Merge branch 'bpf-multi-prog-improvements'

Sandipan Das says:

====================
[1] Support for bpf-to-bpf function calls in the powerpc64 JIT compiler.

[2] Provide a way for resolving function calls because of the way JITed
    images are allocated in powerpc64.

[3] Fix to get JITed instruction dumps for multi-function programs from
    the bpf system call.

[4] Fix for bpftool to show delimited multi-function JITed image dumps.

v4:
 - Incorporate review comments from Jakub.
 - Fix JSON output for bpftool.

v3:
 - Change base tree tag to bpf-next.
 - Incorporate review comments from Alexei, Daniel and Jakub.
 - Make sure that the JITed image does not grow or shrink after
   the last pass due to the way the instruction sequence used
   to load a callee's address maybe optimized.
 - Make additional changes to the bpf system call and bpftool to
   make multi-function JITed dumps easier to correlate.

v2:
 - Incorporate review comments from Jakub.
====================
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

arch/powerpc/net/bpf_jit_comp64.c

@@ -167,25 +167,37 @@ static void bpf_jit_build_epilogue(u32 *image, struct codegen_context *ctx)
 
 static void bpf_jit_emit_func_call(u32 *image, struct codegen_context *ctx, u64 func)
 {
+	unsigned int i, ctx_idx = ctx->idx;
+
+	/* Load function address into r12 */
+	PPC_LI64(12, func);
+
+	/* For bpf-to-bpf function calls, the callee's address is unknown
+	 * until the last extra pass. As seen above, we use PPC_LI64() to
+	 * load the callee's address, but this may optimize the number of
+	 * instructions required based on the nature of the address.
+	 *
+	 * Since we don't want the number of instructions emitted to change,
+	 * we pad the optimized PPC_LI64() call with NOPs to guarantee that
+	 * we always have a five-instruction sequence, which is the maximum
+	 * that PPC_LI64() can emit.
+	 */
+	for (i = ctx->idx - ctx_idx; i < 5; i++)
+		PPC_NOP();
+
 #ifdef PPC64_ELF_ABI_v1
-	/* func points to the function descriptor */
-	PPC_LI64(b2p[TMP_REG_2], func);
-	/* Load actual entry point from function descriptor */
-	PPC_BPF_LL(b2p[TMP_REG_1], b2p[TMP_REG_2], 0);
-	/* ... and move it to LR */
-	PPC_MTLR(b2p[TMP_REG_1]);
 	/*
 	 * Load TOC from function descriptor at offset 8.
 	 * We can clobber r2 since we get called through a
 	 * function pointer (so caller will save/restore r2)
 	 * and since we don't use a TOC ourself.
 	 */
-	PPC_BPF_LL(2, b2p[TMP_REG_2], 8);
-#else
-	/* We can clobber r12 */
-	PPC_FUNC_ADDR(12, func);
-	PPC_MTLR(12);
+	PPC_BPF_LL(2, 12, 8);
+	/* Load actual entry point from function descriptor */
+	PPC_BPF_LL(12, 12, 0);
 #endif
+
+	PPC_MTLR(12);
 	PPC_BLRL();
 }
 
@@ -256,7 +268,7 @@ static void bpf_jit_emit_tail_call(u32 *image, struct codegen_context *ctx, u32
 /* Assemble the body code between the prologue & epilogue */
 static int bpf_jit_build_body(struct bpf_prog *fp, u32 *image,
 			      struct codegen_context *ctx,
-			      u32 *addrs)
+			      u32 *addrs, bool extra_pass)
 {
 	const struct bpf_insn *insn = fp->insnsi;
 	int flen = fp->len;
@@ -712,11 +724,25 @@ static int bpf_jit_build_body(struct bpf_prog *fp, u32 *image,
 			break;
 
 		/*
-		 * Call kernel helper
+		 * Call kernel helper or bpf function
 		 */
 		case BPF_JMP | BPF_CALL:
 			ctx->seen |= SEEN_FUNC;
-			func = (u8 *) __bpf_call_base + imm;
+
+			/* bpf function call */
+			if (insn[i].src_reg == BPF_PSEUDO_CALL)
+				if (!extra_pass)
+					func = NULL;
+				else if (fp->aux->func && off < fp->aux->func_cnt)
+					/* use the subprog id from the off
+					 * field to lookup the callee address
+					 */
+					func = (u8 *) fp->aux->func[off]->bpf_func;
+				else
+					return -EINVAL;
+			/* kernel helper call */
+			else
+				func = (u8 *) __bpf_call_base + imm;
 
 			bpf_jit_emit_func_call(image, ctx, (u64)func);
 
