klibc - Mar 2020 - [PATCH v2 3/5] Kbuild: use "libc.a" with clang

Clang doesn't have a suitable replacement for libgcc readily available.
Supply one that we know exists. Use "-fno-builtin-bcmp" to prevent
clang
from calling a function that doesn't exist.

Signed-off-by: Bill Wendling <morbo at google.com>
---
 scripts/Kbuild.klibc          | 2 ++
 usr/klibc/arch/x86_64/MCONFIG | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/scripts/Kbuild.klibc b/scripts/Kbuild.klibc
index 411d7ea9bea6..ba3f389626b3 100644
--- a/scripts/Kbuild.klibc
+++ b/scripts/Kbuild.klibc
@@ -141,6 +141,8 @@ KLIBCSTRIPFLAGS  += --strip-all -R .comment -R .note
 # Don't attempt to set it if we are using clang.
 ifneq ($(cc-name),clang)
 KLIBCLIBGCC_DEF  := $(shell $(KLIBCCC) $(KLIBCCFLAGS) --print-libgcc)
+else
+KLIBCLIBGCC_DEF  := $(KLIBCOBJ)/libc.a
 endif
 KLIBCLIBGCC	 ?= $(KLIBCLIBGCC_DEF)
 KLIBCCRT0        := $(KLIBCOBJ)/arch/$(KLIBCARCHDIR)/crt0.o
diff --git a/usr/klibc/arch/x86_64/MCONFIG b/usr/klibc/arch/x86_64/MCONFIG
index ac1ac074c00a..a6ba7c85c68b 100644
--- a/usr/klibc/arch/x86_64/MCONFIG
+++ b/usr/klibc/arch/x86_64/MCONFIG
@@ -26,6 +26,8 @@ ifneq ($(cc-name),clang)
 # These options are not supported by clang but only result in a warning
 # so we can't use a cc-option test to filter them.
 KLIBCOPTFLAGS     += -falign-functions=1 -falign-jumps=1 -falign-loops=1
+else
+KLIBCOPTFLAGS     += -fno-builtin-bcmp
 endif
 
 KLIBCBITSIZE      = 64
-- 
2.26.0.rc2.310.g2932bb562d-goog

On Fri, 2020-03-27 at 15:29 -0700, Bill Wendling wrote:
> Clang doesn't have a suitable replacement for libgcc readily available.
> Supply one that we know exists. Use "-fno-builtin-bcmp" to
prevent clang
> from calling a function that doesn't exist.
Doesn't this contradict the comment added for KLIBCLIBGCC_DEF?
> Signed-off-by: Bill Wendling <morbo at google.com>
> ---
>  scripts/Kbuild.klibc          | 2 ++
>  usr/klibc/arch/x86_64/MCONFIG | 2 ++
>  2 files changed, 4 insertions(+)
> 
> diff --git a/scripts/Kbuild.klibc b/scripts/Kbuild.klibc
> index 411d7ea9bea6..ba3f389626b3 100644
> --- a/scripts/Kbuild.klibc
> +++ b/scripts/Kbuild.klibc
> @@ -141,6 +141,8 @@ KLIBCSTRIPFLAGS  += --strip-all -R .comment -R .note
>  # Don't attempt to set it if we are using clang.
>  ifneq ($(cc-name),clang)
>  KLIBCLIBGCC_DEF  := $(shell $(KLIBCCC) $(KLIBCCFLAGS) --print-libgcc)
> +else
> +KLIBCLIBGCC_DEF  := $(KLIBCOBJ)/libc.a
>  endif
>  KLIBCLIBGCC	 ?= $(KLIBCLIBGCC_DEF)
>  KLIBCCRT0        := $(KLIBCOBJ)/arch/$(KLIBCARCHDIR)/crt0.o
> diff --git a/usr/klibc/arch/x86_64/MCONFIG b/usr/klibc/arch/x86_64/MCONFIG
> index ac1ac074c00a..a6ba7c85c68b 100644
> --- a/usr/klibc/arch/x86_64/MCONFIG
> +++ b/usr/klibc/arch/x86_64/MCONFIG
> @@ -26,6 +26,8 @@ ifneq ($(cc-name),clang)
>  # These options are not supported by clang but only result in a warning
>  # so we can't use a cc-option test to filter them.
>  KLIBCOPTFLAGS     += -falign-functions=1 -falign-jumps=1 -falign-loops=1
> +else
> +KLIBCOPTFLAGS     += -fno-builtin-bcmp
This doesn't seem to be an optimisation option.  It belongs in
KLIBCARCHREQFLAGS if it's really arch-specific.  But it appears to be
arch-independent, in which case scripts/Kbuild.klibc should add it to
KLIBCREQFLAGS.

Ben.
>  endif
>  
>  KLIBCBITSIZE      = 64
-- 
Ben Hutchings
This sentence contradicts itself - no actually it doesn't.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL:
<https://lists.zytor.com/archives/klibc/attachments/20200328/4b5380c7/attachment.sig>

>>  # These options are not supported by clang but only result in a
warning
>>  # so we can't use a cc-option test to filter them.
Do they also result in a warning only if you add -Werror?

This is how I test most of the things in mksh, and besides
sarge?s GCC failing to error out on __attribute__((bounded))
it mostly catches those cases? if not, suggest that to the
clang developers.

bye,
//mirabilos
-- 
11:56??liwakura:#!/bin/mksh? also, i wanted to add mksh to my own distro ?
i was disappointed that there is no makefile ? but somehow the Build.sh is
the least painful built system i've ever seen ? honours CC, {CPP,C,LD}FLAGS
properly ? looks cleary like done by someone who knows what they are doing

Hi Ben,

I uploaded "v3" to address all of your and Thorsten's comments.
PTAL.

On Sat, Mar 28, 2020 at 8:46 AM Ben Hutchings <ben at decadent.org.uk>
wrote:
>
> On Fri, 2020-03-27 at 15:29 -0700, Bill Wendling wrote:
> > Clang doesn't have a suitable replacement for libgcc readily
available.
> > Supply one that we know exists. Use "-fno-builtin-bcmp" to
prevent clang
> > from calling a function that doesn't exist.
>
> Doesn't this contradict the comment added for KLIBCLIBGCC_DEF?
>
> > Signed-off-by: Bill Wendling <morbo at google.com>
> > ---
> >  scripts/Kbuild.klibc          | 2 ++
> >  usr/klibc/arch/x86_64/MCONFIG | 2 ++
> >  2 files changed, 4 insertions(+)
> >
> > diff --git a/scripts/Kbuild.klibc b/scripts/Kbuild.klibc
> > index 411d7ea9bea6..ba3f389626b3 100644
> > --- a/scripts/Kbuild.klibc
> > +++ b/scripts/Kbuild.klibc
> > @@ -141,6 +141,8 @@ KLIBCSTRIPFLAGS  += --strip-all -R .comment -R
.note
> >  # Don't attempt to set it if we are using clang.
> >  ifneq ($(cc-name),clang)
> >  KLIBCLIBGCC_DEF  := $(shell $(KLIBCCC) $(KLIBCCFLAGS) --print-libgcc)
> > +else
> > +KLIBCLIBGCC_DEF  := $(KLIBCOBJ)/libc.a
> >  endif
> >  KLIBCLIBGCC   ?= $(KLIBCLIBGCC_DEF)
> >  KLIBCCRT0        := $(KLIBCOBJ)/arch/$(KLIBCARCHDIR)/crt0.o
> > diff --git a/usr/klibc/arch/x86_64/MCONFIG
b/usr/klibc/arch/x86_64/MCONFIG
> > index ac1ac074c00a..a6ba7c85c68b 100644
> > --- a/usr/klibc/arch/x86_64/MCONFIG
> > +++ b/usr/klibc/arch/x86_64/MCONFIG
> > @@ -26,6 +26,8 @@ ifneq ($(cc-name),clang)
> >  # These options are not supported by clang but only result in a
warning
> >  # so we can't use a cc-option test to filter them.
> >  KLIBCOPTFLAGS     += -falign-functions=1 -falign-jumps=1
-falign-loops=1
> > +else
> > +KLIBCOPTFLAGS     += -fno-builtin-bcmp
>
> This doesn't seem to be an optimisation option.  It belongs in
> KLIBCARCHREQFLAGS if it's really arch-specific.  But it appears to be
> arch-independent, in which case scripts/Kbuild.klibc should add it to
> KLIBCREQFLAGS.
>
> Ben.
>
> >  endif
> >
> >  KLIBCBITSIZE      = 64
> --
> Ben Hutchings
> This sentence contradicts itself - no actually it doesn't.
>