Created attachment 876707 [details] archive container emerge -pav output on .56 and .51, emerge --info, Dockerfile I am unable to reproduce this in a fresh docker container, but i can reliably reproduce it on my host system. I get the feeling this is related to the selinux profile. When attempting to reproduce in a container, i did not run -uDN @world before attempting to run the same emerge command.
Created attachment 876708 [details] Debug level emerge logs
it is being investigated. we think it comes from this: features/selinux/package.use.force:12:sys-libs/libselinux python but theres some confusion as to whether that line makes sense or not still
Yeah, this might well be unavoidable - at a glance, the deps look legitimate. The only real question for me is whether some of it can be avoided temporarily given they're all binaries... But tenatively blocking bug 918929 just so we don't forget to look at it.
(all binaries + ROOT, i should say)
This is fallout from bug 918683, since we increased the priority of the runtime_slot_op in this cycle, so that it cannot be ignored: > * Error: circular dependencies: > > (dev-lang/python-3.11.6-1:3.11/3.11::gentoo, binary scheduled for merge to '/tmp/') depends on > (sys-apps/util-linux-2.38.1-r2-1:0/0::gentoo, binary scheduled for merge to '/tmp/') (runtime_slot_op) > (sys-libs/libselinux-3.5-r1-1:0/0::gentoo, binary scheduled for merge to '/tmp/') (runtime) > (dev-lang/python-3.11.6-1:3.11/3.11::gentoo, binary scheduled for merge to '/tmp/') (runtime) If these are all actually soname dependencies, then we'll need to temporarily disable a USE flag in order to break this cycle. If they are not all soname dependencies, then we may be able to adjust portage to temporarily ignore a lower priority dependency. Alternatively, maybe a lower priority dependency can be changed from RDEPEND to PDEPEND.
Since ROOT != /, temporarily ignoring an soname dependency could be an option as well.
Not sure if we should open a separate bug for this, but anyway, this is a similar cycle that I found which is in fact an soname dependency in both directions. It was the trigger for bug 919311, and it doesn't happen with portage-3.0.51: > * Error: circular dependencies: > > (dev-libs/gmp-6.3.0-2:0/10.4::gentoo, binary scheduled for merge to '/tmp/testroot/') depends on > (sys-devel/gcc-13.2.1_p20230826-2:13/13::gentoo, binary scheduled for merge to '/tmp/testroot/') (runtime) > (dev-libs/gmp-6.3.0-2:0/10.4::gentoo, binary scheduled for merge to '/tmp/testroot/') (runtime_slot_op)
(In reply to Zac Medico from comment #7) The gcc dependency is not specified in the gmp dependencies, so that cycle is only a problem with emerge --ignore-soname-deps=n.
We can add a DepPriority attribute to indicate when the parent ROOT != / and then use that to add additional DepPriority*Range ignore_priority levels that behave more like the behavior prior to bug 918683.
(In reply to Zac Medico from comment #9) > We can add a DepPriority attribute to indicate when the parent ROOT != / and > then use that to add additional DepPriority*Range ignore_priority levels > that behave more like the behavior prior to bug 918683. thanks zac, this is my preference for now, as I really want to avoid weakening it for the general case unless we have to I'm going to check each of the deps in the original cycle to see if they're all really soname deps too.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=1d856747ada48f8d32c033091b1156cc655efed3 commit 1d856747ada48f8d32c033091b1156cc655efed3 Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2023-12-06 06:05:46 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2023-12-06 20:23:14 +0000 DepPriority{Normal,Satisfied}Range: weaken _ignore_runtime for cross root When the dependency parent is for a cross root (ROOT != /) package, weaken _ignore_runtime in order to tolerate runtime cycles that are less problematic for cross root packages. The included test case fails with this error without the fix: * Error: circular dependencies: (dev-libs/gmp-6.3.0:0/10.4::test_repo, binary scheduled for merge to '/tmp/tmp25nwdjn7/cross_root/') depends on (sys-devel/gcc-13.2.1_p20230826:0/0::test_repo, binary scheduled for merge to '/tmp/tmp25nwdjn7/cross_root/') (runtime) (dev-libs/gmp-6.3.0:0/10.4::test_repo, binary scheduled for merge to '/tmp/tmp25nwdjn7/cross_root/') (runtime_slot_op) It might be possible to break this cycle by applying the following change: - dev-libs/gmp-6.3.0 (Change USE: -cxx) Bug: https://bugs.gentoo.org/919174 Signed-off-by: Zac Medico <zmedico@gentoo.org> lib/_emerge/DepPriority.py | 4 +- lib/_emerge/DepPriorityNormalRange.py | 4 +- lib/_emerge/DepPrioritySatisfiedRange.py | 1 + lib/_emerge/UnmergeDepPriority.py | 3 +- lib/_emerge/depgraph.py | 46 ++++-- lib/portage/tests/resolver/meson.build | 1 + .../tests/resolver/test_cross_dep_priority.py | 164 +++++++++++++++++++++ 7 files changed, 208 insertions(+), 15 deletions(-)
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0d365d80099d206e49b592abb30030642f8f09f9 commit 0d365d80099d206e49b592abb30030642f8f09f9 Author: Sam James <sam@gentoo.org> AuthorDate: 2023-12-10 22:34:47 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2023-12-10 22:34:52 +0000 sys-apps/portage: add 3.0.57 Closes: https://bugs.gentoo.org/918929 Closes: https://bugs.gentoo.org/913628 Closes: https://bugs.gentoo.org/915474 Closes: https://bugs.gentoo.org/918597 Closes: https://bugs.gentoo.org/919072 Closes: https://bugs.gentoo.org/919105 Closes: https://bugs.gentoo.org/919174 Closes: https://bugs.gentoo.org/919311 Closes: https://bugs.gentoo.org/919419 Closes: https://bugs.gentoo.org/919668 Signed-off-by: Sam James <sam@gentoo.org> sys-apps/portage/Manifest | 1 + sys-apps/portage/portage-3.0.57.ebuild | 242 +++++++++++++++++++++++++++++++++ 2 files changed, 243 insertions(+)
FWIW, this was also committed a few days ago: commit a0ff82d513b952a0cd7764ffe8958806499c981d Author: Andreas K. Hüttel <dilfridge@gentoo.org> Date: Tue Dec 5 11:52:56 2023 +0100 sys-apps/util-linux: Cut dependency loop in stage1 build Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
