esnet
segfault error 4 in libc.so.6 caused by iperf3 on x86_64 Ubuntu LTS
#Context
-
Version of iperf3: 3.16 (cJSON 1.7.15)
-
Hardware:
- I226-IT on x86_64 Intel Raptorlake CPU
-
Operating system (and distribution, if any):
- Ubuntu 24.04 LTS
- Kernel v6.8.0-31
Bug Report
-
Expected Behavior No segfault in libcc.so.6
-
Actual Behavior
iperf3[1525756]: segfault at 2d0 ip 00007b49bd09a8fd sp 00007fff39e19e80 error 4 in libc.so.6[7b49bd028000+188000] likely on CPU 12 (core 24, socket 0)
Code: 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 f3 0f 1e fa 55 48 89 e5 53 48 81 ec a8 00 00 00 64 48 8b 04 25 28 00 00 00 48 89 45 e8 <8b> 87 d0 02 00 00 85 c0 75 21 48 8b 45 e8 64 48 2b 04 25 28 00 00
iperf3[1536411]: segfault at 2d0 ip 000074f9cd69a8fd sp 00007fff9637dbd0 error 4 in libc.so.6[74f9cd628000+188000] likely on CPU 4 (core 8, socket 0)
Code: 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 f3 0f 1e fa 55 48 89 e5 53 48 81 ec a8 00 00 00 64 48 8b 04 25 28 00 00 00 48 89 45 e8 <8b> 87 d0 02 00 00 85 c0 75 21 48 8b 45 e8 64 48 2b 04 25 28 00 00
Repeatedly
Maybe interesting to now, after a fault like this, the used cpu core to the interrupt-lane (4 of them) will change. We currently don't know if this issue could also be hardwarerelated or a bug in iperf3.
-
Steps to Reproduce Just connect two I226-IT on either the same, or to another system. start iperf3 over and over again ~ server is started like this in network namespaces: iperf3 --format m --bind 10.6.1.1 --server --port 5003 the client connets via tcp It takes some time to appear but after that repeatedly.
-
Possible Solution Not found yet..
It takes some time to appear but after that repeatedly.
I assume that the problem is in the server, not the client. When the problem happens, does it happen immediately when starting the server? When the first test request from a client is received? During the first test? During a later test?
Also, when the problem happens, can you run the server using -d -V (debug and verbose) and send the output?
