Ideas Needed for GCC internal Specs
Hi
I have tried the following patch with gcc's master on the sparc/erc32 with
an
empty bsp_specs. It still is linking with crt0.o. I don't see where it is
coming from.
Do I have a subtle typo or is there something else I am missing?
Thanks
--joel
iff --git a/gcc/config/rtems.h b/gcc/config/rtems.h
index 439199d..2f592bb 100644
--- a/gcc/config/rtems.h
+++ b/gcc/config/rtems.h
@@ -30,11 +30,16 @@
* Dummy start/end specification to let linker work as
* needed by autoconf scripts using this compiler.
*/
+#if 0
#undef STARTFILE_SPEC
#define STARTFILE_SPEC "crt0.o%s"
+#endif
+
+#if 0
#undef ENDFILE_SPEC
#define ENDFILE_SPEC ""
+#endif
/*
* Some targets do not set up LIB_SPECS, override it, here.
diff --git a/gcc/config/sparc/rtemself.h b/gcc/config/sparc/rtemself.h
index 0fb56ee..1ad1d42 100644
--- a/gcc/config/sparc/rtemself.h
+++ b/gcc/config/sparc/rtemself.h
@@ -38,3 +38,6 @@
/* Use the default */
#undef LINK_GCC_C_SEQUENCE_SPEC
+
+#undef STARTFILE_SPEC
+#define STARTFILE_SPEC "%{!qrtems: crt0.o%s} crti.o%s crtbegin.o%s"
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
Re: Ideas Needed for GCC internal Specs
On 24/12/17 2:44 am, Joel Sherrill wrote: > > I have tried the following patch with gcc's master on the sparc/erc32 with an > empty bsp_specs. It still is linking with crt0.o. I don't see where it is > coming > from. > Do I have a subtle typo or is there something else I am missing? Is there a default in ld? Chris ___ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
Re: Ideas Needed for GCC internal Specs
On Dec 23, 2017 6:08 PM, "Chris Johns" wrote: On 24/12/17 2:44 am, Joel Sherrill wrote: > > I have tried the following patch with gcc's master on the sparc/erc32 with an > empty bsp_specs. It still is linking with crt0.o. I don't see where it is coming > from. > Do I have a subtle typo or is there something else I am missing? Is there a default in ld? I have no idea. Was going to see if I could remove just the end_file portion of our bsp_specs and then ask on GCC for some advice. I think I am.missing something magic. This may not be the final state of the solution but we have to empty the bsp_specs before we can move on. Chris ___ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
No preinstall and rtems.git master freeze
Hi, I have a branch called 'no-preinstall' in a personal repo: https://git.rtems.org/chrisj/rtems.git/?h=no-preinstall It builds a beagleboneblack without networking and not much else looking at the results of a tier-1 build: https://lists.rtems.org/pipermail/build/2017-December/000324.html It is now a process of working through the issues. A brief review shows a mix of source referencing a local header, ie '#include "cs8900.h"' and the header is now in a fixed part of the header API tree, or build related issues like the Xilinx Zynq's linkercmds file is not being found. The linkcmds files are now referenced in the source and not copied and renamed to the build tree which exposes issues in some BSPs. I will start to push changes onto my personal branch. Please post patches against the branch to this list and I will merge them. I need help or this will take a while to complete. I suggest we freeze the rtems.git master branch if this is possible until we get the no-preinstall changes merged. Should I post the changes for review or can the branch I listed above we OK for review purposes? Thanks Chris ___ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
