[OE-core] [PATCH 2/2] grub2: fix build with gcc-7

Tue Aug 22 17:28:27 UTC 2017

I have sent few patches to meta oe to fix similar issues earlier found with
clang I think changes to code to define the macros in compliant way are
better and upstreamable

On Tue, Aug 22, 2017 at 10:21 AM Martin Jansa <martin.jansa at gmail.com>
wrote:

> Nothing special on my build host, just gcc 7.1 from gentoo. But the option
> was introduced probably only in gcc-7, on 5.4.0 used in Ubuntu 16.04 it's
> also missing:
>
> OE qemux86@ ~/oe/world/shr-core $ gcc --help=warnings | grep
> expansion-to-defined
> OE qemux86@ ~/oe/world/shr-core $ gcc --version
> gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 20160609
> Copyright (C) 2015 Free Software Foundation, Inc.
> This is free software; see the source for copying conditions.  There is NO
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
>
> OE @ ~ $ gcc --help=warnings | grep expansion-to-defined
>   -Wexpansion-to-defined      Warn if "defined" is used outside #if.
> OE @ ~ $ gcc --version
> gcc (Gentoo 7.1.0-r1 p1.1) 7.1.0
> Copyright (C) 2017 Free Software Foundation, Inc.
> This is free software; see the source for copying conditions.  There is NO
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
>
> OE @ ~/build/oe-core/tmp-glibc/work/i586-oe-linux/grub-efi/2.02-r0 $
> recipe-sysroot-native/usr/bin/i586-oe-linux/i586-oe-linux-gcc
> --help=warnings | grep expansion-to-defined
>   -Wexpansion-to-defined      Warn if "defined" is used outside #if.
> OE @ ~/build/oe-core/tmp-glibc/work/i586-oe-linux/grub-efi/2.02-r0 $
> recipe-sysroot-native/usr/bin/i586-oe-linux/i586-oe-linux-gcc --version
> i586-oe-linux-gcc (GCC) 7.2.0
> Copyright (C) 2017 Free Software Foundation, Inc.
> This is free software; see the source for copying conditions.  There is NO
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
>
> Applying the flags based on `gcc --version` output also doesn't seem good
> way forward and we don't have gcc-native recipes.
>
> We can probably use gcc pragma to disable this warning (which IIRC doesn't
> test if the disabled warning is supported or not), but that might be a bit
> complicated as this file is generated in configure. I'll check it.
>
> The target grub-efi build doesn't fail probably because target gcc doesn't
> define _FORTIFY_SOURCE
>
> I've tried to compare native and target build and it's reproducible e.g.
> with this command:
>
> gcc  -DHAVE_CONFIG_H -I. -I../grub-2.02  -Wall -W -DGRUB_UTIL=1
> -D_FILE_OFFSET_BITS=64 -I./include -DGRUB_FILE=\"grub-core/lib/crypto.c\"
> -I. -I../grub-2.02 -I. -I../grub-2.02 -I../grub-2.02/include -I./include
> -I../grub-2.02/grub-core/lib/libgcrypt-grub/src/  -I./grub-core/gnulib
> -I../grub-2.02/grub-core/gnulib -D_FILE_OFFSET_BITS=64 -fno-builtin-gettext
> -Wall -W -Wshadow -Wpointer-arith -Wundef -Wchar-subscripts -Wcomment
> -Wdeprecated-declarations -Wdisabled-optimization -Wdiv-by-zero
> -Wfloat-equal -Wformat-extra-args -Wformat-security -Wformat-y2k -Wimplicit
> -Wimplicit-function-declaration -Wimplicit-int -Wmain -Wmissing-braces
> -Wmissing-format-attribute -Wmultichar -Wparentheses -Wreturn-type
> -Wsequence-point -Wshadow -Wsign-compare -Wswitch -Wtrigraphs
> -Wunknown-pragmas -Wunused -Wunused-function -Wunused-label
> -Wunused-parameter -Wunused-value  -Wunused-variable -Wwrite-strings
> -Wnested-externs -Wstrict-prototypes -Wcast-align  -Wextra -Wattributes
> -Wendif-labels -Winit-self -Wint-to-pointer-cast -Winvalid-pch
> -Wmissing-field-initializers -Wnonnull -Woverflow -Wvla
> -Wpointer-to-int-cast -Wstrict-aliasing -Wvariadic-macros
> -Wvolatile-register-var -Wpointer-sign -Wmissing-include-dirs
> -Wmissing-prototypes -Wmissing-declarations -Wformat=2 -Werror  -Wno-undef
> -Wno-sign-compare -Wno-unused -Wno-unused-parameter -Wno-redundant-decls
> -Wno-unreachable-code -Wno-conversion -Wno-error=trampolines -c -o
> grub-core/lib/libgrubkern_a-crypto.o `test -f 'grub-core/lib/crypto.c' ||
> echo '../grub-2.02/'`grub-core/lib/crypto.c
>
> After adding -dM -E and comparing the output between target and native I
> can see that my native gcc defines _FORTIFY_SOURCE as:
> #define _FORTIFY_SOURCE ((defined __OPTIMIZE__ && __OPTIMIZE__ > 0) ? 2 :
> 0)
>
> while target build doesn't.
>
> Just to confirm it's caused by this I've added the same line:
> #define _FORTIFY_SOURCE ((defined __OPTIMIZE__ && __OPTIMIZE__ > 0) ? 2 :
> 0)
> at the top of grub-2.02/grub-core/lib/crypto.c and the target build fails
> the same.
>
> And the  _FORTIFY_SOURCE in native gcc comes from gentoo gcc patches, this
> one:
>
> https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo/src/patchsets/gcc/7.1.0/gentoo/10_all_default-fortify-source.patch?view=markup
>
> PATCHv2 with #pragma coming soon.
>
>
>
> On Tue, Aug 22, 2017 at 1:52 AM, Khem Raj <raj.khem at gmail.com> wrote:
>
>> On Mon, Aug 21, 2017 at 1:57 PM, Martin Jansa <martin.jansa at gmail.com>
>> wrote:
>> > * build with gcc-7 fails with:
>> >   | ./config-util.h:1504:48: error: this use of "defined" may not be
>>  portable [-Werror=expansion-to-defined]
>> >   |              || (defined _FORTIFY_SOURCE && 0 < _FORTIFY_SOURCE \
>> >   |
>>                                             ^~~~~~~~~~~~~~~
>> >   and we're not building on Apple or BSD, so we should be able
>> >   to just ignoreit until it's resolved properly in some newer gnulib
>> >   and imported from gnulib to grub-2
>> >
>> > Signed-off-by: Martin Jansa <Martin.Jansa at gmail.com>
>> > ---
>> >  meta/recipes-bsp/grub/grub-efi_2.02.bb | 6 ++++++
>> >  1 file changed, 6 insertions(+)
>> >
>> > diff --git a/meta/recipes-bsp/grub/grub-efi_2.02.bb
>> b/meta/recipes-bsp/grub/grub-efi_2.02.bb
>> > index b949bebced..45f4c1ed34 100644
>> > --- a/meta/recipes-bsp/grub/grub-efi_2.02.bb
>> > +++ b/meta/recipes-bsp/grub/grub-efi_2.02.bb
>> > @@ -36,6 +36,12 @@ EXTRA_OECONF += "--enable-efiemu=no"
>> >  # and many other places in the grub code when compiled with some
>> native gcc compilers (specifically, gentoo)
>> >  CFLAGS_append_class-native = " -Wno-error=trampolines"
>> >
>> > +# this comes from gnulib and it's used only for Apple and BSD, so we
>> can ignore it
>> > +# ./config-util.h:1504:48: error: this use of "defined" may not be
>> portable [-Werror=expansion-to-defined]
>> > +#             || (defined _FORTIFY_SOURCE && 0 < _FORTIFY_SOURCE \
>> > +#                                                ^~~~~~~~~~~~~~~
>> > +CFLAGS_append_class-native = " -Wno-error=expansion-to-defined"
>> > +
>>
>> This workaround seems reasonable but I wonder why it fails on build host
>> for you. how is your build hosts toolchain configured to generate this
>> warning. Since you mentioned gcc7 then why doesnt it fail for OE
>> targets
>> in same way.
>>
>> An upstreamable fix would be to break this define into another #if ..
>> #else macros and define the new macro to 1 or 0 depending upon the
>> conditions
>> being tested.
>>
>> >  do_install_class-native() {
>> >         install -d ${D}${bindir}
>> >         install -m 755 grub-mkimage ${D}${bindir}
>> > --
>> > 2.14.1
>> >
>> > --
>> > _______________________________________________
>> > Openembedded-core mailing list
>> > Openembedded-core at lists.openembedded.org
>> > http://lists.openembedded.org/mailman/listinfo/openembedded-core
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openembedded.org/pipermail/openembedded-core/attachments/20170822/2e22fbdd/attachment-0002.html>