Commit e815df29 authored by Hengqi Chen's avatar Hengqi Chen Committed by Huacai Chen
Browse files

LoongArch: BPF: Add fsession support for trampolines 



Implement BPF_TRACE_FSESSION support in LoongArch BPF JIT. The logic
here is almost identical to what has been done in RISC-V JIT.

The key changes are:
  - Allocate stack space for function meta and session cookies
  - Introduce invoke_bpf() as a wrapper around invoke_bpf_prog() that
    populates session cookies before each invocation
  - Implement bpf_jit_supports_fsession() callback

Tested-by: default avatarVincent Li <vincent.mc.li@gmail.com>
Reviewed-by: default avatarMenglong Dong <menglong8.dong@gmail.com>
Signed-off-by: default avatarHengqi Chen <hengqi.chen@gmail.com>
Signed-off-by: default avatarHuacai Chen <chenhuacai@loongson.cn>
parent 6ef04707

arch/loongarch/net/bpf_jit.c

+66 −10
Original line number Diff line number Diff line
@@ -1737,6 +1737,29 @@ static int invoke_bpf_prog(struct jit_ctx *ctx, struct bpf_tramp_link *l, 
	return ret;

}



static int invoke_bpf(struct jit_ctx *ctx, struct bpf_tramp_links *tl,

		      int args_off, int retval_off, int run_ctx_off,

		      int func_meta_off, bool save_ret, u64 func_meta, int cookie_off)

{

	int i, cur_cookie = (cookie_off - args_off) / 8;



	for (i = 0; i < tl->nr_links; i++) {

		int err;



		if (bpf_prog_calls_session_cookie(tl->links[i])) {

			u64 meta = func_meta | ((u64)cur_cookie << BPF_TRAMP_COOKIE_INDEX_SHIFT);



			emit_store_stack_imm64(ctx, LOONGARCH_GPR_T1, -func_meta_off, meta);

			cur_cookie--;

		}

		err = invoke_bpf_prog(ctx, tl->links[i], args_off, retval_off, run_ctx_off, save_ret);

		if (err)

			return err;

	}



	return 0;

}



void *arch_alloc_bpf_trampoline(unsigned int size)

{

	return bpf_prog_pack_alloc(size, jit_fill_hole);
@@ -1788,8 +1811,10 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
					 void *func_addr, u32 flags)

{

	int i, ret, save_ret;

	int cookie_cnt, cookie_off;

	int stack_size, args_off, stk_args_off, nr_arg_slots = 0;

	int retval_off, nregs_off, ip_off, run_ctx_off, sreg_off, tcc_ptr_off;

	int retval_off, func_meta_off, ip_off, run_ctx_off, sreg_off, tcc_ptr_off;

	unsigned long long func_meta;

	bool is_struct_ops = flags & BPF_TRAMP_F_INDIRECT;

	void *orig_call = func_addr;

	struct bpf_tramp_links *fentry = &tlinks[BPF_TRAMP_FENTRY];
@@ -1813,10 +1838,14 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
	 *                   [ ...               ]

	 * FP - args_off     [ arg reg1          ]

	 *

	 * FP - nregs_off    [ arg regs count    ]

	 * FP - func_meta_off [ regs count, etc ]

	 *

	 * FP - ip_off       [ traced func       ] BPF_TRAMP_F_IP_ARG

	 *

	 *                   [ stack cookie N    ]

	 *                   [ ...               ]

	 * FP - cookie_off   [ stack cookie 1    ]

	 *

	 * FP - run_ctx_off  [ bpf_tramp_run_ctx ]

	 *

	 * FP - sreg_off     [ callee saved reg  ]
@@ -1859,9 +1888,9 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
	stack_size += nr_arg_slots * 8;

	args_off = stack_size;



	/* Room of trampoline frame to store args number */

	/* Room of function metadata, such as regs count */

	stack_size += 8;

	nregs_off = stack_size;

	func_meta_off = stack_size;



	/* Room of trampoline frame to store ip address */

	if (flags & BPF_TRAMP_F_IP_ARG) {
@@ -1869,6 +1898,12 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
		ip_off = stack_size;

	}



	cookie_cnt = bpf_fsession_cookie_cnt(tlinks);



	/* Room for session cookies */

	stack_size += cookie_cnt * 8;

	cookie_off = stack_size;



	/* Room of trampoline frame to store struct bpf_tramp_run_ctx */

	stack_size += round_up(sizeof(struct bpf_tramp_run_ctx), 8);

	run_ctx_off = stack_size;
@@ -1929,10 +1964,20 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
		emit_store_stack_imm64(ctx, LOONGARCH_GPR_T1, -ip_off, (u64)func_addr);



	/* store arg regs count */

	emit_store_stack_imm64(ctx, LOONGARCH_GPR_T1, -nregs_off, nr_arg_slots);

	func_meta = nr_arg_slots;

	emit_store_stack_imm64(ctx, LOONGARCH_GPR_T1, -func_meta_off, func_meta);



	store_args(ctx, nr_arg_slots, args_off);



	if (bpf_fsession_cnt(tlinks)) {

		/* clear all session cookies' value */

		for (i = 0; i < cookie_cnt; i++)

			emit_insn(ctx, std, LOONGARCH_GPR_ZERO, LOONGARCH_GPR_FP, -cookie_off + 8 * i);



		/* clear return value to make sure fentry always get 0 */

		emit_insn(ctx, std, LOONGARCH_GPR_ZERO, LOONGARCH_GPR_FP, -retval_off);

	}



	/* To traced function */

	/* Ftrace jump skips 2 NOP instructions */

	if (is_kernel_text((unsigned long)orig_call) ||
@@ -1949,9 +1994,9 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
			return ret;

	}



	for (i = 0; i < fentry->nr_links; i++) {

		ret = invoke_bpf_prog(ctx, fentry->links[i], args_off, retval_off,

				      run_ctx_off, flags & BPF_TRAMP_F_RET_FENTRY_RET);

	if (fentry->nr_links) {

		ret = invoke_bpf(ctx, fentry, args_off, retval_off, run_ctx_off, func_meta_off,

				 flags & BPF_TRAMP_F_RET_FENTRY_RET, func_meta, cookie_off);

		if (ret)

			return ret;

	}
@@ -1995,8 +2040,14 @@ static int __arch_prepare_bpf_trampoline(struct jit_ctx *ctx, struct bpf_tramp_i 
		*branches[i] = larch_insn_gen_bne(LOONGARCH_GPR_T1, LOONGARCH_GPR_ZERO, offset);

	}



	for (i = 0; i < fexit->nr_links; i++) {

		ret = invoke_bpf_prog(ctx, fexit->links[i], args_off, retval_off, run_ctx_off, false);

	/* Set "is_return" flag for fsession */

	func_meta |= (1ULL << BPF_TRAMP_IS_RETURN_SHIFT);

	if (bpf_fsession_cnt(tlinks))

		emit_store_stack_imm64(ctx, LOONGARCH_GPR_T1, -func_meta_off, func_meta);



	if (fexit->nr_links) {

		ret = invoke_bpf(ctx, fexit, args_off, retval_off, run_ctx_off,

				 func_meta_off, false, func_meta, cookie_off);

		if (ret)

			goto out;

	}
@@ -2331,6 +2382,11 @@ bool bpf_jit_supports_arena(void) 
	return true;

}



bool bpf_jit_supports_fsession(void)

{

	return true;

}



/* Indicate the JIT backend supports mixing bpf2bpf and tailcalls. */

bool bpf_jit_supports_subprog_tailcalls(void)

{