[Buildroot] [Bug 5810] New: Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[Buildroot] [Bug 5810] New: Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

           Summary: Buildroot 2012.11: Additional GCC option
                    "-msoft-float" breaks the build
           Product: buildroot
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: major
          Priority: P5
         Component: Other
        AssignedTo: unassigned at buildroot.uclibc.org
        ReportedBy: tbolsh at gmail.com
                CC: buildroot at uclibc.org
   Estimated Hours: 0.0


I need to build an embedded Linux for 486 without coprocessor.
2012.08 works for that reason, 2012.11 breaks.
Here is the error (meanwhile - why it applies arm related patches to x86?):
_____________________________________________________
Applying 810-arm-softfloat-libgcc.patch using patch:
patching file gcc/config/arm/linux-elf.h
patching file libgcc/config/arm/t-linux

Applying 830-arm_unbreak_armv4t.patch using patch:
patching file gcc/config/arm/linux-eabi.h
touch /root/buildroot-2012.11/output/toolchain/gcc-4.7.2/.patched
mkdir -p /root/buildroot-2012.11/output/toolchain/gcc-4.7.2-initial
(cd /root/buildroot-2012.11/output/toolchain/gcc-4.7.2-initial; rm -rf
config.cache; \
               
PATH=/root/buildroot-2012.11/output/host/bin:/root/buildroot-2012.11/output/host/usr/bin:/usr/lib/git-core:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
AR="/usr/bin/ar" AS="/usr/bin/as" LD="/usr/bin/ld" NM="/usr/bin/nm"
CC="/root/buildroot-2012.11/output/host/usr/bin/ccache /usr/bin/gcc"
GCC="/root/buildroot-2012.11/output/host/usr/bin/ccache /usr/bin/gcc"
CXX="/root/buildroot-2012.11/output/host/usr/bin/ccache /usr/bin/g++"
CPP="/usr/bin/cpp" CPPFLAGS="-I/root/buildroot-2012.11/output/host/usr/include"
CFLAGS="-O2 -I/root/buildroot-2012.11/output/host/usr/include" CXXFLAGS="-O2
-I/root/buildroot-2012.11/output/host/usr/include"
LDFLAGS="-L/root/buildroot-2012.11/output/host/lib
-L/root/buildroot-2012.11/output/host/usr/lib
-Wl,-rpath,/root/buildroot-2012.11/output/host/usr/lib"
PKG_CONFIG_ALLOW_SYSTEM_CFLAGS=1 PKG_CONFIG_ALLOW_SYSTEM_LIBS=1
PKG_CONFIG="/root/buildroot-2012.11/output/host/usr/bin/pkg-config"
PKG_CONFIG_SYSROOT_DIR="/"
PKG_CONFIG_LIBDIR="/root/buildroot-2012.11/output/host/usr/lib/pkgconfig:/root/buildroot-2012.11/output/host/usr/share/pkgconfig"
PERLLIB="/root/buildroot-2012.11/output/host/usr/lib/perl"
LD_LIBRARY_PATH="/root/buildroot-2012.11/output/host/usr/lib:" \
                /root/buildroot-2012.11/output/toolchain/gcc-4.7.2/configure  \
                --prefix=/root/buildroot-2012.11/output/host/usr \
                --build=i686-pc-linux-gnu \
                --host=i686-pc-linux-gnu \
                --target=i486-buildroot-linux-uclibc \
                --enable-languages=c \
               
--with-sysroot=/root/buildroot-2012.11/output/toolchain/uClibc_dev/ \
                --disable-__cxa_atexit \
                --enable-target-optspace \
                 \
                --enable-libgomp \
                --with-gnu-ld \
                --disable-shared \
                --disable-libssp \
                --without-headers \
                --with-newlib \
                --disable-multilib \
                --enable-tls \
                --with-gmp=/root/buildroot-2012.11/output/host/usr \
                --with-mpfr=/root/buildroot-2012.11/output/host/usr \
                --with-mpc=/root/buildroot-2012.11/output/host/usr \
                --disable-nls \
                --enable-threads \
                --disable-decimal-float \
                 \
                 --with-arch="i486" --with-tune="i486"  \
                --with-pkgversion="Buildroot 2012.11"
--with-bugurl="http://bugs.buildroot.net/" -msoft-float \
        )
configure: error: unrecognized option: `-msoft-float'
Try `/root/buildroot-2012.11/output/toolchain/gcc-4.7.2/configure --help' for
more information.
make: ***
[/root/buildroot-2012.11/output/toolchain/gcc-4.7.2-initial/.configured] Error
1

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

--- Comment #1 from Thomas Petazzoni <thomas.petazzoni@free-electrons.com> 2012-12-21 18:22:34 UTC ---
Please post your Buildroot .config so we can reproduce the problem.