@@ -864,6 +890,14 @@ static int bpf_jit_build_body(struct bpf_prog *fp, u32 *image,
 	return 0;
 }
 
+struct powerpc64_jit_data {
+	struct bpf_binary_header *header;
+	u32 *addrs;
+	u8 *image;
+	u32 proglen;
+	struct codegen_context ctx;
+};
+
 struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 {
 	u32 proglen;
@@ -871,6 +905,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	u8 *image = NULL;
 	u32 *code_base;
 	u32 *addrs;
+	struct powerpc64_jit_data *jit_data;
 	struct codegen_context cgctx;
 	int pass;
 	int flen;
@@ -878,6 +913,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	struct bpf_prog *org_fp = fp;
 	struct bpf_prog *tmp_fp;
 	bool bpf_blinded = false;
+	bool extra_pass = false;
 
 	if (!fp->jit_requested)
 		return org_fp;
@@ -891,11 +927,32 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 		fp = tmp_fp;
 	}
 
+	jit_data = fp->aux->jit_data;
+	if (!jit_data) {
+		jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL);
+		if (!jit_data) {
+			fp = org_fp;
+			goto out;
+		}
+		fp->aux->jit_data = jit_data;
+	}
+
 	flen = fp->len;
+	addrs = jit_data->addrs;
+	if (addrs) {
+		cgctx = jit_data->ctx;
+		image = jit_data->image;
+		bpf_hdr = jit_data->header;
+		proglen = jit_data->proglen;
+		alloclen = proglen + FUNCTION_DESCR_SIZE;
+		extra_pass = true;
+		goto skip_init_ctx;
+	}
+
 	addrs = kzalloc((flen+1) * sizeof(*addrs), GFP_KERNEL);
 	if (addrs == NULL) {
 		fp = org_fp;
-		goto out;
+		goto out_addrs;
 	}
 
 	memset(&cgctx, 0, sizeof(struct codegen_context));
@@ -904,10 +961,10 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	cgctx.stack_size = round_up(fp->aux->stack_depth, 16);
 
 	/* Scouting faux-generate pass 0 */
-	if (bpf_jit_build_body(fp, 0, &cgctx, addrs)) {
+	if (bpf_jit_build_body(fp, 0, &cgctx, addrs, false)) {
 		/* We hit something illegal or unsupported. */
 		fp = org_fp;
-		goto out;
+		goto out_addrs;
 	}
 
 	/*
@@ -925,9 +982,10 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 			bpf_jit_fill_ill_insns);
 	if (!bpf_hdr) {
 		fp = org_fp;
-		goto out;
+		goto out_addrs;
 	}
 
+skip_init_ctx:
 	code_base = (u32 *)(image + FUNCTION_DESCR_SIZE);
 
 	/* Code generation passes 1-2 */
@@ -935,7 +993,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 		/* Now build the prologue, body code & epilogue for real. */
 		cgctx.idx = 0;
 		bpf_jit_build_prologue(code_base, &cgctx);
-		bpf_jit_build_body(fp, code_base, &cgctx, addrs);
+		bpf_jit_build_body(fp, code_base, &cgctx, addrs, extra_pass);
 		bpf_jit_build_epilogue(code_base, &cgctx);
 
 		if (bpf_jit_enable > 1)
@@ -961,10 +1019,20 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	fp->jited_len = alloclen;
 
 	bpf_flush_icache(bpf_hdr, (u8 *)bpf_hdr + (bpf_hdr->pages * PAGE_SIZE));
+	if (!fp->is_func || extra_pass) {
+out_addrs:
+		kfree(addrs);
+		kfree(jit_data);
+		fp->aux->jit_data = NULL;
+	} else {
+		jit_data->addrs = addrs;
+		jit_data->ctx = cgctx;
+		jit_data->proglen = proglen;
+		jit_data->image = image;
+		jit_data->header = bpf_hdr;
+	}
 
 out:
-	kfree(addrs);
-
 	if (bpf_blinded)
 		bpf_jit_prog_release_other(fp, fp == org_fp ? tmp_fp : org_fp);
 

include/uapi/linux/bpf.h

@@ -2205,6 +2205,10 @@ struct bpf_prog_info {
 	__u32 gpl_compatible:1;
 	__u64 netns_dev;
 	__u64 netns_ino;
+	__u32 nr_jited_ksyms;
+	__u32 nr_jited_func_lens;
+	__aligned_u64 jited_ksyms;
+	__aligned_u64 jited_func_lens;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {

kernel/bpf/syscall.c

@@ -1933,6 +1933,7 @@ static int bpf_prog_get_info_by_fd(struct bpf_prog *prog,
 	if (!capable(CAP_SYS_ADMIN)) {
 		info.jited_prog_len = 0;
 		info.xlated_prog_len = 0;
+		info.nr_jited_ksyms = 0;
 		goto done;
 	}
 
