From: "Jose E. Marchesi" <jose.marchesi@oracle.com>
To: James Hilliard <james.hilliard1@gmail.com>
Cc: Andrii Nakryiko <andrii.nakryiko@gmail.com>,
bpf <bpf@vger.kernel.org>,
david.faust@oracle.com
Subject: Re: bpftool gen object doesn't handle GCC built BPF ELF files
Date: Sun, 10 Jul 2022 00:41:31 +0200 [thread overview]
Message-ID: <87tu7p3o4k.fsf@oracle.com> (raw)
In-Reply-To: <CADvTj4pFQmS6XHpHCVO8jt-8ZRdTd--uny-n9vA0+vm4xUoLzQ@mail.gmail.com> (James Hilliard's message of "Sat, 9 Jul 2022 14:39:31 -0600")
> On Sat, Jul 9, 2022 at 2:32 PM James Hilliard <james.hilliard1@gmail.com> wrote:
>>
>> On Sat, Jul 9, 2022 at 2:21 PM Jose E. Marchesi
>> <jose.marchesi@oracle.com> wrote:
>> >
>> >
>> > > On Sat, Jul 9, 2022 at 11:24 AM Jose E. Marchesi
>> > > <jose.marchesi@oracle.com> wrote:
>> > >>
>> > >>
>> > >> > On Fri, Jul 8, 2022 at 12:33 PM Jose E. Marchesi
>> > >> > <jose.marchesi@oracle.com> wrote:
>> > >> >>
>> > >> >>
>> > >> >> >> On Wed, Jul 6, 2022 at 10:13 AM James Hilliard
>> > >> >> >> <james.hilliard1@gmail.com> wrote:
>> > >> >> >>>
>> > >> >> >>> Note I'm testing with the following patches:
>> > >> >> >>> https://lore.kernel.org/bpf/20220706111839.1247911-1-james.hilliard1@gmail.com/
>> > >> >> >>> https://lore.kernel.org/bpf/20220706140623.2917858-1-james.hilliard1@gmail.com/
>> > >> >> >>>
>> > >> >> >>> It would appear there's some compatibility issues with bpftool gen and
>> > >> >> >>> GCC, not sure what side though is wrong here:
>> > >> >> >>> /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
>> > >> >> >>> gen object src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
>> > >> >> >>> src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.unstripped.o
>> > >> >> >>> libbpf: failed to find BTF info for global/extern symbol 'sd_restrictif_i'
>> > >> >> >>> Error: failed to link
>> > >> >> >>> 'src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.unstripped.o':
>> > >> >> >>> Unknown error -2 (-2)
>> > >> >> >>>
>> > >> >> >>> Relevant difference seems to be this:
>> > >> >> >>> GCC:
>> > >> >> >>> [55] FUNC 'sd_restrictif_i' type_id=47 linkage=static
>> > >> >> >>> Clang:
>> > >> >> >>> [27] FUNC 'sd_restrictif_i' type_id=26 linkage=global
>> > >> >> >
>> > >> >> > For functions GCC generates a BTF_KIND_FUNC entry, which has no linkage
>> > >> >> > information, or so we thought: I just looked at bpftool/btf.c and I
>> > >> >> > found the linkage info for function types is expected to be encoded in
>> > >> >> > the vlen field of BTF_KIND_FUNC entries (why not adding a btf_func
>> > >> >> > instead???) which is surprising to say the least.
>> > >> >> >
>> > >> >> > We are changing GCC to encode the linkage info in vlen for these types.
>> > >> >> > Thanks for reporting this.
>> > >> >>
>> > >> >> Patch sent to GCC upstream:
>> > >> >> https://gcc.gnu.org/pipermail/gcc-patches/2022-July/598090.html
>> > >> >
>> > >> > I applied that patch on top of GCC 12.1.0 and it appears to fix the
>> > >> > bpftool gen object bug.
>> > >> >
>> > >> > I am however now hitting a different error during skeleton generation:
>> > >> > /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
>> > >> > gen skeleton src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
>> > >> > libbpf: elf: skipping unrecognized data section(9) .comment
>> > >> > libbpf: failed to alloc map 'restrict.bss' content buffer: -22
>> > >> > Error: failed to open BPF object file: Invalid argument
>> > >>
>> > >> What is the size of the .bss section in the object file? Try with:
>> > >>
>> > >> $ size restrict-ifaces.bpf.o
>> > >
>> > > $ size
> output/build/systemd-custom/build/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
>> > > text data bss dec hex filename
>> > > 386 25 0 411 19b
>> > > output/build/systemd-custom/build/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
>> >
>> > Right, so the .bss section is empty. I see a `const volatile unsigned
>> > char is_allow_list = 0;' in restrict-ifaces.bpf.c, but that goes to
>> > .data and not to .bss, as expected.
>> >
>> > If you build restrict-ifaces.bpf.o with LLVM, is the bss still empty? I
>> > don't think the code in libbpf.c even checks for this eventuality...
