Merge branch 'allocate-bpf-trampoline-on-bpf_prog_pack'

Song Liu says:

====================
Allocate bpf trampoline on bpf_prog_pack

This set enables allocating bpf trampoline from bpf_prog_pack on x86. The
majority of this work, however, is the refactoring of trampoline code.
This is needed because we need to handle 4 archs and 2 users (trampoline
and struct_ops).

1/7 through 6/7 refactors trampoline code. A few helpers are added.
7/7 finally let bpf trampoline on x86 use bpf_prog_pack.

Changes in v7:
1. Use kvmalloc for rw_image in x86/arch_prepare_bpf_trampoline. (Alexei)
2. Add comment to explain why we cannot use kvmalloc in
   x86/arch_bpf_trampoline_size. (Alexei)

Changes in v6:
1. Rebase.
2. Add Acked-by and Tested-by from Jiri Olsa and Björn Töpel.

Changes in v5:
1. Adjust size of trampoline ksym. (Jiri)
2. Use "unsigned int size" arg in image management helpers.(Daniel)

Changes in v4:
1. Dropped 1/8 in v3, which is already merged in bpf-next.
2. Add Reviewed-by from Björn Töpel.

Changes in v3:
1. Fix bug in s390. (Thanks to Ilya Leoshkevich).
2. Fix build error in riscv. (kernel test robot).

Changes in v2:
1. Add missing changes in net/bpf/bpf_dummy_struct_ops.c.
2. Reduce one dry run in arch_prepare_bpf_trampoline. (Xu Kuohai)
3. Other small fixes.
====================

Link: https://lore.kernel.org/r/20231206224054.492250-1-song@kernel.orgSigned-off-by: Alexei Starovoitov <ast@kernel.org>

arch/arm64/net/bpf_jit_comp.c

@@ -1828,7 +1828,7 @@ static void restore_args(struct jit_ctx *ctx, int args_off, int nregs)
  *
  */
 static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
-			      struct bpf_tramp_links *tlinks, void *orig_call,
+			      struct bpf_tramp_links *tlinks, void *func_addr,
 			      int nregs, u32 flags)
 {
 	int i;
@@ -1926,7 +1926,7 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
 
 	if (flags & BPF_TRAMP_F_IP_ARG) {
 		/* save ip address of the traced function */
-		emit_addr_mov_i64(A64_R(10), (const u64)orig_call, ctx);
+		emit_addr_mov_i64(A64_R(10), (const u64)func_addr, ctx);
 		emit(A64_STR64I(A64_R(10), A64_SP, ip_off), ctx);
 	}
 
@@ -2026,18 +2026,10 @@ static int prepare_trampoline(struct jit_ctx *ctx, struct bpf_tramp_image *im,
 	return ctx->idx;
 }
 
-int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
-				void *image_end, const struct btf_func_model *m,
-				u32 flags, struct bpf_tramp_links *tlinks,
-				void *orig_call)
+static int btf_func_model_nregs(const struct btf_func_model *m)
 {
-	int i, ret;
 	int nregs = m->nr_args;
-	int max_insns = ((long)image_end - (long)image) / AARCH64_INSN_SIZE;
-	struct jit_ctx ctx = {
-		.image = NULL,
-		.idx = 0,
-	};
+	int i;
 
 	/* extra registers needed for struct argument */
 	for (i = 0; i < MAX_BPF_FUNC_ARGS; i++) {
@@ -2046,22 +2038,49 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
 			nregs += (m->arg_size[i] + 7) / 8 - 1;
 	}
 
+	return nregs;
+}
+
+int arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+			     struct bpf_tramp_links *tlinks, void *func_addr)
+{
+	struct jit_ctx ctx = {
+		.image = NULL,
+		.idx = 0,
+	};
+	struct bpf_tramp_image im;
+	int nregs, ret;
+
+	nregs = btf_func_model_nregs(m);
 	/* the first 8 registers are used for arguments */
 	if (nregs > 8)
 		return -ENOTSUPP;
 
-	ret = prepare_trampoline(&ctx, im, tlinks, orig_call, nregs, flags);
+	ret = prepare_trampoline(&ctx, &im, tlinks, func_addr, nregs, flags);
 	if (ret < 0)
 		return ret;
 
-	if (ret > max_insns)
-		return -EFBIG;
+	return ret < 0 ? ret : ret * AARCH64_INSN_SIZE;
+}
 
-	ctx.image = image;
-	ctx.idx = 0;
+int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
+				void *image_end, const struct btf_func_model *m,
+				u32 flags, struct bpf_tramp_links *tlinks,
+				void *func_addr)
+{
+	int ret, nregs;
+	struct jit_ctx ctx = {
+		.image = image,
+		.idx = 0,
+	};
+
+	nregs = btf_func_model_nregs(m);
+	/* the first 8 registers are used for arguments */
+	if (nregs > 8)
+		return -ENOTSUPP;
 
 	jit_fill_hole(image, (unsigned int)(image_end - image));
-	ret = prepare_trampoline(&ctx, im, tlinks, orig_call, nregs, flags);
+	ret = prepare_trampoline(&ctx, im, tlinks, func_addr, nregs, flags);
 
 	if (ret > 0 && validate_code(&ctx) < 0)
 		ret = -EINVAL;

arch/riscv/net/bpf_jit_comp64.c

@@ -1029,23 +1029,28 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
 	return ret;
 }
 
