#15544: Non-deterministic segmentation fault in cryptohash-sha256 testsuite -------------------------------------+------------------------------------- Reporter: bgamari | Owner: (none) Type: bug | Status: new Priority: highest | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by simonmar): Aha, yes I think you've found a problem. The run queue is doubly-linked, and the `block_info` field of the TSO is used as the back link, but we're overwriting that pointer in `unpark_tso`. I'd forgotten about the double use of `block_info` when I wrote that patch. I don't know if this causes the original problem, there might still be a SRT problem, but this queue corruption is definitely a bug. I guess we shouldn't touch the `block_info` field here. Would you like to make a patch? Unfortunately that reintroduces the problem of how to avoid repeated wakeup messages. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15544#comment:18 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler