linux-stable.git/tools/lib/bpf/libbpf.c, branch v5.0

bpf: libbpf: fix memleak by freeing line_info

2018-12-18T00:16:23+00:00

This patch fixes a memory leak in libbpf by freeing up line_info
member of struct bpf_program while unloading a program.

Fixes: 3d65014146c6 ("bpf: libbpf: Add btf_line_info support to libbpf")
Signed-off-by: Prashant Bhole 
Acked-by: Martin KaFai Lau 
Signed-off-by: Daniel Borkmann

bpf: libbpf: Add btf_line_info support to libbpf

2018-12-09T21:54:38+00:00

This patch adds bpf_line_info support to libbpf:
1) Parsing the line_info sec from ".BTF.ext"
2) Relocating the line_info.  If the main prog *_info relocation
   fails, it will ignore the remaining subprog line_info and continue.
   If the subprog *_info relocation fails, it will bail out.
3) BPF_PROG_LOAD a prog with line_info

Signed-off-by: Martin KaFai Lau 
Acked-by: Yonghong Song 
Signed-off-by: Alexei Starovoitov

bpf: libbpf: Refactor and bug fix on the bpf_func_info loading logic

2018-12-09T21:54:38+00:00

This patch refactor and fix a bug in the libbpf's bpf_func_info loading
logic.  The bug fix and refactoring are targeting the same
commit 2993e0515bb4 ("tools/bpf: add support to read .BTF.ext sections")
which is in the bpf-next branch.

1) In bpf_load_program_xattr(), it should retry when errno == E2BIG
   regardless of log_buf and log_buf_sz.  This patch fixes it.

2) btf_ext__reloc_init() and btf_ext__reloc() are essentially
   the same except btf_ext__reloc_init() always has insns_cnt == 0.
   Hence, btf_ext__reloc_init() is removed.

   btf_ext__reloc() is also renamed to btf_ext__reloc_func_info()
   to get ready for the line_info support in the next patch.

3) Consolidate func_info section logic from "btf_ext_parse_hdr()",
   "btf_ext_validate_func_info()" and "btf_ext__new()" to
   a new function "btf_ext_copy_func_info()" such that similar
   logic can be reused by the later libbpf's line_info patch.

4) The next line_info patch will store line_info_cnt instead of
   line_info_len in the bpf_program because the kernel is taking
   line_info_cnt also.  It will save a few "len" to "cnt" conversions
   and will also save some function args.

   Hence, this patch also makes bpf_program to store func_info_cnt
   instead of func_info_len.

5) btf_ext depends on btf.  e.g. the func_info's type_id
   in ".BTF.ext" is not useful when ".BTF" is absent.
   This patch only init the obj->btf_ext pointer after
   it has successfully init the obj->btf pointer.

   This can avoid always checking "obj->btf && obj->btf_ext"
   together for accessing ".BTF.ext".  Checking "obj->btf_ext"
   alone will do.

6) Move "struct btf_sec_func_info" from btf.h to btf.c.
   There is no external usage outside btf.c.

Fixes: 2993e0515bb4 ("tools/bpf: add support to read .BTF.ext sections")
Signed-off-by: Martin KaFai Lau 
Acked-by: Yonghong Song 
Signed-off-by: Alexei Starovoitov

tools/bpf: make libbpf _GNU_SOURCE friendly

2018-11-30T01:41:02+00:00

During porting libbpf to bcc, I got some warnings like below:
  ...
  [  2%] Building C object src/cc/CMakeFiles/bpf-shared.dir/libbpf/src/libbpf.c.o
  /home/yhs/work/bcc2/src/cc/libbpf/src/libbpf.c:12:0:
  warning: "_GNU_SOURCE" redefined [enabled by default]
   #define _GNU_SOURCE
  ...
  [  3%] Building C object src/cc/CMakeFiles/bpf-shared.dir/libbpf/src/libbpf_errno.c.o
  /home/yhs/work/bcc2/src/cc/libbpf/src/libbpf_errno.c: In function ‘libbpf_strerror’:
  /home/yhs/work/bcc2/src/cc/libbpf/src/libbpf_errno.c:45:7:
  warning: assignment makes integer from pointer without a cast [enabled by default]
     ret = strerror_r(err, buf, size);
  ...

bcc is built with _GNU_SOURCE defined and this caused the above warning.
This patch intends to make libpf _GNU_SOURCE friendly by
  . define _GNU_SOURCE in libbpf.c unless it is not defined
  . undefine _GNU_SOURCE as non-gnu version of strerror_r is expected.

Signed-off-by: Yonghong Song 
Acked-by: Jakub Kicinski 
Signed-off-by: Daniel Borkmann

libbpf: make bpf_object__open default to UNSPEC

2018-11-23T21:27:05+00:00

currently by default libbpf's bpf_object__open requires
bpf's program to specify  version in a code because of two things:
1) default prog type is set to KPROBE
2) KPROBE requires (in kernel/bpf/syscall.c) version to be specified

in this patch i'm changing default prog type to UNSPEC and also changing
requirments for version's section to be present in object file.
now it would reflect what we have today in kernel
(only KPROBE prog type requires for version to be explicitly set).

v1 -> v2:
 - RFC tag has been dropped

Signed-off-by: Nikita V. Shirokov 
Signed-off-by: Daniel Borkmann

bpf: adding support for map in map in libbpf

2018-11-21T22:33:21+00:00

idea is pretty simple. for specified map (pointed by struct bpf_map)
we would provide descriptor of already loaded map, which is going to be
used as a prototype for inner map. proposed workflow:
1) open bpf's object (bpf_object__open)
2) create bpf's map which is going to be used as a prototype
3) find (by name) map-in-map which you want to load and update w/
descriptor of inner map w/ a new helper from this patch
4) load bpf program w/ bpf_object__load

Signed-off-by: Nikita V. Shirokov 
Acked-by: Yonghong Song 
Signed-off-by: Daniel Borkmann

bpf: libbpf: don't specify prog name if kernel doesn't support it

2018-11-21T22:26:14+00:00

Use recently added capability check.

See commit 23499442c319 ("bpf: libbpf: retry map creation without
the name") for rationale.

Signed-off-by: Stanislav Fomichev 
Signed-off-by: Daniel Borkmann

bpf: libbpf: remove map name retry from bpf_create_map_xattr

2018-11-21T22:26:04+00:00

Instead, check for a newly created caps.name bpf_object capability.
If kernel doesn't support names, don't specify the attribute.

See commit 23499442c319 ("bpf: libbpf: retry map creation without
the name") for rationale.

Signed-off-by: Stanislav Fomichev 
Signed-off-by: Daniel Borkmann

bpf, libbpf: introduce bpf_object__probe_caps to test BPF capabilities

2018-11-21T22:25:33+00:00

It currently only checks whether kernel supports map/prog names.
This capability check will be used in the next two commits to
skip setting prog/map names.

Suggested-by: Daniel Borkmann 
Signed-off-by: Stanislav Fomichev 
Signed-off-by: Daniel Borkmann

bpf: fix a libbpf loader issue

2018-11-21T21:22:17+00:00

Commit 2993e0515bb4 ("tools/bpf: add support to read .BTF.ext sections")
added support to read .BTF.ext sections from an object file, create
and pass prog_btf_fd and func_info to the kernel.

The program btf_fd (prog->btf_fd) is initialized to be -1 to please
zclose so we do not need special handling dur prog close.
Passing -1 to the kernel, however, will cause loading error.
Passing btf_fd 0 to the kernel if prog->btf_fd is invalid
fixed the problem.

Fixes: 2993e0515bb4 ("tools/bpf: add support to read .BTF.ext sections")
Reported-by: Andrey Ignatov 
Reported-by: Emre Cantimur 
Tested-by: Andrey Ignatov 
Signed-off-by: Yonghong Song 
Acked-by: Martin KaFai Lau 
Signed-off-by: Daniel Borkmann