Quoting Oren Laadan (orenl@xxxxxxxxxxxxxxx): > > > Serge E. Hallyn wrote: > > Quoting Oren Laadan (orenl@xxxxxxxxxxxxxxx): > >> I rebased the most recent patchset to 2.6.33, and pushed to > >> ckpt-v19-rc2.10. only x86-32 is (lightly) tested. > >> > >> it does not include patches from 1/27 and on. working on it > >> now. > >> > >> please review: > >> > >> * arch-specific parts that I didn't screw up anything. > >> btw, I moved checkpoint code from mm/ to kernel/ > >> > >> * [dan] network: net/checkpoint.c now uses sock_alloc_file(); > >> there may or may not be a leak of sock_hold(). > >> > >> * [serge] selinux now auto-generates some include files, see > >> the changes to classmap.h > >> > >> Oren. > > > > For x86-64, I needed the following patch. With this patch, only > > the fileio and futex tests failed. Suka is looking into the fileio > > failure - Matt you might check the futex tests on whatever arch > > you have available... > > > > For s390, I of course still need > > "fix 31-bit s390 checkpoint/restart wrappers" for compile with > > CONFIG_COMPAT to succeed. > > Ok, fixed, and pushed to ckpt-v19-rc2.10. > If this is more or less ok, I'll change it to ckpt-v19-rc3 :) My git tree isn't happy with whatever rebasing you did. Can you push a ckpt-v19-rc2.11 tag or just ckpt-v19-rc3? I haven't looked at the selinux issue at all yet btw, and I'd like to hear if anyone has figured out the issue with a pipe being treated as Unlinked file in the robust futex testcase. Matt? -serge _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers
