* [PATCH v2 bpf-next 1/4] libbpf: skip CO-RE relocations for not loaded BPF programs
2020-10-07 20:29 [PATCH v2 bpf-next 0/4] libbpf: auto-resize relocatable LOAD/STORE instructions Andrii Nakryiko
@ 2020-10-07 20:29 ` Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 2/4] libbpf: support safe subset of load/store instruction resizing with CO-RE Andrii Nakryiko
` (2 subsequent siblings)
3 siblings, 0 replies; 7+ messages in thread
From: Andrii Nakryiko @ 2020-10-07 20:29 UTC (permalink / raw)
To: bpf, netdev, ast, daniel; +Cc: andrii, kernel-team, Luka Perkov, Tony Ambardar
Bypass CO-RE relocations step for BPF programs that are not going to be
loaded. This allows to have BPF programs compiled in and disabled dynamically
if kernel is not supposed to provide enough relocation information. In such
case, there won't be unnecessary warnings about failed relocations.
Fixes: d929758101fc ("libbpf: Support disabling auto-loading BPF programs")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
tools/lib/bpf/libbpf.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index faec389c4849..07d62771472f 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -5765,6 +5765,11 @@ bpf_object__relocate_core(struct bpf_object *obj, const char *targ_btf_path)
err = -EINVAL;
goto out;
}
+ /* no need to apply CO-RE relocation if the program is
+ * not going to be loaded
+ */
+ if (!prog->load)
+ continue;
err = bpf_core_apply_relo(prog, rec, i, obj->btf,
targ_btf, cand_cache);
--
2.24.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH v2 bpf-next 2/4] libbpf: support safe subset of load/store instruction resizing with CO-RE
2020-10-07 20:29 [PATCH v2 bpf-next 0/4] libbpf: auto-resize relocatable LOAD/STORE instructions Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 1/4] libbpf: skip CO-RE relocations for not loaded BPF programs Andrii Nakryiko
@ 2020-10-07 20:29 ` Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 3/4] libbpf: allow specifying both ELF and raw BTF for CO-RE BTF override Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions Andrii Nakryiko
3 siblings, 0 replies; 7+ messages in thread
From: Andrii Nakryiko @ 2020-10-07 20:29 UTC (permalink / raw)
To: bpf, netdev, ast, daniel
Cc: andrii, kernel-team, Luka Perkov, Tony Ambardar, Andrii Nakryiko
From: Andrii Nakryiko <andriin@fb.com>
Add support for patching instructions of the following form:
- rX = *(T *)(rY + <off>);
- *(T *)(rX + <off>) = rY;
- *(T *)(rX + <off>) = <imm>, where T is one of {u8, u16, u32, u64}.
For such instructions, if the actual kernel field recorded in CO-RE relocation
has a different size than the one recorded locally (e.g., from vmlinux.h),
then libbpf will adjust T to an appropriate 1-, 2-, 4-, or 8-byte loads.
In general, such transformation is not always correct and could lead to
invalid final value being loaded or stored. But two classes of cases are
always safe:
- if both local and target (kernel) types are unsigned integers, but of
different sizes, then it's OK to adjust load/store instruction according to
the necessary memory size. Zero-extending nature of such instructions and
unsignedness make sure that the final value is always correct;
- pointer size mismatch between BPF target architecture (which is always
64-bit) and 32-bit host kernel architecture can be similarly resolved
automatically, because pointer is essentially an unsigned integer. Loading
32-bit pointer into 64-bit BPF register with zero extension will leave
correct pointer in the register.
Both cases are necessary to support CO-RE on 32-bit kernels, as `unsigned
long` in vmlinux.h generated from 32-bit kernel is 32-bit, but when compiled
with BPF program for BPF target it will be treated by compiler as 64-bit
integer. Similarly, pointers in vmlinux.h are 32-bit for kernel, but treated
as 64-bit values by compiler for BPF target. Both problems are now resolved by
libbpf for direct memory reads.
But similar transformations are useful in general when kernel fields are
"resized" from, e.g., unsigned int to unsigned long (or vice versa).
Now, similar transformations for signed integers are not safe to perform as
they will result in incorrect sign extension of the value. If such situation
is detected, libbpf will emit helpful message and will poison the instruction.
Not failing immediately means that it's possible to guard the instruction
based on kernel version (or other conditions) and make sure it's not
reachable.
If there is a need to read signed integers that change sizes between different
kernels, it's possible to use BPF_CORE_READ_BITFIELD() macro, which works both
with bitfields and non-bitfield integers of any signedness and handles
sign-extension properly. Also, bpf_core_read() with proper size and/or use of
bpf_core_field_size() relocation could allow to deal with such complicated
situations explicitly, if not so conventiently as direct memory reads.
Selftests added in a separate patch in progs/test_core_autosize.c demonstrate
both direct memory and probed use cases.
