On Mon, Jul 31, 2006 at 03:05:53AM +0200, Christian Aichinger wrote: > On Fri, Jul 21, 2006 at 12:18:37PM +0200, Christian Aichinger wrote: > > However I've now tried to build the resulting tree on mipsel and > > sparc, but the testsuite fails on both (failed: sparc: 3/23, mipsel: 5/23). > > > > I'll look into these when I have time. > > I finally did have time :) > > I've also looked at Eugeniy Meshcheryakov's patches from systemtap. > They clean up the bswap function that was throwing > warnings on mipsel/sparc. Included in the patch-set. > > As for the test-suite failures. > > * asm-tst[456] failed on sparc&mipsel. They simply eat up all memory > and eventually they segfault or get OOM killed by the kernel. > > The test code looks a bit suspicious at a glance because it uses > EM_386, independently of the current arch. > > * run-elflint-self.sh failed on both too. It runs elflint on the just > built elfutils binaries. This fails because the binaries generated > by gcc there aren't elflint-clean. > > * run-native-test.sh also fails on both. This is intentional, > because this test checks the backends for the current arch, > there's no mips/mipsel backend, and the sparc backend misses some > functions. > > So I've disabled asm-tst[456] and run-elflint-self.sh for now and > added a warning to run-native-test.sh that it's expected to fail on > various architectures. > > I expect we'll get usable binary packages for i386, ia64, > ppc, alpha, ia64 and probably s390 on upload. The others will fail > in run-native-test.sh. > > As for sparc, there is some backend code, just > sparc_return_value_location is missing, which run-native-test checks > for. So this could perhaps be fixable sometime in the future. > > IMHO the package is in good shape now, lintian and linda only report > missing manpages and piuparts doesn't find any problems. > > Do you have any objections to uploading the package (patches > attached)?
Did you see Eugeniy Meshcheryakov mail to the ITP bug about systemtap? He seems to have patches too, can you take a look at that? Kurt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
