[PATCH] mpfr: Fix compilation for ARC
MPFR has a couple of things implemented in assembly and in case of
ARC those parts were written long ago when ARC GCC port was not yet
upstreamed. On upstreaming of GCC some constraints were changed and
so we can no longer build MPFR for ARC with up-to-date tools seeing
something like that:
| In file included from ../../mpfr-4.0.1/src/mpfr-impl.h:112,
| from ../../mpfr-4.0.1/src/mul.c:24:
| ../../mpfr-4.0.1/src/mul.c: In function 'mpfr_mul':
| ../../mpfr-4.0.1/src/mpfr-longlong.h:415:3: error: impossible constraint in
'asm'
|__asm__ ("add.f\t%1, %4, %5\n\tadc\t%0, %2, %3" \
|^~~
Signed-off-by: Alexey Brodkin
---
.../0001-Fix-obsolete-ARC-asm-constraints.patch| 45 ++
meta/recipes-support/mpfr/mpfr_4.0.1.bb| 3 +-
2 files changed, 47 insertions(+), 1 deletion(-)
create mode 100644
meta/recipes-support/mpfr/mpfr/0001-Fix-obsolete-ARC-asm-constraints.patch
diff --git
a/meta/recipes-support/mpfr/mpfr/0001-Fix-obsolete-ARC-asm-constraints.patch
b/meta/recipes-support/mpfr/mpfr/0001-Fix-obsolete-ARC-asm-constraints.patch
new file mode 100644
index 00..59d1b0db67
--- /dev/null
+++ b/meta/recipes-support/mpfr/mpfr/0001-Fix-obsolete-ARC-asm-constraints.patch
@@ -0,0 +1,45 @@
+mpfr-longlong.h: Fix obsolete ARC asm constraints
+
+This patch replaces obsolete ARC "J" asm constraint with
+up-to-date "Cal" constraint.
+
+"J" constraint only existed in pre-upstream GCC port for ARC.
+In current upstream port "Cal" constraint is used which leads
+to compile-time error.
+
+Proposed fix is known to work in Buildroot, Crosstool-NG etc.
+
+[1]
https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=09cb6a17e71bd40d2fbfaf82a1502fc210e33c87
+
+Signed-off-by: Claudiu Zissulescu
+Signed-off-by: Vlad Zakharov
+Signed-off-by: Alexey Brodkin
+
+Upstream-Status: Backport
[https://gforge.inria.fr/scm/viewvc.php/mpfr?view=revision&revision=13251]
+---
+Index: src/mpfr-longlong.h
+===
+--- a/src/mpfr-longlong.h (revision 10963)
b/src/mpfr-longlong.h (working copy)
+@@ -416,17 +416,17 @@
+ : "=r" (sh), \
+"=&r" (sl) \
+ : "r" ((USItype) (ah)), \
+- "rIJ" ((USItype) (bh)),\
++ "rICal" ((USItype) (bh)), \
+"%r" ((USItype) (al)), \
+- "rIJ" ((USItype) (bl)))
++ "rICal" ((USItype) (bl)))
+ #define sub_ddmmss(sh, sl, ah, al, bh, bl) \
+ __asm__ ("sub.f\t%1, %4, %5\n\tsbc\t%0, %2, %3" \
+ : "=r" (sh), \
+"=&r" (sl) \
+ : "r" ((USItype) (ah)), \
+- "rIJ" ((USItype) (bh)),\
++ "rICal" ((USItype) (bh)), \
+"r" ((USItype) (al)), \
+- "rIJ" ((USItype) (bl)))
++ "rICal" ((USItype) (bl)))
+ #endif
+
+ #if defined (__arm__) && (defined (__thumb2__) || !defined (__thumb__)) \
diff --git a/meta/recipes-support/mpfr/mpfr_4.0.1.bb
b/meta/recipes-support/mpfr/mpfr_4.0.1.bb
index f85a846e33..2eee76a357 100644
--- a/meta/recipes-support/mpfr/mpfr_4.0.1.bb
+++ b/meta/recipes-support/mpfr/mpfr_4.0.1.bb
@@ -9,7 +9,8 @@ LIC_FILES_CHKSUM =
"file://COPYING;md5=d32239bcb673463ab874e80d47fae504 \
file://COPYING.LESSER;md5=6a6a8e020838b23406c81b19c1d46df6"
DEPENDS = "gmp autoconf-archive"
-SRC_URI = "http://www.mpfr.org/mpfr-${PV}/mpfr-${PV}.tar.xz";
+SRC_URI = "http://www.mpfr.org/mpfr-${PV}/mpfr-${PV}.tar.xz \
+ file://0001-Fix-obsolete-ARC-asm-constraints.patch"
SRC_URI[md5sum] = "b8dd19bd9bb1ec8831a6a582a7308073"
SRC_URI[sha256sum] =
