Ivo De Decker <iv...@debian.org> writes: >> > The failure that happens on the i386 buildd is this one: >> > >> > 1) Failure: >> > clone::nonetwork::local_absolute_path >> > [/«PKGBUILDDIR»/tests/clone/nonetwork.c:91] >> > Function call failed: (git_clone(&g_repo, local_src, "./foo", >> > &g_options)) >> > error -1 - git_path_direach callback returned -1 >> > >> > >> > I can reproduce this in my test environment on i386 and amd64. It only >> > happens >> > when the builddir and /tmp are on different filesystems. It seems the local >> > clone tries to create a hard link, which fails across filesystems (the fact >> > that this happens without fallback is an error in itself, so the test >> > actually >> > discovered a problem here). When setting the TMPDIR to a directory on the >> > same >> > filesystem, the test doesn't hit this issue, and the build works fine. >> > >> > It's unclear to me why this only happens on i386, but I suspect that the >> > setup >> > of the buildd chroots isn't the same everywhere. >> >> Yeah it struck me as odd. This same problem effects kfreebsd too so the >> change of the tmp dir will fix those builds as well. I'll have go at the >> running the clone nonetwork test with TMPDIR set to a mounted file >> system and report a bug upstream. >> >> > In any case, adding this patch fixes it in my environment. I can do an NMU >> > if >> > necessary. >> >> Thanks heaps for the patch, I only yesterday did exactly the same >> thing. It's included in the yet to be uploaded 0.21.3-1 that will also >> address the recent CVE. >> >> 0. >> http://anonscm.debian.org/cgit/users/arrsim-guest/libgit2.git/commit/?id=bd3a1fc82c5af703fe061fb22022eb48fb89be50 > > Does the tmp-test dir get cleaned up in the clean target?
So that is why to put the TMP_DIR in the target dir for each build. Cunning. I'll follow suite. > Also, please note that most of the changes in the latest uploads probably > don't comply with the freeze policy. If you think they do, you need to file an > unblock request explaining why (including a debdiff). > > https://release.debian.org/jessie/freeze_policy.html > > I guess #761170 and #761539 should be unmerged, as they are really different > issues. Bug #761170 should be downgraded back to important. Adding support for > new architectures isn't something that is allowed during the freeze. The only > real issue for the version in testing is #761539, which is about > clone::nonetwork::local_absolute_path, which can be fixed by setting the > tmp dir. Setting TMP_DIR will fix kfreebsd builds too. This can't be avoided unless we explicitly fail when building on those architectures. I'm happy to split the this into 2 bugs and 2 uploads. One revision to fix the TMP_DIR bug that is messing with the i386 build. And a new version that I'll get uploaded once we unfreeze. So is the best course of action: - Split bugs - Create new revision that only fixes the failing test on i386, it will also force and kfreebsd build to fail. -- Cheers, Russell -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
