Port-arm archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: Linksys NSLU2 (evbarm, armeb) userspace regression origin found
On 05/16/15 12:35, Eddy Petrișor wrote:
Hi again,
I have created the attached patch to fix the user space issue
introduced by the commit quoted in the mail below (basically adding
also ' TARGET_ENDIAN_DEFAULT=MASK_BIG_END' to tm_defines).
I think netbsd-elf.h should change - see diff. Can you test?
Thanks for finding the problem.
Using NetBSD 6.1.5 kernel and some more recent user spaces generated
from trunk (for instance, 'build', 'distribution', sets' obtained from
trunk code dated 2013-09-14), I get a lot of messages
'/etc/ld.so.conf: invalid/unknown sysctl for libc.so.12 (22)' after
starting init.
Any idea what this is? Is it due to the fact I am using NetBSD
6.1.5_PATCH kernel with newer userspace (etc/obj/etc-release says
'NetBSD 6.99.23/evbarm')?
Yeah, don't do that. It's not supported.
Backwards compatibility is supported (with correct options in the
kernel), but not forwards.
Nick
Index: external/gpl3/gcc/dist/gcc/config/arm/netbsd-elf.h
===================================================================
RCS file: /cvsroot/src/external/gpl3/gcc/dist/gcc/config/arm/netbsd-elf.h,v
retrieving revision 1.7
diff -u -p -r1.7 netbsd-elf.h
--- external/gpl3/gcc/dist/gcc/config/arm/netbsd-elf.h 1 Mar 2014 09:04:12 -0000 1.7
+++ external/gpl3/gcc/dist/gcc/config/arm/netbsd-elf.h 16 May 2015 12:20:31 -0000
@@ -30,9 +30,12 @@
(ARM_DEFAULT_ABI != ARM_ABI_APCS && ARM_DEFAULT_ABI != ARM_ABI_ATPCS \
? TARGET_CPU_arm926ejs : TARGET_CPU_arm6)
-/* This defaults us to little-endian. */
-#ifndef TARGET_ENDIAN_DEFAULT
-#define TARGET_ENDIAN_DEFAULT 0
+/* TARGET_BIG_ENDIAN_DEFAULT is set in
+ config.gcc for big endian configurations. */
+#if TARGET_BIG_ENDIAN_DEFAULT
+#define TARGET_ENDIAN_DEFAULT MASK_BIG_END
+#else
+#define TARGET_ENDIAN_DEFAULT 0
#endif
#undef MULTILIB_DEFAULTS
Home |
Main Index |
Thread Index |
Old Index