rcorsi
rcorsi
> Is number 2 even a data race? There is only one writer right? One thread is reading and one is writing to the same location. Generally you shouldn't read...
This is still an issue, one of the problems that stands out for me the most in the CI jobs. Does anyone have any clue why the test would do...
In the recent failures, the issue is around `Libraries/LibJS/Bytecode/Interpreter.cpp:346` where the stack overflow has been detected. Stack size limit is checked in `Libraries/LibJS/Runtime/VM.h:109`, and I guess ASAN is only showing...
A few more discussions about increasing the stack size https://clang.llvm.org/docs/AddressSanitizer.html the previous one has the comment: `AddressSanitizer uses more stack memory. We have seen up to 3x increase.` https://github.com/google/sanitizers/issues/98 https://github.com/google/sanitizers/issues/60...
I am able to reproduce the issue locally. I have 6 failures on 36 attempts. 16.6% reproduction. I will try to use the ASAN flag `sleep_before_dying` to connect with gdb...
I got some information about one of the runners by adding some commands. See #6327 Look in the "Test Checks" tab of the run below. https://github.com/LadybirdBrowser/ladybird/actions/runs/18390236159/job/52398641725?pr=6327 I can see that...
Previously I mentioned that I was able to reproduce the issue, but I forgot that I actually lowered the `bash` process stack size to 1MB (with `ulimit -s 1024`) when...
Problem java script is: `Libraries/LibJS/Tests/runtime-error-call-stack-size.js` By doing the following: - repeatedly only running test-js - by dumping the java script file name of the test `path` in `Libraries/LibTest/JavaScriptTestRunner.h:217` in parse_script()...
@trflynn89 you worked this year on `Libraries/LibJS/Tests/runtime-error-call-stack-size.js`. Seems it might be a source of problems. It is most likely the cause of the infamous randomly failing CI test-js with `AddressSanitizer:DEADLYSIGNAL`...
Attaching the full backtrace from gdb [bt_test-js-stack-overflow.txt](https://github.com/user-attachments/files/22850720/bt_test-js-stack-overflow.txt) It look like the code in `Libraries/LibJS/Runtime/VM.h` is working well on frames 11 through 26 and going to recover from the stack overflow,...