Tobias Klauser

> @tklauser any updates? Sorry @uno20001, this somehow fell between the cracks. I've cherry-picked the first two clean-up commits and will try to get to review the rest of this...

Your original commits (lacking the Signed-off-by lines) still appear in the PR. Please squash them with the following commits. I think they can all actually be in one single commit...

Hi @iyuvalk Thanks for your report. I looks like there is no easy way for me to reproduce these without setting up quite some infrastructure :) Did you try looking...

Hi Yuval Thanks a lot for the information. Could you try building netsniff-ng with debug symbols enabled, so we could get a better view of the backtrack and be able...

Thanks for the coredump. It indeed looks like the crash happens in glibc's optimized `memmove` from the `memcpy` call here: https://github.com/netsniff-ng/netsniff-ng/blob/c5b1ed4446f1da55f474c3eb5196825b19a08a05/netsniff-ng.c#L521 The memory areas are the tx/rx ring buffers and...

Kernel version: `cat /proc/version` glibc version: `/lib/x86_64-linux-gnu/libc.so.6` (executing the glibc .so file will print version information)

> BTW: Even if it is a problem with the kernel or glibc, wouldn't it be better to just drop a few packets and log this instead of going kaboom...

No, C doesn't offer try-catch and even if, it wouldn't help in this case because something is accessing memory it shouldn't be accessing. Moreover the memory for the ring buffers...

If that's option, you could try updating them to the latest version, yes. Or maybe @borkmann has an idea about how we could debug this?

Yes, as the segfault happens in the main receive/transmit loop I suspect it would happen more frequently on systems with higher amount of traffic.