/enable

Then perform your MCTP interaction. This would be best with the minimal traffic required (so possibly fix the PROBE in your invocation there):

$ LIBNVME_MI_PROBE_ENABLED=0 nvme smart-log mctp:1,9

Once that has failed:

# cat /sys/kernel/debug/tracing/trace

And once you're done, disable the tracepoints:

# echo 0 > /sys/kernel/debug/tracing/events/i2c/enable
# echo 0 > /sys/kernel/debug/tracing/events/i2c_slave/enable

[sooraj-emb]

Hi jk,
attaching, actually I am using kenrel 6.1.158, is my issue happening due to old kernel?

mctp_trace.txt
Thanks

[jk-ozlabs]

Hi jk, attaching, actually I am using kenrel 6.1.158, is my issue happening due to old kernel?

Probably not; we'll have to work out what's going wrong to determine that.

mctp_trace.txt

You're using grep here:

root@machine:~# cat /sys/kernel/debug/tracing/trace | grep mctp

and so are likely missing the i2c target events. Can you just do a cat on that file?

[sooraj-emb]

closing the ticket as project decide to use intel mctpd approach (https://github.com/[ocp-hm-openbmc-opf-ami/mctpd](https://github.com/ocp-hm-openbmc-opf-ami/mctpd)
, best regards