-int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
-				void *image_end, const struct btf_func_model *m,
-				u32 flags, struct bpf_tramp_links *tlinks,
-				void *func_addr)
+int arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+			     struct bpf_tramp_links *tlinks, void *func_addr)
 {
-	int ret;
+	struct bpf_tramp_image im;
 	struct rv_jit_context ctx;
+	int ret;
 
 	ctx.ninsns = 0;
 	ctx.insns = NULL;
 	ctx.ro_insns = NULL;
-	ret = __arch_prepare_bpf_trampoline(im, m, tlinks, func_addr, flags, &ctx);
-	if (ret < 0)
-		return ret;
+	ret = __arch_prepare_bpf_trampoline(&im, m, tlinks, func_addr, flags, &ctx);
 
-	if (ninsns_rvoff(ret) > (long)image_end - (long)image)
-		return -EFBIG;
+	return ret < 0 ? ret : ninsns_rvoff(ctx.ninsns);
+}
+
+int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
+				void *image_end, const struct btf_func_model *m,
+				u32 flags, struct bpf_tramp_links *tlinks,
+				void *func_addr)
+{
+	int ret;
+	struct rv_jit_context ctx;
 
 	ctx.ninsns = 0;
 	/*

arch/s390/net/bpf_jit_comp.c

@@ -2637,6 +2637,21 @@ static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im,
 	return 0;
 }
 
+int arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+			     struct bpf_tramp_links *tlinks, void *orig_call)
+{
+	struct bpf_tramp_image im;
+	struct bpf_tramp_jit tjit;
+	int ret;
+
+	memset(&tjit, 0, sizeof(tjit));
+
+	ret = __arch_prepare_bpf_trampoline(&im, &tjit, m, flags,
+					    tlinks, orig_call);
+
+	return ret < 0 ? ret : tjit.common.prg;
+}
+
 int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
 				void *image_end, const struct btf_func_model *m,
 				u32 flags, struct bpf_tramp_links *tlinks,
@@ -2644,30 +2659,27 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image,
 {
 	struct bpf_tramp_jit tjit;
 	int ret;
-	int i;
 
-	for (i = 0; i < 2; i++) {
-		if (i == 0) {
-			/* Compute offsets, check whether the code fits. */
-			memset(&tjit, 0, sizeof(tjit));
-		} else {
-			/* Generate the code. */
-			tjit.common.prg = 0;
-			tjit.common.prg_buf = image;
-		}
-		ret = __arch_prepare_bpf_trampoline(im, &tjit, m, flags,
-						    tlinks, func_addr);
-		if (ret < 0)
-			return ret;
-		if (tjit.common.prg > (char *)image_end - (char *)image)
-			/*
-			 * Use the same error code as for exceeding
-			 * BPF_MAX_TRAMP_LINKS.
-			 */
-			return -E2BIG;
-	}
+	/* Compute offsets, check whether the code fits. */
+	memset(&tjit, 0, sizeof(tjit));
+	ret = __arch_prepare_bpf_trampoline(im, &tjit, m, flags,
+					    tlinks, func_addr);
+
+	if (ret < 0)
+		return ret;
+	if (tjit.common.prg > (char *)image_end - (char *)image)
+		/*
+		 * Use the same error code as for exceeding
+		 * BPF_MAX_TRAMP_LINKS.
+		 */
+		return -E2BIG;
+
+	tjit.common.prg = 0;
+	tjit.common.prg_buf = image;
+	ret = __arch_prepare_bpf_trampoline(im, &tjit, m, flags,
+					    tlinks, func_addr);
 
-	return tjit.common.prg;
+	return ret < 0 ? ret : tjit.common.prg;
 }
 
 bool bpf_jit_supports_subprog_tailcalls(void)

arch/x86/net/bpf_jit_comp.c

@@ -2198,7 +2198,8 @@ static void restore_regs(const struct btf_func_model *m, u8 **prog,
 
 static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 			   struct bpf_tramp_link *l, int stack_size,
-			   int run_ctx_off, bool save_ret)
+			   int run_ctx_off, bool save_ret,
+			   void *image, void *rw_image)
 {
 	u8 *prog = *pprog;
 	u8 *jmp_insn;
@@ -2226,7 +2227,7 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 	else
 		EMIT4(0x48, 0x8D, 0x75, -run_ctx_off);
 
-	if (emit_rsb_call(&prog, bpf_trampoline_enter(p), prog))
+	if (emit_rsb_call(&prog, bpf_trampoline_enter(p), image + (prog - (u8 *)rw_image)))
 		return -EINVAL;
 	/* remember prog start time returned by __bpf_prog_enter */
 	emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0);
@@ -2250,7 +2251,7 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 			       (long) p->insnsi >> 32,
 			       (u32) (long) p->insnsi);
 	/* call JITed bpf program or interpreter */
-	if (emit_rsb_call(&prog, p->bpf_func, prog))
+	if (emit_rsb_call(&prog, p->bpf_func, image + (prog - (u8 *)rw_image)))
 		return -EINVAL;
 
 	/*
@@ -2277,7 +2278,7 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 		EMIT3_off32(0x48, 0x8D, 0x95, -run_ctx_off);
 	else
 		EMIT4(0x48, 0x8D, 0x55, -run_ctx_off);
-	if (emit_rsb_call(&prog, bpf_trampoline_exit(p), prog))
+	if (emit_rsb_call(&prog, bpf_trampoline_exit(p), image + (prog - (u8 *)rw_image)))
 		return -EINVAL;
 
 	*pprog = prog;
@@ -2312,14 +2313,15 @@ static int emit_cond_near_jump(u8 **pprog, void *func, void *ip, u8 jmp_cond)
 
 static int invoke_bpf(const struct btf_func_model *m, u8 **pprog,
 		      struct bpf_tramp_links *tl, int stack_size,
-		      int run_ctx_off, bool save_ret)
+		      int run_ctx_off, bool save_ret,
+		      void *image, void *rw_image)
 {
 	int i;
 	u8 *prog = *pprog;
 
 	for (i = 0; i < tl->nr_links; i++) {
 		if (invoke_bpf_prog(m, &prog, tl->links[i], stack_size,
-				    run_ctx_off, save_ret))
+				    run_ctx_off, save_ret, image, rw_image))
 			return -EINVAL;
 	}
 	*pprog = prog;
@@ -2328,7 +2330,8 @@ static int invoke_bpf(const struct btf_func_model *m, u8 **pprog,
 
 static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
 			      struct bpf_tramp_links *tl, int stack_size,
-			      int run_ctx_off, u8 **branches)
+			      int run_ctx_off, u8 **branches,
+			      void *image, void *rw_image)
 {
 	u8 *prog = *pprog;
 	int i;
@@ -2339,7 +2342,8 @@ static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
 	emit_mov_imm32(&prog, false, BPF_REG_0, 0);
 	emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8);
 	for (i = 0; i < tl->nr_links; i++) {
-		if (invoke_bpf_prog(m, &prog, tl->links[i], stack_size, run_ctx_off, true))
+		if (invoke_bpf_prog(m, &prog, tl->links[i], stack_size, run_ctx_off, true,
+				    image, rw_image))
 			return -EINVAL;
 
 		/* mod_ret prog stored return value into [rbp - 8]. Emit:
@@ -2422,10 +2426,11 @@ static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
  * add rsp, 8                      // skip eth_type_trans's frame
  * ret                             // return to its caller
  */
-int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *image_end,
-				const struct btf_func_model *m, u32 flags,
-				struct bpf_tramp_links *tlinks,
-				void *func_addr)
+static int __arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *rw_image,
+					 void *rw_image_end, void *image,
+					 const struct btf_func_model *m, u32 flags,
+					 struct bpf_tramp_links *tlinks,
+					 void *func_addr)
 {
 	int i, ret, nr_regs = m->nr_args, stack_size = 0;
 	int regs_off, nregs_off, ip_off, run_ctx_off, arg_stack_off, rbx_off;
@@ -2521,7 +2526,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		orig_call += X86_PATCH_SIZE;
 	}
 
-	prog = image;
+	prog = rw_image;
 
 	EMIT_ENDBR();
 	/*
@@ -2563,7 +2568,8 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	if (flags & BPF_TRAMP_F_CALL_ORIG) {
 		/* arg1: mov rdi, im */
 		emit_mov_imm64(&prog, BPF_REG_1, (long) im >> 32, (u32) (long) im);
-		if (emit_rsb_call(&prog, __bpf_tramp_enter, prog)) {
+		if (emit_rsb_call(&prog, __bpf_tramp_enter,
+				  image + (prog - (u8 *)rw_image))) {
 			ret = -EINVAL;
 			goto cleanup;
 		}
@@ -2571,7 +2577,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 
 	if (fentry->nr_links)
 		if (invoke_bpf(m, &prog, fentry, regs_off, run_ctx_off,
-			       flags & BPF_TRAMP_F_RET_FENTRY_RET))
+			       flags & BPF_TRAMP_F_RET_FENTRY_RET, image, rw_image))
 			return -EINVAL;
 
 	if (fmod_ret->nr_links) {
@@ -2581,7 +2587,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 			return -ENOMEM;
 
