[GHC] #14880: GHC panic: updateRole

[GHC]

#14880: GHC panic: updateRole
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  goldfire
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler (Type    |              Version:  8.2.2
  checker)                           |
      Resolution:                    |             Keywords:  TypeInType
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  crash or panic                     |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #15076            |  Differential Rev(s):  Phab:D4769
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by tdammers):

 Drilling down some more, I found that the biggest performance sink is
 `tcPolyInfer`, in both the pre- and post-patch versions; in both cases, it
 is responsible for just over 50% of allocations, but in terms of runtime,
 it's 36% before the patch, but 45% after - so that alone would explain a
 5% performance deviation.

 Further data points:

 - On the first run after making a fresh build,
 `withCleanupSession.cleanup` shows up prominently, but from the second run
 onwards, it scores 0 on all metrics.
 - Overall performance for profile builds is such that the patch does
 actually produce a performance improvement, from approx. 1.25 seconds down
 to 1.07.
 - The patch does not touch `tcPolyInfer` at all, the entire module is
 identical between both versions
 - The performance counters suggests that code paths taken inside
 `tcPolyInfer` are the same

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14880#comment:40>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler