On 24/08/04 09:59AM, Jose Fernandez wrote: > Hi folks, > > I have not found BPF arena documentation, beyond the patches and selftests. > I'm using the arena_atomics selftest as reference to create a simple BPF program > that increments a value atomically. > > bpf: https://github.com/jfernandez/bpf-playground/blob/main/arena.bpf.c > userspace: https://github.com/jfernandez/bpf-playground/blob/main/arena.c > common header: https://github.com/jfernandez/bpf-playground/blob/main/bpf_arena_common.h > > I'm using the 6.10.2 kernel and libbpf 1.4.3. I forgot to mention that I was using the latest clang release (18.1.8), and this turned out to be the issue. The arena bpf program loaded after I used clang compiled from the llvm-project master branch. I now realise that __BPF_FEATURE_ADDR_SPACE_CAST flag is only available starting with the 19.1.0 RC: $ git --no-pager tag --contains 65b123e287d1320170bb3317179bc917f21852fa llvmorg-19.1.0-rc1 llvmorg-20-init > > The program does not load when I use the `__arena_global` macro. If fails with: > > libbpf: map 'arena': at sec_idx 7, offset 0. > libbpf: map 'arena': found type = 33. > libbpf: map 'arena': found max_entries = 10. > libbpf: map 'arena': found map_flags = 0x400. > libbpf: map 'arena': found map_extra = 0x100000000000. > libbpf: sec '.relraw_tp/sys_enter': collecting relocation for section(3) 'raw_tp/sys_enter' > libbpf: sec '.relraw_tp/sys_enter': relo #0: insn #1 against 'add64_value' > libbpf: sec '.relraw_tp/sys_enter': relo #1: insn #4 against 'add64_result' > libbpf: object 'arena': failed (-95) to create BPF token from '/sys/fs/bpf', skipping optional step... > libbpf: map 'arena': created successfully, fd=3 > libbpf: prog 'add': BPF program load failed: Permission denied > libbpf: prog 'add': -- BEGIN PROG LOAD LOG -- > arg#0 reference type('UNKNOWN ') size cannot be determined: -22 > 0: R1=ctx() R10=fp0 > ; int add(const void *ctx) @ arena.bpf.c:17 > 0: (b7) r1 = 2 ; R1_w=2 > ; add64_result = __sync_fetch_and_add(&add64_value, 2); @ arena.bpf.c:19 > 1: (18) r2 = 0x100000000000 ; R2_w=scalar() > 3: (db) r1 = atomic64_fetch_add((u64 *)(r2 +0), r1) > misaligned access off (0x0; 0xffffffffffffffff)+0+0 size 8 > processed 3 insns (limit 1000000) max_states_per_insn 0 total_states 0 peak_states 0 mark_read 0 > -- END PROG LOAD LOG -- > libbpf: prog 'add': failed to load: -13 > libbpf: failed to load object 'arena' > libbpf: failed to load BPF skeleton 'arena': -13 > Failed to open BPF skeleton > > I found Yonghong Song's patch that appears to address a similar issue. But I'm > using a return value and not getting the `lock` instruction: > https://lore.kernel.org/bpf/20240803025928.4184433-1-yonghong.song@xxxxxxxxx/ > > I have compared with code with the selftest and I can't spot any signifcant > differences that may cause this issue. I would appreciate any help or guidance. > > Thanks, > Jose
