Re: [PATCH 2/4] Revert "lib/string.c: implement a basic bcmp"

[Nick Desaulniers]

On Mon, Aug 17, 2020 at 10:44 PM Nathan Chancellor
<natechancellor@xxxxxxxxx> wrote:
>
> On Mon, Aug 17, 2020 at 03:02:10PM -0700, Nick Desaulniers wrote:
> > This reverts commit 5f074f3e192f10c9fade898b9b3b8812e3d83342.
> >
> > Use `-fno-builtin-bcmp` instead.
> >
> > The issue with using `-fno-builtin-*` flags was that they were not
> > retained during an LTO link with LLVM.  This was fixed in clang-11 by
> > https://reviews.llvm.org/D71193
> > (0508c994f0b14144041f2cfd3ba9f9a80f03de08), which is also the minimum
> > supported version of clang for LTO.
> >
> > Signed-off-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> > ---
> >  Makefile               |  1 +
> >  include/linux/string.h |  3 ---
> >  lib/string.c           | 20 --------------------
> >  3 files changed, 1 insertion(+), 23 deletions(-)
> >
> > diff --git a/Makefile b/Makefile
> > index 211a1b6f6478..722ff5864275 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -964,6 +964,7 @@ endif
> >  # to provide implementations of these routines, then prevent the compiler from
> >  # emitting calls to what will be undefined symbols.
> >  KBUILD_CFLAGS        += -fno-builtin-stpcpy
> > +KBUILD_CFLAGS        += -fno-builtin-bcmp
>
> I personally think that this hunk should be its own patch before this
> one then have this patch just be the revert, that way there is no
> regression across a bisect (if one were to ever occur) and so the revert
> is a straight 'git revert', rather than have something else mixed in
> that requires reading the actual changelog text.
>
> No objections if you disagree though.

That's a great idea.  I considered it before sending, but I think it
would be interesting to divorce the KBUILD changes which can be picked
up quickly from the latter changes.  Will send a V2.

>
> >  # include additional Makefiles when needed
> >  include-y                    := scripts/Makefile.extrawarn
> > diff --git a/include/linux/string.h b/include/linux/string.h
> > index b1f3894a0a3e..f3bdb74bc230 100644
> > --- a/include/linux/string.h
> > +++ b/include/linux/string.h
> > @@ -155,9 +155,6 @@ extern void * memscan(void *,int,__kernel_size_t);
> >  #ifndef __HAVE_ARCH_MEMCMP
> >  extern int memcmp(const void *,const void *,__kernel_size_t);
> >  #endif
> > -#ifndef __HAVE_ARCH_BCMP
> > -extern int bcmp(const void *,const void *,__kernel_size_t);
> > -#endif
> >  #ifndef __HAVE_ARCH_MEMCHR
> >  extern void * memchr(const void *,int,__kernel_size_t);
> >  #endif
> > diff --git a/lib/string.c b/lib/string.c
> > index 6012c385fb31..69328b8353e1 100644
> > --- a/lib/string.c
> > +++ b/lib/string.c
> > @@ -922,26 +922,6 @@ __visible int memcmp(const void *cs, const void *ct, size_t count)
> >  EXPORT_SYMBOL(memcmp);
> >  #endif
> >
> > -#ifndef __HAVE_ARCH_BCMP
> > -/**
> > - * bcmp - returns 0 if and only if the buffers have identical contents.
> > - * @a: pointer to first buffer.
> > - * @b: pointer to second buffer.
> > - * @len: size of buffers.
> > - *
> > - * The sign or magnitude of a non-zero return value has no particular
> > - * meaning, and architectures may implement their own more efficient bcmp(). So
> > - * while this particular implementation is a simple (tail) call to memcmp, do
> > - * not rely on anything but whether the return value is zero or non-zero.
> > - */
> > -#undef bcmp
> > -int bcmp(const void *a, const void *b, size_t len)
> > -{
> > -     return memcmp(a, b, len);
> > -}
> > -EXPORT_SYMBOL(bcmp);
> > -#endif
> > -
> >  #ifndef __HAVE_ARCH_MEMSCAN
> >  /**
> >   * memscan - Find a character in an area of memory.
> > --
> > 2.28.0.220.ged08abb693-goog
> >
>
> Cheers,
> Nathan



-- 
Thanks,
~Nick Desaulniers