From: Andrii Nakryiko Date: Thu, 4 Apr 2024 03:47:26 +0000 (-0700) Subject: bpf: handle CONFIG_SMP=n configuration in x86 BPF JIT X-Git-Url: http://git.maquefel.me/?a=commitdiff_plain;h=1e9e0b85255e6eca6036b59d8a5fbca6501905ac;p=linux.git bpf: handle CONFIG_SMP=n configuration in x86 BPF JIT On non-SMP systems, there is no "per-CPU" data, it's just global data. So in such case just don't do this_cpu_off-based per-CPU address adjustment. Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202404040951.d4CUx5S6-lkp@intel.com/ Fixes: 7bdbf7446305 ("bpf: add special internal-only MOV instruction to resolve per-CPU addrs") Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20240404034726.2766740-1-andrii@kernel.org Signed-off-by: Alexei Starovoitov --- diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index af89dd117dce4..964e8154da664 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1383,15 +1383,14 @@ static int do_jit(struct bpf_prog *bpf_prog, int *addrs, u8 *image, u8 *rw_image EMIT3(0x0F, 0x44, add_2reg(0xC0, AUX_REG, dst_reg)); break; } else if (insn_is_mov_percpu_addr(insn)) { - u32 off = (u32)(unsigned long)&this_cpu_off; - /* mov , (if necessary) */ EMIT_mov(dst_reg, src_reg); - +#ifdef CONFIG_SMP /* add , gs:[] */ EMIT2(0x65, add_1mod(0x48, dst_reg)); EMIT3(0x03, add_1reg(0x04, dst_reg), 0x25); - EMIT(off, 4); + EMIT((u32)(unsigned long)&this_cpu_off, 4); +#endif break; } fallthrough;