bpf: Use NOP_ATOMIC5 instead of emit_nops(&prog, 5) for BPF_TRAMP_F_C…

…ALL_ORIG

[ Upstream commit b908297 ]

__bpf_arch_text_poke does rewrite only for atomic nop5, emit_nops(xxx, 5)
emits non-atomic one which breaks fentry/fexit with k8 atomics:

P6_NOP5 == P6_NOP5_ATOMIC (0f1f440000 == 0f1f440000)
K8_NOP5 != K8_NOP5_ATOMIC (6666906690 != 6666666690)

Can be reproduced by doing "ideal_nops = k8_nops" in "arch_init_ideal_nops()
and running fexit_bpf2bpf selftest.

Fixes: e21aa34 ("bpf: Fix fexit trampoline.")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210320000001.915366-1-sdf@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

arch/x86/net/bpf_jit_comp.c

@@ -1811,7 +1811,8 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		/* remember return value in a stack for bpf prog to access */
 		emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8);
 		im->ip_after_call = prog;
-		emit_nops(&prog, 5);
+		memcpy(prog, ideal_nops[NOP_ATOMIC5], X86_PATCH_SIZE);
+		prog += X86_PATCH_SIZE;
 	}
 
 	if (fmod_ret->nr_progs) {