>>
>> LLVM version(which skeleton generation works with):
>> $ size restrict-ifaces.bpf.o
>> text data bss dec hex filename
>> 323 24 0 347 15b restrict-ifaces.bpf.o
>>
>> $ /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
>> btf dump file restrict-ifaces.bpf.o format raw
>> [1] PTR '(anon)' type_id=3
>> [2] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED
>> [3] ARRAY '(anon)' type_id=2 index_type_id=4 nr_elems=1
>> [4] INT '__ARRAY_SIZE_TYPE__' size=4 bits_offset=0 nr_bits=32 encoding=(none)
>> [5] PTR '(anon)' type_id=6
>> [6] TYPEDEF '__u32' type_id=7
>> [7] INT 'unsigned int' size=4 bits_offset=0 nr_bits=32 encoding=(none)
>> [8] PTR '(anon)' type_id=9
>> [9] TYPEDEF '__u8' type_id=10
>> [10] INT 'unsigned char' size=1 bits_offset=0 nr_bits=8 encoding=(none)
>> [11] STRUCT '(anon)' size=24 vlen=3
>> 'type' type_id=1 bits_offset=0
>> 'key' type_id=5 bits_offset=64
>> 'value' type_id=8 bits_offset=128
>> [12] VAR 'sd_restrictif' type_id=11, linkage=global
>> [13] PTR '(anon)' type_id=14
>> [14] CONST '(anon)' type_id=15
>> [15] STRUCT '__sk_buff' size=192 vlen=33
>> 'len' type_id=6 bits_offset=0
>> 'pkt_type' type_id=6 bits_offset=32
>> 'mark' type_id=6 bits_offset=64
>> 'queue_mapping' type_id=6 bits_offset=96
>> 'protocol' type_id=6 bits_offset=128
>> 'vlan_present' type_id=6 bits_offset=160
>> 'vlan_tci' type_id=6 bits_offset=192
>> 'vlan_proto' type_id=6 bits_offset=224
>> 'priority' type_id=6 bits_offset=256
>> 'ingress_ifindex' type_id=6 bits_offset=288
>> 'ifindex' type_id=6 bits_offset=320
>> 'tc_index' type_id=6 bits_offset=352
>> 'cb' type_id=16 bits_offset=384
>> 'hash' type_id=6 bits_offset=544
>> 'tc_classid' type_id=6 bits_offset=576
>> 'data' type_id=6 bits_offset=608
>> 'data_end' type_id=6 bits_offset=640
>> 'napi_id' type_id=6 bits_offset=672
>> 'family' type_id=6 bits_offset=704
>> 'remote_ip4' type_id=6 bits_offset=736
>> 'local_ip4' type_id=6 bits_offset=768
>> 'remote_ip6' type_id=17 bits_offset=800
>> 'local_ip6' type_id=17 bits_offset=928
>> 'remote_port' type_id=6 bits_offset=1056
>> 'local_port' type_id=6 bits_offset=1088
>> 'data_meta' type_id=6 bits_offset=1120
>> '(anon)' type_id=18 bits_offset=1152
>> 'tstamp' type_id=20 bits_offset=1216
>> 'wire_len' type_id=6 bits_offset=1280
>> 'gso_segs' type_id=6 bits_offset=1312
>> '(anon)' type_id=22 bits_offset=1344
>> 'gso_size' type_id=6 bits_offset=1408
>> 'hwtstamp' type_id=20 bits_offset=1472
>> [16] ARRAY '(anon)' type_id=6 index_type_id=4 nr_elems=5
>> [17] ARRAY '(anon)' type_id=6 index_type_id=4 nr_elems=4
>> [18] UNION '(anon)' size=8 vlen=1
>> 'flow_keys' type_id=19 bits_offset=0
>> [19] PTR '(anon)' type_id=34
>> [20] TYPEDEF '__u64' type_id=21
>> [21] INT 'unsigned long long' size=8 bits_offset=0 nr_bits=64 encoding=(none)
>> [22] UNION '(anon)' size=8 vlen=1
>> 'sk' type_id=23 bits_offset=0
>> [23] PTR '(anon)' type_id=35
>> [24] FUNC_PROTO '(anon)' ret_type_id=2 vlen=1
>> 'sk' type_id=13
>> [25] FUNC 'sd_restrictif_e' type_id=24 linkage=global
>> [26] FUNC 'sd_restrictif_i' type_id=24 linkage=global
>> [27] CONST '(anon)' type_id=28
>> [28] VOLATILE '(anon)' type_id=9
>> [29] VAR 'is_allow_list' type_id=27, linkage=global
>> [30] CONST '(anon)' type_id=31
>> [31] INT 'char' size=1 bits_offset=0 nr_bits=8 encoding=SIGNED
>> [32] ARRAY '(anon)' type_id=30 index_type_id=4 nr_elems=18
>> [33] VAR '_license' type_id=32, linkage=static
>> [34] FWD 'bpf_flow_keys' fwd_kind=struct
>> [35] FWD 'bpf_sock' fwd_kind=struct
>> [36] DATASEC '.rodata' size=1 vlen=1
>> type_id=29 offset=0 size=1 (VAR 'is_allow_list')
>> [37] DATASEC 'license' size=18 vlen=1
>> type_id=33 offset=0 size=18 (VAR '_license')
>> [38] DATASEC '.maps' size=24 vlen=1
>> type_id=12 offset=0 size=24 (VAR 'sd_restrictif')
>>
>
> Skeleton generation debug output for GCC(failing) and LLVM(working)
> which may be helpful:
Indeed it was helpful :)
The GNU assembler generates an empty .bss section. This is a well
established behavior in GAS that happens in all supported targets.
The LLVM assembler doesn't generate an empty .bss section.
bpftool chokes on the empty .bss section.
In this case I would suggest to fix bpf_object__init_global_data_maps in
order to skip empty sections.
