[GHC] #10062: Codegen on sequential FFI calls is not very good

[GHC]

#10062: Codegen on sequential FFI calls is not very good
-------------------------------------+-------------------------------------
        Reporter:  chadaustin        |                   Owner:
            Type:  bug               |                  Status:  new
        Priority:  normal            |               Milestone:
       Component:  Compiler          |                 Version:  7.8.3
  (CodeGen)                          |                Keywords:
      Resolution:                    |            Architecture:
Operating System:  Unknown/Multiple  |  Unknown/Multiple
 Type of failure:  Runtime           |               Test Case:
  performance bug                    |                Blocking:
      Blocked By:                    |  Differential Revisions:
 Related Tickets:                    |
-------------------------------------+-------------------------------------

Comment (by Yuras):

 > I don't know why rsp is being changed so much.

 It is necessary because `ccall` calling convention requires alignment that
 is different from the one used in `ghc`, see
 [[https://github.com/ghc/ghc/blob/64678e9e8ff0107cac956f0c7b799a1dd317b963/compiler/nativeGen/X86/CodeGen.hs#L2320|here]]
 for details. Though all this `addq`/`subq` probably can be combined...
 IIRC we don't have optimization pass for asm (?)

 > I also can't explain the assignment to eax before the call

 It is required by calling convention
 [[https://github.com/ghc/ghc/blob/64678e9e8ff0107cac956f0c7b799a1dd317b963/compiler/nativeGen/X86/CodeGen.hs#L2357|too]].

 So, the generated code is pretty reasonable, but there definitely is space
 for improvements.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10062#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler