Skip to content

Races in cache of opened security databases [CORE6450] #6683

@firebird-automations

Description

@firebird-automations

Submitted by: @pavel-zotov

When new sec db is added to cache there is small timeframe when empty covering, already added to the list of known DBs, is not locked. An attempt to use it caused segfault when running tests on FB4 but same may happen for 3.0.6 & 3.0.7.

Commits: fb99f96 8b70d33 8f49cd7 41a8fc5

====== Test Details ======

Test can not be implemented because of negligible probability to reproduce this error.
Bug appeared first time at nov-2020 during regular OLTP-EMUL runs on Linux, but source code had it since jan-2020 (letter from Alex, 19.11.2020 11:10).

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions