On Fri, Nov 12, 2021 at 4:10 AM Floh <[email protected]> wrote:

> Ah... that was simply the optimizer resolving the math code into a
> constant expression. Reproduction test cases incoming :)
>

Yes, all these LTO issues related to LLVM creating references to symbols
during LTO that didn't exist prior.

There has been an upstream LLVM issue open for a while now:
https://bugs.llvm.org/show_bug.cgi?id=44353

Hopefully this change is enough to address them:
https://github.com/emscripten-core/emscripten/pull/15497

cheers,
sam



> On Friday, 12 November 2021 at 13:02:26 UTC+1 Floh wrote:
>
>> I started a github issue here:
>> https://github.com/emscripten-core/emscripten/issues/15506
>>
>> ...unfortunately I haven't been able to create a simple reproducer yet,
>> building a simple test program which calls into one of the exp, exp2 or pow
>> functions appears to work, with or without lto.
>>
>> On Friday, 12 November 2021 at 12:23:09 UTC+1 Floh wrote:
>>
>>> > I suspect LTO is the culprit here.
>>>
>>> That's indeed the issue. Building without LTO works. I'll try to come up
>>> with a minimal reproducer, create a github ticket and then see if I can get
>>> your PR working with my local emsdk installation :)
>>>
>>> On Thursday, 11 November 2021 at 23:08:14 UTC+1 [email protected] wrote:
>>>
>>>> I think this should fix your issue:
>>>> https://github.com/emscripten-core/emscripten/pull/15497
>>>>
>>>> On Thu, Nov 11, 2021 at 1:44 PM Sam Clegg <[email protected]> wrote:
>>>>
>>>>>
>>>>>
>>>>> On Thu, Nov 11, 2021 at 4:46 AM Floh <[email protected]> wrote:
>>>>>
>>>>>> These are the related header/source pairs in MUSL, and it seems that
>>>>>> these are new (didn't exist in the stable SDK):
>>>>>>
>>>>>>
>>>>>> https://github.com/emscripten-core/emscripten/blob/main/system/lib/libc/musl/src/math/exp_data.h
>>>>>>
>>>>>> https://github.com/emscripten-core/emscripten/blob/main/system/lib/libc/musl/src/math/exp_data.c
>>>>>>
>>>>>>
>>>>>> https://github.com/emscripten-core/emscripten/blob/main/system/lib/libc/musl/src/math/exp2f_data.h
>>>>>>
>>>>>> https://github.com/emscripten-core/emscripten/blob/main/system/lib/libc/musl/src/math/exp2f_data.c
>>>>>>
>>>>>> When listing the libc.a symbols in the emscripten cache with llvm-nm
>>>>>> I can see the entry for exp_data but nothing for exp2f_data:
>>>>>>
>>>>>> exp_data.o:
>>>>>> -------- D __exp_data
>>>>>>
>>>>>> (which is doubly weird, why does exp_data produce a linker error even
>>>>>> though it's there, and why isn't exp2f_data anywhere to be found)
>>>>>>
>>>>>> ...anyway, that's how far I got :)
>>>>>>
>>>>>
>>>>> I suspect LTO is the culprit here.    Does you non-LTO build work?
>>>>>
>>>>> I think these symbols can get generated at codegen time which can
>>>>> confuse LTO because it likes to know all the symbols needed before running
>>>>> the compiler/LTO.
>>>>>
>>>>>
>>>>>
>>>>>> On Thursday, 11 November 2021 at 12:57:06 UTC+1 Floh wrote:
>>>>>>
>>>>>>> PS: my other WASM projects appear to work fine (the home computer
>>>>>>> emulators here: https://floooh.github.io/tiny8bit/, and the Z80 and
>>>>>>> 6502 netlist simulations: https://floooh.github.io/visualz80remix/,
>>>>>>> https://floooh.github.io/visual6502remix/).
>>>>>>>
>>>>>>> On Thursday, 11 November 2021 at 12:02:12 UTC+1 Floh wrote:
>>>>>>>
>>>>>>>> I'm getting the following linker errors in my sokol-samples project:
>>>>>>>>
>>>>>>>> 5x:
>>>>>>>> wasm-ld: error:
>>>>>>>> /...cache/sysroot/lib/wasm32-emscripten/lto/libc.a(exp2f_data.o): 
>>>>>>>> undefined
>>>>>>>> symbol: __exp2f_data
>>>>>>>>
>>>>>>>> 8x:
>>>>>>>> wasm-ld: error:
>>>>>>>> /...cache/sysroot/lib/wasm32-emscripten/lto/libc.a(exp_data.o): 
>>>>>>>> undefined
>>>>>>>> symbol: __exp_data
>>>>>>>>
>>>>>>>> Full wasm-ld command line:
>>>>>>>>
>>>>>>>> em++: error:
>>>>>>>> '/Users/floh/projects/fips-sdks/emsdk/upstream/bin/wasm-ld -o
>>>>>>>> /Users/floh/projects/fips-deploy/sokol-samples/sapp-webgl2-wasm-ninja-release/modplay-sapp.wasm
>>>>>>>> sapp/CMakeFiles/modplay-sapp.dir/modplay-sapp.c.obj 
>>>>>>>> libs/sokol/libsokol.a
>>>>>>>> fips-libmodplug_libmodplug/liblibmodplug.a
>>>>>>>> -L/Users/floh/projects/fips-sdks/emsdk/upstream/emscripten/cache/sysroot/lib/wasm32-emscripten/lto
>>>>>>>> -lGL-webgl2 -lal -lhtml5 -lstubs -lc -lcompiler_rt -lc++-noexcept
>>>>>>>> -lc++abi-noexcept -lemmalloc -lc_rt_wasm -lsockets -mllvm
>>>>>>>> -combiner-global-alias-analysis=false -mllvm -enable-emscripten-sjlj 
>>>>>>>> -mllvm
>>>>>>>> -disable-lsr --import-undefined --strip-debug --export-if-defined=main
>>>>>>>> --export-if-defined=stackSave --export-if-defined=stackRestore
>>>>>>>> --export-if-defined=stackAlloc --export-if-defined=__wasm_call_ctors
>>>>>>>> --export-if-defined=__errno_location --export-if-defined=malloc
>>>>>>>> --export-if-defined=free --export-if-defined=__start_em_asm
>>>>>>>> --export-if-defined=__stop_em_asm --export-table -z stack-size=5242880
>>>>>>>> --initial-memory=33554432 --no-entry --max-memory=2147483648
>>>>>>>> <(214)%20748-3648> --global-base=1024' failed (returned 1)
>>>>>>>>
>>>>>>>> ...and the em++ command line:
>>>>>>>>
>>>>>>>> /Users/floh/projects/fips-sdks/emsdk/upstream/emscripten/em++ -s
>>>>>>>> DISABLE_EXCEPTION_CATCHING=1  -fno-exceptions -fno-rtti -std=c++11
>>>>>>>> -fstrict-aliasing -Wall -Wno-multichar -Wextra -Wno-unknown-pragmas
>>>>>>>> -Wno-ignored-qualifiers -Wno-long-long -Wno-overloaded-virtual
>>>>>>>> -Wno-deprecated-writable-strings -Wno-unused-volatile-lvalue
>>>>>>>> -Wno-inconsistent-missing-override -Wno-warn-absolute-paths
>>>>>>>> -Wno-expansion-to-defined  -flto -O3 -DNDEBUG -s
>>>>>>>> DISABLE_EXCEPTION_CATCHING=1  --memory-init-file 0 -s
>>>>>>>> INITIAL_MEMORY=33554432 -s ERROR_ON_UNDEFINED_SYMBOLS=1 -s
>>>>>>>> NO_EXIT_RUNTIME=1 -s ALLOW_MEMORY_GROWTH=1 -s USE_WEBGL2=1 -s
>>>>>>>> "MALLOC='emmalloc'" -s NO_FILESYSTEM=1 -s WASM=1  --shell-file
>>>>>>>> /Users/floh/projects/sokol-samples/webpage/shell.html -O3  -flto  
>>>>>>>> --closure
>>>>>>>> 1 -s ASSERTIONS=0 sapp/CMakeFiles/modplay-sapp.dir/modplay-sapp.c.obj 
>>>>>>>> -o
>>>>>>>> /Users/floh/projects/fips-deploy/sokol-samples/sapp-webgl2-wasm-ninja-release/modplay-sapp.html
>>>>>>>> libs/sokol/libsokol.a  fips-libmodplug_libmodplug/liblibmodplug.a
>>>>>>>>
>>>>>>>> Investigating now, I think the most likely culprit is libmodplug (a
>>>>>>>> MOD player library).
>>>>>>>>
>>>>>>>> Should I open a separate github ticket, or append this type of
>>>>>>>> feedback to the PR?
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> -Andre.
>>>>>>>>
>>>>>>>> On Wednesday, 10 November 2021 at 17:33:34 UTC+1 [email protected]
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>>> Yesterday I landed the update to musl v1.2.2:
>>>>>>>>> https://github.com/emscripten-core/emscripten/pull/13006
>>>>>>>>>
>>>>>>>>> This is a fairly large change so we decided that next emscripten
>>>>>>>>> release will bump the version 3.0.0.
>>>>>>>>>
>>>>>>>>> If you are interested in helping us test this new version
>>>>>>>>> before we release it it would be great if you could try installing the
>>>>>>>>> `tot` version of emsdk which now contains this new version musl 
>>>>>>>>> (`./emsdk
>>>>>>>>> install tot`).   We are keen to find and fix any issues that might
>>>>>>>>> arise from this change.
>>>>>>>>>
>>>>>>>>> cheers,
>>>>>>>>> sam
>>>>>>>>>
>>>>>>>> --
>>>>>> 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 [email protected].
>>>>>> To view this discussion on the web visit
>>>>>> https://groups.google.com/d/msgid/emscripten-discuss/ff0f0683-9ed5-44b0-b0be-f51267863559n%40googlegroups.com
>>>>>> <https://groups.google.com/d/msgid/emscripten-discuss/ff0f0683-9ed5-44b0-b0be-f51267863559n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>>>>> .
>>>>>>
>>>>> --
> 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 [email protected].
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/emscripten-discuss/dc1badc0-77ea-4f08-85ba-40298631c025n%40googlegroups.com
> <https://groups.google.com/d/msgid/emscripten-discuss/dc1badc0-77ea-4f08-85ba-40298631c025n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
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 [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/emscripten-discuss/CAL_va28LGW8nRh6QnAB0%2BuxU3dgXbG3ZVo3QsZBo8dYkxUzF4w%40mail.gmail.com.

Reply via email to