Something like this:
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index e89cc9c885b3..e3a6808f0bb6 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -1591,6 +1591,10 @@ static int bpf_object__init_global_data_maps(struct bpf_object *obj)
for (sec_idx = 1; sec_idx < obj->efile.sec_cnt; sec_idx++) {
sec_desc = &obj->efile.secs[sec_idx];
+ /* Skip empty sections. */
+ if (sec_desc->data->d_size == 0)
+ continue;
+
switch (sec_desc->sec_type) {
case SEC_DATA:
sec_name = elf_sec_name(obj, elf_sec_by_idx(obj, sec_idx));
> GCC:
> $ /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
> --debug gen skeleton
> output/build/systemd-custom/build/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
> libbpf: loading object 'restrict_ifaces_bpf' from buffer
> libbpf: elf: section(2) .symtab, size 336, link 1, flags 0, type=2
> libbpf: elf: section(3) .data, size 1, link 0, flags 3, type=1
> libbpf: elf: section(4) .bss, size 0, link 0, flags 3, type=8
> libbpf: elf: section(5) cgroup_skb/egress, size 184, link 0, flags 6, type=1
> libbpf: sec 'cgroup_skb/egress': found program 'sd_restrictif_e' at
> insn offset 0 (0 bytes), code size 23 insns (184 bytes)
> libbpf: elf: section(6) cgroup_skb/ingress, size 184, link 0, flags 6, type=1
> libbpf: sec 'cgroup_skb/ingress': found program 'sd_restrictif_i' at
> insn offset 0 (0 bytes), code size 23 insns (184 bytes)
> libbpf: elf: section(7) license, size 18, link 0, flags 2, type=1
> libbpf: license of restrict_ifaces_bpf is LGPL-2.1-or-later
> libbpf: elf: section(8) .maps, size 24, link 0, flags 3, type=1
> libbpf: elf: section(9) .comment, size 49, link 0, flags 30, type=1
> libbpf: elf: skipping unrecognized data section(9) .comment
> libbpf: elf: section(10) .relcgroup_skb/egress, size 32, link 2, flags
> 40, type=9
> libbpf: elf: section(11) .relcgroup_skb/ingress, size 32, link 2,
> flags 40, type=9
> libbpf: elf: section(12) .BTF, size 3606, link 0, flags 0, type=1
> libbpf: looking for externs among 14 symbols...
> libbpf: collected 0 externs total
> libbpf: map 'sd_restrictif': at sec_idx 8, offset 0.
> libbpf: map 'sd_restrictif': found type = 1.
> libbpf: map 'sd_restrictif': found key [12], sz = 4.
> libbpf: map 'sd_restrictif': found value [3], sz = 1.
> libbpf: map 'restrict.data' (global data): at sec_idx 3, offset 0, flags 400.
> libbpf: map 1 is "restrict.data"
> libbpf: map 'restrict.bss' (global data): at sec_idx 4, offset 0, flags 400.
> libbpf: failed to alloc map 'restrict.bss' content buffer: -22
> Error: failed to open BPF object file: Invalid argument
>
> LLVM:
> $ /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
> --debug gen skeleton restrict-ifaces.bpf.o
> libbpf: loading object 'restrict_ifaces_bpf' from buffer
> libbpf: elf: section(2) .symtab, size 384, link 1, flags 0, type=2
> libbpf: elf: section(3) cgroup_skb/egress, size 152, link 0, flags 6, type=1
> libbpf: sec 'cgroup_skb/egress': found program 'sd_restrictif_e' at
> insn offset 0 (0 bytes), code size 19 insns (152 bytes)
> libbpf: elf: section(4) cgroup_skb/ingress, size 152, link 0, flags 6, type=1
> libbpf: sec 'cgroup_skb/ingress': found program 'sd_restrictif_i' at
> insn offset 0 (0 bytes), code size 19 insns (152 bytes)
> libbpf: elf: section(5) .rodata, size 1, link 0, flags 2, type=1
> libbpf: elf: section(6) license, size 18, link 0, flags 2, type=1
> libbpf: license of restrict_ifaces_bpf is LGPL-2.1-or-later
> libbpf: elf: section(7) .maps, size 24, link 0, flags 3, type=1
> libbpf: elf: section(8) .relcgroup_skb/egress, size 32, link 2, flags 40, type=9
> libbpf: elf: section(9) .relcgroup_skb/ingress, size 32, link 2, flags
> 40, type=9
> libbpf: elf: section(10) .BTF, size 1988, link 0, flags 0, type=1
> libbpf: elf: section(11) .BTF.ext, size 376, link 0, flags 0, type=1
> libbpf: looking for externs among 16 symbols...
> libbpf: collected 0 externs total
> libbpf: map 'sd_restrictif': at sec_idx 7, offset 0.
> libbpf: map 'sd_restrictif': found type = 1.
> libbpf: map 'sd_restrictif': found key [6], sz = 4.
> libbpf: map 'sd_restrictif': found value [9], sz = 1.
> libbpf: map 'restrict.rodata' (global data): at sec_idx 5, offset 0, flags 480.
> libbpf: map 1 is "restrict.rodata"
> libbpf: sec '.relcgroup_skb/egress': collecting relocation for
> section(3) 'cgroup_skb/egress'
> libbpf: sec '.relcgroup_skb/egress': relo #0: insn #4 against 'sd_restrictif'
> libbpf: prog 'sd_restrictif_e': found map 0 (sd_restrictif, sec 7, off
> 0) for insn #4
> libbpf: sec '.relcgroup_skb/egress': relo #1: insn #7 against 'is_allow_list'
> libbpf: prog 'sd_restrictif_e': found data map 1 (restrict.rodata, sec
> 5, off 0) for insn 7
> libbpf: sec '.relcgroup_skb/ingress': collecting relocation for
> section(4) 'cgroup_skb/ingress'
> libbpf: sec '.relcgroup_skb/ingress': relo #0: insn #4 against 'sd_restrictif'
> libbpf: prog 'sd_restrictif_i': found map 0 (sd_restrictif, sec 7, off
> 0) for insn #4
> libbpf: sec '.relcgroup_skb/ingress': relo #1: insn #7 against 'is_allow_list'
> libbpf: prog 'sd_restrictif_i': found data map 1 (restrict.rodata, sec
> 5, off 0) for insn 7
>
>> >
>> > >>
>> > >> Looking at libbpf.c, it seems to me that this may be due of trying to
>> > >> mmap 0 bytes in `bpf_object__init_internal_map':
>> > >>
>> > >> map->mmaped = mmap(NULL, bpf_map_mmap_sz(map), PROT_READ | PROT_WRITE,
>> > >> MAP_SHARED | MAP_ANONYMOUS, -1, 0);
>> > >> if (map->mmaped == MAP_FAILED) {
>> > >> err = -errno;
>> > >> map->mmaped = NULL;
>> > >> pr_warn("failed to alloc map '%s' content buffer: %d\n",
>> > >> map->name, err);
>> > >> zfree(&map->real_name);
>> > >> zfree(&map->name);
>> > >> return err;
>> > >> }
>> > >>
>> > >> I see no check for zero sized sections in
>> > >> bpf_object__init_global_data_maps.
