linux-kernel - Re: [GIT PULL] non-MM commits for 6.2-rc1

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, 13 Dec 2022 13:58:11 +1100
From:   Stephen Rothwell <sfr@...b.auug.org.au>
To:     Andrew Morton <akpm@...ux-foundation.org>
Cc:     Vlastimil Babka <vbabka@...e.cz>,
        "Jason A. Donenfeld" <Jason@...c4.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        linux-mm@...ck.org, mm-commits@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [GIT PULL] non-MM commits for 6.2-rc1

Hi Andrew,

On Mon, 12 Dec 2022 18:35:54 -0800 Andrew Morton <akpm@...ux-foundation.org> wrote:
>
> On Tue, 13 Dec 2022 00:16:07 +0100 Vlastimil Babka <vbabka@...e.cz> wrote:
> 
> > On 12/12/22 23:27, Andrew Morton wrote:  
> > > On Mon, 12 Dec 2022 15:16:43 -0700 "Jason A. Donenfeld" <Jason@...c4.com> wrote:
> > >   
> > >> On Mon, Dec 12, 2022 at 01:44:25PM -0800, Andrew Morton wrote:  
> > >> >       wifi: rt2x00: use explicitly signed or unsigned types  
> > >> 
> > >> Why is this part of your PULL? This was a netdev/wireless tree fix which
> > >> was in 6.1-rc5. Kalle wrote that he took it on 21 Oct 2022 in:
> > >> https://lore.kernel.org/all/166633563389.6242.13987912613257140089.kvalo@kernel.org/
> > >>   
> > > 
> > > Huh.  I guess git quietly accepts the identical commit, so Stephen
> > > never told me and my test-merge-with-Linus-latest also came up clean. 
> > > Which leaves it to me to manually handle these things, and that's
> > > unreliable.
> > > 
> > > Is there some way of telling git to complain about identical changes
> > > when doing a trial merge?  
> >  
> > Probably not exactly that, but there seems to be a way with "git cherry". Dunno
> > how reliable in practice, seems to be based on patch-id so probably can fail
> > easily if the commit or context changes slightly?  
> 
> Well here's a hacky thing - just look for duplicated patch titles after
> stripping away the hashes.
> 
> 
> $ git log --oneline --no-merges v6.0..HEAD | sed -e "s/[^ ]* //" | sort > /tmp/1        
> $ uniq < /tmp/1 > /tmp/2
> $ diff -u /tmp/1 /tmp/2 | grep "^-.*"
> 
> The above turns up 67 duplicates in Linus's tree and 382 in linux-next.
> Maybe this is wrong - I'm just tossing it out there in the hope that
> someone will do the work for me :)

I already do this when I produce my stats after the merge window - I
produce three lines like these (from last time):

Commits with the same SHA1:                        10436
Commits with the same patch_id:                      342
Commits with the same subject line:                   20

The above is pretty easy, but takes a while (looking at 11000+ commits)
because I am just conparing Linus' tree to one of my trees (so 2
branches in my tree).  During out development phase, these commits
could be spread over 350+ branches in my tree :-(

I will have a play and see what I can come up with.  The plan would be
to figure these out when I fetch trees and only report them to people
who care (just you, Andrew, initially) as there are trees out there for
which duplicating patches between the development branches and -fixes
branches (and Linus' tree) is part of the process :-(

-- 
Cheers,
Stephen Rothwell

Content of type "application/pgp-signature" skipped