Dealing with "Maximum call stack size exceeded" on fsanitize=undefined

[Maksim Ivanov]

Hello, does anyone have a tip on how to work around the "Maximum call stack size exceeded" error when running a UBsan-sanitized Emscripten program under Node?

The error is:

"RangeError: Maximum call stack size exceeded

    at _emscripten_return_address (a.out.js:6788:35)

    at __ubsan_handle_dynamic_type_cache_miss (wasm-function[8220]:0x210eb3)

    at std::type_info::operator==(std::type_info const&) const (wasm-function[938]:0x648b3)

    at is_equal(std::type_info const*, std::type_info const*, bool) (wasm-function[8154]:0x20a911)

    at __dynamic_cast (wasm-function[8156]:0x20aa48)

    at __ubsan::checkDynamicType(void*, void*, unsigned long) (wasm-function[8196]:0x20f1b9)

    at HandleDynamicTypeCacheMiss(__ubsan::DynamicTypeCacheMissData*, unsigned long, unsigned long, __ubsan::ReportOptions) (wasm-function[8221]:0x210f03)

    at __ubsan_handle_dynamic_type_cache_miss (wasm-function[8220]:0x210ed7)

    at std::type_info::operator==(std::type_info const&) const (wasm-function[938]:0x648b3)

    at is_equal(std::type_info const*, std::type_info const*, bool) (wasm-function[8154]:0x20a911)"

I'm not 100% sure, but it seems that the error doesn't indicate a bug in the program, because it doesn't appear when compiling a small program (e.g., a Googletest binary with 0 tests) and begins to appear even after adding unused code (e.g., a bunch of disabled Googletest tests).

I tried passing "--stack_size=8192" to Node, but this didn't help; also, when increasing the number (e.g., to 131072) I'm starting to get a different error:

RuntimeError: memory access out of bounds

    at __dynamic_cast (wasm-function[8156]:0x20aa0f)

    at __ubsan::checkDynamicType(void*, void*, unsigned long) (wasm-function[8196]:0x20f1b9)

    at HandleDynamicTypeCacheMiss(__ubsan::DynamicTypeCacheMissData*, unsigned long, unsigned long, __ubsan::ReportOptions) (wasm-function[8221]:0x210f03)

    at __ubsan_handle_dynamic_type_cache_miss (wasm-function[8220]:0x210ed7)

    at std::type_info::operator==(std::type_info const&) const (wasm-function[938]:0x648b3)

    at is_equal(std::type_info const*, std::type_info const*, bool) (wasm-function[8154]:0x20a911)

    at __dynamic_cast (wasm-function[8156]:0x20aa48)

    at __ubsan::checkDynamicType(void*, void*, unsigned long) (wasm-function[8196]:0x20f1b9)

    at HandleDynamicTypeCacheMiss(__ubsan::DynamicTypeCacheMissData*, unsigned long, unsigned long, __ubsan::ReportOptions) (wasm-function[8221]:0x210f03)

    at __ubsan_handle_dynamic_type_cache_miss (wasm-function[8220]:0x210ed7)

Thanks.

[Alon Zakai]

Does the stack trace look like it's an infinite recursion? That could be a toolchain bug.

--
You received this message because you are subscribed to the Google Groups "emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/8c9c6b8b-a4b1-427c-965c-5828f8fa8fe9n%40googlegroups.com.

[Thomas Lively]

I've seen reports of infinite recursions involving __ubsan_handle_dynamic_type_cache_miss before, but I've never been able to reproduce them to investigate. Would you be able to share your reproducer?

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpRAZ87AKvva7HxffLjQa64rGHk%3Dp%3DxfiWjgRO6X0iZzVg%40mail.gmail.com.

[Maksim Ivanov]

@Alon: Hard to say whether it's an infinite recursion - the snippet I attached is all what's output. Can it be that something truncates the stack before printing, without leaving any trace of what was truncated?

My other hypothesis was that some code generated by UBSan allocates huge variables on stack; again, I don't know how to verify that.

@Thomas: Sure; I'll probably be able to do it in a couple of days, after minimizing the repro and removing irrelevant details from it.

Maksim

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAJZD_EWHYYeb6q8TU690PR%2Bk%3DtpBPAEU%2Bp6fp_Xvi%3DDiaHWKyg%40mail.gmail.com.

[Alon Zakai]

Different versions of node may truncate the stack to different defaults. But it is configurable, try

node --stack-trace-limit=1000

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAH%3DwB8wGU1PO2UJw1die2A6xHWAhAeNADvfryG8Jo8nbKdQhdA%40mail.gmail.com.

[Maksim Ivanov]

Thanks, Alon.

I added this flag and, indeed, it's an infinite recursion.

I'll update this thread once I have a minimized repro sample to share.

Maksim

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpTqgwkOxBvYH5539nvZ9JuGePDKi10jmFzEN_nhSe-gnA%40mail.gmail.com.

[Maksim Ivanov]

The minimized repro is:

#include <gmock/gmock.h>
#include <gtest/gtest.h>
class A final {
 public:
  MOCK_METHOD0(Foo, void());
};
TEST(Some, DISABLED_Basic) {
  A a;
  EXPECT_CALL(a, Foo());
}

Googletest/mock are 1.10.0.

Maksim

Maksim Ivanov

Software Engineer

em...@google.com
+49 (0)176 65889070

Google Germany GmbH

Erika-Mann-Straße 33

80331 München

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

Registergericht und -nummer: Hamburg, HRB 86891

Sitz der Gesellschaft: Hamburg

Diese E-Mail ist vertraulich. Wenn Sie nicht der richtige Adressat sind, leiten Sie diese bitte nicht weiter, informieren Sie den Absender und löschen Sie die E-Mail und alle Anhänge. Vielen Dank.

      

This e-mail is confidential. If you are not the right addressee please do not forward it, please inform the sender, and please erase this e-mail including any attachments. Thanks.

[Thomas Lively]

Thanks, I am able to reproduce the issue locally. Looking into it now.

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAH%3DwB8wP71fpX4NxvwFBifZf_Wzqozi-FAMfBbSE2-Un74o%2BAw%40mail.gmail.com.

[Maksim Ivanov]

Thanks, Thomas.

As this is a confirmed issue now, I went ahead and created an issue to track it: https://github.com/emscripten-core/emscripten/issues/13324

Maksim

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAJZD_EVETdfzrchHrqQNoH8WgemDv2HGAJhDPw6QyX-5sZL51Q%40mail.gmail.com.

[Thomas Lively]

I have a fix up for review: https://github.com/emscripten-core/emscripten/pull/13367

To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAH%3DwB8wcmf2zo9pK6qeK%2BU0Ou2ZVTwK1%3D--0wGoJJYtEHUfREQ%40mail.gmail.com.