This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [bootstrap] Fix build for several targets (including pr58242)

On 3/09/2013, at 5:53 PM, Jakub Jelinek wrote:

> On Tue, Sep 03, 2013 at 09:25:31AM +0400, Alexander Ivchenko wrote:
>> Several builds are broken after r201838.
> What targets actually support bionic?  If it is just arm, i?86/x86_64
> and perhaps aarch64 and nothing else, I'd like to question the way where
> you enforce all the
>  # Add Android userspace support to Linux targets.
>  case $target in
>    *linux*)
>      tm_p_file="${tm_p_file} linux-protos.h"
>      tmake_file="${tmake_file} t-linux-android"
>      tm_file="$tm_file linux-android.h"
>      extra_options="$extra_options linux-android.opt"
>      extra_objs="$extra_objs linux-android.o"
>      ;;
>  esac
> stuff onto all Linux targets, without actually testing all of them.
> Wouldn't it be much better not to support Android in any way
> in config/linux.h (say at most #define TARGET_HAS_BIONIC 0 there,
> #define TARGET_LIBC_HAS_FUNCTION hook_bool*true
> #endif
> etc.), and append the android stuff only to targets which you
> support, not before the target specific snippets in config.gcc,
> but after those (so linux-android.h, etc. come last, not first)?


Nominal handling of Bionic by all *linux* targets is my fault, and it seemed like a good thing at the time.  With several iterations of improvements laid on top of this initial decision it now becoming a mess.  I'm thinking along similar lines as you to restrict Bionic/Android handling only to targets that actually support it.

As Alexander said in his email, this rework should be done as a separate patch.


Maxim Kuvyrkov

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]