 		if (invoke_bpf_mod_ret(m, &prog, fmod_ret, regs_off,
-				       run_ctx_off, branches)) {
+				       run_ctx_off, branches, image, rw_image)) {
 			ret = -EINVAL;
 			goto cleanup;
 		}
@@ -2602,14 +2608,14 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 			EMIT2(0xff, 0xd3); /* call *rbx */
 		} else {
 			/* call original function */
-			if (emit_rsb_call(&prog, orig_call, prog)) {
+			if (emit_rsb_call(&prog, orig_call, image + (prog - (u8 *)rw_image))) {
 				ret = -EINVAL;
 				goto cleanup;
 			}
 		}
 		/* 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;
+		im->ip_after_call = image + (prog - (u8 *)rw_image);
 		memcpy(prog, x86_nops[5], X86_PATCH_SIZE);
 		prog += X86_PATCH_SIZE;
 	}
@@ -2625,12 +2631,13 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		 * aligned address of do_fexit.
 		 */
 		for (i = 0; i < fmod_ret->nr_links; i++)
-			emit_cond_near_jump(&branches[i], prog, branches[i],
-					    X86_JNE);
+			emit_cond_near_jump(&branches[i], image + (prog - (u8 *)rw_image),
+					    image + (branches[i] - (u8 *)rw_image), X86_JNE);
 	}
 
 	if (fexit->nr_links)
-		if (invoke_bpf(m, &prog, fexit, regs_off, run_ctx_off, false)) {
+		if (invoke_bpf(m, &prog, fexit, regs_off, run_ctx_off,
+			       false, image, rw_image)) {
 			ret = -EINVAL;
 			goto cleanup;
 		}
@@ -2643,10 +2650,10 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	 * restored to R0.
 	 */
 	if (flags & BPF_TRAMP_F_CALL_ORIG) {
-		im->ip_epilogue = prog;
+		im->ip_epilogue = image + (prog - (u8 *)rw_image);
 		/* arg1: mov rdi, im */
 		emit_mov_imm64(&prog, BPF_REG_1, (long) im >> 32, (u32) (long) im);
-		if (emit_rsb_call(&prog, __bpf_tramp_exit, prog)) {
+		if (emit_rsb_call(&prog, __bpf_tramp_exit, image + (prog - (u8 *)rw_image))) {
 			ret = -EINVAL;
 			goto cleanup;
 		}
@@ -2665,19 +2672,90 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	if (flags & BPF_TRAMP_F_SKIP_FRAME)
 		/* skip our return address and return to parent */
 		EMIT4(0x48, 0x83, 0xC4, 8); /* add rsp, 8 */
-	emit_return(&prog, prog);
+	emit_return(&prog, image + (prog - (u8 *)rw_image));
 	/* Make sure the trampoline generation logic doesn't overflow */
-	if (WARN_ON_ONCE(prog > (u8 *)image_end - BPF_INSN_SAFETY)) {
+	if (WARN_ON_ONCE(prog > (u8 *)rw_image_end - BPF_INSN_SAFETY)) {
 		ret = -EFAULT;
 		goto cleanup;
 	}
-	ret = prog - (u8 *)image;
+	ret = prog - (u8 *)rw_image + BPF_INSN_SAFETY;
 
 cleanup:
 	kfree(branches);
 	return ret;
 }
 
+void *arch_alloc_bpf_trampoline(unsigned int size)
+{
+	return bpf_prog_pack_alloc(size, jit_fill_hole);
+}
+
+void arch_free_bpf_trampoline(void *image, unsigned int size)
+{
+	bpf_prog_pack_free(image, size);
+}
+
+void arch_protect_bpf_trampoline(void *image, unsigned int size)
+{
+}
+
+void arch_unprotect_bpf_trampoline(void *image, unsigned int size)
+{
+}
+
+int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *image_end,
+				const struct btf_func_model *m, u32 flags,
+				struct bpf_tramp_links *tlinks,
+				void *func_addr)
+{
+	void *rw_image, *tmp;
+	int ret;
+	u32 size = image_end - image;
+
+	/* rw_image doesn't need to be in module memory range, so we can
+	 * use kvmalloc.
+	 */
+	rw_image = kvmalloc(size, GFP_KERNEL);
+	if (!rw_image)
+		return -ENOMEM;
+
+	ret = __arch_prepare_bpf_trampoline(im, rw_image, rw_image + size, image, m,
+					    flags, tlinks, func_addr);
+	if (ret < 0)
+		goto out;
+
+	tmp = bpf_arch_text_copy(image, rw_image, size);
+	if (IS_ERR(tmp))
+		ret = PTR_ERR(tmp);
+out:
+	kvfree(rw_image);
+	return ret;
+}
+
+int arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+			     struct bpf_tramp_links *tlinks, void *func_addr)
+{
+	struct bpf_tramp_image im;
+	void *image;
+	int ret;
+
+	/* Allocate a temporary buffer for __arch_prepare_bpf_trampoline().
+	 * This will NOT cause fragmentation in direct map, as we do not
+	 * call set_memory_*() on this buffer.
+	 *
+	 * We cannot use kvmalloc here, because we need image to be in
+	 * module memory range.
+	 */
+	image = bpf_jit_alloc_exec(PAGE_SIZE);
+	if (!image)
+		return -ENOMEM;
+
+	ret = __arch_prepare_bpf_trampoline(&im, image, image + PAGE_SIZE, image,
+					    m, flags, tlinks, func_addr);
+	bpf_jit_free_exec(image);
+	return ret;
+}
+
 static int emit_bpf_dispatcher(u8 **pprog, int a, int b, s64 *progs, u8 *image, u8 *buf)
 {
 	u8 *jg_reloc, *prog = *pprog;

