Subject: Re: [PATCH bpf-next] selftests/bpf: Fix building error when using
 userspace pt_regs
From:   Pu Lehui <pulehui@huawei.com>
To:     Daniel Borkmann <daniel@iogearbox.net>, <ast@kernel.org>,
        <andrii@kernel.org>, <kafai@fb.com>, <songliubraving@fb.com>,
        <yhs@fb.com>, <john.fastabend@gmail.com>, <kpsingh@kernel.org>,
        <paul.walmsley@sifive.com>, <palmer@dabbelt.com>,
        <aou@eecs.berkeley.edu>, <shuah@kernel.org>
CC:     <linux-kselftest@vger.kernel.org>, <netdev@vger.kernel.org>,
        <bpf@vger.kernel.org>, <linux-kernel@vger.kernel.org>
References: <20211214135555.125348-1-pulehui@huawei.com>
 <9063be69-fbd9-c0a5-9271-c6d4281c71ef@iogearbox.net>
 <30aa8ea2-3752-d711-50f8-4b3d49cad56f@huawei.com>
Message-ID: <5525a45e-0f4a-f686-bfa9-c269321a8b86@huawei.com>
Date:   Fri, 17 Dec 2021 10:19:51 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.8.1
MIME-Version: 1.0
In-Reply-To: <30aa8ea2-3752-d711-50f8-4b3d49cad56f@huawei.com>
Content-Type: text/plain; charset="utf-8"; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 2021/12/15 9:20, Pu Lehui wrote:
> On 2021/12/15 4:01, Daniel Borkmann wrote:
>> On 12/14/21 2:55 PM, Pu Lehui wrote:
>>> When building bpf selftests on arm64, the following error will occur:
>>>
>>> progs/loop2.c:20:7: error: incomplete definition of type 'struct
>>> user_pt_regs'
>>>
>>> Some archs, like arm64 and riscv, use userspace pt_regs in
>>> bpf_tracing.h, which causes build failure when bpf prog use
>>> macro in bpf_tracing.h. So let's use vmlinux.h directly.
>>>
>>> Signed-off-by: Pu Lehui <pulehui@huawei.com>
>>
>> Looks like this lets CI fail, did you run the selftests also with 
>> vmtest.sh to
>> double check?
>>
>> https://github.com/kernel-patches/bpf/runs/4521708490?check_suite_focus=true 
>> :
>>
>> [...]
>> #189 verif_scale_loop6:FAIL
>> libbpf: prog 'trace_virtqueue_add_sgs': BPF program load failed: 
>> Argument list too long
>> libbpf: prog 'trace_virtqueue_add_sgs': -- BEGIN PROG LOAD LOG --
>> R1 type=ctx expected=fp
>> BPF program is too large. Processed 1000001 insn
>> verification time 12250995 usec
>> stack depth 88
>> processed 1000001 insns (limit 1000000) max_states_per_insn 107 
>> total_states 21739 peak_states 2271 mark_read 6
>> -- END PROG LOAD LOG --
>> libbpf: failed to load program 'trace_virtqueue_add_sgs'
>> libbpf: failed to load object 'loop6.o'
>> scale_test:FAIL:expect_success unexpected error: -7 (errno 7)
>> Summary: 221/986 PASSED, 8 SKIPPED, 1 FAILED
>> [...]
>>
>> Please take a look and fix in your patch, thanks!
>> .
> Sorry for my negligence, I'll take a look and fix it.
> .
It seems strange that verifier think the loop can execute up to u64_max 
while I just replace the header file.
This looks very similar to the previous llvm issue, 
https://github.com/iovisor/bcc/pull/3270, but I have no idea how to locate.

Back to arm64 bpf selftest compiling problem, we can use header file 
directory generated by "make headers_install" to fix it.

--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -294,7 +294,8 @@ MENDIAN=$(if 
$(IS_LITTLE_ENDIAN),-mlittle-endian,-mbig-endian)
  CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG))
  BPF_CFLAGS = -g -D__TARGET_ARCH_$(SRCARCH) $(MENDIAN) \
  	     -I$(INCLUDE_DIR) -I$(CURDIR) -I$(APIDIR) \
-	     -I$(abspath $(OUTPUT)/../usr/include)
+	     -I$(abspath $(OUTPUT)/../usr/include) \
+	     -I../../../../usr/include