BPF_CORE_READ() is not changed and it won't deal with such situations as
automatically as direct memory reads due to the signedness integer
limitations, which are much harder to detect and control with compiler macro
magic. So it's encouraged to utilize direct memory reads as much as possible.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
tools/lib/bpf/libbpf.c | 144 ++++++++++++++++++++++++++++++++++++++---
1 file changed, 136 insertions(+), 8 deletions(-)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 07d62771472f..032cf0049ddb 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -5040,16 +5040,19 @@ static int bpf_core_spec_match(struct bpf_core_spec *local_spec,
static int bpf_core_calc_field_relo(const struct bpf_program *prog,
const struct bpf_core_relo *relo,
const struct bpf_core_spec *spec,
- __u32 *val, bool *validate)
+ __u32 *val, __u32 *field_sz, __u32 *type_id,
+ bool *validate)
{
const struct bpf_core_accessor *acc;
const struct btf_type *t;
- __u32 byte_off, byte_sz, bit_off, bit_sz;
+ __u32 byte_off, byte_sz, bit_off, bit_sz, field_type_id;
const struct btf_member *m;
const struct btf_type *mt;
bool bitfield;
__s64 sz;
+ *field_sz = 0;
+
if (relo->kind == BPF_FIELD_EXISTS) {
*val = spec ? 1 : 0;
return 0;
@@ -5065,6 +5068,12 @@ static int bpf_core_calc_field_relo(const struct bpf_program *prog,
if (!acc->name) {
if (relo->kind == BPF_FIELD_BYTE_OFFSET) {
*val = spec->bit_offset / 8;
+ /* remember field size for load/store mem size */
+ sz = btf__resolve_size(spec->btf, acc->type_id);
+ if (sz < 0)
+ return -EINVAL;
+ *field_sz = sz;
+ *type_id = acc->type_id;
} else if (relo->kind == BPF_FIELD_BYTE_SIZE) {
sz = btf__resolve_size(spec->btf, acc->type_id);
if (sz < 0)
@@ -5081,7 +5090,7 @@ static int bpf_core_calc_field_relo(const struct bpf_program *prog,
}
m = btf_members(t) + acc->idx;
- mt = skip_mods_and_typedefs(spec->btf, m->type, NULL);
+ mt = skip_mods_and_typedefs(spec->btf, m->type, &field_type_id);
bit_off = spec->bit_offset;
bit_sz = btf_member_bitfield_size(t, acc->idx);
@@ -5101,7 +5110,7 @@ static int bpf_core_calc_field_relo(const struct bpf_program *prog,
byte_off = bit_off / 8 / byte_sz * byte_sz;
}
} else {
- sz = btf__resolve_size(spec->btf, m->type);
+ sz = btf__resolve_size(spec->btf, field_type_id);
if (sz < 0)
return -EINVAL;
byte_sz = sz;
@@ -5119,6 +5128,10 @@ static int bpf_core_calc_field_relo(const struct bpf_program *prog,
switch (relo->kind) {
case BPF_FIELD_BYTE_OFFSET:
*val = byte_off;
+ if (!bitfield) {
+ *field_sz = byte_sz;
+ *type_id = field_type_id;
+ }
break;
case BPF_FIELD_BYTE_SIZE:
*val = byte_sz;
@@ -5219,6 +5232,19 @@ struct bpf_core_relo_res
bool poison;
/* some relocations can't be validated against orig_val */
bool validate;
+ /* for field byte offset relocations or the forms:
+ * *(T *)(rX + <off>) = rY
+ * rX = *(T *)(rY + <off>),
+ * we remember original and resolved field size to adjust direct
+ * memory loads of pointers and integers; this is necessary for 32-bit
+ * host kernel architectures, but also allows to automatically
+ * relocate fields that were resized from, e.g., u32 to u64, etc.
+ */
+ bool fail_memsz_adjust;
+ __u32 orig_sz;
+ __u32 orig_type_id;
+ __u32 new_sz;
+ __u32 new_type_id;
};
/* Calculate original and target relocation values, given local and target
@@ -5240,10 +5266,56 @@ static int bpf_core_calc_relo(const struct bpf_program *prog,
res->new_val = 0;
res->poison = false;
res->validate = true;
+ res->fail_memsz_adjust = false;
+ res->orig_sz = res->new_sz = 0;
+ res->orig_type_id = res->new_type_id = 0;
if (core_relo_is_field_based(relo->kind)) {
- err = bpf_core_calc_field_relo(prog, relo, local_spec, &res->orig_val, &res->validate);
- err = err ?: bpf_core_calc_field_relo(prog, relo, targ_spec, &res->new_val, NULL);
+ err = bpf_core_calc_field_relo(prog, relo, local_spec,
+ &res->orig_val, &res->orig_sz,
+ &res->orig_type_id, &res->validate);
+ err = err ?: bpf_core_calc_field_relo(prog, relo, targ_spec,
+ &res->new_val, &res->new_sz,
+ &res->new_type_id, NULL);
+ if (err)
+ goto done;
+ /* Validate if it's safe to adjust load/store memory size.
+ * Adjustments are performed only if original and new memory
+ * sizes differ.
+ */
+ res->fail_memsz_adjust = false;
+ if (res->orig_sz != res->new_sz) {
+ const struct btf_type *orig_t, *new_t;
+
+ orig_t = btf__type_by_id(local_spec->btf, res->orig_type_id);
+ new_t = btf__type_by_id(targ_spec->btf, res->new_type_id);
+
+ /* There are two use cases in which it's safe to
+ * adjust load/store's mem size:
+ * - reading a 32-bit kernel pointer, while on BPF
+ * size pointers are always 64-bit; in this case
+ * it's safe to "downsize" instruction size due to
+ * pointer being treated as unsigned integer with
+ * zero-extended upper 32-bits;
+ * - reading unsigned integers, again due to
+ * zero-extension is preserving the value correctly.
+ *
+ * In all other cases it's incorrect to attempt to
+ * load/store field because read value will be
+ * incorrect, so we poison relocated instruction.
+ */
+ if (btf_is_ptr(orig_t) && btf_is_ptr(new_t))
+ goto done;
+ if (btf_is_int(orig_t) && btf_is_int(new_t) &&
+ btf_int_encoding(orig_t) != BTF_INT_SIGNED &&
+ btf_int_encoding(new_t) != BTF_INT_SIGNED)
+ goto done;
+
+ /* mark as invalid mem size adjustment, but this will
+ * only be checked for LDX/STX/ST insns
+ */
+ res->fail_memsz_adjust = true;
+ }
} else if (core_relo_is_type_based(relo->kind)) {
err = bpf_core_calc_type_relo(relo, local_spec, &res->orig_val);
err = err ?: bpf_core_calc_type_relo(relo, targ_spec, &res->new_val);
@@ -5252,6 +5324,7 @@ static int bpf_core_calc_relo(const struct bpf_program *prog,
err = err ?: bpf_core_calc_enumval_relo(relo, targ_spec, &res->new_val);
}
+done:
if (err == -EUCLEAN) {
/* EUCLEAN is used to signal instruction poisoning request */
res->poison = true;
@@ -5291,6 +5364,28 @@ static bool is_ldimm64(struct bpf_insn *insn)
return insn->code == (BPF_LD | BPF_IMM | BPF_DW);
}
+static int insn_bpf_size_to_bytes(struct bpf_insn *insn)
+{
+ switch (BPF_SIZE(insn->code)) {
+ case BPF_DW: return 8;
+ case BPF_W: return 4;
+ case BPF_H: return 2;
+ case BPF_B: return 1;
+ default: return -1;
+ }
+}
+
+static int insn_bytes_to_bpf_size(__u32 sz)
+{
+ switch (sz) {
+ case 8: return BPF_DW;
+ case 4: return BPF_W;
+ case 2: return BPF_H;
+ case 1: return BPF_B;
+ default: return -1;
+ }
+}
+
/*
* Patch relocatable BPF instruction.
*
@@ -5300,10 +5395,13 @@ static bool is_ldimm64(struct bpf_insn *insn)
* spec, and is checked before patching instruction. If actual insn->imm value
* is wrong, bail out with error.
*
- * Currently three kinds of BPF instructions are supported:
+ * Currently supported classes of BPF instruction are:
* 1. rX = <imm> (assignment with immediate operand);
* 2. rX += <imm> (arithmetic operations with immediate operand);
- * 3. rX = <imm64> (load with 64-bit immediate value).
+ * 3. rX = <imm64> (load with 64-bit immediate value);
+ * 4. rX = *(T *)(rY + <off>), where T is one of {u8, u16, u32, u64};
+ * 5. *(T *)(rX + <off>) = rY, where T is one of {u8, u16, u32, u64};
+ * 6. *(T *)(rX + <off>) = <imm>, where T is one of {u8, u16, u32, u64}.
