This is the mail archive of the
mailing list for the GCC project.
Re: Question about sysroot and fixincludes
- From: Andrew Hsieh <andrewhsieh at google dot com>
- To: Alexander Ivchenko <aivchenk at gmail dot com>
- Cc: gcc <gcc at gcc dot gnu dot org>, Pavel Chupin <pavel dot v dot chupin at gmail dot com>
- Date: Thu, 17 Jul 2014 08:39:07 +0800
- Subject: Re: Question about sysroot and fixincludes
- Authentication-results: sourceware.org; auth=none
- References: <CACysShj6pMaq4LJ7ZQf6V7k+155oQ1R0SCN4poYp=YKFohg8Sg at mail dot gmail dot com>
Bionic headers prior to android-L (for L-preview) aren't changed
except for bug fixes since last major update in android-9 (gingerbread
era), the API level used to build all 32-bit NDK toolchains, so it
would be interesting to see how fixincluded differs should we use
different API level > 9.
Since JB, bionic is overhauled but not stable enough for NDK until
android-L. It's very possible that fixincluded for andorid-9 is wrong
for android-L. Do you have example? We can fix up bionic headers for
all levels in NDK to make fixincluded consistent if not gone
On Wed, Jul 16, 2014 at 8:08 PM, Alexander Ivchenko <firstname.lastname@example.org> wrote:
> Hi, I have a question about sysroot and fixincludes.
> On Android there are different API levels (like android-9, android-10
> etc) that match different versions of OS. Gcc from NDK is configured
> using sysroot for android-9 and the convenient way for compiling for,
> say, android-19 was by providing the sysroot to android-19 as a
> command line option (--sysroot).
> However, the header from the sysroot with which gcc was configured
> could be "fixincluded", and, when I provide a different sysroot as a
> command line option, "fixincluded" header could replace the actual
> header from the specified sysroot - that is the root-cause of certain
> Should search in 'include-fixed' be disabled when sysroot command line
> option is specified?