encounter Segmentation fault when using ipmitool

[leye7755]

Core was generated by ipmitool -I lanplus -H xxxxxx -U xxxx -P XXXXXXXXX raw 0x3a 0x71 0x01
Program terminated with signal 11, Segmentation fault.

Why does this problem arise? We use popen to use ipmitool

[leye7755]

Looking forward to and thanking you for your help。@AlexanderAmelkin

[leye7755]

using sdr module encountered a problem:
debug details:
ipmi_sdr.c : ipmi_ipmi_sdr_read_sensor_value() call ipmi_sdr_get_unit_string( intf, pct, type, base,modifier)
and base is 255 which exceed the size of unit_desc array . leading to Segmentation fault。

[AlexanderAmelkin]

Can you please run ipmitool with nine -v arguments (that is, ipmitool -vvvvvvvvv ...) and attach (don't copy-paste) the output here? It's unclear why base is 255. Looks like an invalid response from your BMC. I'd like to see the full communication log to understand how to properly handle it.

[leye7755]

@AlexanderAmelkin Can i send # your email? can't find the way to attach log.

[AlexanderAmelkin]

Please check the branch from PR #120

[AlexanderAmelkin]

As for attachments, there is the "Attach files by dragging..." string right below the comment entry box here on this page.

[AlexanderAmelkin]

Sorry, committed the wrong version there. Will fix soon.

[leye7755]

sdr.log
this is sdr log

[AlexanderAmelkin]

@leye7755, I don't see a segmentation fault in the attached log. Is it the log from #120 ?

[leye7755]

sdr-20190423.log
I add comment in ipmi_sdr.c : ipmi_sdr_get_unit_string()
printf("#test:%d#\n",base);

[AlexanderAmelkin]

I now see it hit the wrong string for CPU1 DTS sensor.
That's quite expected with the old code, there is no range check in it.
Have you tried running the code from #120 ?

[AlexanderAmelkin]

It's unclear though why base is 0xFF, or to be more precise, why only 19 bytes of SDR Record are requested by ipmitool (see line 2379 of your sdr-20190423.log) when unit info is in bytes 21-23.
I'll look into it further, but I suspect there may be some invalid response from your BMC involved.

[AlexanderAmelkin]

Ok, those 19 bytes are just a remainder of total 51 bytes minus 32 bytes read earlier. Digging further...

[AlexanderAmelkin]

@leye7755, please tell how the program behaves if built from PR #120.