Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <a5d1512b-2c1b-1713-ae52-84fc148ecb5a@isovalent.com>
Date:   Wed, 7 Jun 2023 11:42:16 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.2
Subject: Re: [PATCH v2 bpf 03/11] bpftool: use a local bpf_perf_event_value to
 fix accessing its fields
Content-Language: en-GB
To:     Nick Desaulniers <ndesaulniers@google.com>,
        Alexander Lobakin <alobakin@pm.me>
Cc:     Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Maciej Fijalkowski <maciej.fijalkowski@intel.com>,
        Song Liu <songliubraving@fb.com>,
        Kumar Kartikeya Dwivedi <memxor@gmail.com>,
        bpf@vger.kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, tpgxyz@gmail.com,
        Jiri Olsa <olsajiri@gmail.com>
References: <20220421003152.339542-1-alobakin@pm.me>
 <20220421003152.339542-4-alobakin@pm.me> <ZH+e9IYk+DIZzUFL@google.com>
From:   Quentin Monnet <quentin@isovalent.com>
In-Reply-To: <ZH+e9IYk+DIZzUFL@google.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

2023-06-06 14:02 UTC-0700 ~ Nick Desaulniers <ndesaulniers@google.com>
> On Thu, Apr 21, 2022 at 12:39:04AM +0000, Alexander Lobakin wrote:
>> Fix the following error when building bpftool:
>>
>>   CLANG   profiler.bpf.o
>>   CLANG   pid_iter.bpf.o
>> skeleton/profiler.bpf.c:18:21: error: invalid application of 'sizeof' to an incomplete type 'struct bpf_perf_event_value'
>>         __uint(value_size, sizeof(struct bpf_perf_event_value));
>>                            ^     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> tools/bpf/bpftool/bootstrap/libbpf/include/bpf/bpf_helpers.h:13:39: note: expanded from macro '__uint'
>> tools/bpf/bpftool/bootstrap/libbpf/include/bpf/bpf_helper_defs.h:7:8: note: forward declaration of 'struct bpf_perf_event_value'
>> struct bpf_perf_event_value;
>>        ^
>>
>> struct bpf_perf_event_value is being used in the kernel only when
>> CONFIG_BPF_EVENTS is enabled, so it misses a BTF entry then.
>> Define struct bpf_perf_event_value___local with the
>> `preserve_access_index` attribute inside the pid_iter BPF prog to
>> allow compiling on any configs. It is a full mirror of a UAPI
>> structure, so is compatible both with and w/o CO-RE.
>> bpf_perf_event_read_value() requires a pointer of the original type,
>> so a cast is needed.
>>
> 
> Hi Alexander,
> What's the status of this series? I wasn't able to find a v3 on lore.
> 
> We received a report that OpenMandriva is carrying around this patch.
> https://github.com/ClangBuiltLinux/linux/issues/1805.
> 
> + Tomasz
> 
> Tomasz, do you have more info which particular configs can reproduce
> this issue? Is this patch still necessary?
> 
>> Fixes: 47c09d6a9f67 ("bpftool: Introduce "prog profile" command")

Hi Nick,

This patch is still necessary if you attempt to compile bpftool with
skeletons support, on a host with a kernel version lower than 5.15.

I took over on the bpftool patches from this series, and sent a new
version last month. Given that it only contains the bpftool patches, the
series has a different title and is not tagged as v3, but you can find
it here:

https://lore.kernel.org/all/20230512103354.48374-1-quentin@isovalent.com/t/#u

Jiri (+Cc) found an issue with this set when CONFIG_BPF_EVENTS is
disabled. I need to replicate and investigate, but I've been short on
time to do that over the last few weeks.

Best,
Quentin