From 1c0c7074fefd769f62dda155e881ca90c9e3e75a Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Fri, 19 Jun 2020 16:16:56 -0700
Subject: libbpf: Add support for extracting kernel symbol addresses

Add support for another (in addition to existing Kconfig) special kind of
externs in BPF code, kernel symbol externs. Such externs allow BPF code to
"know" kernel symbol address and either use it for comparisons with kernel
data structures (e.g., struct file's f_op pointer, to distinguish different
kinds of file), or, with the help of bpf_probe_user_kernel(), to follow
pointers and read data from global variables. Kernel symbol addresses are
found through /proc/kallsyms, which should be present in the system.

Currently, such kernel symbol variables are typeless: they have to be defined
as `extern const void <symbol>` and the only operation you can do (in C code)
with them is to take its address. Such extern should reside in a special
section '.ksyms'. bpf_helpers.h header provides __ksym macro for this. Strong
vs weak semantics stays the same as with Kconfig externs. If symbol is not
found in /proc/kallsyms, this will be a failure for strong (non-weak) extern,
but will be defaulted to 0 for weak externs.

If the same symbol is defined multiple times in /proc/kallsyms, then it will
be error if any of the associated addresses differs. In that case, address is
ambiguous, so libbpf falls on the side of caution, rather than confusing user
with randomly chosen address.

In the future, once kernel is extended with variables BTF information, such
ksym externs will be supported in a typed version, which will allow BPF
program to read variable's contents directly, similarly to how it's done for
fentry/fexit input arguments.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Hao Luo <haoluo@google.com>
Link: https://lore.kernel.org/bpf/20200619231703.738941-3-andriin@fb.com
---
 tools/lib/bpf/btf.h | 5 +++++
 1 file changed, 5 insertions(+)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 70c1b7ec2bd0..06cd1731c154 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -168,6 +168,11 @@ static inline bool btf_kflag(const struct btf_type *t)
 	return BTF_INFO_KFLAG(t->info);
 }
 
+static inline bool btf_is_void(const struct btf_type *t)
+{
+	return btf_kind(t) == BTF_KIND_UNKN;
+}
+
 static inline bool btf_is_int(const struct btf_type *t)
 {
 	return btf_kind(t) == BTF_KIND_INT;
-- 
cgit v1.2.3


From 81372e121802fd57892a0b44d93cc747d9568627 Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Tue, 7 Jul 2020 18:53:14 -0700
Subject: libbpf: Add btf__set_fd() for more control over loaded BTF FD

Add setter for BTF FD to allow application more fine-grained control in more
advanced scenarios. Storing BTF FD inside `struct btf` provides little benefit
and probably would be better done differently (e.g., btf__load() could just
return FD on success), but we are stuck with this due to backwards
compatibility. The main problem is that it's impossible to load BTF and than
free user-space memory, but keep FD intact, because `struct btf` assumes
ownership of that FD upon successful load and will attempt to close it during
btf__free(). To allow callers (e.g., libbpf itself for BTF sanitization) to
have more control over this, add btf__set_fd() to allow to reset FD
arbitrarily, if necessary.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200708015318.3827358-3-andriin@fb.com
---
 tools/lib/bpf/btf.c      | 7 ++++++-
 tools/lib/bpf/btf.h      | 1 +
 tools/lib/bpf/libbpf.map | 1 +
 3 files changed, 8 insertions(+), 1 deletion(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
index bfef3d606b54..c8861c9e3635 100644
--- a/tools/lib/bpf/btf.c
+++ b/tools/lib/bpf/btf.c
@@ -389,7 +389,7 @@ void btf__free(struct btf *btf)
 	if (!btf)
 		return;
 
-	if (btf->fd != -1)
+	if (btf->fd >= 0)
 		close(btf->fd);
 
 	free(btf->data);
@@ -700,6 +700,11 @@ int btf__fd(const struct btf *btf)
 	return btf->fd;
 }
 
+void btf__set_fd(struct btf *btf, int fd)
+{
+	btf->fd = fd;
+}
+
 const void *btf__get_raw_data(const struct btf *btf, __u32 *size)
 {
 	*size = btf->data_size;
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 06cd1731c154..173eff23c472 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -79,6 +79,7 @@ LIBBPF_API __s64 btf__resolve_size(const struct btf *btf, __u32 type_id);
 LIBBPF_API int btf__resolve_type(const struct btf *btf, __u32 type_id);
 LIBBPF_API int btf__align_of(const struct btf *btf, __u32 id);
 LIBBPF_API int btf__fd(const struct btf *btf);
+LIBBPF_API void btf__set_fd(struct btf *btf, int fd);
 LIBBPF_API const void *btf__get_raw_data(const struct btf *btf, __u32 *size);
 LIBBPF_API const char *btf__name_by_offset(const struct btf *btf, __u32 offset);
 LIBBPF_API int btf__get_from_id(__u32 id, struct btf **btf);
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index 6544d2cd1ed6..c5d5c7664c3b 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -288,4 +288,5 @@ LIBBPF_0.1.0 {
 		bpf_map__value_size;
 		bpf_program__autoload;
 		bpf_program__set_autoload;
+		btf__set_fd;
 } LIBBPF_0.0.9;
-- 
cgit v1.2.3


From 5c3320d7fece4612d4a413aa3c8e82cdb5b49fcb Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Thu, 9 Jul 2020 18:10:23 -0700
Subject: libbpf: Fix memory leak and optimize BTF sanitization

Coverity's static analysis helpfully reported a memory leak introduced by
0f0e55d8247c ("libbpf: Improve BTF sanitization handling"). While fixing it,
I realized that btf__new() already creates a memory copy, so there is no need
to do this. So this patch also fixes misleading btf__new() signature to make
data into a `const void *` input parameter. And it avoids unnecessary memory
allocation and copy in BTF sanitization code altogether.

Fixes: 0f0e55d8247c ("libbpf: Improve BTF sanitization handling")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200710011023.1655008-1-andriin@fb.com
---
 tools/lib/bpf/btf.c    |  2 +-
 tools/lib/bpf/btf.h    |  2 +-
 tools/lib/bpf/libbpf.c | 11 +++--------
 3 files changed, 5 insertions(+), 10 deletions(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
index c8861c9e3635..c9e760e120dc 100644
--- a/tools/lib/bpf/btf.c
+++ b/tools/lib/bpf/btf.c
@@ -397,7 +397,7 @@ void btf__free(struct btf *btf)
 	free(btf);
 }
 
-struct btf *btf__new(__u8 *data, __u32 size)
+struct btf *btf__new(const void *data, __u32 size)
 {
 	struct btf *btf;
 	int err;
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 173eff23c472..a3b7ef9b737f 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -63,7 +63,7 @@ struct btf_ext_header {
 };
 
 LIBBPF_API void btf__free(struct btf *btf);
-LIBBPF_API struct btf *btf__new(__u8 *data, __u32 size);
+LIBBPF_API struct btf *btf__new(const void *data, __u32 size);
 LIBBPF_API struct btf *btf__parse_elf(const char *path,
 				      struct btf_ext **btf_ext);
 LIBBPF_API int btf__finalize_data(struct bpf_object *obj, struct btf *btf);
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 6602eb479596..25e4f77be8d7 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -2533,17 +2533,12 @@ static int bpf_object__sanitize_and_load_btf(struct bpf_object *obj)
 
 	sanitize = btf_needs_sanitization(obj);
 	if (sanitize) {
-		const void *orig_data;
-		void *san_data;
+		const void *raw_data;
 		__u32 sz;
 
 		/* clone BTF to sanitize a copy and leave the original intact */
-		orig_data = btf__get_raw_data(obj->btf, &sz);
-		san_data = malloc(sz);
-		if (!san_data)
-			return -ENOMEM;
-		memcpy(san_data, orig_data, sz);
-		kern_btf = btf__new(san_data, sz);
+		raw_data = btf__get_raw_data(obj->btf, &sz);
+		kern_btf = btf__new(raw_data, sz);
 		if (IS_ERR(kern_btf))
 			return PTR_ERR(kern_btf);
 
-- 
cgit v1.2.3


From 7c819e701382d4969ca4837b8cbe157895f5d0bf Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Mon, 13 Jul 2020 16:24:08 -0700
Subject: libbpf: Support stripping modifiers for btf_dump

One important use case when emitting const/volatile/restrict is undesirable is
BPF skeleton generation of DATASEC layout. These are further memory-mapped and
can be written/read from user-space directly.

For important case of .rodata variables, bpftool strips away first-level
modifiers, to make their use on user-space side simple and not requiring extra
type casts to override compiler complaining about writing to const variables.

This logic works mostly fine, but breaks in some more complicated cases. E.g.:

    const volatile int params[10];

Because in BTF it's a chain of ARRAY -> CONST -> VOLATILE -> INT, bpftool
stops at ARRAY and doesn't strip CONST and VOLATILE. In skeleton this variable
will be emitted as is. So when used from user-space, compiler will complain
about writing to const array. This is problematic, as also mentioned in [0].

To solve this for arrays and other non-trivial cases (e.g., inner
const/volatile fields inside the struct), teach btf_dump to strip away any
modifier, when requested. This is done as an extra option on
btf_dump__emit_type_decl() API.

Reported-by: Anton Protopopov <a.s.protopopov@gmail.com>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200713232409.3062144-2-andriin@fb.com
---
 tools/lib/bpf/btf.h      |  2 ++
 tools/lib/bpf/btf_dump.c | 10 ++++++++--
 2 files changed, 10 insertions(+), 2 deletions(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index a3b7ef9b737f..227d3f844801 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -144,6 +144,8 @@ struct btf_dump_emit_type_decl_opts {
 	 * necessary indentation already
 	 */
 	int indent_level;
+	/* strip all the const/volatile/restrict mods */
+	bool strip_mods;
 };
 #define btf_dump_emit_type_decl_opts__last_field indent_level
 
diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c
index bbb430317260..e1c344504cae 100644
--- a/tools/lib/bpf/btf_dump.c
+++ b/tools/lib/bpf/btf_dump.c
@@ -60,6 +60,7 @@ struct btf_dump {
 	const struct btf_ext *btf_ext;
 	btf_dump_printf_fn_t printf_fn;
 	struct btf_dump_opts opts;
+	bool strip_mods;
 
 	/* per-type auxiliary state */
 	struct btf_dump_type_aux_state *type_states;
@@ -1032,7 +1033,9 @@ int btf_dump__emit_type_decl(struct btf_dump *d, __u32 id,
 
 	fname = OPTS_GET(opts, field_name, "");
 	lvl = OPTS_GET(opts, indent_level, 0);
+	d->strip_mods = OPTS_GET(opts, strip_mods, false);
 	btf_dump_emit_type_decl(d, id, fname, lvl);
+	d->strip_mods = false;
 	return 0;
 }
 
@@ -1045,6 +1048,10 @@ static void btf_dump_emit_type_decl(struct btf_dump *d, __u32 id,
 
 	stack_start = d->decl_stack_cnt;
 	for (;;) {
+		t = btf__type_by_id(d->btf, id);
+		if (d->strip_mods && btf_is_mod(t))
+			goto skip_mod;
+
 		err = btf_dump_push_decl_stack_id(d, id);
 		if (err < 0) {
 			/*
@@ -1056,12 +1063,11 @@ static void btf_dump_emit_type_decl(struct btf_dump *d, __u32 id,
 			d->decl_stack_cnt = stack_start;
 			return;
 		}
-
+skip_mod:
 		/* VOID */
 		if (id == 0)
 			break;
 
-		t = btf__type_by_id(d->btf, id);
 		switch (btf_kind(t)) {
 		case BTF_KIND_PTR:
 		case BTF_KIND_VOLATILE:
-- 
cgit v1.2.3


From 0b20933d8cfe2bf6473e9b581b5d1ed9a2117ecc Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Mon, 13 Jul 2020 16:24:09 -0700
Subject: tools/bpftool: Strip away modifiers from global variables

Reliably remove all the type modifiers from read-only (.rodata) global
variable definitions, including cases of inner field const modifiers and
arrays of const values.

Also modify one of selftests to ensure that const volatile struct doesn't
prevent user-space from modifying .rodata variable.

Fixes: 985ead416df3 ("bpftool: Add skeleton codegen command")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200713232409.3062144-3-andriin@fb.com
---
 tools/bpf/bpftool/gen.c                           | 23 ++++++++++-------------
 tools/lib/bpf/btf.h                               |  2 +-
 tools/testing/selftests/bpf/prog_tests/skeleton.c |  6 +++---
 tools/testing/selftests/bpf/progs/test_skeleton.c |  6 ++++--
 4 files changed, 18 insertions(+), 19 deletions(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/bpf/bpftool/gen.c b/tools/bpf/bpftool/gen.c
index 10de76b296ba..b59d26e89367 100644
--- a/tools/bpf/bpftool/gen.c
+++ b/tools/bpf/bpftool/gen.c
@@ -88,7 +88,7 @@ static const char *get_map_ident(const struct bpf_map *map)
 		return NULL;
 }
 
-static void codegen_btf_dump_printf(void *ct, const char *fmt, va_list args)
+static void codegen_btf_dump_printf(void *ctx, const char *fmt, va_list args)
 {
 	vprintf(fmt, args);
 }
@@ -104,17 +104,20 @@ static int codegen_datasec_def(struct bpf_object *obj,
 	int i, err, off = 0, pad_cnt = 0, vlen = btf_vlen(sec);
 	const char *sec_ident;
 	char var_ident[256];
+	bool strip_mods = false;
 
-	if (strcmp(sec_name, ".data") == 0)
+	if (strcmp(sec_name, ".data") == 0) {
 		sec_ident = "data";
-	else if (strcmp(sec_name, ".bss") == 0)
+	} else if (strcmp(sec_name, ".bss") == 0) {
 		sec_ident = "bss";
-	else if (strcmp(sec_name, ".rodata") == 0)
+	} else if (strcmp(sec_name, ".rodata") == 0) {
 		sec_ident = "rodata";
-	else if (strcmp(sec_name, ".kconfig") == 0)
+		strip_mods = true;
+	} else if (strcmp(sec_name, ".kconfig") == 0) {
 		sec_ident = "kconfig";
-	else
+	} else {
 		return 0;
+	}
 
 	printf("	struct %s__%s {\n", obj_name, sec_ident);
 	for (i = 0; i < vlen; i++, sec_var++) {
@@ -123,16 +126,10 @@ static int codegen_datasec_def(struct bpf_object *obj,
 		DECLARE_LIBBPF_OPTS(btf_dump_emit_type_decl_opts, opts,
 			.field_name = var_ident,
 			.indent_level = 2,
+			.strip_mods = strip_mods,
 		);
 		int need_off = sec_var->offset, align_off, align;
 		__u32 var_type_id = var->type;
-		const struct btf_type *t;
-
-		t = btf__type_by_id(btf, var_type_id);
-		while (btf_is_mod(t)) {
-			var_type_id = t->type;
-			t = btf__type_by_id(btf, var_type_id);
-		}
 
 		if (off > need_off) {
 			p_err("Something is wrong for %s's variable #%d: need offset %d, already at %d.\n",
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 227d3f844801..491c7b41ffdc 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -147,7 +147,7 @@ struct btf_dump_emit_type_decl_opts {
 	/* strip all the const/volatile/restrict mods */
 	bool strip_mods;
 };
-#define btf_dump_emit_type_decl_opts__last_field indent_level
+#define btf_dump_emit_type_decl_opts__last_field strip_mods
 
 LIBBPF_API int
 btf_dump__emit_type_decl(struct btf_dump *d, __u32 id,
diff --git a/tools/testing/selftests/bpf/prog_tests/skeleton.c b/tools/testing/selftests/bpf/prog_tests/skeleton.c
index fa153cf67b1b..fe87b77af459 100644
--- a/tools/testing/selftests/bpf/prog_tests/skeleton.c
+++ b/tools/testing/selftests/bpf/prog_tests/skeleton.c
@@ -41,7 +41,7 @@ void test_skeleton(void)
 	CHECK(bss->in4 != 0, "in4", "got %lld != exp %lld\n", bss->in4, 0LL);
 	CHECK(bss->out4 != 0, "out4", "got %lld != exp %lld\n", bss->out4, 0LL);
 
-	CHECK(rodata->in6 != 0, "in6", "got %d != exp %d\n", rodata->in6, 0);
+	CHECK(rodata->in.in6 != 0, "in6", "got %d != exp %d\n", rodata->in.in6, 0);
 	CHECK(bss->out6 != 0, "out6", "got %d != exp %d\n", bss->out6, 0);
 
 	/* validate we can pre-setup global variables, even in .bss */
@@ -49,7 +49,7 @@ void test_skeleton(void)
 	data->in2 = 11;
 	bss->in3 = 12;
 	bss->in4 = 13;
-	rodata->in6 = 14;
+	rodata->in.in6 = 14;
 
 	err = test_skeleton__load(skel);
 	if (CHECK(err, "skel_load", "failed to load skeleton: %d\n", err))
@@ -60,7 +60,7 @@ void test_skeleton(void)
 	CHECK(data->in2 != 11, "in2", "got %lld != exp %lld\n", data->in2, 11LL);
 	CHECK(bss->in3 != 12, "in3", "got %d != exp %d\n", bss->in3, 12);
 	CHECK(bss->in4 != 13, "in4", "got %lld != exp %lld\n", bss->in4, 13LL);
-	CHECK(rodata->in6 != 14, "in6", "got %d != exp %d\n", rodata->in6, 14);
+	CHECK(rodata->in.in6 != 14, "in6", "got %d != exp %d\n", rodata->in.in6, 14);
 
 	/* now set new values and attach to get them into outX variables */
 	data->in1 = 1;
diff --git a/tools/testing/selftests/bpf/progs/test_skeleton.c b/tools/testing/selftests/bpf/progs/test_skeleton.c
index 77ae86f44db5..374ccef704e1 100644
--- a/tools/testing/selftests/bpf/progs/test_skeleton.c
+++ b/tools/testing/selftests/bpf/progs/test_skeleton.c
@@ -20,7 +20,9 @@ long long in4 __attribute__((aligned(64))) = 0;
 struct s in5 = {};
 
 /* .rodata section */
-const volatile int in6 = 0;
+const volatile struct {
+	const int in6;
+} in = {};
 
 /* .data section */
 int out1 = -1;
@@ -46,7 +48,7 @@ int handler(const void *ctx)
 	out3 = in3;
 	out4 = in4;
 	out5 = in5;
-	out6 = in6;
+	out6 = in.in6;
 
 	bpf_syscall = CONFIG_BPF_SYSCALL;
 	kern_ver = LINUX_KERNEL_VERSION;
-- 
cgit v1.2.3


From 94a1fedd63edb672933bef44ca9213937e377c05 Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Sat, 1 Aug 2020 18:32:17 -0700
Subject: libbpf: Add btf__parse_raw() and generic btf__parse() APIs

Add public APIs to parse BTF from raw data file (e.g.,
/sys/kernel/btf/vmlinux), as well as generic btf__parse(), which will try to
determine correct format, currently either raw or ELF.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200802013219.864880-2-andriin@fb.com
---
 tools/lib/bpf/btf.c      | 114 ++++++++++++++++++++++++++++++++---------------
 tools/lib/bpf/btf.h      |   5 ++-
 tools/lib/bpf/libbpf.map |   2 +
 3 files changed, 83 insertions(+), 38 deletions(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
index ded5b29965f9..856b09a04563 100644
--- a/tools/lib/bpf/btf.c
+++ b/tools/lib/bpf/btf.c
@@ -562,6 +562,83 @@ done:
 	return btf;
 }
 
+struct btf *btf__parse_raw(const char *path)
+{
+	void *data = NULL;
+	struct btf *btf;
+	FILE *f = NULL;
+	__u16 magic;
+	int err = 0;
+	long sz;
+
+	f = fopen(path, "rb");
+	if (!f) {
+		err = -errno;
+		goto err_out;
+	}
+
+	/* check BTF magic */
+	if (fread(&magic, 1, sizeof(magic), f) < sizeof(magic)) {
+		err = -EIO;
+		goto err_out;
+	}
+	if (magic != BTF_MAGIC) {
+		/* definitely not a raw BTF */
+		err = -EPROTO;
+		goto err_out;
+	}
+
+	/* get file size */
+	if (fseek(f, 0, SEEK_END)) {
+		err = -errno;
+		goto err_out;
+	}
+	sz = ftell(f);
+	if (sz < 0) {
+		err = -errno;
+		goto err_out;
+	}
+	/* rewind to the start */
+	if (fseek(f, 0, SEEK_SET)) {
+		err = -errno;
+		goto err_out;
+	}
+
+	/* pre-alloc memory and read all of BTF data */
+	data = malloc(sz);
+	if (!data) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+	if (fread(data, 1, sz, f) < sz) {
+		err = -EIO;
+		goto err_out;
+	}
+
+	/* finally parse BTF data */
+	btf = btf__new(data, sz);
+
+err_out:
+	free(data);
+	if (f)
+		fclose(f);
+	return err ? ERR_PTR(err) : btf;
+}
+
+struct btf *btf__parse(const char *path, struct btf_ext **btf_ext)
+{
+	struct btf *btf;
+
+	if (btf_ext)
+		*btf_ext = NULL;
+
+	btf = btf__parse_raw(path);
+	if (!IS_ERR(btf) || PTR_ERR(btf) != -EPROTO)
+		return btf;
+
+	return btf__parse_elf(path, btf_ext);
+}
+
 static int compare_vsi_off(const void *_a, const void *_b)
 {
 	const struct btf_var_secinfo *a = _a;
@@ -2951,41 +3028,6 @@ static int btf_dedup_remap_types(struct btf_dedup *d)
 	return 0;
 }
 
-static struct btf *btf_load_raw(const char *path)
-{
-	struct btf *btf;
-	size_t read_cnt;
-	struct stat st;
-	void *data;
-	FILE *f;
-
-	if (stat(path, &st))
-		return ERR_PTR(-errno);
-
-	data = malloc(st.st_size);
-	if (!data)
-		return ERR_PTR(-ENOMEM);
-
-	f = fopen(path, "rb");
-	if (!f) {
-		btf = ERR_PTR(-errno);
-		goto cleanup;
-	}
-
-	read_cnt = fread(data, 1, st.st_size, f);
-	fclose(f);
-	if (read_cnt < st.st_size) {
-		btf = ERR_PTR(-EBADF);
-		goto cleanup;
-	}
-
-	btf = btf__new(data, read_cnt);
-
-cleanup:
-	free(data);
-	return btf;
-}
-
 /*
  * Probe few well-known locations for vmlinux kernel image and try to load BTF
  * data out of it to use for target BTF.
@@ -3021,7 +3063,7 @@ struct btf *libbpf_find_kernel_btf(void)
 			continue;
 
 		if (locations[i].raw_btf)
-			btf = btf_load_raw(path);
+			btf = btf__parse_raw(path);
 		else
 			btf = btf__parse_elf(path, NULL);
 
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 491c7b41ffdc..f4a1a1d2b9a3 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -64,8 +64,9 @@ struct btf_ext_header {
 
 LIBBPF_API void btf__free(struct btf *btf);
 LIBBPF_API struct btf *btf__new(const void *data, __u32 size);
-LIBBPF_API struct btf *btf__parse_elf(const char *path,
-				      struct btf_ext **btf_ext);
+LIBBPF_API struct btf *btf__parse(const char *path, struct btf_ext **btf_ext);
+LIBBPF_API struct btf *btf__parse_elf(const char *path, struct btf_ext **btf_ext);
+LIBBPF_API struct btf *btf__parse_raw(const char *path);
 LIBBPF_API int btf__finalize_data(struct bpf_object *obj, struct btf *btf);
 LIBBPF_API int btf__load(struct btf *btf);
 LIBBPF_API __s32 btf__find_by_name(const struct btf *btf,
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index 099863411f7d..0c4722bfdd0a 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -293,5 +293,7 @@ LIBBPF_0.1.0 {
 		bpf_program__is_sk_lookup;
 		bpf_program__set_autoload;
 		bpf_program__set_sk_lookup;
+		btf__parse;
+		btf__parse_raw;
 		btf__set_fd;
 } LIBBPF_0.0.9;
-- 
cgit v1.2.3


From 44ad23dfbccbcd26d6ca504eba1ac55755864969 Mon Sep 17 00:00:00 2001
From: Andrii Nakryiko <andriin@fb.com>
Date: Thu, 13 Aug 2020 13:49:40 -0700
Subject: libbpf: Handle BTF pointer sizes more carefully

With libbpf and BTF it is pretty common to have libbpf built for one
architecture, while BTF information was generated for a different architecture
(typically, but not always, BPF). In such case, the size of a pointer might
differ betweem architectures. libbpf previously was always making an
assumption that pointer size for BTF is the same as native architecture
pointer size, but that breaks for cases where libbpf is built as 32-bit
library, while BTF is for 64-bit architecture.

To solve this, add heuristic to determine pointer size by searching for `long`
or `unsigned long` integer type and using its size as a pointer size. Also,
allow to override the pointer size with a new API btf__set_pointer_size(), for
cases where application knows which pointer size should be used. User
application can check what libbpf "guessed" by looking at the result of
btf__pointer_size(). If it's not 0, then libbpf successfully determined a
pointer size, otherwise native arch pointer size will be used.

For cases where BTF is parsed from ELF file, use ELF's class (32-bit or
64-bit) to determine pointer size.

Fixes: 8a138aed4a80 ("bpf: btf: Add BTF support to libbpf")
Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200813204945.1020225-5-andriin@fb.com
---
 tools/lib/bpf/btf.c      | 83 ++++++++++++++++++++++++++++++++++++++++++++++--
 tools/lib/bpf/btf.h      |  2 ++
 tools/lib/bpf/btf_dump.c |  4 ++-
 tools/lib/bpf/libbpf.map |  2 ++
 4 files changed, 87 insertions(+), 4 deletions(-)

(limited to 'tools/lib/bpf/btf.h')

diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
index 4843e44916f7..7dfca7016aaa 100644
--- a/tools/lib/bpf/btf.c
+++ b/tools/lib/bpf/btf.c
@@ -41,6 +41,7 @@ struct btf {
 	__u32 types_size;
 	__u32 data_size;
 	int fd;
+	int ptr_sz;
 };
 
 static inline __u64 ptr_to_u64(const void *ptr)
@@ -221,6 +222,70 @@ const struct btf_type *btf__type_by_id(const struct btf *btf, __u32 type_id)
 	return btf->types[type_id];
 }
 
+static int determine_ptr_size(const struct btf *btf)
+{
+	const struct btf_type *t;
+	const char *name;
+	int i;
+
+	for (i = 1; i <= btf->nr_types; i++) {
+		t = btf__type_by_id(btf, i);
+		if (!btf_is_int(t))
+			continue;
+
+		name = btf__name_by_offset(btf, t->name_off);
+		if (!name)
+			continue;
+
+		if (strcmp(name, "long int") == 0 ||
+		    strcmp(name, "long unsigned int") == 0) {
+			if (t->size != 4 && t->size != 8)
+				continue;
+			return t->size;
+		}
+	}
+
+	return -1;
+}
+
+static size_t btf_ptr_sz(const struct btf *btf)
+{
+	if (!btf->ptr_sz)
+		((struct btf *)btf)->ptr_sz = determine_ptr_size(btf);
+	return btf->ptr_sz < 0 ? sizeof(void *) : btf->ptr_sz;
+}
+
+/* Return pointer size this BTF instance assumes. The size is heuristically
+ * determined by looking for 'long' or 'unsigned long' integer type and
+ * recording its size in bytes. If BTF type information doesn't have any such
+ * type, this function returns 0. In the latter case, native architecture's
+ * pointer size is assumed, so will be either 4 or 8, depending on
+ * architecture that libbpf was compiled for. It's possible to override
+ * guessed value by using btf__set_pointer_size() API.
+ */
+size_t btf__pointer_size(const struct btf *btf)
+{
+	if (!btf->ptr_sz)
+		((struct btf *)btf)->ptr_sz = determine_ptr_size(btf);
+
+	if (btf->ptr_sz < 0)
+		/* not enough BTF type info to guess */
+		return 0;
+
+	return btf->ptr_sz;
+}
+
+/* Override or set pointer size in bytes. Only values of 4 and 8 are
+ * supported.
+ */
+int btf__set_pointer_size(struct btf *btf, size_t ptr_sz)
+{
+	if (ptr_sz != 4 && ptr_sz != 8)
+		return -EINVAL;
+	btf->ptr_sz = ptr_sz;
+	return 0;
+}
+
 static bool btf_type_is_void(const struct btf_type *t)
 {
 	return t == &btf_void || btf_is_fwd(t);
@@ -253,7 +318,7 @@ __s64 btf__resolve_size(const struct btf *btf, __u32 type_id)
 			size = t->size;
 			goto done;
 		case BTF_KIND_PTR:
-			size = sizeof(void *);
+			size = btf_ptr_sz(btf);
 			goto done;
 		case BTF_KIND_TYPEDEF:
 		case BTF_KIND_VOLATILE:
@@ -293,9 +358,9 @@ int btf__align_of(const struct btf *btf, __u32 id)
 	switch (kind) {
 	case BTF_KIND_INT:
 	case BTF_KIND_ENUM:
-		return min(sizeof(void *), (size_t)t->size);
+		return min(btf_ptr_sz(btf), (size_t)t->size);
 	case BTF_KIND_PTR:
-		return sizeof(void *);
+		return btf_ptr_sz(btf);
 	case BTF_KIND_TYPEDEF:
 	case BTF_KIND_VOLATILE:
 	case BTF_KIND_CONST:
@@ -533,6 +598,18 @@ struct btf *btf__parse_elf(const char *path, struct btf_ext **btf_ext)
 	if (IS_ERR(btf))
 		goto done;
 
