The ABI hash changed for s390 (32-bit only) between ghc-7.8.4-41.fc22 and ghc-7.8.4-42.fc22 and then again between ghc-7.8.4-42.2.fc22 and ghc-7.8.4-43.fc22
Change between 42.2 and 43 is to base. For 41 and 42 at least bytestring changed.
the subpackage I checked is ghc-bytestring-devel ghc-bytestring-devel = 0.10.4.0-43.fc22 ghc-bytestring-devel(s390-32) = 0.10.4.0-43.fc22 ghc-devel(bytestring-0.10.4.0-a38eb69cabfbe89e8d8f729239293977) ghc-bytestring-devel = 0.10.4.0-42.2.fc22 ghc-bytestring-devel(s390-32) = 0.10.4.0-42.2.fc22 ghc-devel(bytestring-0.10.4.0-ed911e7019ca71e97666b106a852ff81) ghc-bytestring-devel = 0.10.4.0-42.fc22 ghc-bytestring-devel(s390-32) = 0.10.4.0-42.fc22 ghc-devel(bytestring-0.10.4.0-ed911e7019ca71e97666b106a852ff81) ghc-bytestring-devel = 0.10.4.0-41.fc22 ghc-bytestring-devel(s390-32) = 0.10.4.0-41.fc22 ghc-devel(bytestring-0.10.4.0-0b5b88348decbdcebbd54cb05aef5611)
Your "make -j1" build has same base ABI hash as -42.2. http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1777914 Whereas your "make -j4" rebuild has same base hash as -43. http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1777911 It looks to me as if -j4 is unstable wrt ABI hashes on s390 but perhaps the -43 hash happens more often. The -j4 build takes less than 2 hours, whereas the -j1 build takes almost 4 hours. Might be worth trying -j2 or -j3 too. My guess is that -j3 would be stable: somehow s390 is just unlucky with unstable -j4. I can do a -44 build soon which forces -j2 say for s390. Maybe another factor is how many packages have been built with -42.2 vs -43 say. Assuming more with -42.2 then better to use that hash I guess for F22.
Dan did -j2 and -j3, to compare with -j4. http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1780836 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1780842
$ s390-koji rpminfo --buildroots ghc-compiler-7.8.4-41.fc22.s390| grep Used\ in Used in 256 buildroots: $ s390-koji rpminfo --buildroots ghc-compiler-7.8.4-42.fc22.s390| grep Used\ in Used in 21 buildroots: $ s390-koji rpminfo --buildroots ghc-compiler-7.8.4-42.2.fc22.s390| grep Used\ in Used in 1 buildroots: $ s390-koji rpminfo --buildroots ghc-compiler-7.8.4-43.fc22.s390| grep Used\ in Used in 0 buildroots:
Here is a summary of the base and bytestring hashes: ghc(base-4.7.0.2-df338830bd5d424971e14b689f5c72dd) ghc(bytestring-0.10.4.0-ed911e7019ca71e97666b106a852ff81) - ghc-7.8.4-42.fc22 - ghc-7.8.4-42.2.fc22 -j2 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1780836 [5:07] -j3 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1780842 [4:20] ghc(base-4.7.0.2-f17d569edb6d8cdf3c82780f71c69a57) ghc(bytestring-0.10.4.0-a38eb69cabfbe89e8d8f729239293977) - ghc-7.8.4-43.fc22 -j4 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1777911 ghc(base-4.7.0.2-df338830bd5d424971e14b689f5c72dd) ghc(bytestring-0.10.4.0-a38eb69cabfbe89e8d8f729239293977) -j1 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1777914 ghc(base-4.7.0.2-df338830bd5d424971e14b689f5c72dd) ghc(bytestring-0.10.4.0-0b5b88348decbdcebbd54cb05aef5611) - ghc-7.8.4-41.fc22 So the -41 (used a lot) ABI hashes seems hard to reproduce. I think we should chose one of -j1, -j2 or -j3 and do another test build to increase our confidence that their hashes are stable.
another 2 scratch builds with -j2: http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1782219 http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1782371
(In reply to Dan Horák from comment #7) > another 2 scratch builds with -j2: > http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1782219 > http://s390.koji.fedoraproject.org/koji/taskinfo?taskID=1782371 They look good to me: both s390 and s390x. http://pkgs.fedoraproject.org/cgit/ghc.git/commit/?id=8b0348fcf1f55252a90773196a1c9b513927f9d0
ghc-7.8.4-44.fc23 has the change to -j2 for s390. I will build it soon for f22 too.
ghc-7.8.4-44.fc22, ghc-srpm-macros-1.4.2-1.fc22 has been submitted as an update for Fedora 22. https://admin.fedoraproject.org/updates/ghc-7.8.4-44.fc22,ghc-srpm-macros-1.4.2-1.fc22
Package ghc-7.8.4-44.fc22, ghc-srpm-macros-1.4.2-1.fc22: * should fix your issue, * was pushed to the Fedora 22 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing ghc-7.8.4-44.fc22 ghc-srpm-macros-1.4.2-1.fc22' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/FEDORA-2015-7395/ghc-7.8.4-44.fc22,ghc-srpm-macros-1.4.2-1.fc22 then log in and leave karma (feedback).
ghc-7.8.4-44.fc22, ghc-srpm-macros-1.4.2-1.fc22 has been pushed to the Fedora 22 stable repository. If problems still persist, please make note of it in this bug report.