And for the "meanwhile - why it applies arm related patches to x86?" question,
this is because we don't do architecture-specific patches. Those patches are
always applied, regardless of which architecture you are building for, but in
practice they only touch ARM-specific parts of gcc, so they are totally
harmless if you build for x86, PowerPC or MIPS. Having a single stack of
patches that are always applied regardless of the selected architecture is much
better than having per-architecture patches.

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

--- Comment #2 from Timofei Bolshakov <tbolsh@gmail.com> 2012-12-21 18:42:55 UTC ---
(In reply to comment #1)
> Please post your Buildroot .config so we can reproduce the problem.
> 
> And for the "meanwhile - why it applies arm related patches to x86?" question,
> this is because we don't do architecture-specific patches. Those patches are
> always applied, regardless of which architecture you are building for, but in
> practice they only touch ARM-specific parts of gcc, so they are totally
> harmless if you build for x86, PowerPC or MIPS. Having a single stack of
> patches that are always applied regardless of the selected architecture is much
> better than having per-architecture patches.

OK.
Especially if it is harmless.

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

--- Comment #3 from Timofei Bolshakov <tbolsh@gmail.com> 2012-12-21 18:52:40 UTC ---
Created attachment 4706
  --> https://bugs.busybox.net/attachment.cgi?id=4706
Zip file contains .config and kernel defconfig used in .config

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

--- Comment #4 from Thomas Petazzoni <thomas.petazzoni@free-electrons.com> 2012-12-21 19:00:13 UTC ---
So, you're explicitly passing:

BR2_EXTRA_GCC_CONFIG_OPTIONS="-msoft-float"

in your Buildroot configuration, and then you complain because GCC doesn't
understand your special flag? If you pass a special flag, aren't you supposed
to know what you're doing?

This seems like a curious bug report, isn't it ?

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

--- Comment #5 from Timofei Bolshakov <tbolsh@gmail.com> 2012-12-21 20:44:10 UTC ---
(In reply to comment #4)
> So, you're explicitly passing:
> 
> BR2_EXTRA_GCC_CONFIG_OPTIONS="-msoft-float"

No, I am setting "Additional gcc options". That leads to 
BR2_EXTRA_GCC_CONFIG_OPTIONS="-msoft-float"
I surely was not understanding that it will lead to CONFIG option. 
My fault. But that is curious conversion from "Additional gcc options" to 
BR2_EXTRA_GCC_CONFIG_OPTIONS - it is not clear until one goes into 
the .config in text form. Meanwhile - using it in "Target optimizations" also 
leads to break. 

And that was not breaking the build in 2012.8

So, where should I put my -msoft-float then? I want everything to be compiled 
with this flag.

> 
> in your Buildroot configuration, and then you complain because GCC doesn't
> understand your special flag? 

What should I complain about then? Unclear language?
If one will change "Additional gcc options" to "Additional gcc config options"
there will be no confusion. Or to use anything I should go and check the source
code? 

> If you pass a special flag, aren't you supposed
> to know what you're doing?
> 
> This seems like a curious bug report, isn't it ?

That is clear enough:

-msoft-float
           Use (do not use) the hardware floating-point instructions for
floating-point operations.  When -msoft-float is specified, functions in
libgcc.a will be used to perform floating-point operations. Unless they are
replaced by routines that emulate the floating-point operations, or compiled in
such a way as to call such emulations routines, these routines will issue
floating-point operations. If you are compiling for an Alpha without
floating-point operations, you must ensure that the library is built so as not
to call them. Note that Alpha implementations without floating-point operations
are required to have floating-point registers.

That is what I want - "functions in libgcc.a will be used to perform
floating-point operations". I was wrong to use config option labelled as
"additional" for that. Where can I put my "-msoft-float"?

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Buildroot] [Bug 5810] Buildroot 2012.11: Additional GCC option "-msoft-float" breaks the build

[bugzilla at busybox.net]

https://bugs.busybox.net/show_bug.cgi?id=5810

Thomas Petazzoni <thomas.petazzoni@free-electrons.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #6 from Thomas Petazzoni <thomas.petazzoni@free-electrons.com> 2014-01-07 07:37:43 UTC ---
The ARM soft-float / hard-float support has been significantly improved in the
2013.08 release. There is now a clear way of choosing between soft-float,
hard-float with floating point arguments passed in integer registers (so-called
softfp) and hard-float with floating point arguments passed in floating point
registers (so called hardfp).

Therefore, I believe this bug can now be considered as fixed.

-- 
Configure bugmail: https://bugs.busybox.net/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.