>> > >>
>> > >> Is maybe GCC failing to allocate stuff in BSS that is supposed to be
>> > >> there?
>> > >>
>> > >> > Stripped file passed to gen skeleton:
>> > >> > /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
>> > >> > btf dump file
>> > >> > output/build/systemd-custom/build/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.o
>> > >> > format raw
>> > >> > [1] INT 'signed char' size=1 bits_offset=0 nr_bits=8 encoding=UNKN
>> > >> > [2] INT 'unsigned char' size=1 bits_offset=0 nr_bits=8 encoding=CHAR
>> > >> > [3] TYPEDEF '__u8' type_id=2
>> > >> > [4] CONST '(anon)' type_id=3
>> > >> > [5] VOLATILE '(anon)' type_id=4
>> > >> > [6] INT 'short int' size=2 bits_offset=0 nr_bits=16 encoding=SIGNED
>> > >> > [7] INT 'short unsigned int' size=2 bits_offset=0 nr_bits=16 encoding=(none)
>> > >> > [8] TYPEDEF '__u16' type_id=7
>> > >> > [9] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED
>> > >> > [10] TYPEDEF '__s32' type_id=9
>> > >> > [11] INT 'unsigned int' size=4 bits_offset=0 nr_bits=32 encoding=(none)
>> > >> > [12] TYPEDEF '__u32' type_id=11
>> > >> > [13] INT 'long long int' size=8 bits_offset=0 nr_bits=64 encoding=SIGNED
>> > >> > [14] INT 'long long unsigned int' size=8 bits_offset=0 nr_bits=64
>> > >> > encoding=(none)
>> > >> > [15] TYPEDEF '__u64' type_id=14
>> > >> > [16] INT 'long unsigned int' size=8 bits_offset=0 nr_bits=64 encoding=(none)
>> > >> > [17] INT 'long int' size=8 bits_offset=0 nr_bits=64 encoding=SIGNED
>> > >> > [18] INT 'char' size=1 bits_offset=0 nr_bits=8 encoding=UNKN
>> > >> > [19] CONST '(anon)' type_id=18
>> > >> > [20] TYPEDEF '__be16' type_id=8
>> > >> > [21] TYPEDEF '__be32' type_id=12
>> > >> > [22] ENUM 'bpf_map_type' encoding=UNSIGNED size=4 vlen=31
>> > >> > 'BPF_MAP_TYPE_UNSPEC' val=0
>> > >> > 'BPF_MAP_TYPE_HASH' val=1
>> > >> > 'BPF_MAP_TYPE_ARRAY' val=2
>> > >> > 'BPF_MAP_TYPE_PROG_ARRAY' val=3
>> > >> > 'BPF_MAP_TYPE_PERF_EVENT_ARRAY' val=4
>> > >> > 'BPF_MAP_TYPE_PERCPU_HASH' val=5
>> > >> > 'BPF_MAP_TYPE_PERCPU_ARRAY' val=6
>> > >> > 'BPF_MAP_TYPE_STACK_TRACE' val=7
>> > >> > 'BPF_MAP_TYPE_CGROUP_ARRAY' val=8
>> > >> > 'BPF_MAP_TYPE_LRU_HASH' val=9
>> > >> > 'BPF_MAP_TYPE_LRU_PERCPU_HASH' val=10
>> > >> > 'BPF_MAP_TYPE_LPM_TRIE' val=11
>> > >> > 'BPF_MAP_TYPE_ARRAY_OF_MAPS' val=12
>> > >> > 'BPF_MAP_TYPE_HASH_OF_MAPS' val=13
>> > >> > 'BPF_MAP_TYPE_DEVMAP' val=14
>> > >> > 'BPF_MAP_TYPE_SOCKMAP' val=15
>> > >> > 'BPF_MAP_TYPE_CPUMAP' val=16
>> > >> > 'BPF_MAP_TYPE_XSKMAP' val=17
>> > >> > 'BPF_MAP_TYPE_SOCKHASH' val=18
>> > >> > 'BPF_MAP_TYPE_CGROUP_STORAGE' val=19
>> > >> > 'BPF_MAP_TYPE_REUSEPORT_SOCKARRAY' val=20
>> > >> > 'BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE' val=21
>> > >> > 'BPF_MAP_TYPE_QUEUE' val=22
>> > >> > 'BPF_MAP_TYPE_STACK' val=23
>> > >> > 'BPF_MAP_TYPE_SK_STORAGE' val=24
>> > >> > 'BPF_MAP_TYPE_DEVMAP_HASH' val=25
>> > >> > 'BPF_MAP_TYPE_STRUCT_OPS' val=26
>> > >> > 'BPF_MAP_TYPE_RINGBUF' val=27
>> > >> > 'BPF_MAP_TYPE_INODE_STORAGE' val=28
>> > >> > 'BPF_MAP_TYPE_TASK_STORAGE' val=29
>> > >> > 'BPF_MAP_TYPE_BLOOM_FILTER' val=30
>> > >> > [23] UNION '(anon)' size=8 vlen=1
>> > >> > 'flow_keys' type_id=29 bits_offset=0
>> > >> > [24] STRUCT 'bpf_flow_keys' size=56 vlen=13
>> > >> > 'nhoff' type_id=8 bits_offset=0
>> > >> > 'thoff' type_id=8 bits_offset=16
>> > >> > 'addr_proto' type_id=8 bits_offset=32
>> > >> > 'is_frag' type_id=3 bits_offset=48
>> > >> > 'is_first_frag' type_id=3 bits_offset=56
>> > >> > 'is_encap' type_id=3 bits_offset=64
>> > >> > 'ip_proto' type_id=3 bits_offset=72
>> > >> > 'n_proto' type_id=20 bits_offset=80
>> > >> > 'sport' type_id=20 bits_offset=96
>> > >> > 'dport' type_id=20 bits_offset=112
>> > >> > '(anon)' type_id=25 bits_offset=128
>> > >> > 'flags' type_id=12 bits_offset=384
>> > >> > 'flow_label' type_id=21 bits_offset=416
>> > >> > [25] UNION '(anon)' size=32 vlen=2
>> > >> > '(anon)' type_id=26 bits_offset=0
>> > >> > '(anon)' type_id=27 bits_offset=0
>> > >> > [26] STRUCT '(anon)' size=8 vlen=2
>> > >> > 'ipv4_src' type_id=21 bits_offset=0
>> > >> > 'ipv4_dst' type_id=21 bits_offset=32
>> > >> > [27] STRUCT '(anon)' size=32 vlen=2
>> > >> > 'ipv6_src' type_id=28 bits_offset=0
>> > >> > 'ipv6_dst' type_id=28 bits_offset=128
>> > >> > [28] ARRAY '(anon)' type_id=12 index_type_id=16 nr_elems=4
>> > >> > [29] PTR '(anon)' type_id=24
>> > >> > [30] UNION '(anon)' size=8 vlen=1
>> > >> > 'sk' type_id=32 bits_offset=0
>> > >> > [31] STRUCT 'bpf_sock' size=80 vlen=14
>> > >> > 'bound_dev_if' type_id=12 bits_offset=0
>> > >> > 'family' type_id=12 bits_offset=32
>> > >> > 'type' type_id=12 bits_offset=64
>> > >> > 'protocol' type_id=12 bits_offset=96
>> > >> > 'mark' type_id=12 bits_offset=128
>> > >> > 'priority' type_id=12 bits_offset=160
>> > >> > 'src_ip4' type_id=12 bits_offset=192
>> > >> > 'src_ip6' type_id=28 bits_offset=224
>> > >> > 'src_port' type_id=12 bits_offset=352
>> > >> > 'dst_port' type_id=20 bits_offset=384
>> > >> > 'dst_ip4' type_id=12 bits_offset=416
>> > >> > 'dst_ip6' type_id=28 bits_offset=448
>> > >> > 'state' type_id=12 bits_offset=576
>> > >> > 'rx_queue_mapping' type_id=10 bits_offset=608
>> > >> > [32] PTR '(anon)' type_id=31
>> > >> > [33] STRUCT '__sk_buff' size=192 vlen=33
>> > >> > 'len' type_id=12 bits_offset=0
>> > >> > 'pkt_type' type_id=12 bits_offset=32
>> > >> > 'mark' type_id=12 bits_offset=64
>> > >> > 'queue_mapping' type_id=12 bits_offset=96
>> > >> > 'protocol' type_id=12 bits_offset=128
>> > >> > 'vlan_present' type_id=12 bits_offset=160
>> > >> > 'vlan_tci' type_id=12 bits_offset=192
>> > >> > 'vlan_proto' type_id=12 bits_offset=224
>> > >> > 'priority' type_id=12 bits_offset=256
>> > >> > 'ingress_ifindex' type_id=12 bits_offset=288
>> > >> > 'ifindex' type_id=12 bits_offset=320
>> > >> > 'tc_index' type_id=12 bits_offset=352
>> > >> > 'cb' type_id=34 bits_offset=384
>> > >> > 'hash' type_id=12 bits_offset=544
>> > >> > 'tc_classid' type_id=12 bits_offset=576
>> > >> > 'data' type_id=12 bits_offset=608
>> > >> > 'data_end' type_id=12 bits_offset=640
>> > >> > 'napi_id' type_id=12 bits_offset=672
>> > >> > 'family' type_id=12 bits_offset=704
>> > >> > 'remote_ip4' type_id=12 bits_offset=736
>> > >> > 'local_ip4' type_id=12 bits_offset=768
>> > >> > 'remote_ip6' type_id=28 bits_offset=800
>> > >> > 'local_ip6' type_id=28 bits_offset=928
>> > >> > 'remote_port' type_id=12 bits_offset=1056
>> > >> > 'local_port' type_id=12 bits_offset=1088
>> > >> > 'data_meta' type_id=12 bits_offset=1120
>> > >> > '(anon)' type_id=23 bits_offset=1152
>> > >> > 'tstamp' type_id=15 bits_offset=1216
>> > >> > 'wire_len' type_id=12 bits_offset=1280
>> > >> > 'gso_segs' type_id=12 bits_offset=1312
>> > >> > '(anon)' type_id=30 bits_offset=1344
>> > >> > 'gso_size' type_id=12 bits_offset=1408
>> > >> > 'hwtstamp' type_id=15 bits_offset=1472
>> > >> > [34] ARRAY '(anon)' type_id=12 index_type_id=16 nr_elems=5
>> > >> > [35] CONST '(anon)' type_id=33
>> > >> > [36] PTR '(anon)' type_id=0
>> > >> > [37] STRUCT '(anon)' size=24 vlen=3
>> > >> > 'type' type_id=39 bits_offset=0
>> > >> > 'key' type_id=40 bits_offset=64
>> > >> > 'value' type_id=41 bits_offset=128
>> > >> > [38] ARRAY '(anon)' type_id=9 index_type_id=16 nr_elems=1
>> > >> > [39] PTR '(anon)' type_id=38
>> > >> > [40] PTR '(anon)' type_id=12
>> > >> > [41] PTR '(anon)' type_id=3
>> > >> > [42] ARRAY '(anon)' type_id=19 index_type_id=16 nr_elems=18
>> > >> > [43] CONST '(anon)' type_id=42
>> > >> > [44] FUNC_PROTO '(anon)' ret_type_id=36 vlen=2
>> > >> > '(anon)' type_id=36
>> > >> > '(anon)' type_id=46
>> > >> > [45] CONST '(anon)' type_id=0
>> > >> > [46] PTR '(anon)' type_id=45
>> > >> > [47] FUNC_PROTO '(anon)' ret_type_id=9 vlen=1
>> > >> > 'sk' type_id=48
>> > >> > [48] PTR '(anon)' type_id=35
>> > >> > [49] VAR '_license' type_id=43, linkage=static
