[PATCH 2 of 2] cc/gcc: add support for a statcally linked shared core for static toolchain

[Bryan Hundven]

On Wed, Dec 8, 2010 at 2:58 PM, Yann E. MORIN
<yann.morin.1998@anciens.enib.fr> wrote:
> Bryan, All,
>
> On Monday 18 October 2010 23:12:04 Bryan Hundven wrote:
>> # HG changeset patch
>> # User Bryan Hundven <bryanhundven@gmail.com>
>> # Date 1287435325 25200
>> # Node ID fa5bd6f1ea1aa117771e4c53fefaba16ba603d82
>> # Parent  8091950d99a8b10bbff4b7a35518380afeefc5f6
>> cc/gcc: add support for a statcally linked shared core for static toolchain
>
>> diff -r 8091950d99a8 -r fa5bd6f1ea1a scripts/build/cc/gcc.sh
>> --- a/scripts/build/cc/gcc.sh Mon Oct 18 12:15:08 2010 -0700
>> +++ b/scripts/build/cc/gcc.sh Mon Oct 18 13:55:25 2010 -0700
>> @@ -66,9 +66,18 @@
>>          y,*,*)  do_cc_core mode=baremetal build_libgcc=yes build_libstdcxx=yes;;
>>          ,y,*)   ;;
>>          ,,nptl)
>> -            do_cc_core mode=shared build_libgcc=yes
>> +            if [ "${CT_STATIC_TOOLCHAIN}" = "y" ]; then
>> +                do_cc_core mode=shared build_libgcc=yes build_staticlinked=yes
>> +            else
>> +                do_cc_core mode=shared build_libgcc=yes
>> +            fi
>>              ;;
>> -        ,,win32) do_cc_core mode=static build_libgcc=yes
>> +        ,,win32)
>> +            if [ "${CT_STATIC_TOOLCHAIN}" = "y" ]; then
>> +                do_cc_core mode=static build_libgcc=yes build_staticlinked=yes
>> +            else
>> +                do_cc_core mode=static build_libgcc=yes
>> +            fi
>
> Unfortunately, you got it wrong:
>
> - when not bare-metal, the core pass 1 & 2 gcc are discarded, and only the
>  final gcc is part of the toolchain. And that is the final gcc we want to
>  be statically linked. You got it right in your first patch.
>
> - but when doing bare-metal, the final gcc is coming from the core_cc_pass_2
>  (yes, that's odd, but that's the way it goes). And that is that pass 2 gcc
>  we want to be statically linked.
>
> Unfortunately, your patch #2 only builds static core gcc when not bare-metal.
> So, for bare-metal (the first match in the case statement) the useful gcc
> will never be statically linked.
>
> Don't worry, I've fixed this in my local copy. ;-)
>
>>              ;;
>>          *)  if [ "${CT_CC_GCC_4_3_or_later}" = "y" ]; then
>>                  do_cc_core mode=static build_libgcc=yes
>> @@ -84,18 +93,20 @@
>>  # This function is used to build both the static and the shared core C conpiler,
>>  # with or without the target libgcc. We need to know wether:
>>  #  - we're building static, shared or bare metal: mode=[static|shared|baremetal]
>> -#  - we need to build libgcc or not             : build_libgcc=[yes|no]     (default: no)
>> -#  - we need to build libstdc++ or not          : build_libstdcxx=[yes|no]  (default: no)
>> -# Usage: do_cc_core_static mode=[static|shared|baremetal] build_libgcc=[yes|no]
>> +#  - we need to build libgcc or not             : build_libgcc=[yes|no]       (default: no)
>> +#  - we need to build libstdc++ or not          : build_libstdcxx=[yes|no]    (default: no)
>> +#  - we need to build statically linked or not  : build_staticlinked=[yes|no] (default: no)
>> +# Usage: do_cc_core mode=[static|shared|baremetal] build_libgcc=[yes|no] build_staticlinked=[yes|no]
>>  do_cc_core() {
>>      local mode
>>      local build_libgcc=no
>>      local build_libstdcxx=no
>> +    local build_staticlinked=no
>>      local core_prefix_dir
>>      local lang_opt
>>      local tmp
>>      local -a extra_config
>> -    local core_LDFLAGS
>> +    local -a core_LDFLAGS
>>      local -a core_targets
>>
>>      while [ $# -ne 0 ]; do
>> @@ -163,7 +174,12 @@
>>      # is needed. Shoe-horn it in our LDFLAGS
>>      # Ditto libm on some Fedora boxen
>>      if [ "${CT_COMPLIBS_SHARED}" != "y" ]; then
>> -        core_LDFLAGS='-lstdc++ -lm'
>> +        if [ "${build_staticlinked}" = "yes" ]; then
>> +            core_LDFLAGS+=("-static")
>> +            extra_config+=("--with-host-libstdcxx=-static-libgcc -Wl,-Bstatic,-lstdc++ -lm")
>> +        fi
>> +        core_LDFLAGS+=("-lstdc++")
>> +        core_LDFLAGS+=("-lm")
>
> Well, not your fault, but this part of the code is incomplete.
> I've fixed it as well.
>
> Regards,
> Yann E. MORIN.
>
> --
> .-----------------.--------------------.------------------.--------------------.
> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
> | +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
> '------------------------------^-------^------------------^--------------------'
>

I'd love to see your patch.

-- 
Bryan Hundven
bryanhundven@gmail.com

--
For unsubscribe information see http://sourceware.org/lists.html#faq