Hello everyone!
My question will be primarily addressed to the Milk-V staff and people envolved in the low-level programming. I have a Milk-V Titan, and I’ve encountered a very nasty bug related to concurrency or something similar. From time to time (the issue is floating/non-deterministic), under unclear circumstances, the following error occurs, which causes to some user-level application failure:
Fatal glibc error: ../nptl/pthread_mutex_lock.c:94 (__pthread_mutex_cond_lock): assertion failed: mutex->__data.__owner == 0
There is a patch from UltraRISC which could be applied at least on glibc-2.41 which seems to fix (or workaround?) the issue. The patch could not be applied to glibc-2.43+.
Could you please so kind to clarify the situation regarding this issue? Could it be fixed somehow or a support in glibc or whatever is needed, if so, is the mainstreaming planned? Is it a hardware bug or what?
Any suggestions will be appriciated!
Related links (where the issue is mentioned):
- Getting Started with RHEL 10 on RISC-V Platforms - Red Hat Customer Portal
- #11128 - guix pull on riscv64 machine gives error - guix/guix - Codeberg.org
Thanks in advance.