>> > >> > [50] VAR 'is_allow_list' type_id=5, linkage=global
>> > >> > [51] VAR 'sd_restrictif' type_id=37, linkage=global
>> > >> > [52] FUNC 'sd_restrictif_i' type_id=47 linkage=global
>> > >> > [53] FUNC 'sd_restrictif_e' type_id=47 linkage=global
>> > >> > [54] FUNC 'restrict_network_interfaces_impl' type_id=47 linkage=static
>> > >> > [55] DATASEC '.data' size=1 vlen=1
>> > >> > type_id=50 offset=0 size=1 (VAR 'is_allow_list')
>> > >> > [56] DATASEC 'license' size=18 vlen=1
>> > >> > type_id=49 offset=0 size=18 (VAR '_license')
>> > >> > [57] DATASEC '.maps' size=24 vlen=1
>> > >> > type_id=51 offset=0 size=24 (VAR 'sd_restrictif')
>> > >> >
>> > >> > File before being stripped using bpftool gen object:
>> > >> > /home/buildroot/buildroot/output/per-package/systemd/host/sbin/bpftool
>> > >> > btf dump file
>> > >> >
> output/build/systemd-custom/build/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.unstripped.o
>> > >> > format raw
>> > >> > [1] INT 'signed char' size=1 bits_offset=0 nr_bits=8 encoding=UNKN
>> > >> > [2] INT 'unsigned char' size=1 bits_offset=0 nr_bits=8 encoding=CHAR
>> > >> > [3] TYPEDEF '__u8' type_id=2
>> > >> > [4] CONST '(anon)' type_id=3
>> > >> > [5] VOLATILE '(anon)' type_id=4
>> > >> > [6] INT 'short int' size=2 bits_offset=0 nr_bits=16 encoding=SIGNED
>> > >> > [7] INT 'short unsigned int' size=2 bits_offset=0 nr_bits=16 encoding=(none)
>> > >> > [8] TYPEDEF '__u16' type_id=7
>> > >> > [9] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED
>> > >> > [10] TYPEDEF '__s32' type_id=9
>> > >> > [11] INT 'unsigned int' size=4 bits_offset=0 nr_bits=32 encoding=(none)
>> > >> > [12] TYPEDEF '__u32' type_id=11
>> > >> > [13] INT 'long long int' size=8 bits_offset=0 nr_bits=64 encoding=SIGNED
>> > >> > [14] INT 'long long unsigned int' size=8 bits_offset=0 nr_bits=64
>> > >> > encoding=(none)
>> > >> > [15] TYPEDEF '__u64' type_id=14
>> > >> > [16] INT 'long unsigned int' size=8 bits_offset=0 nr_bits=64 encoding=(none)
>> > >> > [17] INT 'long int' size=8 bits_offset=0 nr_bits=64 encoding=SIGNED
>> > >> > [18] INT 'char' size=1 bits_offset=0 nr_bits=8 encoding=UNKN
>> > >> > [19] CONST '(anon)' type_id=18
>> > >> > [20] TYPEDEF '__be16' type_id=8
>> > >> > [21] TYPEDEF '__be32' type_id=12
>> > >> > [22] ENUM 'bpf_map_type' encoding=UNSIGNED size=4 vlen=31
>> > >> > 'BPF_MAP_TYPE_UNSPEC' val=0
>> > >> > 'BPF_MAP_TYPE_HASH' val=1
>> > >> > 'BPF_MAP_TYPE_ARRAY' val=2
>> > >> > 'BPF_MAP_TYPE_PROG_ARRAY' val=3
>> > >> > 'BPF_MAP_TYPE_PERF_EVENT_ARRAY' val=4
>> > >> > 'BPF_MAP_TYPE_PERCPU_HASH' val=5
>> > >> > 'BPF_MAP_TYPE_PERCPU_ARRAY' val=6
>> > >> > 'BPF_MAP_TYPE_STACK_TRACE' val=7
>> > >> > 'BPF_MAP_TYPE_CGROUP_ARRAY' val=8
>> > >> > 'BPF_MAP_TYPE_LRU_HASH' val=9
>> > >> > 'BPF_MAP_TYPE_LRU_PERCPU_HASH' val=10
>> > >> > 'BPF_MAP_TYPE_LPM_TRIE' val=11
>> > >> > 'BPF_MAP_TYPE_ARRAY_OF_MAPS' val=12
>> > >> > 'BPF_MAP_TYPE_HASH_OF_MAPS' val=13
>> > >> > 'BPF_MAP_TYPE_DEVMAP' val=14
>> > >> > 'BPF_MAP_TYPE_SOCKMAP' val=15
>> > >> > 'BPF_MAP_TYPE_CPUMAP' val=16
>> > >> > 'BPF_MAP_TYPE_XSKMAP' val=17
>> > >> > 'BPF_MAP_TYPE_SOCKHASH' val=18
>> > >> > 'BPF_MAP_TYPE_CGROUP_STORAGE' val=19
>> > >> > 'BPF_MAP_TYPE_REUSEPORT_SOCKARRAY' val=20
>> > >> > 'BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE' val=21
>> > >> > 'BPF_MAP_TYPE_QUEUE' val=22
>> > >> > 'BPF_MAP_TYPE_STACK' val=23
>> > >> > 'BPF_MAP_TYPE_SK_STORAGE' val=24
>> > >> > 'BPF_MAP_TYPE_DEVMAP_HASH' val=25
>> > >> > 'BPF_MAP_TYPE_STRUCT_OPS' val=26
>> > >> > 'BPF_MAP_TYPE_RINGBUF' val=27
>> > >> > 'BPF_MAP_TYPE_INODE_STORAGE' val=28
>> > >> > 'BPF_MAP_TYPE_TASK_STORAGE' val=29
>> > >> > 'BPF_MAP_TYPE_BLOOM_FILTER' val=30
>> > >> > [23] UNION '(anon)' size=8 vlen=1
>> > >> > 'flow_keys' type_id=29 bits_offset=0
>> > >> > [24] STRUCT 'bpf_flow_keys' size=56 vlen=13
>> > >> > 'nhoff' type_id=8 bits_offset=0
>> > >> > 'thoff' type_id=8 bits_offset=16
>> > >> > 'addr_proto' type_id=8 bits_offset=32
