trap invalid opcode / illegal instruction (core dumped)

[thorstenhirsch]

Describe the bug

When starting reth on an Intel i5-4278U it crashes du to an invalid opcode. Is the CPU too old?
Here's how the start looks like:

2024-01-24T20:35:33.589013Z  INFO reth::cli: reth 0.1.0-alpha.16 (d8a9b58) starting
2024-01-24T20:35:33.589052Z  INFO reth::cli: Opening database path="/some/reth-eth-mainnet-data/db"
2024-01-24T20:35:33.592394Z  INFO reth::cli: Configuration loaded path="/some/reth-eth-mainnet-data/reth.toml"
2024-01-24T20:35:33.593035Z  INFO reth::cli: Database opened
2024-01-24T20:35:33.593296Z  INFO reth::cli: Pre-merge hard forks (block based):
- Frontier                         @0
- Homestead                        @1150000
- Dao                              @1920000
- Tangerine                        @2463000
- SpuriousDragon                   @2675000
- Byzantium                        @4370000
- Constantinople                   @7280000
- Petersburg                       @7280000
- Istanbul                         @9069000
- MuirGlacier                      @9200000
- Berlin                           @12244000
- London                           @12965000
- ArrowGlacier                     @13773000
- GrayGlacier                      @15050000
Merge hard forks:
- Paris                            @58750000000000000000000 (network is not known to be merged)

Post-merge hard forks (timestamp based):
- Shanghai                         @1681338455

bash: line 1:  8314 Illegal instruction     (core dumped) reth node --datadir=/some/reth-eth-mainnet-data --nat=extip:(my-ip-removed) --full

dmesg:

[ 3670.918117] traps: reth[8077] trap invalid opcode ip:55e4ca2a212e sp:7ffe9930fca8 error:0 in reth[55e4c8000000+2d1a000]

Steps to reproduce

1. Download reth alpha16 for Linux/x86
2. Run "reth node ..."

Node logs

No response

Platform(s)

Linux (x86)

What version/commit are you on?

alpha16

What database version are you on?

(default)

What type of node are you running?

Full via --full flag

What prune config do you use, if any?

No response

If you've built Reth from source, provide the full command you used

No response

Code of Conduct

• I agree to follow the Code of Conduct

[Rjected]

Could you try running make maxperf-no-asm and let us know if it still errors?

[DaniPopes]

Before that can you open it in a debugger and print the stacktrace?

Is the CPU too old?

I believe we build dist binaries with RUSTFLAGS=native so it depends on the CI runners' CPU

[mattsse]

I believe we build dist binaries with RUSTFLAGS=native

oh? I don't think we should do that

[thorstenhirsch]

Is the backtrace of gdb what you're looking for?

Thread 1 "reth" received signal SIGILL, Illegal instruction.
0x00005555576a212e in __mulx_mont_384 ()
(gdb) bt
#0  0x00005555576a212e in __mulx_mont_384 ()
#1  0x00005555576a20ab in mulx_mont_384 ()
#2  0x000055555769be94 in POINTonE1_Uncompress_Z ()
#3  0x000055555769342e in load_trusted_setup.part ()
#4  0x00005555572dda9e in core::ops::function::FnOnce::call_once ()
#5  0x000055555741127a in once_cell::imp::OnceCell<T>::initialize::{{closure}} ()
#6  0x0000555556b5f2cf in once_cell::imp::initialize_or_wait ()
#7  0x000055555570414b in once_cell::imp::OnceCell<T>::initialize ()
#8  0x0000555557412039 in reth_transaction_pool::validate::task::TransactionValidationTaskExecutor<()>::eth_builder ()
#9  0x000055555608c126 in reth::builder::NodeConfig::build_and_spawn_txpool ()
#10 0x00005555560be788 in reth::builder::NodeBuilderWithDatabase<DB>::launch::{{closure}} ()
#11 0x000055555607d9b3 in reth::runner::run_to_completion_or_panic::{{closure}} ()
#12 0x000055555605dc36 in reth::cli::Cli<Ext>::run ()
#13 0x00005555565dd1a3 in reth::main ()
#14 0x0000555555f3ad13 in std::sys_common::backtrace::__rust_begin_short_backtrace ()
#15 0x0000555555f3b799 in std::rt::lang_start ()
#16 0x00007ffff7c280d0 in __libc_start_call_main (main=main@entry=0x5555565dd930 <main>, argc=argc@entry=2, 
    argv=argv@entry=0x7fffffffdec8) at ../sysdeps/nptl/libc_start_call_main.h:58
#17 0x00007ffff7c28189 in __libc_start_main_impl (main=0x5555565dd930 <main>, argc=2, argv=0x7fffffffdec8, init=<optimized out>, 
    fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffdeb8) at ../csu/libc-start.c:360
#18 0x0000555555747cc9 in _start ()

[Rjected]

awesome, this helped narrow it down to blst, which conditions on the platform
supranational/blst@83d78c2

Let's introduce a portable feature to c-kzg-4844 which enables the blst portable feature. the binary built by the release process will use this feature. compiling from source however, should not.

[Rjected]

PR here: ethereum/c-kzg-4844#389

[github-actions]

This issue is stale because it has been open for 21 days with no activity.