@@ -1969,18 +1970,93 @@ static int bpf_prog_get_info_by_fd(struct bpf_prog *prog,
 	 * for offload.
 	 */
 	ulen = info.jited_prog_len;
-	info.jited_prog_len = prog->jited_len;
+	if (prog->aux->func_cnt) {
+		u32 i;
+
+		info.jited_prog_len = 0;
+		for (i = 0; i < prog->aux->func_cnt; i++)
+			info.jited_prog_len += prog->aux->func[i]->jited_len;
+	} else {
+		info.jited_prog_len = prog->jited_len;
+	}
+
 	if (info.jited_prog_len && ulen) {
 		if (bpf_dump_raw_ok()) {
 			uinsns = u64_to_user_ptr(info.jited_prog_insns);
 			ulen = min_t(u32, info.jited_prog_len, ulen);
-			if (copy_to_user(uinsns, prog->bpf_func, ulen))
-				return -EFAULT;
+
+			/* for multi-function programs, copy the JITed
+			 * instructions for all the functions
+			 */
+			if (prog->aux->func_cnt) {
+				u32 len, free, i;
+				u8 *img;
+
+				free = ulen;
+				for (i = 0; i < prog->aux->func_cnt; i++) {
+					len = prog->aux->func[i]->jited_len;
+					len = min_t(u32, len, free);
+					img = (u8 *) prog->aux->func[i]->bpf_func;
+					if (copy_to_user(uinsns, img, len))
+						return -EFAULT;
+					uinsns += len;
+					free -= len;
+					if (!free)
+						break;
+				}
+			} else {
+				if (copy_to_user(uinsns, prog->bpf_func, ulen))
+					return -EFAULT;
+			}
 		} else {
 			info.jited_prog_insns = 0;
 		}
 	}
 
+	ulen = info.nr_jited_ksyms;
+	info.nr_jited_ksyms = prog->aux->func_cnt;
+	if (info.nr_jited_ksyms && ulen) {
+		if (bpf_dump_raw_ok()) {
+			u64 __user *user_ksyms;
+			ulong ksym_addr;
+			u32 i;
+
+			/* copy the address of the kernel symbol
+			 * corresponding to each function
+			 */
+			ulen = min_t(u32, info.nr_jited_ksyms, ulen);
+			user_ksyms = u64_to_user_ptr(info.jited_ksyms);
+			for (i = 0; i < ulen; i++) {
+				ksym_addr = (ulong) prog->aux->func[i]->bpf_func;
+				ksym_addr &= PAGE_MASK;
+				if (put_user((u64) ksym_addr, &user_ksyms[i]))
+					return -EFAULT;
+			}
+		} else {
+			info.jited_ksyms = 0;
+		}
+	}
+
+	ulen = info.nr_jited_func_lens;
+	info.nr_jited_func_lens = prog->aux->func_cnt;
+	if (info.nr_jited_func_lens && ulen) {
+		if (bpf_dump_raw_ok()) {
+			u32 __user *user_lens;
+			u32 func_len, i;
+
+			/* copy the JITed image lengths for each function */
+			ulen = min_t(u32, info.nr_jited_func_lens, ulen);
+			user_lens = u64_to_user_ptr(info.jited_func_lens);
+			for (i = 0; i < ulen; i++) {
+				func_len = prog->aux->func[i]->jited_len;
+				if (put_user(func_len, &user_lens[i]))
+					return -EFAULT;
+			}
+		} else {
+			info.jited_func_lens = 0;
+		}
+	}
+
 done:
 	if (copy_to_user(uinfo, &info, info_len) ||
 	    put_user(info_len, &uattr->info.info_len))

kernel/bpf/verifier.c

@@ -5383,11 +5383,24 @@ static int jit_subprogs(struct bpf_verifier_env *env)
 			    insn->src_reg != BPF_PSEUDO_CALL)
 				continue;
 			subprog = insn->off;
-			insn->off = 0;
 			insn->imm = (u64 (*)(u64, u64, u64, u64, u64))
 				func[subprog]->bpf_func -
 				__bpf_call_base;
 		}
+
+		/* we use the aux data to keep a list of the start addresses
+		 * of the JITed images for each function in the program
+		 *
+		 * for some architectures, such as powerpc64, the imm field
+		 * might not be large enough to hold the offset of the start
+		 * address of the callee's JITed image from __bpf_call_base
+		 *
+		 * in such cases, we can lookup the start address of a callee
+		 * by using its subprog id, available from the off field of
+		 * the call instruction, as an index for this list
+		 */
+		func[i]->aux->func = func;
+		func[i]->aux->func_cnt = env->subprog_cnt;
 	}
 	for (i = 0; i < env->subprog_cnt; i++) {
 		old_bpf_func = func[i]->bpf_func;
@@ -5413,17 +5426,12 @@ static int jit_subprogs(struct bpf_verifier_env *env)
 	 * later look the same as if they were interpreted only.
 	 */
 	for (i = 0, insn = prog->insnsi; i < prog->len; i++, insn++) {
-		unsigned long addr;
-
 		if (insn->code != (BPF_JMP | BPF_CALL) ||
 		    insn->src_reg != BPF_PSEUDO_CALL)
 			continue;
 		insn->off = env->insn_aux_data[i].call_imm;
 		subprog = find_subprog(env, i + insn->off + 1);
-		addr  = (unsigned long)func[subprog]->bpf_func;
-		addr &= PAGE_MASK;
-		insn->imm = (u64 (*)(u64, u64, u64, u64, u64))
-			    addr - __bpf_call_base;
+		insn->imm = subprog;
 	}
 
 	prog->jited = 1;

tools/bpf/bpftool/prog.c

@@ -420,7 +420,11 @@ static int do_show(int argc, char **argv)
 
 static int do_dump(int argc, char **argv)
 {
+	unsigned long *func_ksyms = NULL;
 	struct bpf_prog_info info = {};
+	unsigned int *func_lens = NULL;
+	unsigned int nr_func_ksyms;
+	unsigned int nr_func_lens;
 	struct dump_data dd = {};
 	__u32 len = sizeof(info);
 	unsigned int buf_size;
@@ -496,10 +500,34 @@ static int do_dump(int argc, char **argv)
 		return -1;
 	}
 
+	nr_func_ksyms = info.nr_jited_ksyms;
+	if (nr_func_ksyms) {
+		func_ksyms = malloc(nr_func_ksyms * sizeof(__u64));
+		if (!func_ksyms) {
+			p_err("mem alloc failed");
+			close(fd);
+			goto err_free;
+		}
+	}
+
+	nr_func_lens = info.nr_jited_func_lens;
+	if (nr_func_lens) {
+		func_lens = malloc(nr_func_lens * sizeof(__u32));
+		if (!func_lens) {
+			p_err("mem alloc failed");
+			close(fd);
+			goto err_free;
+		}
+	}
+
 	memset(&info, 0, sizeof(info));
 
 	*member_ptr = ptr_to_u64(buf);
 	*member_len = buf_size;
+	info.jited_ksyms = ptr_to_u64(func_ksyms);
+	info.nr_jited_ksyms = nr_func_ksyms;
+	info.jited_func_lens = ptr_to_u64(func_lens);
+	info.nr_jited_func_lens = nr_func_lens;
 
 	err = bpf_obj_get_info_by_fd(fd, &info, &len);
 	close(fd);
@@ -513,6 +541,16 @@ static int do_dump(int argc, char **argv)
 		goto err_free;
 	}
 
+	if (info.nr_jited_ksyms > nr_func_ksyms) {
+		p_err("too many addresses returned");
+		goto err_free;
+	}
+
+	if (info.nr_jited_func_lens > nr_func_lens) {
+		p_err("too many values returned");
+		goto err_free;
+	}
+
 	if ((member_len == &info.jited_prog_len &&
 	     info.jited_prog_insns == 0) ||
 	    (member_len == &info.xlated_prog_len &&
@@ -550,7 +588,57 @@ static int do_dump(int argc, char **argv)
 				goto err_free;
 		}
 
