In https://bugs.chromium.org/p/chromium/issues/detail?id=627529#c59 I noted:

"""
/chrome.debug: warning: failed to demangle _ZN4base8internal14CheckedNumericIjE4castIiEES2_RKNS1_IT_EENSt9enable_ifIXntsr3std7is_sameIS4_jEE5valueEiE4typeE

I don't immediately see why dump_syms is failing on those as it just calls

  abi::__cxa_demangle(mangled.c_str(), NULL, NULL, &status);

, but https://github.com/nico/hack/tree/master/demumble which does something similar demangles that to

  base::internal::CheckedNumeric<unsigned int> base::internal::CheckedNumeric<unsigned int>::cast<int>(base::internal::CheckedNumeric<int> const&, std::enable_if<!(std::is_same<int, unsigned int>::value), int>::type)
.
"""


This might cause bucketing of crashes to be not great for cross-platform code.

I looked into it a bit, and it's returning -2 in dump_syms, which means "mangled_name is not a valid name under the C++ ABI mangling rules.". I guess this means that whatever runtime it's using has an ABI that doesn't match the Chrome binaries? But I built dump_syms and Chrome in the same tree with what I would have expected were the same compiler, so I'm not really sure.

https://bugs.chromium.org/p/chromium/issues/detail?id=606626 also seems to indicate that demangling is up there with cache invalidation and off-by-one, so we're writing a third try. So maybe that will help or hurt fixing this problem.