On 2025-10-24 12:36:07 [+0800], kernel test robot wrote: > kernel/trace/trace_syscalls.c:877:7: error: use of undeclared identifier > '__NR_mq_timedsend' > 877 | case __NR_mq_timedsend: > | ^ > kernel/trace/trace_syscalls.c:898:7: error: use of undeclared identifier > '__NR_access' > 898 | case __NR_access: > | ^ > kernel/trace/trace_syscalls.c:901:7: error: use of undeclared identifier > '__NR_chown' > 901 | case __NR_chown:
You need to exclude those if the numbers are not available. It does not compare about all of them, like it is fine with __NR_openat2. Risc-V probably did not wire them all up. So they started to use __NR_faccessat2 in favour of __NR_access. "Current" glibc probably uses just the latest one so why bother. x86 supported it since day one so they need to keep supporting it. The newcomer just started with the last variant. Sebastian
