Stream: git-wasmtime

Topic: wasmtime / issue #6003 Out-of-memory on wasmtime aarch64 CI

Wasmtime GitHub notifications bot (Mar 13 2023 at 16:07):

bnjbvr opened issue #6003:

In the latest daily runs of wasmtime testing on our internal Aarch64 machine, I see new failures related to multiple memories tests:

---- wast::Cranelift::spec::multi_memory::traps0_pooling stdout ----
thread 'wast::Cranelift::spec::multi_memory::traps0_pooling' panicked at 'called `Result::unwrap()` on an `Err` value: failed to create memory pool mapping

Caused by:
    0: mmap failed to allocate 0x17bb80000000 bytes
    1: Cannot allocate memory (os error 12)', /opt/homebrew/var/buildkite-agent/builds/mac-aarch64-4/embark-studios/wasmtime-aarch64-apple-darwin/wasmtime/target/debug/build/wasmtime-cli-5998d6d2b565964d/out/wast_testsuite_tests.rs:6063:18

failures:
    wast::Cranelift::spec::multi_memory::data_pooling
    wast::Cranelift::spec::multi_memory::exports0_pooling
    wast::Cranelift::spec::multi_memory::float_exprs0_pooling
    wast::Cranelift::spec::multi_memory::float_exprs1_pooling
    wast::Cranelift::spec::multi_memory::float_memory0_pooling
    wast::Cranelift::spec::multi_memory::imports0_pooling
    wast::Cranelift::spec::multi_memory::imports1_pooling
    wast::Cranelift::spec::multi_memory::imports2_pooling
    wast::Cranelift::spec::multi_memory::imports3_pooling
    wast::Cranelift::spec::multi_memory::imports4_pooling
    wast::Cranelift::spec::multi_memory::imports_pooling
    wast::Cranelift::spec::multi_memory::linking1_pooling
    wast::Cranelift::spec::multi_memory::linking2_pooling
    wast::Cranelift::spec::multi_memory::load0_pooling
    wast::Cranelift::spec::multi_memory::memory_init0_pooling
    wast::Cranelift::spec::multi_memory::memory_multi_pooling
    wast::Cranelift::spec::multi_memory::memory_pooling
    wast::Cranelift::spec::multi_memory::memory_size2_pooling
    wast::Cranelift::spec::multi_memory::memory_size3_pooling
    wast::Cranelift::spec::multi_memory::memory_size_pooling
    wast::Cranelift::spec::multi_memory::memory_trap0_pooling
    wast::Cranelift::spec::multi_memory::simd_memory_multi_pooling
    wast::Cranelift::spec::multi_memory::store0_pooling
    wast::Cranelift::spec::multi_memory::store1_pooling
    wast::Cranelift::spec::multi_memory::store_pooling
    wast::Cranelift::spec::multi_memory::traps0_pooling

Are these large memory allocations expected to fail?

Wasmtime GitHub notifications bot (Mar 13 2023 at 16:20):

alexcrichton commented on issue #6003:

I've been seeing this as well locally and I think it's related to running out of virtual address space given the number of memories being created. I've been fixing temporarily with --test-threads 2 but I'll work on implementing a default solution for these tests.

Wasmtime GitHub notifications bot (Mar 13 2023 at 20:55):

alexcrichton closed issue #6003:

In the latest daily runs of wasmtime testing on our internal Aarch64 machine, I see new failures related to multiple memories tests:

---- wast::Cranelift::spec::multi_memory::traps0_pooling stdout ----
thread 'wast::Cranelift::spec::multi_memory::traps0_pooling' panicked at 'called `Result::unwrap()` on an `Err` value: failed to create memory pool mapping

Caused by:
    0: mmap failed to allocate 0x17bb80000000 bytes
    1: Cannot allocate memory (os error 12)', /opt/homebrew/var/buildkite-agent/builds/mac-aarch64-4/embark-studios/wasmtime-aarch64-apple-darwin/wasmtime/target/debug/build/wasmtime-cli-5998d6d2b565964d/out/wast_testsuite_tests.rs:6063:18

failures:
    wast::Cranelift::spec::multi_memory::data_pooling
    wast::Cranelift::spec::multi_memory::exports0_pooling
    wast::Cranelift::spec::multi_memory::float_exprs0_pooling
    wast::Cranelift::spec::multi_memory::float_exprs1_pooling
    wast::Cranelift::spec::multi_memory::float_memory0_pooling
    wast::Cranelift::spec::multi_memory::imports0_pooling
    wast::Cranelift::spec::multi_memory::imports1_pooling
    wast::Cranelift::spec::multi_memory::imports2_pooling
    wast::Cranelift::spec::multi_memory::imports3_pooling
    wast::Cranelift::spec::multi_memory::imports4_pooling
    wast::Cranelift::spec::multi_memory::imports_pooling
    wast::Cranelift::spec::multi_memory::linking1_pooling
    wast::Cranelift::spec::multi_memory::linking2_pooling
    wast::Cranelift::spec::multi_memory::load0_pooling
    wast::Cranelift::spec::multi_memory::memory_init0_pooling
    wast::Cranelift::spec::multi_memory::memory_multi_pooling
    wast::Cranelift::spec::multi_memory::memory_pooling
    wast::Cranelift::spec::multi_memory::memory_size2_pooling
    wast::Cranelift::spec::multi_memory::memory_size3_pooling
    wast::Cranelift::spec::multi_memory::memory_size_pooling
    wast::Cranelift::spec::multi_memory::memory_trap0_pooling
    wast::Cranelift::spec::multi_memory::simd_memory_multi_pooling
    wast::Cranelift::spec::multi_memory::store0_pooling
    wast::Cranelift::spec::multi_memory::store1_pooling
    wast::Cranelift::spec::multi_memory::store_pooling
    wast::Cranelift::spec::multi_memory::traps0_pooling

Are these large memory allocations expected to fail?

Wasmtime GitHub notifications bot (Mar 14 2023 at 09:54):

bnjbvr commented on issue #6003:

Wow, thanks @alexcrichton, you rock :tada:

Last updated: Oct 23 2024 at 20:03 UTC