#10676: silly assembly for comparing the result of comparisons that return Int# against 0# -------------------------------------+------------------------------------- Reporter: rwbarton | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.1 (CodeGen) | Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: x86_64 Type of failure: Runtime | (amd64) performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by rwbarton): Yes, I think so. But there is a problem besides the fact that the heap check is duplicated. Referring to the Cmm listing in #8326, something seems to think that `_sEV` (the 0#-or-1# result of the primop) is live in the alternatives, even to the point of passing it through the GC, when really it will never be used after the branch. I think that's the reason that later passes are unable to avoid this ugly `cmp/setCC/movzbl/test/jne` sequence. Otherwise the definition of `_sEV` would be inlined at its only use site, the branch, and the code generator would then generate a nice `cmp/jCC` sequence. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10676#comment:3 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler