bpf, arm64: remove prefetch insn in xadd mapping

Prefetch-with-intent-to-write is currently part of the XADD mapping in
the AArch64 JIT and follows the kernel's implementation of atomic_add.
This may interfere with other threads executing the LDXR/STXR loop,
leading to potential starvation and fairness issues. Drop the optional
prefetch instruction.

Fixes: 85f68fe ("bpf, arm64: implement jiting of BPF_XADD")
Reported-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

arch/arm64/net/bpf_jit.h

@@ -100,12 +100,6 @@
 #define A64_STXR(sf, Rt, Rn, Rs) \
 	A64_LSX(sf, Rt, Rn, Rs, STORE_EX)
 
-/* Prefetch */
-#define A64_PRFM(Rn, type, target, policy) \
-	aarch64_insn_gen_prefetch(Rn, AARCH64_INSN_PRFM_TYPE_##type, \
-				  AARCH64_INSN_PRFM_TARGET_##target, \
-				  AARCH64_INSN_PRFM_POLICY_##policy)
-
 /* Add/subtract (immediate) */
 #define A64_ADDSUB_IMM(sf, Rd, Rn, imm12, type) \
 	aarch64_insn_gen_add_sub_imm(Rd, Rn, imm12, \

arch/arm64/net/bpf_jit_comp.c

@@ -762,7 +762,6 @@ static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx,
 	case BPF_STX | BPF_XADD | BPF_DW:
 		emit_a64_mov_i(1, tmp, off, ctx);
 		emit(A64_ADD(1, tmp, tmp, dst), ctx);
-		emit(A64_PRFM(tmp, PST, L1, STRM), ctx);
 		emit(A64_LDXR(isdw, tmp2, tmp), ctx);
 		emit(A64_ADD(isdw, tmp2, tmp2, src), ctx);
 		emit(A64_STXR(isdw, tmp2, tmp, tmp3), ctx);