In arraymap.c:

In bpf_array_map_seq_start() and bpf_array_map_seq_next()
cast return values from the __percpu address space to
the generic address space via uintptr_t [1].

Correct the declaration of pptr pointer in __bpf_array_map_seq_show()
to void __percpu * and cast the value from the generic address
space to the __percpu address space via uintptr_t [1].

In hashtab.c:

Assign the return value from bpf_mem_cache_alloc() to void pointer
and cast the value to void __percpu ** (void pointer to percpu void
pointer) before dereferencing.

In memalloc.c:

Explicitly declare __percpu variables.

Cast obj to void __percpu **.

In helpers.c:

Cast ptr in BPF_CALL_1 and BPF_CALL_2 from generic address space
to __percpu address space via const uintptr_t [1].

Found by GCC's named address space checks.

There were no changes in the resulting object files.

[1] https://sparse.docs.kernel.org/en/latest/annotations.html#address-space-nameSigned-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Martin KaFai Lau <martin.lau@linux.dev>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Song Liu <song@kernel.org>
Cc: Yonghong Song <yonghong.song@linux.dev>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: KP Singh <kpsingh@kernel.org>
Cc: Stanislav Fomichev <sdf@fomichev.me>
Cc: Hao Luo <haoluo@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240811161414.56744-1-ubizjak@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Eduard Zingerman says:

====================
bpf: fix null pointer access for malformed BPF_CORE_TYPE_ID_LOCAL relos

Liu RuiTong reported an in-kernel null pointer derefence when
processing BPF_CORE_TYPE_ID_LOCAL relocations referencing non-existing
BTF types. Fix this by adding proper id checks.

Changes v2->v3:
- selftest update suggested by Andrii:
  avoid memset(0) for log buffer and do memset(0) for bpf_attr.

Changes v1->v2:
- moved check from bpf_core_calc_relo_insn() to bpf_core_apply()
  now both in kernel and in libbpf relocation type id is guaranteed
  to exist when bpf_core_calc_relo_insn() is called;
- added a test case.

v1: https://lore.kernel.org/bpf/20240821164620.1056362-1-eddyz87@gmail.com/
v2: https://lore.kernel.org/bpf/20240822001837.2715909-1-eddyz87@gmail.com/
====================

Link: https://lore.kernel.org/r/20240822080124.2995724-1-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Check that verifier rejects BPF program containing relocation
pointing to non-existent BTF type.

To force relocation resolution on kernel side test case uses
bpf_attr->core_relos field. This field is not exposed by libbpf,
so directly do BPF system call in the test.
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240822080124.2995724-3-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

In case of malformed relocation record of kind BPF_CORE_TYPE_ID_LOCAL
referencing a non-existing BTF type, function bpf_core_calc_relo_insn
would cause a null pointer deference.

Fix this by adding a proper check upper in call stack, as malformed
relocation records could be passed from user space.

Simplest reproducer is a program:

    r0 = 0
    exit

With a single relocation record:

    .insn_off = 0,          /* patch first instruction */
    .type_id = 100500,      /* this type id does not exist */
    .access_str_off = 6,    /* offset of string "0" */
    .kind = BPF_CORE_TYPE_ID_LOCAL,

See the link for original reproducer or next commit for a test case.

Fixes: 74753e14 ("libbpf: Replace btf__type_by_id() with btf_type_by_id().")
Reported-by: Liu RuiTong <cnitlrt@gmail.com>
Closes: https://lore.kernel.org/bpf/CAK55_s6do7C+DVwbwY_7nKfUz0YLDoiA1v6X3Y9+p0sWzipFSA@mail.gmail.com/Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240822080124.2995724-2-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Let the kmemdup_array() take care about multiplication and possible
overflows.
Signed-off-by: Yu Jiaoliang <yujiaoliang@vivo.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20240821073709.4067177-1-yujiaoliang@vivo.com

Eduard Zingerman says:

====================
__jited test tag to check disassembly after jit

Some of the logic in the BPF jits might be non-trivial.
It might be useful to allow testing this logic by comparing
generated native code with expected code template.
This patch set adds a macro __jited() that could be used for
test_loader based tests in a following manner:

    SEC("tp")
    __arch_x86_64
    __jited("   endbr64")
    __jited("   nopl    (%rax,%rax)")
    __jited("   xorq    %rax, %rax")
    ...
    __naked void some_test(void) { ... }

Also add a test for jit code generated for tail calls handling to
demonstrate the feature.

