RFC: first patch for getting arch-specific stuff out of asm-generic/barebox.lds.h

Robert P. J. Day rpjday at crashcourse.ca
Mon Nov 19 08:53:45 EST 2012

  looking for comments on this patch as simply the *first* tweak to
start cleaning arch-specific content out of that barebox.lds.h.
here's the incredibly simple change to handle just x86:

===== start =====

diff --git a/arch/x86/lib/barebox.lds.S b/arch/x86/lib/barebox.lds.S
index 8bd2a7b..500fa91 100644
--- a/arch/x86/lib/barebox.lds.S
+++ b/arch/x86/lib/barebox.lds.S
@@ -16,6 +16,7 @@

 #undef i386
+#include <mach/barebox.lds.h>
 #include <asm-generic/barebox.lds.h>

 OUTPUT_FORMAT("elf32-i386", "elf32-i386", "elf32-i386")
diff --git a/include/asm-generic/barebox.lds.h b/include/asm-generic/barebox.lds.h
index b2bd19e..6e09de1 100644
--- a/include/asm-generic/barebox.lds.h
+++ b/include/asm-generic/barebox.lds.h
@@ -4,7 +4,6 @@
 	defined CONFIG_ARCH_IMX51 || \
 	defined CONFIG_ARCH_IMX53 || \
 	defined CONFIG_ARCH_IMX6 || \
-	defined CONFIG_X86 || \
 	defined CONFIG_ARCH_EP93XX
 #include <mach/barebox.lds.h>

===== end =====

  of all the tests in the asm-generic file, the x86-related test is
unique in that it would end up including nothing but macro
definitions, so all the above does is remove that check from the
asm-generic file and explicitly put it into the x86 file -- it
*appears* that this has the same effect, and i tested it by building a
regular sandbox, which worked fine.  (does that represent a reasonable

  there's more that can be done, but the above at least gets rid of
one silly test in asm-generic/barebox.lds.h that doesn't need to be

  thoughts?  if it looks reasonable, i can submit it as an official



Robert P. J. Day                                 Ottawa, Ontario, CANADA

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday

More information about the barebox mailing list