Re: [PATCH v2 00/33] locking/atomic: convert all architectures to ARCH_ATOMIC

From: Peter Zijlstra <peterz@infradead.org>
To: Junio C Hamano <gitster@pobox.com>
Cc: Mark Rutland <mark.rutland@arm.com>,
	"Linux-Kernel@Vger. Kernel. Org" <linux-kernel@vger.kernel.org>,
	will@kernel.org, boqun.feng@gmail.com, aou@eecs.berkeley.edu,
	arnd@arndb.de, bcain@codeaurora.org, benh@kernel.crashing.org,
	chris@zankel.net, dalias@libc.org, davem@davemloft.net,
	deanbo422@gmail.com, deller@gmx.de, geert@linux-m68k.org,
	gerg@linux-m68k.org, green.hu@gmail.com, guoren@kernel.org,
	ink@jurassic.park.msu.ru, James.Bottomley@hansenpartnership.com,
	jcmvbkbc@gmail.com, jonas@southpole.se, ley.foon.tan@intel.com,
	linux@armlinux.org.uk, mattst88@gmail.com, monstr@monstr.eu,
	mpe@ellerman.id.au, nickhu@andestech.com,
	palmerdabbelt@google.com, paulus@samba.org,
	paul.walmsley@sifive.com, rth@twiddle.net, shorne@gmail.com,
	stefan.kristiansson@saunalahti.fi, tsbogend@alpha.franken.de,
	vgupta@synopsys.com, ysato@users.sourceforge.jp
Subject: Re: [PATCH v2 00/33] locking/atomic: convert all architectures to ARCH_ATOMIC
Date: Wed, 26 May 2021 16:19:58 +0200	[thread overview]
Message-ID: <YK5ZDtEDmwpxisH1@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <CAPc5daVLg2jVEMW6eAZJQTK7YmMmReDNrNWPnLeX2nckbHhmfw@mail.gmail.com>

On Wed, May 26, 2021 at 09:29:12PM +0900, Junio C Hamano wrote:
> > On Tue, May 25, 2021 at 03:01:59PM +0100, Mark Rutland wrote:
> > > Note: I've generated the patches with:
> > >
> > >   git format-patch -C -M -D
> > >
> > > ... so the preimage of include/linux/atomic-fallback.h is not included
> > > in the diff when it is deleted.
> > ...
> > Junio; can we get something like the below sorted?
> 
> Funny, I recently looked at "git format-patch --help" and read that "-D"
> is explicitly described as "The resulting patch is not meant to be applied
> with patch or git apply; this is solely for people who want to just
> concentrate on reviewing the text after the change."

Yeah, I read that too; but like I said, I feel strongly that anything
generated by git-format-patch should be accepted by git-apply (albeit
with a --do-what-I-want argument, I agree that -D output should never be
applied without express permission).

> So, no, I think it is anti-social to send such a patch generated with -D
> unless it is purely for discussion. The side that generates the patch
> should be told to drop "-D" when running the format-patch command,
> or perhaps format-patch should be taught to error out when "-D" is
> given, perhaps?

Removing -D from git-format-patch works too, although I rather like not
seeing the '-' for the entire files. I really don't care much for the
reversible thing, I already have the content, I can zap the commit to
get it back.

I'll leave it up to you, either remove -D from format-patch, or add an
option to apply to force accept it.