-		disasm_print_insn(buf, *member_len, opcodes, name);
+		if (info.nr_jited_func_lens && info.jited_func_lens) {
+			struct kernel_sym *sym = NULL;
+			char sym_name[SYM_MAX_NAME];
+			unsigned char *img = buf;
+			__u64 *ksyms = NULL;
+			__u32 *lens;
+			__u32 i;
+
+			if (info.nr_jited_ksyms) {
+				kernel_syms_load(&dd);
+				ksyms = (__u64 *) info.jited_ksyms;
+			}
+
+			if (json_output)
+				jsonw_start_array(json_wtr);
+
+			lens = (__u32 *) info.jited_func_lens;
+			for (i = 0; i < info.nr_jited_func_lens; i++) {
+				if (ksyms) {
+					sym = kernel_syms_search(&dd, ksyms[i]);
+					if (sym)
+						sprintf(sym_name, "%s", sym->name);
+					else
+						sprintf(sym_name, "0x%016llx", ksyms[i]);
+				} else {
+					strcpy(sym_name, "unknown");
+				}
+
+				if (json_output) {
+					jsonw_start_object(json_wtr);
+					jsonw_name(json_wtr, "name");
+					jsonw_string(json_wtr, sym_name);
+					jsonw_name(json_wtr, "insns");
+				} else {
+					printf("%s:\n", sym_name);
+				}
+
+				disasm_print_insn(img, lens[i], opcodes, name);
+				img += lens[i];
+
+				if (json_output)
+					jsonw_end_object(json_wtr);
+				else
+					printf("\n");
+			}
+
+			if (json_output)
+				jsonw_end_array(json_wtr);
+		} else {
+			disasm_print_insn(buf, *member_len, opcodes, name);
+		}
 	} else if (visual) {
 		if (json_output)
 			jsonw_null(json_wtr);
@@ -558,6 +646,9 @@ static int do_dump(int argc, char **argv)
 			dump_xlated_cfg(buf, *member_len);
 	} else {
 		kernel_syms_load(&dd);
+		dd.nr_jited_ksyms = info.nr_jited_ksyms;
+		dd.jited_ksyms = (__u64 *) info.jited_ksyms;
+
 		if (json_output)
 			dump_xlated_json(&dd, buf, *member_len, opcodes);
 		else
@@ -566,10 +657,14 @@ static int do_dump(int argc, char **argv)
 	}
 
 	free(buf);
+	free(func_ksyms);
+	free(func_lens);
 	return 0;
 
 err_free:
 	free(buf);
+	free(func_ksyms);
+	free(func_lens);
 	return -1;
 }
 

tools/bpf/bpftool/xlated_dumper.c

@@ -102,8 +102,8 @@ void kernel_syms_destroy(struct dump_data *dd)
 	free(dd->sym_mapping);
 }
 
-static struct kernel_sym *kernel_syms_search(struct dump_data *dd,
-					     unsigned long key)
+struct kernel_sym *kernel_syms_search(struct dump_data *dd,
+				      unsigned long key)
 {
 	struct kernel_sym sym = {
 		.address = key,
@@ -174,7 +174,11 @@ static const char *print_call_pcrel(struct dump_data *dd,
 				    unsigned long address,
 				    const struct bpf_insn *insn)
 {
-	if (sym)
+	if (!dd->nr_jited_ksyms)
+		/* Do not show address for interpreted programs */
+		snprintf(dd->scratch_buff, sizeof(dd->scratch_buff),
+			"%+d", insn->off);
+	else if (sym)
 		snprintf(dd->scratch_buff, sizeof(dd->scratch_buff),
 			 "%+d#%s", insn->off, sym->name);
 	else
@@ -203,6 +207,10 @@ static const char *print_call(void *private_data,
 	unsigned long address = dd->address_call_base + insn->imm;
 	struct kernel_sym *sym;
 
+	if (insn->src_reg == BPF_PSEUDO_CALL &&
+	    (__u32) insn->imm < dd->nr_jited_ksyms)
+		address = dd->jited_ksyms[insn->imm];
+
 	sym = kernel_syms_search(dd, address);
 	if (insn->src_reg == BPF_PSEUDO_CALL)
 		return print_call_pcrel(dd, sym, address, insn);

tools/bpf/bpftool/xlated_dumper.h

@@ -49,11 +49,14 @@ struct dump_data {
 	unsigned long address_call_base;
 	struct kernel_sym *sym_mapping;
 	__u32 sym_count;
+	__u64 *jited_ksyms;
+	__u32 nr_jited_ksyms;
 	char scratch_buff[SYM_MAX_NAME + 8];
 };
 
 void kernel_syms_load(struct dump_data *dd);
 void kernel_syms_destroy(struct dump_data *dd);
+struct kernel_sym *kernel_syms_search(struct dump_data *dd, unsigned long key);
 void dump_xlated_json(struct dump_data *dd, void *buf, unsigned int len,
 		      bool opcodes);
 void dump_xlated_plain(struct dump_data *dd, void *buf, unsigned int len,

tools/include/uapi/linux/bpf.h

@@ -2205,6 +2205,10 @@ struct bpf_prog_info {
 	__u32 gpl_compatible:1;
 	__u64 netns_dev;
 	__u64 netns_ino;
+	__u32 nr_jited_ksyms;
+	__u32 nr_jited_func_lens;
+	__aligned_u64 jited_ksyms;
+	__aligned_u64 jited_func_lens;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {