Closed effofxprime closed 3 years ago
I have been googling around and came across someone who used gdb to help figure out whats going on. Here was the output from the SIGILL:
Thread 1 "cortex" received signal SIGILL, Illegal instruction. 0x000000000191908e in CuckooVerifyProof_cuckaroo () (gdb) x/4i $pc => 0x191908e <CuckooVerifyProof_cuckaroo+14>: vmovdqu (%rdi),%xmm0 0x1919092 <CuckooVerifyProof_cuckaroo+18>: vmovdqu 0x10(%rdi),%xmm1 0x1919097 <CuckooVerifyProof_cuckaroo+23>: mov %fs:0x28,%rax 0x19190a0 <CuckooVerifyProof_cuckaroo+32>: mov %rax,0x48(%rsp)
I have good news. I was able to resolve my issue. This will probably be a thing for anyone running a VM, including paid for solutions online or from your own PC. This is a CPU instruction problem. I narrowed my own issue down to missing AVX when running the following command on my Cortex VM:
cat /proc/cpuinfo
I had Hyper-V installed and enabled, I switched to KVM and removed Hyper-V. I also ensured that all my Intel VT was enabled. I used ICH over PIIX in virtualbox settings.
After doing that, I reinstalled from scratch and have a working full node now. If anyone is having a SIGILL termination like above, you are likely having the same issue but possibly with a different instruction problem. When I was googling around, it seemed like most of these kinds of errors were related to that and the gdb output helped narrow down which instruction.
When I compiled Cortex Theseus, I received the following warnings. Are they an issue? I am having an issue with starting a full node.
My cortex start script:
Crashing of the node. **I didn't include everything from the error, it was well over 200000 lines and I've assumed that I have included the relevant info. I apologize if you need the whole output. Where would it output this information to a file?
I am running Ubuntu 20.04.2 LTS server. I am running as a virtual machine on a windows host in Virtualbox. My host is a 12core, 64GB ram, also running as my miner. The VM has a 2TB drive that is actually a physical extension and not a vmdiskfile. I've followed the steps outlined in the readme, and I even updated golang to 1.16.3 since the errors are something to do with go. What's interesting is, prior to a host OS reinstall, I had a VM of ubuntu up and running as a full node. It had sync'd the entire chain and I was seeing all the pools hit my node. I am unsure why I am unable to get it to run this time.