Re: Pango cross-compile failure (ARM)

[Behdad Esfahbod <behdad behdad org>]

On 12/14/2009 09:00 PM, William To wrote:
> Dear Behdad,
>
> In the case where we cannot actually work around the compiler limitation
> (its ARM-GCC) is it possible that we can #undef the static asserts and
> the library still builds and runs correctly?

No.  It would build, but not run correctly.  There is a reason why the asserts 
are there.  Better fix the compiler.

behdad

> Thanks very much and Best Regards
>
> William To
>
>
>
> On Tue, Dec 15, 2009 at 9:43 AM, Behdad Esfahbod <behdad behdad org
> <mailto:behdad behdad org>> wrote:
>
>     On 12/14/2009 08:36 PM, William To wrote:
>
>         Dear Behdad,
>
>         I have found out that the compiler automatically aligns the data to
>         double word alignment,
>
>
>     Sure.  But does it do that to every member of a struct?  That
>     doesn't make sense.
>
>
>       and that means its screwing up the static asserts
>
>         because the assumed sizes are no longer valid. Have tried
>         adding __attribute__((packed)) to the affected structures, but still
>         there are places still not working. Is there any other way of
>         working
>         around it?
>
>
>     Not that I know of.  You may want to investigate why
>     __attribute__((packed)) is not working.
>
>     behdad
>
>         Thanks very much and Best Regards
>
>         William To
>
>
>
>         On Wed, Dec 9, 2009 at 4:16 PM, Behdad Esfahbod
>         <behdad behdad org <mailto:behdad behdad org>
>         <mailto:behdad behdad org <mailto:behdad behdad org>>> wrote:
>
>             On 12/09/2009 01:32 AM, William To wrote:
>
>                 Dear All,
>
>                 First of all, I am new to the world of Pango and
>         gtk-i18n, so
>                 excuse my
>                 lack of knowledge.
>
>
>             Welcome.
>
>
>
>                 This is a problem I am facing right now with
>         pango-1.26.1 when
>                 cross-compiling to arm-hardfloat-linux-gnu:
>
>             [...]
>
>                 I tracked down to the source and it seems that a static
>                 assertion on the
>                 size of the USHORT/SHORT failed (too small or too large,
>         I don't
>                 know)
>
>
>             That's a deal breaker.  It's basically this struct:
>
>             struct USHORT {
>               char v[2];
>             }
>
>             So if the size is not 2 bytes, you need to debug and see why
>         your
>             compiler is doing that.  And possibly find a solution.  I
>         can't help
>             much with that.
>
>             behdad
>
>
>                 has failed and the build is stuck here. It is possible
>         to remove the
>                 ASSERT_SIZE in the header file, but then it won't build
>         either.
>
>                 Here are my environmental variables:
>
>                 export
>
>           CC_PATH=/opt/crosstool/gcc-4.3.2-glibc-2.6.1/hardfloat/arm-hardfloat-linux-gnu/bin
>                 export COMPILER=$CC_PATH/arm-hardfloat-linux-gnu-gcc
>                 export CC=$CC_PATH/arm-hardfloat-linux-gnu-gcc
>                 export CPP=$CC_PATH/arm-hardfloat-linux-gnu-cpp
>                 export CXX=$CC_PATH/arm-hardfloat-linux-gnu-g++
>                 export AR=$CC_PATH/arm-hardfloat-linux-gnu-ar
>                 export RANLIB=$CC_PATH/arm-hardfloat-linux-gnu-ranlib
>                 export AS=$CC_PATH/arm-hardfloat-linux-gnu-as
>                 export LD=$CC_PATH/arm-hardfloat-linux-gnu-ld
>                 export NM=$CC_PATH/arm-hardfloat-linux-gnu-nm
>                 export PREFIX=/home/william/1234/libraries/prefix/webapps/
>                 export EXEC_PREFIX=/home/william/1234/libraries/mnt/webapps/
>                 export PKG_CONFIG_PATH=$PREFIX/lib/pkgconfig
>
>                 I use the ubuntu gcc 4.3.2 compiler to build and there
>         is no such
>                 problem. Previous versions of pango compiles normally
>         with the same
>                 cross compiler. I wonder has anybody had hit the same
>         problem as me.
>
>                 Thanks very much and Best Regards
>
>                 William To
>
>
>
>                 _______________________________________________
>                 gtk-i18n-list mailing list
>         gtk-i18n-list gnome org <mailto:gtk-i18n-list gnome org>
>         <mailto:gtk-i18n-list gnome org <mailto:gtk-i18n-list gnome org>>
>
>         http://mail.gnome.org/mailman/listinfo/gtk-i18n-list