Re: [PATCH 6/9] lib/bpgit.py: add support for git paranoia

["Luis R. Rodriguez" <mcgrof@do-not-panic.com>]

On Thu, May 23, 2013 at 8:25 AM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> On Wed, 2013-05-22 at 17:37 -0700, Luis R. Rodriguez wrote:
>> On Wed, May 22, 2013 at 5:05 AM, Johannes Berg
>> <johannes@sipsolutions.net> wrote:
>> > On Wed, 2013-05-22 at 04:34 -0700, Luis R. Rodriguez wrote:
>> >> From: "Luis R. Rodriguez" <mcgrof@do-not-panic.com>
>> >>
>> >> This cleans sanitizes a source tree and ensures
>> >> no content is present from what was intended.
>> >
>> > All of this (including the previous patches) makes me think
>> > "--git-revision HEAD" should just be the default instead of trying to
>> > muck with the working directory of the kernel tree too much?
>>
>> Not sure I follow, but in any case, I just want to start being able to
>> start making daily releases daily and fast, can we address this later?
>
> I'm suggesting that you use --git-revision to do so, then you only have
> to verify the tag you passed to --git-revision is signed and not have to
> muck around with the tree. Technically, all the stuff you'd been doing
> there is racy too, if you edit the checkout while the script is running,
> and I think it's a bit pointless. I doubt it's even faster than using
> --git-revision if it checks everything.
>
> IOW, why go to all this complexity? I'd say all you need to do is
> require that not passing --git-revision when --kup is passed is an
> error. Then you can make --kup verify the tag passed to --git-revision
> by just calling "git tag -v ..." for the --git-revision argument. This
> will error out if it's not a tag.
>
> Basically this means you only need patch 4, 6, 7 and a modified version
> of 8.

I see what you're saying now and I actually did consider it, the
reason I didn't use --git-revision is speed. Git reset --hard tag, and
then copying the code directly is a lot faster. The race issues you
point out though are worth reconsidering this though.

  Luis