Created attachment 587812 [details] debug log > WARNING: One or more updates/rebuilds have been skipped due to a dependency conflict: > > dev-db/mysql-connector-c:0 > > (dev-db/mysql-connector-c-8.0.17-r2:0/21::gentoo, ebuild scheduled for merge) conflicts with > dev-db/mysql-connector-c:0/18[abi_x86_64(-)] required by (virtual/libmysqlclient-18-r1:0/18::gentoo, installed) > ^^^^^ > > > !!! The following update(s) have been skipped due to unsatisfied dependencies > !!! triggered by backtracking: > > virtual/libmysqlclient:0
Created attachment 587966 [details] debug log for successful calculation with similar configuration I've attempted to reproduce the problem, but the calculation succeeds like this: > # emerge -pvuD @world > > These are the packages that would be merged, in order: > > Calculating dependencies... done! > [ebuild U ~] dev-db/mysql-connector-c-8.0.17-r3:0/21::gentoo [6.1.11-r2:0/18::gentoo] USE="-ldap% -libressl -static-libs" ABI_X86="(64) -32 (-x32)" 184885 KiB > [ebuild r U ~] virtual/libmysqlclient-21:0/21::gentoo [18-r1:0/18::gentoo] USE="-static-libs" ABI_X86="(64) -32 (-x32)" 0 KiB > [ebuild rR ] dev-perl/DBD-mysql-4.44.0::gentoo USE="ssl -test" 0 KiB > [ebuild U ~] dev-db/percona-server-5.7.26.29-r1:0/18::gentoo [5.7.26.29:0/18::gentoo] USE="pam perl server -cjk (-client-libs) -cracklib -debug -experimental -jemalloc -latin1 -libressl -numa -profiling -rocksdb (-selinux) -static -static-libs -systemtap -tcmalloc -test -tokudb -tokudb-backup-plugin -yassl" ABI_X86="(64) -32 (-x32)" 0 KiB > [blocks b ] <dev-db/percona-server-5.7.26.29-r1 ("<dev-db/percona-server-5.7.26.29-r1" is blocking dev-db/mysql-connector-c-8.0.17-r3) > > Total: 4 packages (3 upgrades, 1 reinstall), Size of downloads: 184885 KiB > Conflict: 1 block > > The following packages are causing rebuilds: > > (virtual/libmysqlclient-21:0/21::gentoo, ebuild scheduled for merge) causes rebuilds for: > (dev-perl/DBD-mysql-4.44.0:0/0::gentoo, ebuild scheduled for merge)
I can reproduce the problem after I removed virtual/libmysqlclient from my world file.
This slot conflict occurs: > backtracking due to slot conflict: > first package: (dev-db/mysql-connector-c-6.1.11-r2:0/18::gentoo, installed) > package to mask: (dev-db/mysql-connector-c-8.0.17-r3:0/21::gentoo, ebuild scheduled for merge) > slot: dev-db/mysql-connector-c:0 > parents: ((dev-db/percona-server-5.7.26.29-r1:0/18::gentoo, ebuild scheduled for merge), 'dev-db/mysql-connector-c[abi_x86_64(-)]'), ((virtual/libmysqlclient-18-r1:0/18::gentoo, installed), 'dev-db/mysql-connector-c:0/18[abi_x86_64(-)]'), ((dev-db/mysql-init-scripts-2.3:0/0::gentoo, installed), 'dev-db/mysql-connector-c') Since all of the parent atoms match dev-db/mysql-connector-c-6.1.11-r2:0/18, it masks dev-db/mysql-connector-c-8.0.17-r3:0/21 and prevents the update. It doesn't recognize that it should trigger a rebuild because none of the parent atoms are := slot operators at this point in the calculation. The slot conflict is triggered solely by the dev-db/mysql-connector-c:0/18[abi_x86_64(-)] atom.
In fact there are no dev-db/mysql-connector-c:= dependencies, the closest thing is the virtual/libmysqlclient:= dependency of dev-perl/DBD-mysql.
Created attachment 587974 [details, diff] patch which causes some unit test to fail due to slot conflicts that would be solved by downgrades This patch currently causes the following unit test to fail due to slot conflicts that would be solved by downgrades: > ====================================================================== > FAIL: testSlotOperatorCompleteGraph > (portage.tests.resolver.test_slot_operator_complete_graph.SlotOperatorCompleteGraphTestCase) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "lib/portage/tests/__init__.py, line 211, in run > testMethod() > File "lib/portage/tests/resolver/test_slot_operator_complete_graph.py", line 137, in testSlotOperatorCompleteGraph > test_case.fail_msg) > AssertionError: False != True : atoms: (=app-misc/meta-pkg-2, app-misc/C), key: success, expected: True, got: False > atoms: (=app-misc/meta-pkg-2, app-misc/C), key: slot_collision_solutions, expected: None, got: [] > > ====================================================================== > FAIL: testSlotOperatorRuntimePkgMask > (portage.tests.resolver.test_slot_operator_runtime_pkg_mask.SlotOperatorRuntimePkgMaskTestCase) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "lib/portage/tests/__init__.py", line 211, in run > testMethod() > File "lib/portage/tests/resolver/test_slot_operator_runtime_pkg_mask.py", line 132, in testSlotOperatorRuntimePkgMask > test_case.fail_msg) > AssertionError: False != True : atoms: (=app-misc/meta-pkg-2), key: success, expected: True, got: False > atoms: (=app-misc/meta-pkg-2), key: slot_collision_solutions, expected: None, got: [] > > ====================================================================== > FAIL: testSonameSkipUpdate (portage.tests.resolver.soname.test_skip_update.SonameSkipUpdateTestCase) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "lib/portage/tests/__init__.py", line 211, in run > testMethod() > File "lib/portage/tests/resolver/soname/test_skip_update.py", line 82, in testSonameSkipUpdate > test_case.fail_msg) > AssertionError: False != True : atoms: (@world), key: success, expected: True, got: False > atoms: (@world), key: mergelist, expected: [], got: ['[binary]dev-libs/B-1', '[binary]dev-libs/B-2', '[binary]app-misc/A-1'] > atoms: (@world), key: slot_collision_solutions, expected: None, got: []
This is similar to bug 531656 in the sense that we need to be more aggressive about trying to prefer updates in the event of a slot conflict.
This unit test reproduces the issue: lib/portage/tests/resolver/test_slot_conflict_update_virt.py https://github.com/zmedico/portage/compare/master..bug_692746_missed_mysql-connector-c_update With the patch for this bug, the debug log for the slot conflict with impossible upgrade in lib/portage/tests/resolver/soname/test_skip_update.py shows that the backtracker masks installed instances and corresponding binary packages, and never considers unmasking them: backtracking try 2 runtime_pkg_mask: {<Package ('binary', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'merge', 'None', '208', '0', '1566696256')>: {'slot conflict': set()}} backtracking try 3 runtime_pkg_mask: {<Package ('binary', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'merge', 'None', '208', '0', '1566696256')>: {'slot conflict': set()}, <Package ('installed', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'nomerge', 'installed')>: {'slot conflict': set()}} backtracking try 4 runtime_pkg_mask: {<Package ('binary', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'merge', 'None', '208', '0', '1566696256')>: {'slot conflict': set()}, <Package ('installed', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'nomerge', 'installed')>: {'slot conflict': set()}, <Package ('binary', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'merge', 'None', '265', '0', '1566696256')>: {'missing dependency': {(<Package ('binary', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'merge', 'None', '265', '0', '1566696256')>, '/tmp/tmpwsp8hrj_/', SonameAtom('x86_32', 'libB.so.1'))}}} backtracking try 5 runtime_pkg_mask: {<Package ('binary', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'merge', 'None', '208', '0', '1566696256')>: {'slot conflict': set()}, <Package ('installed', '/tmp/tmpwsp8hrj_/', 'dev-libs/B-1', 'nomerge', 'installed')>: {'slot conflict': set()}, <Package ('binary', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'merge', 'None', '265', '0', '1566696256')>: {'missing dependency': {(<Package ('binary', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'merge', 'None', '265', '0', '1566696256')>, '/tmp/tmpwsp8hrj_/', SonameAtom('x86_32', 'libB.so.1'))}}, <Package ('installed', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'nomerge', 'installed')>: {'missing dependency': {(<Package ('installed', '/tmp/tmpwsp8hrj_/', 'app-misc/A-1', 'nomerge', 'installed')>, '/tmp/tmpwsp8hrj_/', SonameAtom('x86_32', 'libB.so.1'))}}} backtracking aborted after 5 tries
Patch posted for review: https://archives.gentoo.org/gentoo-portage-dev/message/4d6c9814135359fd47e28240ac2b73b6 https://github.com/gentoo/portage/pull/455
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=994ac00aa764615ec6d319c7c1cb8123cf9f2aa1 commit 994ac00aa764615ec6d319c7c1cb8123cf9f2aa1 Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2019-08-24 22:30:29 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2019-09-01 17:33:54 +0000 _slot_confict_backtrack: consider masking a package matched by all parent atoms (bug 692746) When a slot conflict occurs involving a package that is matched by all involved parent atoms, consider masking the package in order to avoid a possible missed update. The included unit test demonstrates the case fixed by this patch. There are 2 previously existing unit tests that require larger backtracking values in order to succeed with this patch, since more possible solutions are now considered. Bug: https://bugs.gentoo.org/692746 Signed-off-by: Zac Medico <zmedico@gentoo.org> lib/_emerge/depgraph.py | 5 ++ lib/_emerge/resolver/backtracking.py | 9 +++ .../resolver/test_slot_conflict_update_virt.py | 79 ++++++++++++++++++++++ .../resolver/test_slot_operator_complete_graph.py | 2 +- .../test_slot_operator_runtime_pkg_mask.py | 2 +- 5 files changed, 95 insertions(+), 2 deletions(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=fc1ee19f135a3446d16deac8f783472b464e3a2e commit fc1ee19f135a3446d16deac8f783472b464e3a2e Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2019-09-01 21:12:24 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2019-09-01 21:20:36 +0000 sys-apps/portage: Bump to version 2.3.75 #235970 glsa-check: add --reverse option #692134 glsa-check: fix truncated CVE ids in listmode #692746 missed virtual/libmysqlclient update #693242 premature backtracking termination Bug: https://bugs.gentoo.org/691278 Bug: https://bugs.gentoo.org/235970 Bug: https://bugs.gentoo.org/692134 Bug: https://bugs.gentoo.org/692746 Bug: https://bugs.gentoo.org/693242 Package-Manager: Portage-2.3.75, Repoman-2.3.17 Signed-off-by: Zac Medico <zmedico@gentoo.org> sys-apps/portage/Manifest | 1 + sys-apps/portage/portage-2.3.75.ebuild | 261 +++++++++++++++++++++++++++++++++ 2 files changed, 262 insertions(+)
