We've just run into this problem again: java loading JNI code instrumented for profiling created thread stack blocks that were too small.
It would be *really* nice for glibc to just do the right thing and account for __static_tls without users having to explicitly work around it (by adding __static_tls to their requested stack size).
Carlos, any update on this?
We've just run into this problem again: java loading JNI code instrumented for profiling created thread stack blocks that were too small.
It would be *really* nice for glibc to just do the right thing and account for __static_tls without users having to explicitly work around it (by adding __static_tls to their requested stack size).