summaryrefslogtreecommitdiffstats
path: root/package/gnuconfig/patches/config.sub.sh.patch
diff options
context:
space:
mode:
Diffstat (limited to 'package/gnuconfig/patches/config.sub.sh.patch')
-rw-r--r--package/gnuconfig/patches/config.sub.sh.patch25
1 files changed, 25 insertions, 0 deletions
diff --git a/package/gnuconfig/patches/config.sub.sh.patch b/package/gnuconfig/patches/config.sub.sh.patch
new file mode 100644
index 000000000..c1ccf1440
--- /dev/null
+++ b/package/gnuconfig/patches/config.sub.sh.patch
@@ -0,0 +1,25 @@
+gnuconfig seems intent on only supporting superfluous targets that don't
+represent any real hardware (such as sh2a LE). GCC in its infinite wisdom is
+also incapable of supporting tuning for other variants in a sensible fashion.
+
+As such, we still need to be able to support such useful targets as
+sh2a_nofpueb in order to not only get the byte order right, but also to
+support -m2a-nofpu tuning from the kernel (though we tend to do this through
+-Wa,-isa= instead, as the binutils people do a much better job of not screwing
+up their config code every other day. Way to go config.gcc..).
+
+The fact that the sh variant matching is the ugliest out of any of the other
+architectures doesn't seem to deter GCC folk from their well thought out and
+brilliantly managed config target list.
+
+--- config.sub 2005-11-21 21:18:16.000000000 +0200
++++ config.sub 2005-11-21 21:39:48.000000000 +0200
+@@ -264,7 +264,7 @@
+ | pdp10 | pdp11 | pj | pjl \
+ | powerpc | powerpc64 | powerpc64le | powerpcle | ppcbe \
+ | pyramid \
+- | sh | sh[1234] | sh[24]a | sh[23]e | sh[34]eb | shbe | shle | sh[1234]le | sh3ele \
++ | sh | sh[1234] | sh[24]a | sh[24]a*eb | sh[23]e | sh[34]eb | shbe | shle | sh[1234]le | sh3ele \
+ | sh64 | sh64le \
+ | sparc | sparc64 | sparc64b | sparc86x | sparclet | sparclite \
+ | sparcv8 | sparcv9 | sparcv9b \