include/linux/bpf.h

@@ -1098,10 +1098,17 @@ struct bpf_tramp_run_ctx;
  *      fexit = a set of program to run after original function
  */
 struct bpf_tramp_image;
-int arch_prepare_bpf_trampoline(struct bpf_tramp_image *tr, void *image, void *image_end,
+int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *image_end,
 				const struct btf_func_model *m, u32 flags,
 				struct bpf_tramp_links *tlinks,
-				void *orig_call);
+				void *func_addr);
+void *arch_alloc_bpf_trampoline(unsigned int size);
+void arch_free_bpf_trampoline(void *image, unsigned int size);
+void arch_protect_bpf_trampoline(void *image, unsigned int size);
+void arch_unprotect_bpf_trampoline(void *image, unsigned int size);
+int arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+			     struct bpf_tramp_links *tlinks, void *func_addr);
+
 u64 notrace __bpf_prog_enter_sleepable_recur(struct bpf_prog *prog,
 					     struct bpf_tramp_run_ctx *run_ctx);
 void notrace __bpf_prog_exit_sleepable_recur(struct bpf_prog *prog, u64 start,
@@ -1134,6 +1141,7 @@ enum bpf_tramp_prog_type {
 
 struct bpf_tramp_image {
 	void *image;
+	int size;
 	struct bpf_ksym ksym;
 	struct percpu_ref pcref;
 	void *ip_after_call;
@@ -1318,7 +1326,7 @@ int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_func
 void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from,
 				struct bpf_prog *to);
 /* Called only from JIT-enabled code, so there's no need for stubs. */
-void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym);
+void bpf_image_ksym_add(void *data, unsigned int size, struct bpf_ksym *ksym);
 void bpf_image_ksym_del(struct bpf_ksym *ksym);
 void bpf_ksym_add(struct bpf_ksym *ksym);
 void bpf_ksym_del(struct bpf_ksym *ksym);

include/linux/filter.h

@@ -1067,7 +1067,7 @@ struct bpf_binary_header *
 bpf_jit_binary_pack_hdr(const struct bpf_prog *fp);
 
 void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insns);
-void bpf_prog_pack_free(struct bpf_binary_header *hdr);
+void bpf_prog_pack_free(void *ptr, u32 size);
 
 static inline bool bpf_prog_kallsyms_verify_off(const struct bpf_prog *fp)
 {

kernel/bpf/bpf_struct_ops.c

@@ -355,6 +355,7 @@ int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks,
 				      void *image, void *image_end)
 {
 	u32 flags;
+	int size;
 
 	tlinks[BPF_TRAMP_FENTRY].links[0] = link;
 	tlinks[BPF_TRAMP_FENTRY].nr_links = 1;
@@ -362,6 +363,12 @@ int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks,
 	 * and it must be used alone.
 	 */
 	flags = model->ret_size > 0 ? BPF_TRAMP_F_RET_FENTRY_RET : 0;
+
+	size = arch_bpf_trampoline_size(model, flags, tlinks, NULL);
+	if (size < 0)
+		return size;
+	if (size > (unsigned long)image_end - (unsigned long)image)
+		return -E2BIG;
 	return arch_prepare_bpf_trampoline(NULL, image, image_end,
 					   model, flags, tlinks, NULL);
 }
@@ -515,7 +522,7 @@ static long bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 			if (err)
 				goto reset_unlock;
 		}
-		set_memory_rox((long)st_map->image, 1);
+		arch_protect_bpf_trampoline(st_map->image, PAGE_SIZE);
 		/* Let bpf_link handle registration & unregistration.
 		 *
 		 * Pair with smp_load_acquire() during lookup_elem().
@@ -524,7 +531,7 @@ static long bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 		goto unlock;
 	}
 