+	switch (gelf_getclass(elf)) {
+	case ELFCLASS32:
+		btf__set_pointer_size(btf, 4);
+		break;
+	case ELFCLASS64:
+		btf__set_pointer_size(btf, 8);
+		break;
+	default:
+		pr_warn("failed to get ELF class (bitness) for %s\n", path);
+		break;
+	}
+
 	if (btf_ext && btf_ext_data) {
 		*btf_ext = btf_ext__new(btf_ext_data->d_buf,
 					btf_ext_data->d_size);
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index f4a1a1d2b9a3..1ca14448df4c 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -76,6 +76,8 @@ LIBBPF_API __s32 btf__find_by_name_kind(const struct btf *btf,
 LIBBPF_API __u32 btf__get_nr_types(const struct btf *btf);
 LIBBPF_API const struct btf_type *btf__type_by_id(const struct btf *btf,
 						  __u32 id);
+LIBBPF_API size_t btf__pointer_size(const struct btf *btf);
+LIBBPF_API int btf__set_pointer_size(struct btf *btf, size_t ptr_sz);
 LIBBPF_API __s64 btf__resolve_size(const struct btf *btf, __u32 type_id);
 LIBBPF_API int btf__resolve_type(const struct btf *btf, __u32 type_id);
 LIBBPF_API int btf__align_of(const struct btf *btf, __u32 id);
diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c
index ac81f3f8957a..fe39bd774697 100644
--- a/tools/lib/bpf/btf_dump.c
+++ b/tools/lib/bpf/btf_dump.c
@@ -61,6 +61,7 @@ struct btf_dump {
 	const struct btf_ext *btf_ext;
 	btf_dump_printf_fn_t printf_fn;
 	struct btf_dump_opts opts;
+	int ptr_sz;
 	bool strip_mods;
 
 	/* per-type auxiliary state */
@@ -139,6 +140,7 @@ struct btf_dump *btf_dump__new(const struct btf *btf,
 	d->btf_ext = btf_ext;
 	d->printf_fn = printf_fn;
 	d->opts.ctx = opts ? opts->ctx : NULL;
+	d->ptr_sz = btf__pointer_size(btf) ? : sizeof(void *);
 
 	d->type_names = hashmap__new(str_hash_fn, str_equal_fn, NULL);
 	if (IS_ERR(d->type_names)) {
@@ -804,7 +806,7 @@ static void btf_dump_emit_bit_padding(const struct btf_dump *d,
 				      int align, int lvl)
 {
 	int off_diff = m_off - cur_off;
-	int ptr_bits = sizeof(void *) * 8;
+	int ptr_bits = d->ptr_sz * 8;
 
 	if (off_diff <= 0)
 		/* no gap */
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index 0c4722bfdd0a..e35bd6cdbdbf 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -295,5 +295,7 @@ LIBBPF_0.1.0 {
 		bpf_program__set_sk_lookup;
 		btf__parse;
 		btf__parse_raw;
+		btf__pointer_size;
 		btf__set_fd;
+		btf__set_pointer_size;
 } LIBBPF_0.0.9;
-- 
cgit v1.2.3