*/
static int bpf_core_patch_insn(struct bpf_program *prog,
const struct bpf_core_relo *relo,
@@ -5327,6 +5425,7 @@ static int bpf_core_patch_insn(struct bpf_program *prog,
class = BPF_CLASS(insn->code);
if (res->poison) {
+poison:
/* poison second part of ldimm64 to avoid confusing error from
* verifier about "unknown opcode 00"
*/
@@ -5369,10 +5468,39 @@ static int bpf_core_patch_insn(struct bpf_program *prog,
prog->name, relo_idx, insn_idx, new_val);
return -ERANGE;
}
+ if (res->fail_memsz_adjust) {
+ pr_warn("prog '%s': relo #%d: insn #%d (LDX/ST/STX) accesses field incorrectly. "
+ "Make sure you are accessing pointers, unsigned integers, or fields of matching type and size.\n",
+ prog->name, relo_idx, insn_idx);
+ goto poison;
+ }
+
orig_val = insn->off;
insn->off = new_val;
pr_debug("prog '%s': relo #%d: patched insn #%d (LDX/ST/STX) off %u -> %u\n",
prog->name, relo_idx, insn_idx, orig_val, new_val);
+
+ if (res->new_sz != res->orig_sz) {
+ int insn_bytes_sz, insn_bpf_sz;
+
+ insn_bytes_sz = insn_bpf_size_to_bytes(insn);
+ if (insn_bytes_sz != res->orig_sz) {
+ pr_warn("prog '%s': relo #%d: insn #%d (LDX/ST/STX) unexpected mem size: got %d, exp %u\n",
+ prog->name, relo_idx, insn_idx, insn_bytes_sz, res->orig_sz);
+ return -EINVAL;
+ }
+
+ insn_bpf_sz = insn_bytes_to_bpf_size(res->new_sz);
+ if (insn_bpf_sz < 0) {
+ pr_warn("prog '%s': relo #%d: insn #%d (LDX/ST/STX) invalid new mem size: %u\n",
+ prog->name, relo_idx, insn_idx, res->new_sz);
+ return -EINVAL;
+ }
+
+ insn->code = BPF_MODE(insn->code) | insn_bpf_sz | BPF_CLASS(insn->code);
+ pr_debug("prog '%s': relo #%d: patched insn #%d (LDX/ST/STX) mem_sz %u -> %u\n",
+ prog->name, relo_idx, insn_idx, res->orig_sz, res->new_sz);
+ }
break;
case BPF_LD: {
__u64 imm;
--
2.24.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH v2 bpf-next 3/4] libbpf: allow specifying both ELF and raw BTF for CO-RE BTF override
2020-10-07 20:29 [PATCH v2 bpf-next 0/4] libbpf: auto-resize relocatable LOAD/STORE instructions Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 1/4] libbpf: skip CO-RE relocations for not loaded BPF programs Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 2/4] libbpf: support safe subset of load/store instruction resizing with CO-RE Andrii Nakryiko
@ 2020-10-07 20:29 ` Andrii Nakryiko
2020-10-07 20:29 ` [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions Andrii Nakryiko
3 siblings, 0 replies; 7+ messages in thread
From: Andrii Nakryiko @ 2020-10-07 20:29 UTC (permalink / raw)
To: bpf, netdev, ast, daniel
Cc: andrii, kernel-team, Luka Perkov, Tony Ambardar, Andrii Nakryiko
From: Andrii Nakryiko <andriin@fb.com>
Use generalized BTF parsing logic, making it possible to parse BTF both from
ELF file, as well as a raw BTF dump. This makes it easier to write custom
tests with manually generated BTFs.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
tools/lib/bpf/libbpf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 032cf0049ddb..105cb8ad1d75 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -5842,7 +5842,7 @@ bpf_object__relocate_core(struct bpf_object *obj, const char *targ_btf_path)
return 0;
if (targ_btf_path)
- targ_btf = btf__parse_elf(targ_btf_path, NULL);
+ targ_btf = btf__parse(targ_btf_path, NULL);
else
targ_btf = obj->btf_vmlinux;
if (IS_ERR_OR_NULL(targ_btf)) {
--
2.24.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions
2020-10-07 20:29 [PATCH v2 bpf-next 0/4] libbpf: auto-resize relocatable LOAD/STORE instructions Andrii Nakryiko
` (2 preceding siblings ...)
2020-10-07 20:29 ` [PATCH v2 bpf-next 3/4] libbpf: allow specifying both ELF and raw BTF for CO-RE BTF override Andrii Nakryiko
@ 2020-10-07 20:29 ` Andrii Nakryiko
2020-10-07 23:29 ` Alexei Starovoitov
3 siblings, 1 reply; 7+ messages in thread
From: Andrii Nakryiko @ 2020-10-07 20:29 UTC (permalink / raw)
To: bpf, netdev, ast, daniel
Cc: andrii, kernel-team, Luka Perkov, Tony Ambardar, Andrii Nakryiko
From: Andrii Nakryiko <andriin@fb.com>
Add selftests validating libbpf's auto-resizing of load/store instructions
when used with CO-RE relocations. An explicit and manual approach with using
bpf_core_read() is also demonstrated and tested. Separate BPF program is
supposed to fail due to using signed integers of sizes that differ from
kernel's sizes.
To reliably simulate 32-bit BTF (i.e., the one with sizeof(long) ==
sizeof(void *) == 4), selftest generates its own custom BTF and passes it as
a replacement for real kernel BTF. This allows to test 32/64-bitness mix on
all architectures.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
.../selftests/bpf/prog_tests/core_autosize.c | 225 ++++++++++++++++++
.../selftests/bpf/progs/test_core_autosize.c | 172 +++++++++++++
2 files changed, 397 insertions(+)
create mode 100644 tools/testing/selftests/bpf/prog_tests/core_autosize.c
create mode 100644 tools/testing/selftests/bpf/progs/test_core_autosize.c
diff --git a/tools/testing/selftests/bpf/prog_tests/core_autosize.c b/tools/testing/selftests/bpf/prog_tests/core_autosize.c
new file mode 100644
index 000000000000..981c251453d9
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/core_autosize.c
@@ -0,0 +1,225 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+
+#include <test_progs.h>
+#include <bpf/btf.h>
+
+/* real layout and sizes according to test's (32-bit) BTF
+ * needs to be defined before skeleton is included */
+struct test_struct___real {
+ unsigned int ptr; /* can't use `void *`, it is always 8 byte in BPF target */
+ unsigned int val2;
+ unsigned long long val1;
+ unsigned short val3;
+ unsigned char val4;
+ unsigned char _pad;
+};
+
+#include "test_core_autosize.skel.h"
+
+static int duration = 0;
+
+static struct {
+ unsigned long long ptr_samesized;
+ unsigned long long val1_samesized;
+ unsigned long long val2_samesized;
+ unsigned long long val3_samesized;
+ unsigned long long val4_samesized;
+ struct test_struct___real output_samesized;
+
+ unsigned long long ptr_downsized;
+ unsigned long long val1_downsized;
+ unsigned long long val2_downsized;
+ unsigned long long val3_downsized;
+ unsigned long long val4_downsized;
+ struct test_struct___real output_downsized;
+
+ unsigned long long ptr_probed;
+ unsigned long long val1_probed;
+ unsigned long long val2_probed;
+ unsigned long long val3_probed;
+ unsigned long long val4_probed;
+
+ unsigned long long ptr_signed;
+ unsigned long long val1_signed;
+ unsigned long long val2_signed;
+ unsigned long long val3_signed;
+ unsigned long long val4_signed;
+ struct test_struct___real output_signed;
+} out;
+
+void test_core_autosize(void)
+{
+ char btf_file[] = "/tmp/core_autosize.btf.XXXXXX";
+ int err, fd = -1, zero = 0;
+ int char_id, short_id, int_id, long_long_id, void_ptr_id, id;
+ struct test_core_autosize* skel = NULL;
+ struct bpf_object_load_attr load_attr = {};
+ struct bpf_program *prog;
+ struct bpf_map *bss_map;
+ struct btf *btf = NULL;
+ size_t written;
+ const void *raw_data;
+ __u32 raw_sz;
+ FILE *f = NULL;
+
+ btf = btf__new_empty();
+ if (!ASSERT_OK_PTR(btf, "empty_btf"))
+ return;
+ /* Emit the following struct with 32-bit pointer size:
+ *
+ * struct test_struct {
+ * void *ptr;
+ * unsigned long val2;
+ * unsigned long long val1;
+ * unsigned short val3;
+ * unsigned char val4;
+ * char: 8;
+ * };
+ *
+ * This struct is going to be used as the "kernel BTF" for this test.
+ * It's equivalent memory-layout-wise to test_struct__real above.
+ */
+
+ /* force 32-bit pointer size */
+ btf__set_pointer_size(btf, 4);
+
+ char_id = btf__add_int(btf, "unsigned char", 1, 0);
+ ASSERT_EQ(char_id, 1, "char_id");
+ short_id = btf__add_int(btf, "unsigned short", 2, 0);
+ ASSERT_EQ(short_id, 2, "short_id");
+ /* "long unsigned int" of 4 byte size tells BTF that sizeof(void *) == 4 */
+ int_id = btf__add_int(btf, "long unsigned int", 4, 0);
+ ASSERT_EQ(int_id, 3, "int_id");
+ long_long_id = btf__add_int(btf, "unsigned long long", 8, 0);
+ ASSERT_EQ(long_long_id, 4, "long_long_id");
+ void_ptr_id = btf__add_ptr(btf, 0);
+ ASSERT_EQ(void_ptr_id, 5, "void_ptr_id");
+
+ id = btf__add_struct(btf, "test_struct", 20 /* bytes */);
+ ASSERT_EQ(id, 6, "struct_id");
+ err = btf__add_field(btf, "ptr", void_ptr_id, 0, 0);
+ err = err ?: btf__add_field(btf, "val2", int_id, 32, 0);
+ err = err ?: btf__add_field(btf, "val1", long_long_id, 64, 0);
+ err = err ?: btf__add_field(btf, "val3", short_id, 128, 0);
+ err = err ?: btf__add_field(btf, "val4", char_id, 144, 0);
+ ASSERT_OK(err, "struct_fields");
+
+ fd = mkstemp(btf_file);
+ if (CHECK(fd < 0, "btf_tmp", "failed to create file: %d\n", fd))
+ goto cleanup;
+ f = fdopen(fd, "w");
+ if (!ASSERT_OK_PTR(f, "btf_fdopen"))
+ goto cleanup;
+
+ raw_data = btf__get_raw_data(btf, &raw_sz);
+ if (!ASSERT_OK_PTR(raw_data, "raw_data"))
+ goto cleanup;
+ written = fwrite(raw_data, 1, raw_sz, f);
+ if (CHECK(written != raw_sz, "btf_write", "written: %zu, errno: %d\n", written, errno))
+ goto cleanup;
+ fflush(f);
+ fclose(f);
+ f = NULL;
+ close(fd);
+ fd = -1;
+
+ /* open and load BPF program with custom BTF as the kernel BTF */
+ skel = test_core_autosize__open();
+ if (!ASSERT_OK_PTR(skel, "skel_open"))
+ return;
+
+ /* disable handle_signed() for now */
+ prog = bpf_object__find_program_by_name(skel->obj, "handle_signed");
+ if (!ASSERT_OK_PTR(prog, "prog_find"))
+ goto cleanup;
+ bpf_program__set_autoload(prog, false);
+
+ load_attr.obj = skel->obj;
+ load_attr.target_btf_path = btf_file;
+ err = bpf_object__load_xattr(&load_attr);
+ if (!ASSERT_OK(err, "prog_load"))
+ goto cleanup;
+
+ prog = bpf_object__find_program_by_name(skel->obj, "handle_samesize");
+ if (!ASSERT_OK_PTR(prog, "prog_find"))
+ goto cleanup;
+ skel->links.handle_samesize = bpf_program__attach(prog);
+ if (!ASSERT_OK_PTR(skel->links.handle_samesize, "prog_attach"))
+ goto cleanup;
+
+ prog = bpf_object__find_program_by_name(skel->obj, "handle_downsize");
+ if (!ASSERT_OK_PTR(prog, "prog_find"))
+ goto cleanup;
+ skel->links.handle_downsize = bpf_program__attach(prog);
+ if (!ASSERT_OK_PTR(skel->links.handle_downsize, "prog_attach"))
+ goto cleanup;
+
+ prog = bpf_object__find_program_by_name(skel->obj, "handle_probed");
+ if (!ASSERT_OK_PTR(prog, "prog_find"))
+ goto cleanup;
+ skel->links.handle_probed = bpf_program__attach(prog);
+ if (!ASSERT_OK_PTR(skel->links.handle_probed, "prog_attach"))
+ goto cleanup;
+
+ usleep(1);
+
+ bss_map = bpf_object__find_map_by_name(skel->obj, "test_cor.bss");
+ if (!ASSERT_OK_PTR(bss_map, "bss_map_find"))
+ goto cleanup;
+
+ err = bpf_map_lookup_elem(bpf_map__fd(bss_map), &zero, (void *)&out);
+ if (!ASSERT_OK(err, "bss_lookup"))
+ goto cleanup;
+
+ ASSERT_EQ(out.ptr_samesized, 0x01020304, "ptr_samesized");
+ ASSERT_EQ(out.val1_samesized, 0x1020304050607080, "val1_samesized");
+ ASSERT_EQ(out.val2_samesized, 0x0a0b0c0d, "val2_samesized");
+ ASSERT_EQ(out.val3_samesized, 0xfeed, "val3_samesized");
+ ASSERT_EQ(out.val4_samesized, 0xb9, "val4_samesized");
+ ASSERT_EQ(out.output_samesized.ptr, 0x01020304, "ptr_samesized");
+ ASSERT_EQ(out.output_samesized.val1, 0x1020304050607080, "val1_samesized");
+ ASSERT_EQ(out.output_samesized.val2, 0x0a0b0c0d, "val2_samesized");
+ ASSERT_EQ(out.output_samesized.val3, 0xfeed, "val3_samesized");
+ ASSERT_EQ(out.output_samesized.val4, 0xb9, "val4_samesized");
+
+ ASSERT_EQ(out.ptr_downsized, 0x01020304, "ptr_downsized");
+ ASSERT_EQ(out.val1_downsized, 0x1020304050607080, "val1_downsized");
+ ASSERT_EQ(out.val2_downsized, 0x0a0b0c0d, "val2_downsized");
+ ASSERT_EQ(out.val3_downsized, 0xfeed, "val3_downsized");
+ ASSERT_EQ(out.val4_downsized, 0xb9, "val4_downsized");
+ ASSERT_EQ(out.output_downsized.ptr, 0x01020304, "ptr_downsized");
+ ASSERT_EQ(out.output_downsized.val1, 0x1020304050607080, "val1_downsized");
+ ASSERT_EQ(out.output_downsized.val2, 0x0a0b0c0d, "val2_downsized");
+ ASSERT_EQ(out.output_downsized.val3, 0xfeed, "val3_downsized");
+ ASSERT_EQ(out.output_downsized.val4, 0xb9, "val4_downsized");
+
+ ASSERT_EQ(out.ptr_probed, 0x01020304, "ptr_probed");
+ ASSERT_EQ(out.val1_probed, 0x1020304050607080, "val1_probed");
+ ASSERT_EQ(out.val2_probed, 0x0a0b0c0d, "val2_probed");
+ ASSERT_EQ(out.val3_probed, 0xfeed, "val3_probed");
+ ASSERT_EQ(out.val4_probed, 0xb9, "val4_probed");
+
+ test_core_autosize__destroy(skel);
+ skel = NULL;
+
+ /* now re-load with handle_signed() enabled, it should fail loading */
+ skel = test_core_autosize__open();
+ if (!ASSERT_OK_PTR(skel, "skel_open"))
+ return;
+
+ load_attr.obj = skel->obj;
+ load_attr.target_btf_path = btf_file;
+ err = bpf_object__load_xattr(&load_attr);
+ if (!ASSERT_ERR(err, "bad_prog_load"))
+ goto cleanup;
+
+cleanup:
+ if (f)
+ fclose(f);
+ if (fd >= 0)
+ close(fd);
+ remove(btf_file);
+ btf__free(btf);
+ test_core_autosize__destroy(skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/test_core_autosize.c b/tools/testing/selftests/bpf/progs/test_core_autosize.c
new file mode 100644
index 000000000000..e999f2e2ea22
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_core_autosize.c
@@ -0,0 +1,172 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2019 Facebook
+
+#include <linux/bpf.h>
+#include <stdint.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_core_read.h>
+
+char _license[] SEC("license") = "GPL";
+
+/* fields of exactly the same size */
+struct test_struct___samesize {
+ void *ptr;
+ unsigned long long val1;
+ unsigned int val2;
+ unsigned short val3;
+ unsigned char val4;
+} __attribute((preserve_access_index));
+
+/* unsigned fields that have to be downsized by libbpf */
+struct test_struct___downsize {
+ void *ptr;
+ unsigned long val1;
+ unsigned long val2;
+ unsigned long val3;
+ unsigned long val4;
+ /* total sz: 40 */
+} __attribute__((preserve_access_index));
+
+/* fields with signed integers of wrong size, should be rejected */
+struct test_struct___signed {
+ void *ptr;
+ long val1;
+ long val2;
+ long val3;
+ long val4;
+} __attribute((preserve_access_index));
+
+/* real layout and sizes according to test's (32-bit) BTF */
+struct test_struct___real {
+ unsigned int ptr; /* can't use `void *`, it is always 8 byte in BPF target */
+ unsigned int val2;
+ unsigned long long val1;
+ unsigned short val3;
+ unsigned char val4;
+ unsigned char _pad;
+ /* total sz: 20 */
+};
+
+struct test_struct___real input = {
+ .ptr = 0x01020304,
+ .val1 = 0x1020304050607080,
+ .val2 = 0x0a0b0c0d,
+ .val3 = 0xfeed,
+ .val4 = 0xb9,
+ ._pad = 0xff, /* make sure no accidental zeros are present */
+};
+
+unsigned long long ptr_samesized = 0;
+unsigned long long val1_samesized = 0;
+unsigned long long val2_samesized = 0;
+unsigned long long val3_samesized = 0;
+unsigned long long val4_samesized = 0;
+struct test_struct___real output_samesized = {};
+
+unsigned long long ptr_downsized = 0;
+unsigned long long val1_downsized = 0;
+unsigned long long val2_downsized = 0;
+unsigned long long val3_downsized = 0;
+unsigned long long val4_downsized = 0;
+struct test_struct___real output_downsized = {};
+
+unsigned long long ptr_probed = 0;
+unsigned long long val1_probed = 0;
+unsigned long long val2_probed = 0;
+unsigned long long val3_probed = 0;
+unsigned long long val4_probed = 0;
+
+unsigned long long ptr_signed = 0;
+unsigned long long val1_signed = 0;
+unsigned long long val2_signed = 0;
+unsigned long long val3_signed = 0;
+unsigned long long val4_signed = 0;
+struct test_struct___real output_signed = {};
+
+SEC("raw_tp/sys_exit")
+int handle_samesize(void *ctx)
+{
+ struct test_struct___samesize *in = (void *)&input;
+ struct test_struct___samesize *out = (void *)&output_samesized;
+
+ ptr_samesized = (unsigned long long)in->ptr;
+ val1_samesized = in->val1;
+ val2_samesized = in->val2;
+ val3_samesized = in->val3;
+ val4_samesized = in->val4;
+
+ out->ptr = in->ptr;
+ out->val1 = in->val1;
+ out->val2 = in->val2;
+ out->val3 = in->val3;
+ out->val4 = in->val4;
+
+ return 0;
+}
+
+SEC("raw_tp/sys_exit")
+int handle_downsize(void *ctx)
+{
+ struct test_struct___downsize *in = (void *)&input;
+ struct test_struct___downsize *out = (void *)&output_downsized;
+
+ ptr_downsized = (unsigned long long)in->ptr;
+ val1_downsized = in->val1;
+ val2_downsized = in->val2;
+ val3_downsized = in->val3;
+ val4_downsized = in->val4;
+
+ out->ptr = in->ptr;
+ out->val1 = in->val1;
+ out->val2 = in->val2;
+ out->val3 = in->val3;
+ out->val4 = in->val4;
+
+ return 0;
+}
+
+SEC("raw_tp/sys_enter")
+int handle_probed(void *ctx)
+{
+ struct test_struct___downsize *in = (void *)&input;
+ __u64 tmp;
+
+ tmp = 0;
+ bpf_core_read(&tmp, bpf_core_field_size(in->ptr), &in->ptr);
+ ptr_probed = tmp;
+
+ tmp = 0;
+ bpf_core_read(&tmp, bpf_core_field_size(in->val1), &in->val1);
+ val1_probed = tmp;
+
+ tmp = 0;
+ bpf_core_read(&tmp, bpf_core_field_size(in->val2), &in->val2);
+ val2_probed = tmp;
+
+ tmp = 0;
+ bpf_core_read(&tmp, bpf_core_field_size(in->val3), &in->val3);
+ val3_probed = tmp;
+
+ tmp = 0;
+ bpf_core_read(&tmp, bpf_core_field_size(in->val4), &in->val4);
+ val4_probed = tmp;
+
+ return 0;
+}
+
+SEC("raw_tp/sys_enter")
+int handle_signed(void *ctx)
+{
+ struct test_struct___signed *in = (void *)&input;
+ struct test_struct___signed *out = (void *)&output_signed;
+
+ val2_signed = in->val2;
+ val3_signed = in->val3;
+ val4_signed = in->val4;
+
+ out->val2= in->val2;
+ out->val3= in->val3;
+ out->val4= in->val4;
+
+ return 0;
+}
--
2.24.1
^ permalink raw reply related [flat|nested] 7+ messages in thread* Re: [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions
2020-10-07 20:29 ` [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions Andrii Nakryiko
@ 2020-10-07 23:29 ` Alexei Starovoitov
2020-10-07 23:33 ` Andrii Nakryiko
0 siblings, 1 reply; 7+ messages in thread
From: Alexei Starovoitov @ 2020-10-07 23:29 UTC (permalink / raw)
To: Andrii Nakryiko
Cc: bpf, netdev, ast, daniel, kernel-team, Luka Perkov,
Tony Ambardar, Andrii Nakryiko
On Wed, Oct 07, 2020 at 01:29:46PM -0700, Andrii Nakryiko wrote:
> diff --git a/tools/testing/selftests/bpf/progs/test_core_autosize.c b/tools/testing/selftests/bpf/progs/test_core_autosize.c
> new file mode 100644
> index 000000000000..e999f2e2ea22
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/progs/test_core_autosize.c
> @@ -0,0 +1,172 @@
> +// SPDX-License-Identifier: GPL-2.0
> +// Copyright (c) 2019 Facebook
The year is wrong.
Also Copyright should be in /* */
The // exception is only for SPDX and only in .c
.h should have both in /* */
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2 bpf-next 4/4] selftests/bpf: validate libbpf's auto-sizing of LD/ST/STX instructions
2020-10-07 23:29 ` Alexei Starovoitov
@ 2020-10-07 23:33 ` Andrii Nakryiko
0 siblings, 0 replies; 7+ messages in thread
From: Andrii Nakryiko @ 2020-10-07 23:33 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Andrii Nakryiko, bpf, Networking, Alexei Starovoitov,
Daniel Borkmann, Kernel Team, Luka Perkov, Tony Ambardar,
Andrii Nakryiko
On Wed, Oct 7, 2020 at 4:30 PM Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
>
> On Wed, Oct 07, 2020 at 01:29:46PM -0700, Andrii Nakryiko wrote:
> > diff --git a/tools/testing/selftests/bpf/progs/test_core_autosize.c b/tools/testing/selftests/bpf/progs/test_core_autosize.c
> > new file mode 100644
> > index 000000000000..e999f2e2ea22
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/progs/test_core_autosize.c
> > @@ -0,0 +1,172 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +// Copyright (c) 2019 Facebook
>
> The year is wrong.
> Also Copyright should be in /* */
> The // exception is only for SPDX and only in .c
> .h should have both in /* */
argh, copy/paste from test_core_reloc.c
We have plenty of those '//' vs '/* */' issues with SPDX and
Copyrights in selftests. Might be a good idea to finally fix all that.
^ permalink raw reply [flat|nested] 7+ messages in thread