-	set_memory_rox((long)st_map->image, 1);
+	arch_protect_bpf_trampoline(st_map->image, PAGE_SIZE);
 	err = st_ops->reg(kdata);
 	if (likely(!err)) {
 		/* This refcnt increment on the map here after
@@ -547,8 +554,7 @@ static long bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 	 * there was a race in registering the struct_ops (under the same name) to
 	 * a sub-system through different struct_ops's maps.
 	 */
-	set_memory_nx((long)st_map->image, 1);
-	set_memory_rw((long)st_map->image, 1);
+	arch_unprotect_bpf_trampoline(st_map->image, PAGE_SIZE);
 
 reset_unlock:
 	bpf_struct_ops_map_put_progs(st_map);
@@ -616,7 +622,7 @@ static void __bpf_struct_ops_map_free(struct bpf_map *map)
 		bpf_struct_ops_map_put_progs(st_map);
 	bpf_map_area_free(st_map->links);
 	if (st_map->image) {
-		bpf_jit_free_exec(st_map->image);
+		arch_free_bpf_trampoline(st_map->image, PAGE_SIZE);
 		bpf_jit_uncharge_modmem(PAGE_SIZE);
 	}
 	bpf_map_area_free(st_map->uvalue);
@@ -691,7 +697,7 @@ static struct bpf_map *bpf_struct_ops_map_alloc(union bpf_attr *attr)
 		return ERR_PTR(ret);
 	}
 
-	st_map->image = bpf_jit_alloc_exec(PAGE_SIZE);
+	st_map->image = arch_alloc_bpf_trampoline(PAGE_SIZE);
 	if (!st_map->image) {
 		/* __bpf_struct_ops_map_free() uses st_map->image as flag
 		 * for "charged or not". In this case, we need to unchange
@@ -711,7 +717,6 @@ static struct bpf_map *bpf_struct_ops_map_alloc(union bpf_attr *attr)
 	}
 
 	mutex_init(&st_map->lock);
-	set_vm_flush_reset_perms(st_map->image);
 	bpf_map_init_from_attr(map, attr);
 
 	return map;

kernel/bpf/core.c

@@ -928,20 +928,20 @@ void *bpf_prog_pack_alloc(u32 size, bpf_jit_fill_hole_t bpf_fill_ill_insns)
 	return ptr;
 }
 
-void bpf_prog_pack_free(struct bpf_binary_header *hdr)
+void bpf_prog_pack_free(void *ptr, u32 size)
 {
 	struct bpf_prog_pack *pack = NULL, *tmp;
 	unsigned int nbits;
 	unsigned long pos;
 
 	mutex_lock(&pack_mutex);
-	if (hdr->size > BPF_PROG_PACK_SIZE) {
-		bpf_jit_free_exec(hdr);
+	if (size > BPF_PROG_PACK_SIZE) {
+		bpf_jit_free_exec(ptr);
 		goto out;
 	}
 
 	list_for_each_entry(tmp, &pack_list, list) {
-		if ((void *)hdr >= tmp->ptr && (tmp->ptr + BPF_PROG_PACK_SIZE) > (void *)hdr) {
+		if (ptr >= tmp->ptr && (tmp->ptr + BPF_PROG_PACK_SIZE) > ptr) {
 			pack = tmp;
 			break;
 		}
@@ -950,10 +950,10 @@ void bpf_prog_pack_free(struct bpf_binary_header *hdr)
 	if (WARN_ONCE(!pack, "bpf_prog_pack bug\n"))
 		goto out;
 
-	nbits = BPF_PROG_SIZE_TO_NBITS(hdr->size);
-	pos = ((unsigned long)hdr - (unsigned long)pack->ptr) >> BPF_PROG_CHUNK_SHIFT;
+	nbits = BPF_PROG_SIZE_TO_NBITS(size);
+	pos = ((unsigned long)ptr - (unsigned long)pack->ptr) >> BPF_PROG_CHUNK_SHIFT;
 
-	WARN_ONCE(bpf_arch_text_invalidate(hdr, hdr->size),
+	WARN_ONCE(bpf_arch_text_invalidate(ptr, size),
 		  "bpf_prog_pack bug: missing bpf_arch_text_invalidate?\n");
 
 	bitmap_clear(pack->bitmap, pos, nbits);
@@ -1100,8 +1100,7 @@ bpf_jit_binary_pack_alloc(unsigned int proglen, u8 **image_ptr,
 
 	*rw_header = kvmalloc(size, GFP_KERNEL);
 	if (!*rw_header) {
-		bpf_arch_text_copy(&ro_header->size, &size, sizeof(size));
-		bpf_prog_pack_free(ro_header);
+		bpf_prog_pack_free(ro_header, size);
 		bpf_jit_uncharge_modmem(size);
 		return NULL;
 	}
@@ -1132,7 +1131,7 @@ int bpf_jit_binary_pack_finalize(struct bpf_prog *prog,
 	kvfree(rw_header);
 
 	if (IS_ERR(ptr)) {
-		bpf_prog_pack_free(ro_header);
+		bpf_prog_pack_free(ro_header, ro_header->size);
 		return PTR_ERR(ptr);
 	}
 	return 0;
@@ -1153,7 +1152,7 @@ void bpf_jit_binary_pack_free(struct bpf_binary_header *ro_header,
 {
 	u32 size = ro_header->size;
 