>> > >> > 'is_frag' type_id=3 bits_offset=48
>> > >> > 'is_first_frag' type_id=3 bits_offset=56
>> > >> > 'is_encap' type_id=3 bits_offset=64
>> > >> > 'ip_proto' type_id=3 bits_offset=72
>> > >> > 'n_proto' type_id=20 bits_offset=80
>> > >> > 'sport' type_id=20 bits_offset=96
>> > >> > 'dport' type_id=20 bits_offset=112
>> > >> > '(anon)' type_id=25 bits_offset=128
>> > >> > 'flags' type_id=12 bits_offset=384
>> > >> > 'flow_label' type_id=21 bits_offset=416
>> > >> > [25] UNION '(anon)' size=32 vlen=2
>> > >> > '(anon)' type_id=26 bits_offset=0
>> > >> > '(anon)' type_id=27 bits_offset=0
>> > >> > [26] STRUCT '(anon)' size=8 vlen=2
>> > >> > 'ipv4_src' type_id=21 bits_offset=0
>> > >> > 'ipv4_dst' type_id=21 bits_offset=32
>> > >> > [27] STRUCT '(anon)' size=32 vlen=2
>> > >> > 'ipv6_src' type_id=28 bits_offset=0
>> > >> > 'ipv6_dst' type_id=28 bits_offset=128
>> > >> > [28] ARRAY '(anon)' type_id=12 index_type_id=16 nr_elems=4
>> > >> > [29] PTR '(anon)' type_id=24
>> > >> > [30] UNION '(anon)' size=8 vlen=1
>> > >> > 'sk' type_id=32 bits_offset=0
>> > >> > [31] STRUCT 'bpf_sock' size=80 vlen=14
>> > >> > 'bound_dev_if' type_id=12 bits_offset=0
>> > >> > 'family' type_id=12 bits_offset=32
>> > >> > 'type' type_id=12 bits_offset=64
>> > >> > 'protocol' type_id=12 bits_offset=96
>> > >> > 'mark' type_id=12 bits_offset=128
>> > >> > 'priority' type_id=12 bits_offset=160
>> > >> > 'src_ip4' type_id=12 bits_offset=192
>> > >> > 'src_ip6' type_id=28 bits_offset=224
>> > >> > 'src_port' type_id=12 bits_offset=352
>> > >> > 'dst_port' type_id=20 bits_offset=384
>> > >> > 'dst_ip4' type_id=12 bits_offset=416
>> > >> > 'dst_ip6' type_id=28 bits_offset=448
>> > >> > 'state' type_id=12 bits_offset=576
>> > >> > 'rx_queue_mapping' type_id=10 bits_offset=608
>> > >> > [32] PTR '(anon)' type_id=31
>> > >> > [33] STRUCT '__sk_buff' size=192 vlen=33
>> > >> > 'len' type_id=12 bits_offset=0
>> > >> > 'pkt_type' type_id=12 bits_offset=32
>> > >> > 'mark' type_id=12 bits_offset=64
>> > >> > 'queue_mapping' type_id=12 bits_offset=96
>> > >> > 'protocol' type_id=12 bits_offset=128
>> > >> > 'vlan_present' type_id=12 bits_offset=160
>> > >> > 'vlan_tci' type_id=12 bits_offset=192
>> > >> > 'vlan_proto' type_id=12 bits_offset=224
>> > >> > 'priority' type_id=12 bits_offset=256
>> > >> > 'ingress_ifindex' type_id=12 bits_offset=288
>> > >> > 'ifindex' type_id=12 bits_offset=320
>> > >> > 'tc_index' type_id=12 bits_offset=352
>> > >> > 'cb' type_id=34 bits_offset=384
>> > >> > 'hash' type_id=12 bits_offset=544
>> > >> > 'tc_classid' type_id=12 bits_offset=576
>> > >> > 'data' type_id=12 bits_offset=608
>> > >> > 'data_end' type_id=12 bits_offset=640
>> > >> > 'napi_id' type_id=12 bits_offset=672
>> > >> > 'family' type_id=12 bits_offset=704
>> > >> > 'remote_ip4' type_id=12 bits_offset=736
>> > >> > 'local_ip4' type_id=12 bits_offset=768
>> > >> > 'remote_ip6' type_id=28 bits_offset=800
>> > >> > 'local_ip6' type_id=28 bits_offset=928
>> > >> > 'remote_port' type_id=12 bits_offset=1056
>> > >> > 'local_port' type_id=12 bits_offset=1088
>> > >> > 'data_meta' type_id=12 bits_offset=1120
>> > >> > '(anon)' type_id=23 bits_offset=1152
>> > >> > 'tstamp' type_id=15 bits_offset=1216
>> > >> > 'wire_len' type_id=12 bits_offset=1280
>> > >> > 'gso_segs' type_id=12 bits_offset=1312
>> > >> > '(anon)' type_id=30 bits_offset=1344
>> > >> > 'gso_size' type_id=12 bits_offset=1408
>> > >> > 'hwtstamp' type_id=15 bits_offset=1472
>> > >> > [34] ARRAY '(anon)' type_id=12 index_type_id=16 nr_elems=5
>> > >> > [35] CONST '(anon)' type_id=33
>> > >> > [36] PTR '(anon)' type_id=0
>> > >> > [37] STRUCT '(anon)' size=24 vlen=3
>> > >> > 'type' type_id=39 bits_offset=0
>> > >> > 'key' type_id=40 bits_offset=64
>> > >> > 'value' type_id=41 bits_offset=128
>> > >> > [38] ARRAY '(anon)' type_id=9 index_type_id=16 nr_elems=1
>> > >> > [39] PTR '(anon)' type_id=38
>> > >> > [40] PTR '(anon)' type_id=12
