linux-kernel - Re: [PATCH] bpftool: exclude bash-completion/bpftool from .gitignore pattern

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Tue, 30 Apr 2019 10:21:52 +0100
From:   Quentin Monnet <quentin.monnet@...ronome.com>
To:     Masahiro Yamada <yamada.masahiro@...ionext.com>
Cc:     Daniel Borkmann <daniel@...earbox.net>,
        Sirio Balmelli <sirio@...d.ch>,
        Song Liu <songliubraving@...com>,
        Alexei Starovoitov <ast@...nel.org>,
        Networking <netdev@...r.kernel.org>, Yonghong Song <yhs@...com>,
        Taeung Song <treeze.taeung@...il.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Jakub Kicinski <jakub.kicinski@...ronome.com>,
        Martin KaFai Lau <kafai@...com>, bpf@...r.kernel.org
Subject: Re: [PATCH] bpftool: exclude bash-completion/bpftool from .gitignore
 pattern

2019-04-30 09:15 UTC+0900 ~ Masahiro Yamada <yamada.masahiro@...ionext.com>
> Hi Quentin,
> 
> 
> On Tue, Apr 30, 2019 at 12:33 AM Quentin Monnet
> <quentin.monnet@...ronome.com> wrote:
>>
>> 2019-04-29 23:47 UTC+0900 ~ Masahiro Yamada <yamada.masahiro@...ionext.com>
>>> tools/bpf/bpftool/.gitignore has the "bpftool" pattern, which is
>>> intended to ignore the following build artifact:
>>>
>>>    tools/bpf/bpftool/bpftool
>>>
>>> However, the .gitignore entry is effective not only for the current
>>> directory, but also for any sub-directories.
>>>
>>> So, the following file is also considered to be ignored:
>>>
>>>    tools/bpf/bpftool/bash-completion/bpftool
>>>
>>> It is obviously version-controlled, so should be excluded from the
>>> .gitignore pattern.
>>>
>>> You can fix it by prefixing the pattern with '/', which means it is
>>> only effective in the current directory.
>>>
>>> I prefixed the other patterns consistently. IMHO, '/' prefixing is
>>> safer when you intend to ignore specific files.
>>>
>>> Signed-off-by: Masahiro Yamada <yamada.masahiro@...ionext.com>
>>> ---
>>
>> Hi,
>>
>> “Files already tracked by Git are not affected” by the .gitignore (says
>> the relevant man page), so bash completion file is not ignored. It would
>> be if we were to add the sources to the index of a new Git repo. But
>> sure, it does not cost much to make the .gitignore cleaner.
> 
> Right, git seems to be flexible enough.
> 
> 
> But, .gitignore is useful to identify
> build artifacts in general.
> In fact, other than git, some projects
> already parse this.
> 
> For example, tar(1) supports:
> 
>       --exclude-vcs-ignores
>             read exclude patterns from the VCS ignore files
> 
> 
> As of writing, this option works only to some extent,
> but I thought this would be useful to create a source
> package without relying on "git archive".
> 
> When I tried "tar --exclude-vcs-ignores", I noticed
> tools/bpf/bpftool/bash-completion/bpftool was not
> contained in the tarball.
> 
> That's why I sent this patch.

Ok, thanks for explaining! Makes sense to me now.

> 
> I can add more info in v2 to clarify
> my motivation though.

Sounds good, yes please.

> 
> 
>>>
>>>   tools/bpf/bpftool/.gitignore | 8 ++++----
>>>   1 file changed, 4 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/tools/bpf/bpftool/.gitignore b/tools/bpf/bpftool/.gitignore
>>> index 67167e4..19efcc8 100644
>>> --- a/tools/bpf/bpftool/.gitignore
>>> +++ b/tools/bpf/bpftool/.gitignore
>>> @@ -1,5 +1,5 @@
>>>   *.d
>>> -bpftool
>>> -bpftool*.8
>>> -bpf-helpers.*
>>> -FEATURE-DUMP.bpftool
>>> +/bpftool
>>> +/bpftool*.8
>>> +/bpf-helpers.*
>>
>> Careful when you add all those slashes, however. "bpftool*.8" and
>> "bpf-helpers.*" should match files under Documentation/, so you do NOT
>> want to prefix them with just a "/".
> 
> OK, I should not have touched what I was unsure about.
> Will fix in v2.

Thanks!
Quentin