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;
Date:   Thu, 31 Mar 2022 10:33:44 -0700
From:   Andres Freund <andres@anarazel.de>
To:     Matthieu Baerts <matthieu.baerts@tessares.net>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Jiri Olsa <jolsa@kernel.org>,
        Namhyung Kim <namhyung@kernel.org>,
        John Garry <john.garry@huawei.com>, mptcp@lists.linux.dev,
        Arnaldo Carvalho de Melo <acme@redhat.com>,
        linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH perf] perf tools: .git/ORIG_HEAD might not exist
Message-ID: <20220331173344.xvhp4yi67fv6p6t5@alap3.anarazel.de>
References: <20220329093120.4173283-1-matthieu.baerts@tessares.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220329093120.4173283-1-matthieu.baerts@tessares.net>
Precedence: bulk

Hi,

On 2022-03-29 11:31:20 +0200, Matthieu Baerts wrote:
> It seems it cannot be assumed .git/ORIG_HEAD exists if .git/HEAD is
> there.
> 
> Indeed, recently our public CI reported[1] the following error when
> compiling Perf tool:
> 
>   $ cd tools/perf
>   $ make -j4 -l4 O=/tmp/(...)/perf DESTDIR=/usr install
>   (...)
>   make[2]: *** No rule to make target '../../.git/ORIG_HEAD', needed by '/tmp/(...)/perf/PERF-VERSION-FILE'.
> 
> This is because apparently[2] Cirrus Ci uses a Git client implemented
> purely in Go[3] to perform a clone. Most likely, this tool doesn't
> create any .git/ORIG_HEAD file but .git/HEAD is there. The error can
> indeed be reproduced by renaming this .git/ORIG_HEAD file while keeping
> .git/HEAD. In other words, it means it is not enough to check the
> presence of .git/HEAD to assume .git/ORIG_HEAD exists as well.

FWIW, It's not just custom git implementations, stock git doesn't ensure it's
there either. I build a nightly VM image with Linus' kernel for postgres
testing, and as part of that I do a minimal clone:
  git clone --single-branch --depth 1 git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git /usr/src/linux
and then build the kernel. The build recently started failing like this:
https://cirrus-ci.com/task/4648999113195520?logs=build_image#L3121

It's not a question of "--single-branch --depth 1" - ORIG_HEAD just isn't
there in a new clone. Which makes sense, because there's no previous value for
HEAD.

Greetings,

Andres Freund