The feature uses LLVM libraries to do the disassembly.
At selftests compilation time Makefile detects if these libraries are
available. When libraries are not available tests using __jit_x86()
are skipped.
Current CI environment does not include llvm development libraries,
but changes to add these are trivial.

This was previously discussed here:
https://lore.kernel.org/bpf/20240718205158.3651529-1-yonghong.song@linux.dev/

Patch-set includes a few auxiliary steps:
- patches #2 and #3 fix a few bugs in test_loader behaviour;
- patch #4 replaces __regex macro with ability to specify regular
  expressions in __msg and __xlated using "{{" "}}" escapes;
- patch #8 updates __xlated to match disassembly lines consequently,
  same way as __jited does.

Changes v2->v3:
- changed macro name from __jit_x86 to __jited with __arch_* to
  specify disassembly arch (Yonghong);
- __jited matches disassembly lines consequently with "..."
  allowing to skip some number of lines (Andrii);
- __xlated matches disassembly lines consequently, same as __jited;
- "{{...}}" regex brackets instead of __regex macro;
- bug fixes for old commits.

Changes v1->v2:
- stylistic changes suggested by Yonghong;
- fix for -Wformat-truncation related warning when compiled with
  llvm15 (Yonghong).

v1: https://lore.kernel.org/bpf/20240809010518.1137758-1-eddyz87@gmail.com/
v2: https://lore.kernel.org/bpf/20240815205449.242556-1-eddyz87@gmail.com/
====================

Link: https://lore.kernel.org/r/20240820102357.3372779-1-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Both __xlated and __jited work with disassembly.
It is logical to have both work in a similar manner.

This commit updates __xlated macro handling in test_loader.c by making
it expect matches on sequential lines, same way as __jited operates.
For example:

    __xlated("1: *(u64 *)(r10 -16) = r1")      ;; matched on line N
    __xlated("3: r0 = &(void __percpu *)(r0)") ;; matched on line N+1

Also:

    __xlated("1: *(u64 *)(r10 -16) = r1")      ;; matched on line N
    __xlated("...")                            ;; not matched
    __xlated("3: r0 = &(void __percpu *)(r0)") ;; mantched on any
                                               ;; line >= N
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-10-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

A program calling sub-program which does a tail call.
The idea is to verify instructions generated by jit for tail calls:
- in program and sub-program prologues;
- for subprogram call instruction;
- for tail call itself.
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-9-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Allow to verify jit behaviour by writing tests as below:

    SEC("tp")
    __arch_x86_64
    __jited("   endbr64")
    __jited("   nopl    (%rax,%rax)")
    __jited("   xorq    %rax, %rax")
    ...
    __naked void some_test(void)
    {
        asm volatile (... ::: __clobber_all);
    }

Allow regular expressions in patterns, same way as in __msg.
By default assume that each __jited pattern has to be matched on the
next consecutive line of the disassembly, e.g.:

    __jited("   endbr64")             # matched on line N
    __jited("   nopl    (%rax,%rax)") # matched on line N+1

If match occurs on a wrong line an error is reported.
To override this behaviour use __jited("..."), e.g.:

    __jited("   endbr64")             # matched on line N
    __jited("...")                    # not matched
    __jited("   nopl    (%rax,%rax)") # matched on any line >= N
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-7-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

This commit adds a utility function to get disassembled text for jited
representation of a BPF program designated by file descriptor.
Function prototype looks as follows:

    int get_jited_program_text(int fd, char *text, size_t text_sz)

Where 'fd' is a file descriptor for the program, 'text' and 'text_sz'
refer to a destination buffer for disassembled text.
Output format looks as follows:

    18:	77 06                               	ja	L0
    1a:	50                                  	pushq	%rax
    1b:	48 89 e0                            	movq	%rsp, %rax
    1e:	eb 01                               	jmp	L1
    20:	50                                  L0:	pushq	%rax
    21:	50                                  L1:	pushq	%rax
     ^  ^^^^^^^^                             ^  ^^^^^^^^^^^^^^^^^^
     |  binary insn                          |  textual insn
     |  representation                       |  representation
     |                                       |
    instruction offset              inferred local label name

The code and makefile changes are inspired by jit_disasm.c from bpftool.
Use llvm libraries to disassemble BPF program instead of libbfd to avoid
issues with disassembly output stability pointed out in [1].