"67874a60826303ee2fb6affc6dc0ddd3e749e9bfcb4c8655e3953d0458a6e16e"
--
2.16.2
___
linux-snps-arc mailing list
linux-snps-arc@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-snps-arc
[PATCH] packagegroup-core-tools-profile: disable valgrind on arc
As of today there's no port of Valgrind for ARC so disabling it. Signed-off-by: Alexey Brodkin --- meta/recipes-core/packagegroups/packagegroup-core-tools-profile.bb | 1 + 1 file changed, 1 insertion(+) diff --git a/meta/recipes-core/packagegroups/packagegroup-core-tools-profile.bb b/meta/recipes-core/packagegroups/packagegroup-core-tools-profile.bb index 520d907714..4ec1f9284f 100644 --- a/meta/recipes-core/packagegroups/packagegroup-core-tools-profile.bb +++ b/meta/recipes-core/packagegroups/packagegroup-core-tools-profile.bb @@ -55,6 +55,7 @@ VALGRIND = "valgrind" VALGRIND_libc-musl = "" VALGRIND_mipsarch = "" VALGRIND_nios2 = "" +VALGRIND_arc = "" VALGRIND_armv4 = "" VALGRIND_armv5 = "" VALGRIND_armv6 = "" -- 2.16.2 ___ linux-snps-arc mailing list linux-snps-arc@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-snps-arc
Re: [PATCH] ARC: Get rid of toolchain check
On Wed, Oct 17, 2018 at 02:28:41PM +, Alexey Brodkin wrote:
> Hello,
>
> > -Original Message-
> > From: Rob Herring [mailto:r...@kernel.org]
> > Sent: Friday, September 14, 2018 12:04 AM
> > To: alexey.brod...@synopsys.com
> > Cc: linux-snps-arc@lists.infradead.org; Linux Kernel Mailing List
> > ; vineet.gup...@synopsys.com
> > Subject: Re: [PATCH] ARC: Get rid of toolchain check
> >
> > On Thu, Sep 13, 2018 at 3:24 PM Alexey Brodkin
> > wrote:
> > >
> > > This check is very naive: we simply test if GCC invoked without
> > > "-mcpu=XXX" has ARC700 define set. In that case we think that GCC
> > > was built with "--with-cpu=arc700" and has libgcc built for ARC700.
> > >
> > > Otherwise if ARC700 is not defined we think that everythng was built
> > > for ARCv2.
> > >
> > > But in reality our life is much more interesting.
> > >
> > > 1. Regardless of GCC configuration (i.e. what we pass in "--with-cpu"
> > >it may generate code for any ARC core).
> > >
> > > 2. libgcc might be built with explicitly specified "--mcpu=YYY"
> > >
> > > That's exactly what happens in case of multilibbed toolchains:
> > > - GCC is configured with default settings
> > > - All the libs built for many different CPU flavors
> > >
> > > I.e. that check gets in the way of usage of multilibbed
> > > toolchains. And even non-multilibbed toolchains are affected.
> > > OpenEmbedded also builds GCC without "--with-cpu" because
> > > each and every target component later is compiled with explicitly
> > > set "-mcpu=ZZZ".
> > >
> > > Signed-off-by: Alexey Brodkin
> > > ---
> > > arch/arc/Makefile | 14 --
> > > 1 file changed, 14 deletions(-)
> >
> > +1 for this. Removing it also helps with my work to be able to build
> > all the .dts files with only a host compiler. That also needs the hunk
> > setting CROSS_COMPILE removed and not having a built-in dtb by
> > default, but this is a step in the right direction.
> >
> > Acked-by: Rob Herring
>
> May we get this one back-ported to stable trees?
>
> Upstream commit in Linus' tree is
> 615f64458ad8 ("ARC: build: Get rid of toolchain check").
>
> This fixes kernel configuration for ARC in case of missing
> ARC cross-tools in current PATH.
Now queued up, thanks.
greg k-h
___
linux-snps-arc mailing list
linux-snps-arc@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-snps-arc
Re: [PATCH] ARC: Don't set CROSS_COMPILE in arch's Makefile
On Wed, Oct 17, 2018 at 02:33:02PM +, Alexey Brodkin wrote:
> Hello,
>
> > -Original Message-
> > From: Alexey Brodkin [mailto:abrod...@synopsys.com]
> > Sent: Sunday, September 16, 2018 11:48 PM
> > To: linux-snps-arc@lists.infradead.org
> > Cc: linux-ker...@vger.kernel.org; Vineet Gupta ;
> > Alexey Brodkin ; Masahiro
> > Yamada ; Rob Herring
> > Subject: [PATCH] ARC: Don't set CROSS_COMPILE in arch's Makefile
> >
> > There's not much sense in doing that because if user or
> > his build-system didn't set CROSS_COMPILE we still may
> > very well make incorrect guess.
> >
> > But as it turned out setting CROSS_COMPILE is not as harmless
> > as one may think: with recent changes that implemented automatic
> > discovery of __host__ gcc features unconditional setup of
> > CROSS_COMPILE leads to failures on execution of "make xxx_defconfig"
> > with absent cross-compiler, for more info see [1].
> >
> > Set CROSS_COMPILE as well gets in the way if we want only to build
> > .dtb's (again with absent cross-compiler which is not really needed
> > for building .dtb's), see [2].
> >
> > Note, we had to change LIBGCC assignment type from ":=" to "="
> > so that is is resolved on its usage, otherwise if it is resolved
> > at declaration time with missing CROSS_COMPILE we're getting this
> > error message from host GCC:
> > ->8-
> > gcc: error: unrecognized command line option ‘-mmedium-calls’
> > gcc: error: unrecognized command line option ‘-mno-sdata’; did you mean
> > ‘-fno-stats’?
> > ->8-
> >
> > [1]
> > http://lists.infradead.org/pipermail/linux-snps-arc/2018-September/004308.html
> > [2]
> > http://lists.infradead.org/pipermail/linux-snps-arc/2018-September/004320.html
> >
> > Signed-off-by: Alexey Brodkin
> > Cc: Masahiro Yamada
> > Cc: Rob Herring
> > ---
> > arch/arc/Makefile | 10 +-
> > 1 file changed, 1 insertion(+), 9 deletions(-)
> >
> > diff --git a/arch/arc/Makefile b/arch/arc/Makefile
> > index 99cce77ab98f..5f6b67917dc2 100644
> > --- a/arch/arc/Makefile
> > +++ b/arch/arc/Makefile
> > @@ -6,14 +6,6 @@
> > # published by the Free Software Foundation.
> > #
> >
> > -ifeq ($(CROSS_COMPILE),)
> > -ifndef CONFIG_CPU_BIG_ENDIAN
> > -CROSS_COMPILE := arc-linux-
> > -else
> > -CROSS_COMPILE := arceb-linux-
> > -endif
> > -endif
> > -
> > KBUILD_DEFCONFIG := nsim_700_defconfig
> >
> > cflags-y += -fno-common -pipe -fno-builtin -mmedium-calls -D__linux__
> > @@ -79,7 +71,7 @@ cflags-$(disable_small_data) += -mno-sdata
> > -fcall-used-gp
> > cflags-$(CONFIG_CPU_BIG_ENDIAN)+= -mbig-endian
> > ldflags-$(CONFIG_CPU_BIG_ENDIAN) += -EB
> >
> > -LIBGCC := $(shell $(CC) $(cflags-y) --print-libgcc-file-name)
> > +LIBGCC = $(shell $(CC) $(cflags-y) --print-libgcc-file-name)
> >
> > # Modules with short calls might break for calls into builtin-kernel
> > KBUILD_CFLAGS_MODULE += -mlong-calls -mno-millicode
> > --
> > 2.17.1
>
> May we have this one back-ported to stable branches?
>
> Upstream commit in Linus' tree is:
> 40660f1fcee8 ("ARC: build: Don't set CROSS_COMPILE in arch's Makefile").
Applied to 4.9.y, 4.14.y, and 4.18.y, thanks.
greg k-h
___
linux-snps-arc mailing list
linux-snps-arc@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-snps-arc
