[OE-core] [PATCH 3/4] gcc-runtime: Disable gcc version of libssp

[Khem Raj]

On Tue, Apr 17, 2018 at 3:42 AM, Richard Purdie
<richard.purdie at linuxfoundation.org> wrote:
> On Sun, 2018-04-15 at 23:50 -0700, Khem Raj wrote:
>> diff --git a/meta/recipes-devtools/gcc/gcc-runtime.inc
>> b/meta/recipes-devtools/gcc/gcc-runtime.inc
>> index 97365a21dc..b4e5539575 100644
>> --- a/meta/recipes-devtools/gcc/gcc-runtime.inc
>> +++ b/meta/recipes-devtools/gcc/gcc-runtime.inc
>> @@ -24,8 +24,10 @@ RUNTIMELIBITM_nios2 = ""
>>  RUNTIMELIBITM_microblaze = ""
>>  RUNTIMELIBITM_riscv32 = ""
>>  RUNTIMELIBITM_riscv64 = ""
>> +RUNTIMELIBSSP ?= ""
>> +RUNTIMELIBSSP_mingw ?= "libssp"
>>
>> -RUNTIMETARGET = "libssp libstdc++-v3 libgomp libatomic
>> ${RUNTIMELIBITM} \
>> +RUNTIMETARGET = "libstdc++-v3 libgomp libatomic ${RUNTIMELIBITM} \
>>      ${@bb.utils.contains_any('FORTRAN', [',fortran',',f77'],
>> 'libquadmath', '', d)} \
>>  "
>>  RUNTIMETARGET_append_x86 = " libmpx"
>
> This last piece looks like you removed libssp for everywhere, glibc
> included except mingw?
>

yes, since the needed bits for ssp are already provided via libc.so there should
not be a need for it needing libssp on glibc based systems, I think
its due to the libssp
patch we are carrying in gcc, I will take a look at that and limit it
to just musl tuples

> Is the source file you included in the new recipe the same as the one
> in the gcc source but linked differently? If so, I'd be tempted to
> build that gcc-runtime with a musl specific do_compile addition...
>
> I assume this is post 2.5 material as its fairly invasive changes at
> the tail end of a release...
>

yes this is for 2.6 we may backport it later if needed.

> Cheers,
>
> Richard