tags 645822 +patch
thanks
When discussing the sparc bug* Eric Botcazou from gcc upstream claimed
...you should never configure the compiler with --disable-checking, even in a
cross configuration. This will save a few percents in compilation times but
disables critical internal checking; you risk silently generating wrong code.
As a matter of fact, remove this switch is a workaround for the problem.
The attatched patch removes this switch from debian/rules I have tested this
on mips and mipsel in qemu (unfortunately debian sparc won't run in qemu anymore
so I can't try that) and they are still building gcc-avr over 8 hours later
(which is far longer than it takes the unmodfied version to fail) so it
seems they are fixed by this change.
Combining my qemu tests with upstreams assertion I belive this patch will
make the package build on mips, mipsel and sparc and hence make it able to
transition to testing.
* I do not know for sure whether the sparc and mipsel failures have the same
underlying cause but they seem suspiciously similar to me.
diff -ur gcc-avr-4.5.3/debian/rules gcc-avr-4.5.3.new/debian/rules
--- gcc-avr-4.5.3/debian/rules 2011-07-10 09:36:45.000000000 +0000
+++ gcc-avr-4.5.3.new/debian/rules 2011-11-18 09:26:24.000000000 +0000
@@ -23,7 +23,6 @@
--enable-long-long \
--enable-nls \
--without-included-gettext \
- --disable-checking \
--disable-libssp \
--build=$(DEB_BUILD_GNU_TYPE) \
--host=$(DEB_HOST_GNU_TYPE) \