>> > >> > [41] PTR '(anon)' type_id=3
>> > >> > [42] ARRAY '(anon)' type_id=19 index_type_id=16 nr_elems=18
>> > >> > [43] CONST '(anon)' type_id=42
>> > >> > [44] FUNC_PROTO '(anon)' ret_type_id=36 vlen=2
>> > >> > '(anon)' type_id=36
>> > >> > '(anon)' type_id=46
>> > >> > [45] CONST '(anon)' type_id=0
>> > >> > [46] PTR '(anon)' type_id=45
>> > >> > [47] FUNC_PROTO '(anon)' ret_type_id=9 vlen=1
>> > >> > 'sk' type_id=48
>> > >> > [48] PTR '(anon)' type_id=35
>> > >> > [49] FUNC_PROTO '(anon)' ret_type_id=9 vlen=1
>> > >> > 'sk' type_id=48
>> > >> > [50] FUNC_PROTO '(anon)' ret_type_id=9 vlen=1
>> > >> > 'sk' type_id=48
>> > >> > [51] VAR '_license' type_id=43, linkage=static
>> > >> > [52] VAR 'is_allow_list' type_id=5, linkage=global
>> > >> > [53] VAR 'sd_restrictif' type_id=37, linkage=global
>> > >> > [54] FUNC 'bpf_map_lookup_elem' type_id=44 linkage=global
>> > >> > [55] FUNC 'sd_restrictif_i' type_id=47 linkage=global
>> > >> > [56] FUNC 'sd_restrictif_e' type_id=49 linkage=global
>> > >> > [57] FUNC 'restrict_network_interfaces_impl' type_id=50 linkage=static
>> > >> > [58] DATASEC 'license' size=0 vlen=1
>> > >> > type_id=51 offset=0 size=18 (VAR '_license')
>> > >> > [59] DATASEC '.maps' size=0 vlen=1
>> > >> > type_id=53 offset=0 size=24 (VAR 'sd_restrictif')
>> > >> > [60] DATASEC '.data' size=0 vlen=1
>> > >> > type_id=52 offset=0 size=1 (VAR 'is_allow_list')
>> > >> >
>> > >> >>
>> > >> >> >> GCC is wrong, clearly. This function is global ([0]) and libbpf
>> > >> >> >> expects it to be marked as such in BTF.
>> > >> >> >>
>> > >> >> >>
>> > >> >
> https://github.com/systemd/systemd/blob/main/src/core/bpf/restrict_ifaces/restrict-ifaces.bpf.c#L42-L50
>> > >> >> >>
>> > >> >> >>
>> > >> >> >>> GCC:
>> > >> >> >>>
>> > >> >> >>> [1] INT 'signed char' size=1 bits_offset=0 nr_bits=8 encoding=UNKN
>> > >> >> >>> [2] INT 'unsigned char' size=1 bits_offset=0 nr_bits=8 encoding=CHAR
>> > >> >> >>> [3] TYPEDEF '__u8' type_id=2
>> > >> >> >>> [4] CONST '(anon)' type_id=3
>> > >> >> >>
>> > >> >> >> [...]
next prev parent reply other threads:[~2022-07-09 22:41 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-07-06 17:11 bpftool gen object doesn't handle GCC built BPF ELF files James Hilliard
2022-07-06 17:20 ` Andrii Nakryiko
2022-07-06 17:49 ` James Hilliard
2022-07-08 11:55 ` Jose E. Marchesi
2022-07-08 12:23 ` James Hilliard
2022-07-08 13:46 ` Jose E. Marchesi
2022-07-08 13:53 ` James Hilliard
2022-07-08 14:39 ` Jose E. Marchesi
2022-07-08 14:51 ` Jose E. Marchesi
2022-07-08 18:33 ` Jose E. Marchesi
2022-07-08 20:59 ` James Hilliard
2022-07-09 17:24 ` Jose E. Marchesi
2022-07-09 17:28 ` James Hilliard
2022-07-09 20:20 ` Jose E. Marchesi
2022-07-09 20:32 ` James Hilliard
2022-07-09 20:39 ` James Hilliard
2022-07-09 22:41 ` Jose E. Marchesi [this message]
2022-07-09 22:54 ` James Hilliard
2022-07-10 9:38 ` Jose E. Marchesi
2022-07-10 12:16 ` James Hilliard
2022-07-10 13:06 ` Jose E. Marchesi
2022-07-10 19:49 ` Jose E. Marchesi
2022-07-10 20:22 ` Jose E. Marchesi
2022-07-11 1:55 ` Jose E. Marchesi
[not found] ` <CADvTj4o-36iuru665BW0XnEauXBeszW438QTtpt4_VUEjf5nXg@mail.gmail.com>
2022-08-01 22:51 ` Andrii Nakryiko
2022-08-01 23:35 ` James Hilliard
2022-08-02 21:28 ` Andrii Nakryiko
2022-08-02 22:05 ` James Hilliard
2022-08-03 0:29 ` Andrii Nakryiko
2022-08-03 1:49 ` James Hilliard
2022-08-03 16:57 ` Andrii Nakryiko
2022-08-03 19:24 ` James Hilliard
2022-09-30 21:52 ` James Hilliard
2022-09-30 22:50 ` Andrii Nakryiko
2022-08-04 17:29 ` Alexei Starovoitov
2022-08-02 15:57 ` Jose E. Marchesi
2022-07-12 0:34 ` Indu Bhagat
2022-07-13 4:28 ` Yonghong Song
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87tu7p3o4k.fsf@oracle.com \
--to=jose.marchesi@oracle.com \
--cc=andrii.nakryiko@gmail.com \
--cc=bpf@vger.kernel.org \
--cc=david.faust@oracle.com \
--cc=james.hilliard1@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox