Re: [PATCH v2] block-sha1: remove use of assembly

From: Taylor Blau <me@ttaylorr.com>
Date: Mar 10, 2022, 02:34 UTC
Message-ID: <Yilj0qMqk5eC4TKo@nand.local>
In-Reply-To: <220310.86cziulls6.gmgdl@evledraar.gmail.com>

On Thu, Mar 10, 2022 at 12:52:31AM +0100, Ævar Arnfjörð Bjarmason wrote:

Show 15 quoted lines

> >> I suspect it has the same problem.  My inclination is to just remove it,
> >> because my guess is that the compiler has gotten smarter between 2009
> >> and now.
> >
> > Almost certainly. I don't have a machine to test it on, either, but I
> > would be shocked if `make BLK_SHA=YesPlease DEVELOPER=1` worked on
> > master today on an arm machine.
>
> Why is that? The -pedantic error is specifically about
> "gnu-statement-expression", i.e. the bracket syntax, not the inline
> assembly per-se.
>
> The ARM assembly isn't using that, and we have other code __asm__ code
> compiled with -pedantic. E.g. I get the __asm__ in "compat/bswap.h" by
> default, and it passes -pedantic (the code starting around line 38).

You're right, I had this completely mixed up in my mind. In GitHub's fork there is a spot I have been working near for the past couple of days where there is inline assembly right below a statement expression.

The statement expression is what causes the -pedantic builds to fail, not the inline __asm__. Indeed, if you just stick a memory barrier anywhere in Git's codebase, we'll still compile under the DEVELOPER=1 builds.

> Isn't that __extension__ only needed *if* it warns under -pedantic,
> which AFAICT doesn't apply to all uses of __asm__ (but your compiler
> version etc. may be different...).

Yes, disregard that last suggestion :-).

Thanks, Taylor