-	bpf_prog_pack_free(ro_header);
+	bpf_prog_pack_free(ro_header, size);
 	kvfree(rw_header);
 	bpf_jit_uncharge_modmem(size);
 }

kernel/bpf/dispatcher.c

@@ -150,14 +150,11 @@ void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from,
 			goto out;
 		d->rw_image = bpf_jit_alloc_exec(PAGE_SIZE);
 		if (!d->rw_image) {
-			u32 size = PAGE_SIZE;
-
-			bpf_arch_text_copy(d->image, &size, sizeof(size));
-			bpf_prog_pack_free((struct bpf_binary_header *)d->image);
+			bpf_prog_pack_free(d->image, PAGE_SIZE);
 			d->image = NULL;
 			goto out;
 		}
-		bpf_image_ksym_add(d->image, &d->ksym);
+		bpf_image_ksym_add(d->image, PAGE_SIZE, &d->ksym);
 	}
 
 	prev_num_progs = d->num_progs;

kernel/bpf/trampoline.c

@@ -115,10 +115,10 @@ bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
 		(ptype == BPF_PROG_TYPE_LSM && eatype == BPF_LSM_MAC);
 }
 
-void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym)
+void bpf_image_ksym_add(void *data, unsigned int size, struct bpf_ksym *ksym)
 {
 	ksym->start = (unsigned long) data;
-	ksym->end = ksym->start + PAGE_SIZE;
+	ksym->end = ksym->start + size;
 	bpf_ksym_add(ksym);
 	perf_event_ksymbol(PERF_RECORD_KSYMBOL_TYPE_BPF, ksym->start,
 			   PAGE_SIZE, false, ksym->name);
@@ -254,8 +254,8 @@ bpf_trampoline_get_progs(const struct bpf_trampoline *tr, int *total, bool *ip_a
 static void bpf_tramp_image_free(struct bpf_tramp_image *im)
 {
 	bpf_image_ksym_del(&im->ksym);
-	bpf_jit_free_exec(im->image);
-	bpf_jit_uncharge_modmem(PAGE_SIZE);
+	arch_free_bpf_trampoline(im->image, im->size);
+	bpf_jit_uncharge_modmem(im->size);
 	percpu_ref_exit(&im->pcref);
 	kfree_rcu(im, rcu);
 }
@@ -349,7 +349,7 @@ static void bpf_tramp_image_put(struct bpf_tramp_image *im)
 	call_rcu_tasks_trace(&im->rcu, __bpf_tramp_image_put_rcu_tasks);
 }
 
-static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key)
+static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key, int size)
 {
 	struct bpf_tramp_image *im;
 	struct bpf_ksym *ksym;
@@ -360,15 +360,15 @@ static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key)
 	if (!im)
 		goto out;
 
-	err = bpf_jit_charge_modmem(PAGE_SIZE);
+	err = bpf_jit_charge_modmem(size);
 	if (err)
 		goto out_free_im;
+	im->size = size;
 
 	err = -ENOMEM;
-	im->image = image = bpf_jit_alloc_exec(PAGE_SIZE);
+	im->image = image = arch_alloc_bpf_trampoline(size);
 	if (!image)
 		goto out_uncharge;
-	set_vm_flush_reset_perms(image);
 
 	err = percpu_ref_init(&im->pcref, __bpf_tramp_image_release, 0, GFP_KERNEL);
 	if (err)
@@ -377,13 +377,13 @@ static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key)
 	ksym = &im->ksym;
 	INIT_LIST_HEAD_RCU(&ksym->lnode);
 	snprintf(ksym->name, KSYM_NAME_LEN, "bpf_trampoline_%llu", key);
-	bpf_image_ksym_add(image, ksym);
+	bpf_image_ksym_add(image, size, ksym);
 	return im;
 
 out_free_image:
-	bpf_jit_free_exec(im->image);
+	arch_free_bpf_trampoline(im->image, im->size);
 out_uncharge:
-	bpf_jit_uncharge_modmem(PAGE_SIZE);
+	bpf_jit_uncharge_modmem(size);
 out_free_im:
 	kfree(im);
 out:
@@ -396,7 +396,7 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut
 	struct bpf_tramp_links *tlinks;
 	u32 orig_flags = tr->flags;
 	bool ip_arg = false;
-	int err, total;
+	int err, total, size;
 
 	tlinks = bpf_trampoline_get_progs(tr, &total, &ip_arg);
 	if (IS_ERR(tlinks))
@@ -409,12 +409,6 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut
 		goto out;
 	}
 
-	im = bpf_tramp_image_alloc(tr->key);
-	if (IS_ERR(im)) {
-		err = PTR_ERR(im);
-		goto out;
-	}
-
 	/* clear all bits except SHARE_IPMODIFY and TAIL_CALL_CTX */
 	tr->flags &= (BPF_TRAMP_F_SHARE_IPMODIFY | BPF_TRAMP_F_TAIL_CALL_CTX);
 
@@ -438,13 +432,31 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut
 		tr->flags |= BPF_TRAMP_F_ORIG_STACK;
 #endif
 
-	err = arch_prepare_bpf_trampoline(im, im->image, im->image + PAGE_SIZE,
+	size = arch_bpf_trampoline_size(&tr->func.model, tr->flags,
+					tlinks, tr->func.addr);
+	if (size < 0) {
+		err = size;
+		goto out;
+	}
+
+	if (size > PAGE_SIZE) {
+		err = -E2BIG;
+		goto out;
+	}
+
+	im = bpf_tramp_image_alloc(tr->key, size);
+	if (IS_ERR(im)) {
+		err = PTR_ERR(im);
+		goto out;
+	}
+
+	err = arch_prepare_bpf_trampoline(im, im->image, im->image + size,
 					  &tr->func.model, tr->flags, tlinks,
 					  tr->func.addr);
 	if (err < 0)
 		goto out_free;
 
-	set_memory_rox((long)im->image, 1);
+	arch_protect_bpf_trampoline(im->image, im->size);
 
 	WARN_ON(tr->cur_image && total == 0);
 	if (tr->cur_image)
@@ -464,9 +476,8 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut
 		tr->fops->func = NULL;
 		tr->fops->trampoline = 0;
 
-		/* reset im->image memory attr for arch_prepare_bpf_trampoline */
-		set_memory_nx((long)im->image, 1);
-		set_memory_rw((long)im->image, 1);
+		/* free im memory and reallocate later */
+		bpf_tramp_image_free(im);
 		goto again;
 	}
 #endif
@@ -1032,10 +1043,50 @@ bpf_trampoline_exit_t bpf_trampoline_exit(const struct bpf_prog *prog)
 }
 
 int __weak
-arch_prepare_bpf_trampoline(struct bpf_tramp_image *tr, void *image, void *image_end,
+arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *image_end,
 			    const struct btf_func_model *m, u32 flags,
 			    struct bpf_tramp_links *tlinks,
-			    void *orig_call)
+			    void *func_addr)
+{
+	return -ENOTSUPP;
+}
+
+void * __weak arch_alloc_bpf_trampoline(unsigned int size)
+{
+	void *image;
+
+	if (WARN_ON_ONCE(size > PAGE_SIZE))
+		return NULL;
+	image = bpf_jit_alloc_exec(PAGE_SIZE);
+	if (image)
+		set_vm_flush_reset_perms(image);
+	return image;
+}
+
+void __weak arch_free_bpf_trampoline(void *image, unsigned int size)
+{
+	WARN_ON_ONCE(size > PAGE_SIZE);
+	/* bpf_jit_free_exec doesn't need "size", but
+	 * bpf_prog_pack_free() needs it.
+	 */
+	bpf_jit_free_exec(image);
+}
+
+void __weak arch_protect_bpf_trampoline(void *image, unsigned int size)
+{
+	WARN_ON_ONCE(size > PAGE_SIZE);
+	set_memory_rox((long)image, 1);
+}
+
+void __weak arch_unprotect_bpf_trampoline(void *image, unsigned int size)
+{
+	WARN_ON_ONCE(size > PAGE_SIZE);
+	set_memory_nx((long)image, 1);
+	set_memory_rw((long)image, 1);
+}
+
+int __weak arch_bpf_trampoline_size(const struct btf_func_model *m, u32 flags,
+				    struct bpf_tramp_links *tlinks, void *func_addr)
 {
 	return -ENOTSUPP;
 }

net/bpf/bpf_dummy_struct_ops.c

@@ -101,12 +101,11 @@ int bpf_struct_ops_test_run(struct bpf_prog *prog, const union bpf_attr *kattr,
 		goto out;
 	}
 
-	image = bpf_jit_alloc_exec(PAGE_SIZE);
+	image = arch_alloc_bpf_trampoline(PAGE_SIZE);
 	if (!image) {
 		err = -ENOMEM;
 		goto out;
 	}
-	set_vm_flush_reset_perms(image);
 
 	link = kzalloc(sizeof(*link), GFP_USER);
 	if (!link) {
@@ -124,7 +123,7 @@ int bpf_struct_ops_test_run(struct bpf_prog *prog, const union bpf_attr *kattr,
 	if (err < 0)
 		goto out;
 
-	set_memory_rox((long)image, 1);
+	arch_protect_bpf_trampoline(image, PAGE_SIZE);
 	prog_ret = dummy_ops_call_op(image, args);
 
 	err = dummy_ops_copy_args(args);
@@ -134,7 +133,7 @@ int bpf_struct_ops_test_run(struct bpf_prog *prog, const union bpf_attr *kattr,
 		err = -EFAULT;
 out:
 	kfree(args);
-	bpf_jit_free_exec(image);
+	arch_free_bpf_trampoline(image, PAGE_SIZE);
 	if (link)
 		bpf_link_put(&link->link);
 	kfree(tlinks);