Selftests makefile uses Makefile.feature to detect if LLVM libraries
are available. If that is not the case selftests build proceeds but
the function returns -EOPNOTSUPP at runtime.

[1] commit eb9d1acf ("bpftool: Add LLVM as default library for disassembling JIT-ed programs")
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-6-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Upcoming changes require a notation to specify regular expression
matches for regular verifier log messages, disassembly of BPF
instructions, disassembly of jited instructions.

Neither basic nor extended POSIX regular expressions w/o additional
escaping are good for this role because of wide use of special
characters in disassembly, for example:

    movq -0x10(%rbp), %rax  ;; () are special characters
    cmpq $0x21, %rax        ;; $ is a special character

    *(u64 *)(r10 -16) = r1  ;; * and () are special characters

This commit borrows syntax from LLVM's FileCheck utility.
It replaces __regex macro with ability to embed regular expressions
in __msg patters using "{{" "}}" pairs for escaping.
Syntax for __msg patterns:

    pattern := (<verbatim text> | regex)*
    regex := "{{" <posix extended regular expression> "}}"

For example, pattern "foo{{[0-9]+}}" matches strings like
"foo0", "foo007", etc.
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-5-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

__msg, __regex and __xlated tags are based on
__attribute__((btf_decl_tag("..."))) annotations.

Clang de-duplicates such annotations, e.g. the following
two sequences of tags are identical in final BTF:

    /* seq A */            /* seq B */
    __tag("foo")           __tag("foo")
    __tag("bar")           __tag("bar")
    __tag("foo")

Fix this by adding a unique suffix for each tag using __COUNTER__
pre-processor macro. E.g. here is a new definition for __msg:

    #define __msg(msg) \
      __attribute__((btf_decl_tag("comment:test_expect_msg=" XSTR(__COUNTER__) "=" msg)))

Using this definition the "seq A" from example above is translated to
BTF as follows:

    [..] DECL_TAG 'comment:test_expect_msg=0=foo' type_id=X component_idx=-1
    [..] DECL_TAG 'comment:test_expect_msg=1=bar' type_id=X component_idx=-1
    [..] DECL_TAG 'comment:test_expect_msg=2=foo' type_id=X component_idx=-1

Surprisingly, this bug affects a single existing test:
verifier_spill_fill/old_stack_misc_vs_cur_ctx_ptr,
where sequence of identical messages was expected in the log.

Fixes: 537c3f66 ("selftests/bpf: add generic BPF program tester-loader")
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-4-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Suppose log="foo bar buz" and msg->substr="bar".
In such case current match processing logic would update 'log' as
follows: log += strlen(msg->substr); -> log += 3 -> log=" bar".
However, the intent behind the 'log' update is to make it point after
the successful match, e.g. to make log=" buz" in the example above.

Fixes: 4ef5d6af ("selftests/bpf: no need to track next_match_pos in struct test_loader")
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-3-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

When running test_loader based tests in the verbose mode each matched
message leaves a trace in the stderr, e.g.:

    ./test_progs -vvv -t ...
    validate_msgs:PASS:expect_msg 0 nsec
    validate_msgs:PASS:expect_msg 0 nsec
    validate_msgs:PASS:expect_msg 0 nsec
    validate_msgs:PASS:expect_msg 0 nsec
    validate_msgs:PASS:expect_msg 0 nsec

This is not very helpful when debugging such tests and clobbers the
log a lot.
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240820102357.3372779-2-eddyz87@gmail.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Andrii Nakryiko says:

====================
Support passing BPF iterator to kfuncs

Add support for passing BPF iterator state to any kfunc. Such kfunc has to
declare such argument with valid `struct bpf_iter_<type> *` type and should
use "__iter" suffix in argument name, following the established suffix-based
convention. We add a simple test/demo iterator getter in bpf_testmod.
====================

Link: https://lore.kernel.org/r/20240808232230.2848712-1-andrii@kernel.orgSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Define BPF iterator "getter" kfunc, which accepts iterator pointer as
one of the arguments. Make sure that argument passed doesn't have to be
the very first argument (unlike new-next-destroy combo).
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240808232230.2848712-4-andrii@kernel.orgSigned-off-by: Alexei Starovoitov <ast@kernel.org>

There are potentially useful cases where a specific iterator type might
need to be passed into some kfunc. So, in addition to existing
bpf_iter_<type>_{new,next,destroy}() kfuncs, allow to pass iterator
pointer to any kfunc.

We employ "__iter" naming suffix for arguments that are meant to accept
iterators. We also enforce that they accept PTR -> STRUCT btf_iter_<type>
type chain and point to a valid initialized on-the-stack iterator state.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240808232230.2848712-3-andrii@kernel.orgSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Verifier enforces that all iterator structs are named `bpf_iter_<name>`
and that whenever iterator is passed to a kfunc it's passed as a valid PTR ->
STRUCT chain (with potentially const modifiers in between).

We'll need this check for upcoming changes, so instead of duplicating
the logic, extract it into a helper function.
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240808232230.2848712-2-andrii@kernel.orgSigned-off-by: Alexei Starovoitov <ast@kernel.org>

Cupertino Miranda says:

====================
Correct recent GCC incompatible changes.

Hi everyone,

Apologies for the previous patches which did not include a cover letter.
My wish was to send 3 indepepdent patches but after the initial mistake lets keep
this as a series although they are all independent from themselves.

The changes in this patch series is related to recovering GCC support to
build the selftests.
A few tests and a makefile change have broken the support for GCC in the
last few months.

Looking forward to your comments.

Best regards,
Cupertino
====================

Link: https://lore.kernel.org/r/20240819151129.1366484-1-cupertino.miranda@oracle.comSigned-off-by: Andrii Nakryiko <andrii@kernel.org>

GCC errors when compiling tailcall_bpf2bpf_hierarchy2.c and
tailcall_bpf2bpf_hierarchy3.c with the following error:

progs/tailcall_bpf2bpf_hierarchy2.c: In function 'tailcall_bpf2bpf_hierarchy_2':
progs/tailcall_bpf2bpf_hierarchy2.c:66:9: error: input operand constraint contains '+'
   66 |         asm volatile (""::"r+"(ret));
      |         ^~~

Changed implementation to make use of __sink macro that abstracts the
desired behaviour.

The proposed change seems valid for both GCC and CLANG.
Signed-off-by: Cupertino Miranda <cupertino.miranda@oracle.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20240819151129.1366484-4-cupertino.miranda@oracle.com

verfifier_nocsr.c fails to compile in GCC. The reason behind it was
initially explained in commit 27a90b14.

"A few BPF selftests perform type punning and they may break strict
aliasing rules, which are exploited by both GCC and clang by default
while optimizing.  This can lead to broken compiled programs."
Signed-off-by: Cupertino Miranda <cupertino.miranda@oracle.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20240819151129.1366484-2-cupertino.miranda@oracle.com

Matteo Croce says:

====================
bpf: enable some functions in cgroup programs

From: Matteo Croce <teknoraver@meta.com>

Enable some BPF kfuncs and the helper bpf_current_task_under_cgroup()
for program types BPF_CGROUP_*.
These will be used by systemd-networkd:
https://github.com/systemd/systemd/pull/32212

v5->v6:
Called register_btf_kfunc_id_set() only once
Fixed build error with !CONFIG_CGROUPS

v4->v5:
Same code, but v4 had an old cover letter

v3->v4:
Reset all the acked-by tags because the code changed a bit.
Signed-off-by: Matteo Croce <teknoraver@meta.com>
====================

Link: https://lore.kernel.org/r/20240819162805.78235-1-technoboy85@gmail.comSigned-off-by: Andrii Nakryiko <andrii@kernel.org>

The helper bpf_current_task_under_cgroup() currently is only allowed for
tracing programs, allow its usage also in the BPF_CGROUP_* program types.

Move the code from kernel/trace/bpf_trace.c to kernel/bpf/helpers.c,
so it compiles also without CONFIG_BPF_EVENTS.

This will be used in systemd-networkd to monitor the sysctl writes,
and filter it's own writes from others:
https://github.com/systemd/systemd/pull/32212Signed-off-by: Matteo Croce <teknoraver@meta.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20240819162805.78235-3-technoboy85@gmail.com

These kfuncs are enabled even in BPF_PROG_TYPE_TRACING, so they
should be safe also in BPF_CGROUP_* programs.
Since all BPF_CGROUP_* programs share the same hook,
call register_btf_kfunc_id_set() only once.

In enum btf_kfunc_hook, rename BTF_KFUNC_HOOK_CGROUP_SKB to a more
generic BTF_KFUNC_HOOK_CGROUP, since it's used for all the cgroup
related program types.
Signed-off-by: Matteo Croce <teknoraver@meta.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20240819162805.78235-2-technoboy85@gmail.com

Currently, compiling the bpf programs will result the compilation errors
with the cf-protection option as follows in arm64 and loongarch64 machine
when using gcc 12.3.1 and clang 17.0.6. This commit fixes the compilation
errors by limited the cf-protection option only used in x86 platform.

[root@localhost linux]# make M=samples/bpf
	......
  CLANG-bpf  samples/bpf/xdp2skb_meta_kern.o
error: option 'cf-protection=return' cannot be specified on this target
error: option 'cf-protection=branch' cannot be specified on this target
2 errors generated.
  CLANG-bpf  samples/bpf/syscall_tp_kern.o
error: option 'cf-protection=return' cannot be specified on this target
error: option 'cf-protection=branch' cannot be specified on this target
2 errors generated.
	......

Fixes: 34f6e38f ("samples/bpf: fix warning with ignored-attributes")
Reported-by: Jiangshan Yi <yijiangshan@kylinos.cn>
Signed-off-by: Jiangshan Yi <yijiangshan@kylinos.cn>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Tested-by: Qiang Wang <wangqiang1@kylinos.cn>
Link: https://lore.kernel.org/bpf/20240815135524.140675-1-13667453960@163.com

In `elf_close`, we get this with GCC 15 -O3 (at least):
```
In function ‘elf_close’,
    inlined from ‘elf_close’ at elf.c:53:6,
    inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2:
elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized]
   57 |         elf_end(elf_fd->elf);
      |         ^~~~~~~~~~~~~~~~~~~~
elf.c: In function ‘elf_find_func_offset_from_file’:
elf.c:377:23: note: ‘elf_fd.elf’ was declared here
  377 |         struct elf_fd elf_fd;
      |                       ^~~~~~
In function ‘elf_close’,
    inlined from ‘elf_close’ at elf.c:53:6,
    inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2:
elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized]
   58 |         close(elf_fd->fd);
      |         ^~~~~~~~~~~~~~~~~
elf.c: In function ‘elf_find_func_offset_from_file’:
elf.c:377:23: note: ‘elf_fd.fd’ was declared here
  377 |         struct elf_fd elf_fd;
      |                       ^~~~~~
```

In reality, our use is fine, it's just that GCC doesn't model errno
here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly
by initializing elf_fd.fd to -1 and elf_fd.elf to NULL.

I've done this in two other functions as well given it could easily
occur there too (same access/use pattern).
Signed-off-by: Sam James <sam@gentoo.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://gcc.gnu.org/PR114952
Link: https://lore.kernel.org/bpf/14ec488a1cac02794c2fa2b83ae0cef1bce2cb36.1723578546.git.sam@gentoo.org

__btf_name_valid() can be completely replaced with
btf_name_valid_identifier, and since most of the time you already call
btf_name_valid_identifier instead of __btf_name_valid , it would be
appropriate to rename the __btf_name_valid function to
btf_name_valid_identifier and remove __btf_name_valid.
Signed-off-by: Jeongjun Park <aha310510@gmail.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Link: https://lore.kernel.org/bpf/20240807143110.181497-1-aha310510@gmail.com

Kui-Feng Lee says:

====================
Capture packets in the background for flaky test cases related to
network features.

We have some flaky test cases that are difficult to debug without
knowing what the traffic looks like. Capturing packets, the CI log and
packet files may help developers to fix these flaky test cases.

This patch set monitors a few test cases. Recently, they have been
showing flaky behavior.

    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 68, SYN
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 60, SYN, ACK
    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 60, ACK
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 52, ACK
    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 52, FIN, ACK
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 52, RST, ACK
    Packet file: packets-2173-86-select_reuseport:sockhash_IPv4_TCP_LOOPBACK_test_detach_bpf-test.log
    #280/87 select_reuseport/sockhash IPv4/TCP LOOPBACK test_detach_bpf:OK

The above block is the log of a test case. It shows every packet of a
connection. The captured packets are stored in the file called
packets-2173-86-select_reuseport:sockhash_IPv4_TCP_LOOPBACK_test_detach_bpf-test.log.

We have a set of high-level helpers and a test_progs option to
simplify the process of enabling the traffic monitor. netns_new() and
netns_free() are helpers used to create and delete namespaces while
also enabling the traffic monitor for the namespace based on the
patterns provided by the "-m" option of test_progs. The value of the
"-m" option is a list of patterns used to enable the traffic monitor
for a group of tests or a file containing patterns. CI can utilize
this option to enable monitoring.

traffic_monitor_start() and traffic_monitor_stop() are low-level
functions to start monitoring explicitly. You can have more controls,
however high-level helpers are preferred.

The following block is an example that monitors the network traffic of
a test case in a network namespace.

    struct netns_obj *netns;

    ...
    netns = netns_new("test", true);
    if (!ASSERT_TRUE(netns, "netns_new"))
        goto err;

    ... test ...

    netns_free(netns);

netns_new() will create a network namespace named "test" and bring up
"lo" in the namespace. By passing "true" as the 2nd argument, it will
set the network namespace of the current process to
"test".netns_free() will destroy the namespace, and the process will
leave the "test" namespace if the struct netns_obj returned by
netns_new() is created with "true" as the 2nd argument. If the name of
the test matches the patterns given by the "-m" option, the traffic
monitor will be enabled for the "test" namespace as well.

The packet files are located in the directory "/tmp/tmon_pcap/". The
directory is intended to be compressed as a file so that developers
can download it from the CI.

This feature is enabled only if libpcap is available when building
selftests.
---

Changes from v7:

 - Remove ":" with "__" from the file names of traffic logs. ':' would
   cause an error of the upload-artifact action of github.

 - Move remove_netns() to avoid a forward declaration.

Changes from v6:

 - Remove unnecessary memcpy for addresses.

 - Make packet messages similar to what tcpdump prints.

 - Check return value of inet_ntop().

 - Remove duplicated errno in messages.

 - Print arphdr_type for not handled packets.

 - Set dev "lo" in make_netns().

 - Avoid stacking netns by moving traffic_monitor_start() to earlier
   position.

 - Remove the word "packet" from packet messages.

 - Replace pipe with eventfd (wake_fd) to synchronize background threads.

Changes from v5:

 - Remove "-m" completely if traffic monitor is not enabled.

Changes from v4:

 - Use pkg-config to detect libpcap, and enable traffic monitor if
   there is libpcap.

 - Move traffic monitor functions back to network_helper.c, and pass
   extra parameters to traffic_monitor_start().

 - Use flockfile() & funlockfile() to avoid log interleaving.

 - Show "In", "Out", "M" ... for captured packets.

 - Print a warning message if the user pass a "-m" when libpcap is not
   available.

 - Bring up dev lo in netns_new().

Changes from v3:

 - Rebase to the latest tip of bpf-next/for-next

 - Change verb back to C string.

Changes from v2:

 - Include pcap header files conditionally.

 - Move the implementation of traffic monitor to test_progs.c.

 - Include test name and namespace as a part of names of packet files.

 - Parse and print ICMP(v4|v6) packets.

 - Add netns_new() and netns_free() to create and delete network
   namespaces.

   - Make tc_redirect, sockmap_listen and select_reuseport test in a
     network namespace.

 - Add the "-m" option to test_progs to enable traffic monitor for the
   tests matching the pattern. CI may use this option to enable
   monitoring for a given set of tests.

Changes from v1:

 - Move to calling libpcap directly to capture packets in a background
   thread.

 - Print parsed packet information for TCP and UDP packets.

v1: https://lore.kernel.org/all/20240713055552.2482367-5-thinker.li@gmail.com/
v2: https://lore.kernel.org/all/20240723182439.1434795-1-thinker.li@gmail.com/
v3: https://lore.kernel.org/all/20240730002745.1484204-1-thinker.li@gmail.com/
v4: https://lore.kernel.org/all/20240731193140.758210-1-thinker.li@gmail.com/
v5: https://lore.kernel.org/all/20240806221243.1806879-1-thinker.li@gmail.com/
v6: https://lore.kernel.org/all/20240807183149.764711-1-thinker.li@gmail.com/
v7: https://lore.kernel.org/all/20240810023534.2458227-2-thinker.li@gmail.com/
====================
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Enable traffic monitoring for the subtests of select_reuseport.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-7-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Enable traffic monitor for each subtest of sockmap_listen.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-6-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Enable traffic monitoring for the test case tc_redirect.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-5-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

netns_new()/netns_free() create/delete network namespaces. They support the
option '-m' of test_progs to start/stop traffic monitor for the network
namespace being created for matched tests.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-4-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Add option '-m' to test_progs to accept names and patterns of test cases.
This option will be used later to enable traffic monitor that capture
network packets generated by test cases.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-3-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Add functions that capture packets and print log in the background. They
are supposed to be used for debugging flaky network test cases. A monitored
test case should call traffic_monitor_start() to start a thread to capture
packets in the background for a given namespace and call
traffic_monitor_stop() to stop capturing. (Or, option '-m' implemented by
the later patches.)

    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 68, SYN
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 60, SYN, ACK
    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 60, ACK
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 52, ACK
    lo      In  IPv4 127.0.0.1:40265 > 127.0.0.1:55907: TCP, length 52, FIN, ACK
    lo      In  IPv4 127.0.0.1:55907 > 127.0.0.1:40265: TCP, length 52, RST, ACK
    Packet file: packets-2173-86-select_reuseport:sockhash_IPv4_TCP_LOOPBACK_test_detach_bpf-test.log
    #280/87 select_reuseport/sockhash IPv4/TCP LOOPBACK test_detach_bpf:OK

The above is the output of an example. It shows the packets of a connection
and the name of the file that contains captured packets in the directory
/tmp/tmon_pcap. The file can be loaded by tcpdump or wireshark.

This feature only works if libpcap is available. (Could be found by pkg-config)
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Kui-Feng Lee <thinker.li@gmail.com>
Link: https://lore.kernel.org/r/20240815053254.470944-2-thinker.li@gmail.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

Alexis Lothoré (eBPF Foundation) says:

====================
Hello,
this series brings a new set of test converted to the test_progs framework.
Since the tests are quite small, I chose to group three tests conversion in
the same series, but feel free to let me know if I should keep one series
per test. The series focuses on cgroup testing and converts the following
tests:
- get_cgroup_id_user
- cgroup_storage
- test_skb_cgroup_id_user

Changes in v4:
- Fix test after netns addition by making sure loopack interface is up
- Link to v3: https://lore.kernel.org/r/20240812-convert_cgroup_tests-v3-0-47ac6ce4e88b@bootlin.com

Changes in v3:
- Fixed multiple leaks on cgroup file descriptors and sockets
- Used dedicated network namespaces for tests involving network
- Link to v2: https://lore.kernel.org/r/20240806-convert_cgroup_tests-v2-0-180c57e5b710@bootlin.com

Changes in v2:
- Use global variables instead of maps when possible
- Collect review tags from Alan
- Link to v1: https://lore.kernel.org/r/20240731-convert_cgroup_tests-v1-0-14cbc51b6947@bootlin.com
====================
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

test_skb_cgroup_id_user allows testing skb cgroup id retrieval at different
levels, but is not integrated in test_progs, so it is not run
automatically in CI. The test overlaps a bit with
cgroup_skb_sk_lookup_kern, which is integrated in test_progs and test
extensively skb cgroup helpers, but there is still one major difference
between the two tests which justifies the conversion:
cgroup_skb_sk_lookup_kern deals with a BPF_PROG_TYPE_CGROUP_SKB (attached
on a cgroup), while test_skb_cgroup_id_user deals with a
BPF_PROG_TYPE_SCHED_CLS (attached on a qdisc)

Convert test_skb_cgroup_id_user into test_progs framework in order to run
it automatically in CI. The main differences with the original test are the
following:
- rename the test to make it shorter and more straightforward regarding
  tested feature
- the wrapping shell script has been dropped since every setup step is now
  handled in the main C test file
- the test has been renamed for a shorter name and reflecting the tested
  API
- add dedicated assert log per level to ease test failure debugging
- use global variables instead of maps to access bpf prog data
Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com>
Link: https://lore.kernel.org/r/20240813-convert_cgroup_tests-v4-4-a33c03458cf6@bootlin.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

test_skb_cgroup_id_kern.c is currently involved in a manual test. In its
current form, it can not be used with the auto-generated skeleton APIs,
because the section name is not valid to allow libbpf to deduce the program
type.

Update section name to allow skeleton APIs usage. Also rename the program
name to make it shorter and more straighforward regarding the API it is
testing. While doing so, make sure that test_skb_cgroup_id.sh passes to get
a working reference before converting it to test_progs
- update the obj name
- fix loading issue (verifier rejecting the program when loaded through tc,
  because of map not found), by preloading the whole obj with bpftool
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com>
Link: https://lore.kernel.org/r/20240813-convert_cgroup_tests-v4-3-a33c03458cf6@bootlin.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

test_cgroup_storage is currently a standalone program which is not run
when executing test_progs.

Convert it to the test_progs framework so it can be automatically executed
in CI. The conversion led to the following changes:
- converted the raw bpf program in the userspace test file into a dedicated
  test program in progs/ dir
- reduced the scope of cgroup_storage test: the content from this test
  overlaps with some other tests already present in test_progs, most
  notably netcnt and cgroup_storage_multi*. Those tests already check
  extensively local storage, per-cpu local storage, cgroups interaction,
  etc. So the new test only keep the part testing that the program return
  code (based on map content) properly leads to packet being passed or
  dropped.
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com>
Link: https://lore.kernel.org/r/20240813-convert_cgroup_tests-v4-2-a33c03458cf6@bootlin.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

get_current_cgroup_id_user allows testing for bpf_get_current_cgroup_id()
bpf API but is not integrated into test_progs, and so is not tested
automatically in CI.

Convert it to the test_progs framework to allow running it automatically.
The most notable differences with the old test are the following:
- the new test relies on autoattach instead of manually hooking/enabling
  the targeted tracepoint through perf_event, which reduces quite a lot the
  test code size
- it also accesses bpf prog data through global variables instead of maps
- sleep duration passed to nanosleep syscall has been reduced to its
  minimum to not impact overall CI duration (we only care about the syscall
  being properly triggered, not about the passed duration)
Signed-off-by: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com>
Link: https://lore.kernel.org/r/20240813-convert_cgroup_tests-v4-1-a33c03458cf6@bootlin.comSigned-off-by: Martin KaFai Lau <martin.lau@kernel.org>

License should be

// SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)

...as with other libbpf files.

Fixes: 19e00c89 ("libbpf: Split BTF relocation")
Reported-by: Neill Kapron <nkapron@google.com>
Signed-off-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/bpf/20240810093504.2111134-1-alan.maguire@oracle.com