]> granicus.if.org Git - strace/commitdiff
tests: check BPF_OBJ_GET_INFO_BY_FD decoding
authorEugene Syromyatnikov <evgsyr@gmail.com>
Fri, 1 Jun 2018 00:55:21 +0000 (02:55 +0200)
committerDmitry V. Levin <ldv@altlinux.org>
Mon, 4 Jun 2018 01:11:05 +0000 (01:11 +0000)
* tests/bpf-obj_get_info_by_fd-prog-v.c: New file.
* tests/bpf-obj_get_info_by_fd-prog.c: Likewise.
* tests/bpf-obj_get_info_by_fd-v.c: Likewise.
* tests/bpf-obj_get_info_by_fd.c: Likewise.
* configure.ac (AC_CHECK_HEADERS): Check for struct bpf_insn,
struct bpf_map_info, and struct bpf_prog_info.
* tests/pure_executables.list: Add bpf-obj_get_info_by_fd,
bpf-obj_get_info_by_fd-v, bpf-obj_get_info_by_fd-prog, and
bpf-obj_get_info_by_fd-prog-v.
* tests/.gitignore: Likewise.
* tests/gen_tests.in (bpf-obj_get_info_by_fd,
bpf-obj_get_info_by_fd-v, bpf-obj_get_info_by_fd-prog,
bpf-obj_get_info_by_fd-prog-v): New tests.

Co-Authored-by: Dmitry V. Levin <ldv@altlinux.org>
configure.ac
tests/.gitignore
tests/Makefile.am
tests/bpf-obj_get_info_by_fd-prog-v.c [new file with mode: 0644]
tests/bpf-obj_get_info_by_fd-prog.c [new file with mode: 0644]
tests/bpf-obj_get_info_by_fd-v.c [new file with mode: 0644]
tests/bpf-obj_get_info_by_fd.c [new file with mode: 0644]
tests/gen_tests.in
tests/pure_executables.list

index 43197687aee152709145cc5d758691ff56254049..f6868f2192045da16e519eae836434044ea427f0 100644 (file)
@@ -466,7 +466,11 @@ AC_CHECK_HEADERS([linux/input.h], [
 ])
 
 AC_CHECK_HEADERS([linux/bpf.h], [
-       AC_CHECK_TYPES([struct bpf_insn],,, [#include <linux/bpf.h>])
+       AC_CHECK_TYPES(m4_normalize([
+               struct bpf_insn,
+               struct bpf_map_info,
+               struct bpf_prog_info
+       ]),,, [#include <linux/bpf.h>])
        st_BPF_ATTR
 ])
 
index 13fa77b78f1cbf5afb3506eb9cb4adb6828ae40c..a754f7886be203b57726414b447b1db08ec48044 100644 (file)
@@ -21,6 +21,10 @@ attach-p-cmd-p
 block_reset_raise_run
 bpf
 bpf-v
+bpf-obj_get_info_by_fd
+bpf-obj_get_info_by_fd-v
+bpf-obj_get_info_by_fd-prog
+bpf-obj_get_info_by_fd-prog-v
 bpf-success
 bpf-success-v
 brk
index 17a823367de20a0174f42d14ef2a583c86c925f7..3cae26c8b416a761f111e8c94bd99b39fa54c89c 100644 (file)
@@ -93,6 +93,10 @@ check_PROGRAMS = $(PURE_EXECUTABLES) \
        attach-p-cmd-cmd \
        attach-p-cmd-p \
        block_reset_raise_run \
+       bpf-obj_get_info_by_fd \
+       bpf-obj_get_info_by_fd-v \
+       bpf-obj_get_info_by_fd-prog \
+       bpf-obj_get_info_by_fd-prog-v \
        bpf-success \
        bpf-success-v \
        caps-abbrev \
diff --git a/tests/bpf-obj_get_info_by_fd-prog-v.c b/tests/bpf-obj_get_info_by_fd-prog-v.c
new file mode 100644 (file)
index 0000000..d7e10b4
--- /dev/null
@@ -0,0 +1,3 @@
+#define CHECK_OBJ_PROG 1
+#define VERBOSE 1
+#include "bpf-obj_get_info_by_fd.c"
diff --git a/tests/bpf-obj_get_info_by_fd-prog.c b/tests/bpf-obj_get_info_by_fd-prog.c
new file mode 100644 (file)
index 0000000..19a5f75
--- /dev/null
@@ -0,0 +1,2 @@
+#define CHECK_OBJ_PROG 1
+#include "bpf-obj_get_info_by_fd.c"
diff --git a/tests/bpf-obj_get_info_by_fd-v.c b/tests/bpf-obj_get_info_by_fd-v.c
new file mode 100644 (file)
index 0000000..2debc5f
--- /dev/null
@@ -0,0 +1,2 @@
+#define VERBOSE 1
+#include "bpf-obj_get_info_by_fd.c"
diff --git a/tests/bpf-obj_get_info_by_fd.c b/tests/bpf-obj_get_info_by_fd.c
new file mode 100644 (file)
index 0000000..67b5122
--- /dev/null
@@ -0,0 +1,499 @@
+/*
+ * Check bpf(BPF_OBJ_GET_INFO_BY_FD) decoding.
+ *
+ * Copyright (c) 2018 The strace developers.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. The name of the author may not be used to endorse or promote products
+ *    derived from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "tests.h"
+
+#ifndef CHECK_OBJ_PROG
+# define CHECK_OBJ_PROG 0
+#endif
+
+#include <inttypes.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <sys/sysmacros.h>
+#include <asm/unistd.h>
+
+#include "print_fields.h"
+#include "scno.h"
+
+#ifdef HAVE_LINUX_BPF_H
+# include <linux/bpf.h>
+#endif
+
+#include "bpf_attr.h"
+
+#include "xlat.h"
+#include "xlat/bpf_map_flags.h"
+#include "xlat/bpf_map_types.h"
+#include "xlat/bpf_prog_types.h"
+
+#define XLAT_MACROS_ONLY
+#include "xlat/bpf_commands.h"
+#include "xlat/bpf_op_alu.h"
+#include "xlat/bpf_op_jmp.h"
+#include "xlat/bpf_size.h"
+#include "xlat/bpf_src.h"
+#include "xlat/ebpf_class.h"
+#include "xlat/ebpf_mode.h"
+#include "xlat/ebpf_op_alu.h"
+#include "xlat/ebpf_regs.h"
+#include "xlat/ebpf_size.h"
+
+#ifndef HAVE_STRUCT_BPF_INSN
+struct bpf_insn {
+       uint8_t code;
+       uint8_t dst_reg:4;
+       uint8_t src_reg:4;
+       int16_t off;
+       int32_t imm;
+};
+#endif
+
+static const char *errstr;
+
+static long
+sys_bpf(kernel_ulong_t cmd, void *attr, kernel_ulong_t size)
+{
+       long rc = syscall(__NR_bpf, cmd, attr, size);
+       errstr = sprintrc(rc);
+       return rc;
+}
+
+static void
+print_map_create(void *attr_void, size_t size, long rc)
+{
+       /* struct BPF_MAP_CREATE_struct *attr = attr_void; */
+
+       printf("bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_ARRAY, key_size=4"
+              ", value_size=8, max_entries=1");
+       if (size > offsetof(struct BPF_MAP_CREATE_struct, map_flags))
+               printf(", map_flags=0");
+       if (size > offsetof(struct BPF_MAP_CREATE_struct, inner_map_fd))
+               printf(", inner_map_fd=0</dev/null>");
+       if (size > offsetof(struct BPF_MAP_CREATE_struct, map_name))
+               printf(", map_name=\"test_map\"");
+       if (size > offsetof(struct BPF_MAP_CREATE_struct, map_ifindex))
+               printf(", map_ifindex=0");
+       printf("}, %zu) = ", size);
+       if (rc >= 0)
+               printf("%ld<anon_inode:bpf-map>\n", rc);
+       else
+               puts(errstr);
+}
+
+#if CHECK_OBJ_PROG
+static struct bpf_insn socket_prog[] = {
+       { /* 0 */
+               .code    = BPF_ALU64 | BPF_K | BPF_MOV,
+               .dst_reg = BPF_REG_1,
+               .imm     = 0,
+       },
+       { /* 1 */
+               .code    = BPF_STX | BPF_W | BPF_MEM,
+               .dst_reg = BPF_REG_10,
+               .src_reg = BPF_REG_1,
+               .off     = -4,
+       },
+       { /* 2 */
+               .code = BPF_ALU64 | BPF_X | BPF_MOV,
+               .dst_reg = BPF_REG_2,
+               .src_reg = BPF_REG_10,
+       },
+       { /* 3 */
+               .code    = BPF_ALU64 | BPF_K | BPF_ADD,
+               .dst_reg = BPF_REG_2,
+               .imm     = -4,
+       },
+       { /* 4 */
+               .code    = BPF_LD | BPF_DW | BPF_IMM,
+               .dst_reg = BPF_REG_1,
+               .src_reg = 1 /* BPF_PSEUDO_MAP_FD */,
+               .imm     = 0, /* to be set to map fd */
+       },
+       { /* 5 */
+               .imm     = 0,
+       },
+       { /* 6 */
+               .code    = BPF_JMP | BPF_K | BPF_CALL,
+               .imm     = 0x1, /* BPF_FUNC_map_lookup_elem */
+       },
+       { /* 7 */
+               .code    = BPF_ALU64 | BPF_K | BPF_MOV,
+               .dst_reg = BPF_REG_0,
+               .imm     = 0,
+       },
+       { /* 8 */
+               .code    = BPF_JMP | BPF_K | BPF_EXIT,
+       },
+};
+
+# if VERBOSE
+static const char *socket_prog_fmt =
+       "[{code=BPF_ALU64|BPF_K|BPF_MOV"
+               ", dst_reg=BPF_REG_1, src_reg=BPF_REG_0, off=0, imm=0}"
+       ", {code=BPF_STX|BPF_W|BPF_MEM"
+               ", dst_reg=BPF_REG_10, src_reg=BPF_REG_1, off=-4, imm=0}"
+       ", {code=BPF_ALU64|BPF_X|BPF_MOV"
+               ", dst_reg=BPF_REG_2, src_reg=BPF_REG_10, off=0, imm=0}"
+       ", {code=BPF_ALU64|BPF_K|BPF_ADD"
+               ", dst_reg=BPF_REG_2, src_reg=BPF_REG_0, off=0, imm=0xfffffffc}"
+       ", {code=BPF_LD|BPF_DW|BPF_IMM"
+               ", dst_reg=BPF_REG_1, src_reg=BPF_REG_1, off=0, imm=%#x}"
+       ", {code=BPF_LD|BPF_W|BPF_IMM"
+               ", dst_reg=BPF_REG_0, src_reg=BPF_REG_0, off=0, imm=0}"
+       ", {code=BPF_JMP|BPF_K|BPF_CALL"
+               ", dst_reg=BPF_REG_0, src_reg=BPF_REG_0, off=0, imm=0x1}"
+       ", {code=BPF_ALU64|BPF_K|BPF_MOV"
+               ", dst_reg=BPF_REG_0, src_reg=BPF_REG_0, off=0, imm=0}"
+       ", {code=BPF_JMP|BPF_K|BPF_EXIT"
+               ", dst_reg=BPF_REG_0, src_reg=BPF_REG_0, off=0, imm=0}"
+       "]";
+# endif /* VERBOSE */
+
+static const char *license = "BSD";
+static char log_buf[4096];
+
+static void
+print_prog_load(void *attr_void, size_t size, long rc)
+{
+       printf("bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_SOCKET_FILTER"
+              ", insn_cnt=%zu, insns=", ARRAY_SIZE(socket_prog));
+# if VERBOSE
+       printf(socket_prog_fmt, socket_prog[4].imm);
+# else
+       printf("%p", socket_prog);
+# endif
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, license))
+               printf(", license=\"BSD\"");
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, log_buf))
+               printf(", log_level=42, log_size=%zu, log_buf=\"\"",
+                      sizeof(log_buf));
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, kern_version))
+               printf(", kern_version=KERNEL_VERSION(57005, 192, 222)");
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, prog_flags))
+               printf(", prog_flags=0");
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, prog_name))
+               printf(", prog_name=\"test_prog\"");
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, prog_ifindex))
+               printf(", prog_ifindex=0");
+       if (size > offsetof(struct BPF_PROG_LOAD_struct, expected_attach_type))
+               printf(", expected_attach_type=BPF_CGROUP_INET_INGRESS");
+       printf("}, %zu) = ", size);
+       if (rc >= 0)
+               printf("%ld<anon_inode:bpf-prog>\n", rc);
+       else
+               puts(errstr);
+}
+#endif /* CHECK_OBJ_PROG */
+
+static long
+try_bpf(kernel_ulong_t cmd, void (*printer)(void *attr, size_t size, long rc),
+       void *attr, size_t **sizes)
+{
+       long rc;
+
+       for (rc = -1; **sizes; (*sizes)++) {
+               rc = sys_bpf(cmd, attr, **sizes);
+               printer(attr, **sizes, rc);
+
+               if (rc >= 0)
+                       break;
+       }
+
+       return rc;
+}
+
+int
+main(void)
+{
+       struct BPF_MAP_CREATE_struct bpf_map_create_attr = {
+               .map_type    = BPF_MAP_TYPE_ARRAY,
+               .key_size    = 4,
+               .value_size  = 8,
+               .max_entries = 1,
+               .map_name    = "test_map",
+       };
+       size_t bpf_map_create_attr_sizes[] = {
+               sizeof(bpf_map_create_attr),
+               offsetofend(struct BPF_MAP_CREATE_struct, max_entries),
+               0,
+       };
+
+#if CHECK_OBJ_PROG
+       struct BPF_PROG_LOAD_struct bpf_prog_load_attr = {
+               .prog_type    = BPF_PROG_TYPE_SOCKET_FILTER,
+               .insn_cnt     = ARRAY_SIZE(socket_prog),
+               .insns        = (uintptr_t) socket_prog,
+               .license      = (uintptr_t) license,
+               .log_level    = 42,
+               .log_size     = sizeof(log_buf),
+               .log_buf      = (uintptr_t) log_buf,
+               .kern_version = 0xdeadc0de,
+               .prog_name    = "test_prog",
+       };
+       size_t bpf_prog_load_attr_sizes[] = {
+               sizeof(bpf_prog_load_attr),
+               offsetofend(struct BPF_PROG_LOAD_struct, prog_name),
+               offsetofend(struct BPF_PROG_LOAD_struct, prog_flags),
+               offsetofend(struct BPF_PROG_LOAD_struct, kern_version),
+               offsetofend(struct BPF_PROG_LOAD_struct, log_buf),
+               offsetofend(struct BPF_PROG_LOAD_struct, license),
+               offsetofend(struct BPF_PROG_LOAD_struct, insns),
+               0,
+       };
+#endif /* CHECK_OBJ_PROG */
+
+       size_t *bpf_map_create_attr_size = bpf_map_create_attr_sizes;
+       int map_fd = try_bpf(BPF_MAP_CREATE, print_map_create,
+                            &bpf_map_create_attr, &bpf_map_create_attr_size);
+       if (map_fd < 0)
+               perror_msg_and_skip("BPF_MAP_CREATE failed");
+
+#if CHECK_OBJ_PROG
+       socket_prog[4].imm = map_fd;
+
+       size_t *bpf_prog_load_attr_size = bpf_prog_load_attr_sizes;
+       int prog_fd = try_bpf(BPF_PROG_LOAD, print_prog_load,
+                             &bpf_prog_load_attr, &bpf_prog_load_attr_size);
+       if (prog_fd < 0)
+               perror_msg_and_skip("BPF_PROG_LOAD failed (log: \"%s\")",
+                                   log_buf);
+#endif /* CHECK_OBJ_PROG */
+
+       /*
+        * This has to be a macro, otherwise the compiler complains that
+        * initializer element is not constant.
+        */
+       #define  MAP_INFO_SZ (sizeof(*map_info) + 64)
+       struct bpf_map_info_struct *map_info = calloc(1, MAP_INFO_SZ);
+       struct BPF_OBJ_GET_INFO_BY_FD_struct bpf_map_get_info_attr = {
+               .bpf_fd   = map_fd,
+               .info_len = MAP_INFO_SZ,
+               .info     = (uintptr_t) map_info,
+       };
+
+       int ret = sys_bpf(BPF_OBJ_GET_INFO_BY_FD, &bpf_map_get_info_attr,
+                         sizeof(bpf_map_get_info_attr));
+       if (ret < 0)
+               perror_msg_and_skip("BPF_OBJ_GET_INFO_BY_FD map failed");
+
+       printf("bpf(BPF_OBJ_GET_INFO_BY_FD"
+              ", {info={bpf_fd=%d<anon_inode:bpf-map>, info_len=%zu",
+              map_fd, MAP_INFO_SZ);
+       if (bpf_map_get_info_attr.info_len != MAP_INFO_SZ)
+               printf(" => %u", bpf_map_get_info_attr.info_len);
+
+       printf(", info=");
+#if VERBOSE
+       printf("{type=");
+       printxval(bpf_map_types, map_info->type, "BPF_MAP_TYPE_???");
+       PRINT_FIELD_U(", ", *map_info, id);
+       PRINT_FIELD_U(", ", *map_info, key_size);
+       PRINT_FIELD_U(", ", *map_info, value_size);
+       PRINT_FIELD_U(", ", *map_info, max_entries);
+       printf(", map_flags=");
+       printflags(bpf_map_flags, map_info->map_flags, "BPF_F_???");
+
+       if (bpf_map_get_info_attr.info_len >
+           offsetof(struct bpf_map_info_struct, name)) {
+               printf(", name=");
+               print_quoted_cstring(map_info->name, sizeof(map_info->name));
+       }
+       if (bpf_map_get_info_attr.info_len >
+           offsetof(struct bpf_map_info_struct, ifindex))
+               printf(", ifindex=%u", map_info->ifindex);
+       if (bpf_map_get_info_attr.info_len >
+           offsetof(struct bpf_map_info_struct, netns_dev))
+               printf(", netns_dev=makedev(%u, %u)",
+                      major(map_info->netns_dev), minor(map_info->netns_dev));
+       if (bpf_map_get_info_attr.info_len >
+           offsetof(struct bpf_map_info_struct, netns_ino))
+               printf(", netns_ino=%" PRIu64, map_info->netns_ino);
+       printf("}");
+#else /* !VERBOSE */
+       printf("%p", map_info);
+#endif /* VERBOSE */
+       printf("}}, %zu) = %s\n", sizeof(bpf_map_get_info_attr), errstr);
+
+#if CHECK_OBJ_PROG
+       /*
+        * This has to be a macro, otherwise the compiler complains that
+        * initializer element is not constant.
+        */
+       #define  PROG_INFO_SZ (sizeof(*prog_info) + 64)
+       struct bpf_prog_info_struct *prog_info = calloc(1, PROG_INFO_SZ);
+       struct bpf_insn *xlated_prog = tail_alloc(sizeof(*xlated_prog) * 42);
+       uint32_t *map_ids = tail_alloc(sizeof(*map_ids) * 2);
+       struct BPF_OBJ_GET_INFO_BY_FD_struct bpf_prog_get_info_attr = {
+               .bpf_fd   = prog_fd,
+               .info_len = PROG_INFO_SZ,
+               .info     = (uintptr_t) prog_info,
+       };
+       size_t old_prog_info_len = PROG_INFO_SZ;
+
+       for (unsigned int i = 0; i < 4; i++) {
+               prog_info->jited_prog_len = 0;
+               switch (i) {
+               case 1:
+                       prog_info->xlated_prog_insns =
+                               (uintptr_t) (xlated_prog + 42);
+                       prog_info->xlated_prog_len = 336;
+                       prog_info->map_ids = (uintptr_t) (map_ids + 2);
+                       prog_info->nr_map_ids = 2;
+                       break;
+               case 2:
+                       prog_info->xlated_prog_insns = (uintptr_t) xlated_prog;
+                       /* TODO: check xlated_prog output */
+                       prog_info->xlated_prog_len = 0;
+                       prog_info->map_ids = (uintptr_t) map_ids;
+                       prog_info->nr_map_ids = 0;
+                       break;
+               case 3:
+                       prog_info->xlated_prog_insns = (uintptr_t) xlated_prog;
+                       prog_info->xlated_prog_len = 0;
+                       prog_info->map_ids = (uintptr_t) map_ids;
+                       prog_info->nr_map_ids = 2;
+                       break;
+               }
+
+               ret = sys_bpf(BPF_OBJ_GET_INFO_BY_FD, &bpf_prog_get_info_attr,
+                             sizeof(bpf_prog_get_info_attr));
+               if (i != 1 && ret < 0)
+                       perror_msg_and_skip("BPF_OBJ_GET_INFO_BY_FD"
+                                           " prog %u failed", i);
+
+               printf("bpf(BPF_OBJ_GET_INFO_BY_FD"
+                      ", {info={bpf_fd=%d<anon_inode:bpf-prog>, info_len=%zu",
+                      prog_fd, old_prog_info_len);
+               if (!i && bpf_prog_get_info_attr.info_len != PROG_INFO_SZ)
+                       printf(" => %u", bpf_prog_get_info_attr.info_len);
+               old_prog_info_len = bpf_prog_get_info_attr.info_len;
+
+               printf(", info=");
+# if VERBOSE
+               printf("{type=");
+               printxval(bpf_prog_types, prog_info->type, "BPF_PROG_TYPE_???");
+               PRINT_FIELD_U(", ", *prog_info, id);
+               printf(", tag=");
+               print_quoted_hex(prog_info->tag, sizeof(prog_info->tag));
+               printf(", jited_prog_len=0");
+               if (prog_info->jited_prog_len)
+                       printf(" => %u", prog_info->jited_prog_len);
+               printf(", jited_prog_insns=NULL");
+               switch (i) {
+               case 0:
+                       printf(", xlated_prog_len=0");
+                       if (prog_info->xlated_prog_len)
+                               printf(" => %u", prog_info->xlated_prog_len);
+                       printf(", xlated_prog_insns=NULL");
+                       break;
+               case 1:
+                       printf(", xlated_prog_len=336");
+                       if (prog_info->xlated_prog_len != 336)
+                               printf(" => %u", prog_info->xlated_prog_len);
+                       if (prog_info->xlated_prog_len)
+                               printf(", xlated_prog_insns=%p", xlated_prog + 42);
+                       else
+                               printf(", xlated_prog_insns=[]");
+                       break;
+               case 2:
+               case 3:
+                       printf(", xlated_prog_len=0");
+                       if (prog_info->xlated_prog_len)
+                               printf(" => %u", prog_info->xlated_prog_len);
+                       printf(", xlated_prog_insns=[]");
+                       break;
+               }
+
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, load_time))
+                       printf(", load_time=%" PRIu64, prog_info->load_time);
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, created_by_uid))
+                       printf(", created_by_uid=%u",
+                              prog_info->created_by_uid);
+
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, map_ids)) {
+                       switch (i) {
+                       case 0:
+                               printf(", nr_map_ids=0");
+                               if (prog_info->nr_map_ids)
+                                       printf(" => 1");
+                               printf(", map_ids=NULL");
+                               break;
+                       case 1:
+                               printf(", nr_map_ids=2, map_ids=%p",
+                                      map_ids + 2);
+                               break;
+                       case 2:
+                               printf(", nr_map_ids=0");
+                               if (prog_info->nr_map_ids)
+                                       printf(" => 1");
+                               printf(", map_ids=[]");
+                               break;
+                       case 3:
+                               printf(", nr_map_ids=2");
+                               if (prog_info->nr_map_ids != 2)
+                                       printf(" => 1");
+                               printf(", map_ids=[%u]", map_info->id);
+                               break;
+                       }
+               }
+
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, name))
+                       printf(", name=\"test_prog\"");
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, ifindex))
+                       printf(", ifindex=%u", prog_info->ifindex);
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, netns_dev))
+                       printf(", netns_dev=makedev(%u, %u)",
+                              major(prog_info->netns_dev),
+                              minor(prog_info->netns_dev));
+               if (bpf_prog_get_info_attr.info_len >
+                   offsetof(struct bpf_prog_info_struct, netns_ino))
+                       printf(", netns_ino=%" PRIu64, prog_info->netns_ino);
+
+               printf("}");
+# else /* !VERBOSE */
+               printf("%p", prog_info);
+# endif /* VERBOSE */
+               printf("}}, %zu) = %s\n",
+                      sizeof(bpf_prog_get_info_attr), errstr);
+       }
+#endif /* CHECK_OBJ_PROG */
+
+       puts("+++ exited with 0 +++");
+       return 0;
+}
index 4400c485b2d25ddf4626622b6ef7745719778c07..f3518d1985f534f0466eac08ffe5341649a4be62 100644 (file)
@@ -37,6 +37,10 @@ aio  -a14 -e trace=io_setup,io_submit,io_getevents,io_cancel,io_destroy
 alarm  -a10
 bpf    -a20
 bpf-v  -a20 -v -e trace=bpf
+bpf-obj_get_info_by_fd -a20 -y -e trace=bpf
+bpf-obj_get_info_by_fd-v       -a20 -y -v -e trace=bpf
+bpf-obj_get_info_by_fd-prog    -a20 -y -e trace=bpf
+bpf-obj_get_info_by_fd-prog-v  -a20 -y -v -e trace=bpf
 btrfs  +ioctl.test
 chmod  -a28
 chown  -a28
index 80c8ae75ac013ecb49e560c7f96a34906b688504..df84fe6698812181bf66dcad8f1a56323294e525 100755 (executable)
@@ -10,6 +10,10 @@ aio
 alarm
 bpf
 bpf-v
+bpf-obj_get_info_by_fd
+bpf-obj_get_info_by_fd-v
+bpf-obj_get_info_by_fd-prog
+bpf-obj_get_info_by_fd-prog-v
 brk
 btrfs
 caps