Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 9 Nov 2021 18:17:08 +0000
From:   Joe Burton <jevburton@google.com>
To:     Hou Tao <houtao1@huawei.com>
Cc:     Joe Burton <jevburton.kernel@gmail.com>,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@kernel.org>, linux-kernel@vger.kernel.org,
        netdev@vger.kernel.org, bpf@vger.kernel.org,
        Petar Penkov <ppenkov@google.com>,
        Stanislav Fomichev <sdf@google.com>
Subject: Re: [RFC PATCH v3 2/3] bpf: Add selftests
Message-ID: <YYq7JHdIAdc2bU55@google.com>
References: <20211102021432.2807760-1-jevburton.kernel@gmail.com>
 <20211102021432.2807760-3-jevburton.kernel@gmail.com>
 <98178f0f-ff43-b996-f78b-778f74b44a6b@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <98178f0f-ff43-b996-f78b-778f74b44a6b@huawei.com>
Precedence: bulk

Thu, Nov 04, 2021 at 02:32:37PM +0800, Hou Tao wrote:

> In fentry__x64_sys_write(), you just do trigger updates to maps, so for the
> portability of the test
> (e.g. run-able for arm64)

Agreed that the test should be runnable on arm64. I haven't tested there
yet but I'll do that before sending out v4.

> and minimal dependency (e.g. don't depends on /tmp),
> why do you
> using nanosleep() and replacing fentry_x64_sys_write by
> tp/syscalls/sys_enter_nanosleep instead.

As written, the example actually modifies the return of write(), so I
don't think I can switch to tp/syscalls/* without significantly
reworking the example. To minimize the amount of reworking while
improving compatibility, how does this sound:

1. Add #ifdefs to support arm64
2. Instead of opening /tmp/map_trace_test_file, open /dev/null

Of course this isn't as portable as your proposal but I think it might
be an acceptable compromise.

Best,
Joe