Opened 5 years ago

Closed 5 years ago

#7046 closed bug (fixed)

Simplifier failing to unbox join point

Reported by: simonpj Owned by: simonpj
Priority: high Milestone: 7.8.1
Component: Compiler Version:
Keywords: Cc:
Operating System: Unknown/Multiple Architecture: Unknown/Multiple
Type of failure: None/Unknown Test Case: perf/should_run/T5113
Blocked By: Blocking:
Related Tickets: Differential Rev(s):
Wiki Page:


A consequence of the recent "silent superclass" patch is that T5113 performance test is failing badly:

=====> T5113(normal) 2794 of 3389 [0, 786, 0] cd ./perf/should_run && '/64playpen/simonmar/nightly/HEAD-cam-04-unx/x86_64-unknown-linux/inplace/bin/ghc-stage2' -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-db -rtsopts  -fno-ghci-history -o T5113 T5113.hs  -O  >T5113.comp.stderr 2>&1
cd ./perf/should_run && ./T5113  +RTS -V0 -tT5113.stats --machine-readable -RTS   </dev/null > 2>
bytes allocated 822112880 is more than maximum allowed 9000000
*** unexpected failure for T5113(normal)

My preliminary diagnosis is that it was only working nicely before by a fluke, and there's a problem here we must look at. In effect, because of a join point we are getting boxing and unboxing in an inner loop.

Change History (6)

comment:1 Changed 5 years ago by igloo


comment:2 Changed 5 years ago by simonmar

Simon - the test case for this is now passing (perf/should_run/T5113). Is that because the bug is fixed, or by accident? In the meantime I'll comment out the expect_broken for this test.

comment:3 Changed 5 years ago by igloo

It's failing again now. Possibly related to the cabal tests, that were failing due to forcing DynFlags too much, but seem to have stopped failing again?

comment:4 Changed 5 years ago by simonpj

Owner: set to simonpj

I think I know what is going on here.

comment:5 Changed 5 years ago by simonmar


comment:6 Changed 5 years ago by simonpj

Resolution: fixed
Status: newclosed

Fixed: see #5113

Note: See TracTickets for help on using tickets.