Mon May 19 20:24:54 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* documentation for ZipDataflow
M ./compiler/cmm/ZipDataflow.hs -53 +237
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20080520032454-39dff-472ad292ee0e176c1ca379c13e774774130b9457.gz
_
Mon May 12 15:40:02 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* added a test for eliminating dead assignment i++ in loop
A ./tests/ghc-regress/cps/cps028.cmm
View patch online:
http://darcs.haskell.org/testsuite/_darcs/patches/20080512224002-39dff-70abb93d1cf7a7f138ffb6650d7b5f041a122bd0.
Sat May 3 15:47:20 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* replace hints with kinds in parser as well
M ./compiler/cmm/CmmParse.y r4
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20080503224720-39dff-9a6141bad9c5f8d5dc925f8eb570d8833bb560d9.gz
__
Sat May 3 15:45:14 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* replace Cmm 'hint' with 'kind'
C-- no longer has 'hints'; to guide parameter passing, it
has 'kinds'. Renamed type constructor, data constructor, and record
fields accordingly
M ./compiler/cmm/CLabel.hs r4
M ./compi
Sat May 3 15:42:08 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* new version of ZipDataflow
This version combines forward/backard into a type class
(actually two classes) of analysis and transformation.
These type classes will always be expanded away at the client,
so SLPJ may wonder why
Sat May 3 15:34:52 PDT 2008 Norman Ramsey <[EMAIL PROTECTED]>
* minor changes to Cmm left over from September 2007
Nothing too deep here; primarily tinking with prettyprinting
and names. Also eliminated some warnings. This patch covers
most (but not all) of the code NR changed
Fri Sep 21 06:41:24 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* massive convulsion in ZipDataflow
After my talk, I got the idea of 'shallow rewriting' for the
dataflow framework. Here it is implemented, along with
some related ideas late making Graph and not LGraph primary.
The o
Fri Sep 21 06:40:35 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* incomplete start on set of intervals for stack model
A ./compiler/utils/Interval.hs
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Wed Sep 19 08:05:44 PDT 2007 [EMAIL PROTECTED]
* Small changes to mk-ing flow graphs
- ZipCfg: add mkBlockId :: Unique -> BlockId
- MkZipCfg: change sequence --> catAGrpahs
- MkZipCfgCmm: add mkCmmIfThen
Not fully validated, but I don't think anything will break
M ./comp
Wed Sep 19 08:03:45 PDT 2007 [EMAIL PROTECTED]
* Mostly comments, following NR/SPJ meeting
M ./compiler/cmm/CmmLiveZ.hs -1 +1
M ./compiler/cmm/ZipDataflow.hs -17 +44
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman
Mon Sep 17 10:29:39 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* added node to push a closure onto the current call context
M ./compiler/cmm/CmmCvt.hs -2 +3
M ./compiler/cmm/CmmLiveZ.hs +1
M ./compiler/cmm/CmmSpillReload.hs +1
M ./compiler/cmm/MkZipCfgCmm.hs -1 +5
M ./compil
Mon Sep 17 09:17:15 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* tightened some dataflow code as part of preparing a talk
M ./compiler/cmm/CmmSpillReload.hs -17 +28
M ./compiler/cmm/ZipDataflow.hs -2 +1
___
Cvs-ghc mailing list
Cvs-ghc@haskel
Mon Sep 17 09:16:34 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* added 'filterRegsUsed' to CmmExpr
M ./compiler/cmm/CmmExpr.hs -1 +6
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 14:54:14 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* reloads are now sunk as deep as possible
M ./compiler/cmm/CmmCPSZ.hs -5 +9
M ./compiler/cmm/CmmSpillReload.hs -3 +65
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.hask
Sat Sep 15 14:53:37 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* added instance declarations so we can fold over local registers used in
Middle and Last nodes
M ./compiler/cmm/ZipCfgCmmRep.hs +26
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http
Sat Sep 15 14:53:03 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* added monadic mapM_blocks. the fear, the fear...
M ./compiler/cmm/ZipCfg.hs -2 +12
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 14:52:38 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* fix misspelled constructor
M ./compiler/cmm/MkZipCfgCmm.hs -1 +1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 14:51:38 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* add another way to run in the fuel monad (this is a mess right now)
M ./compiler/cmm/DFMonad.hs +6
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listin
Sat Sep 15 13:18:49 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* lay ground for more readable dumping of CmmGraph
M ./compiler/main/DynFlags.hs +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:17:46 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* add a function to help identify unique predecessors
M ./compiler/cmm/CmmZipUtil.hs +22
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:17:19 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* scrub away remaining MidNop
M ./compiler/cmm/CmmCvt.hs -1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:16:51 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* changes needed to get map_blocks to actually compile :-(
M ./compiler/cmm/ZipCfg.hs -2 +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-gh
Sat Sep 15 13:15:25 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* remove an unwanted language extension
M ./compiler/cmm/ZipCfg.hs -1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:14:48 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* drop the old, redundant implementation of postorder_dfs
M ./compiler/cmm/ZipCfg.hs -33 +9
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-gh
Sat Sep 15 13:14:14 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* eliminate the last vestige of UniqSM from ZipCfg
M ./compiler/cmm/ZipCfg.hs -5 +6
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:13:55 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* add map_blocks to ZipCfg
M ./compiler/cmm/ZipCfg.hs +4
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 15 13:12:43 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* get rid of MidNop
M ./compiler/cmm/CmmLiveZ.hs -2 +1
M ./compiler/cmm/CmmSpillReload.hs -2 +1
M ./compiler/cmm/MkZipCfgCmm.hs -1 +1
M ./compiler/cmm/ZipCfgCmmRep.hs -5 +1
__
Sat Sep 15 13:10:30 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* get freshBlockId out of ZipCfg and bury it in MkZipCfg where it belongs
M ./compiler/cmm/Cmm.hs -2 +2
M ./compiler/cmm/DFMonad.hs -2 +2
M ./compiler/cmm/MkZipCfg.hs +9
M ./compiler/cmm/ZipCfg.hs -17 +8
M ./comp
Sat Sep 15 12:06:17 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* added foldUFM_Directly, used where appropriate, killed all warnings
M ./compiler/utils/UniqFM.lhs -45 +53
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman
Sat Sep 15 06:39:04 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* (temporarily) check consistency of new postorder DFS no matter what DEBUG is
M ./compiler/cmm/ZipCfg.hs -1 +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mai
Fri Sep 14 10:01:19 PDT 2007 [EMAIL PROTECTED]
* Wibbles to MkZipCfgCmm stuff
Watch out: I've added MkZipCfg.sequence, which clashes with the
Prelude (like some other names in MkZipCfg. Maybe you can think
of a better name for it.
M ./compiler/cmm/MkZipCfg.hs -2 +7
M ./comp
Fri Sep 14 07:54:35 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* make remove_entry_label actually work inside ZipDataflow
M ./compiler/cmm/ZipDataflow.hs -1 +7
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/c
Fri Sep 14 06:52:11 PDT 2007 [EMAIL PROTECTED]
* Remove export of remove_entry_label
M ./compiler/cmm/ZipCfg.hs -1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Fri Sep 14 07:54:17 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* export 'entry' from ZipCfg, at least for now
M ./compiler/cmm/ZipCfg.hs +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Fri Sep 14 06:42:02 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* replace #ifdef DEBUG with debugIsOn in cmm/MkZipCfg
M ./compiler/cmm/MkZipCfg.hs -5 +5
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Thu Sep 13 10:36:53 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* new signatures for splicing functions, new postorder_dfs
M ./compiler/cmm/ZipCfg.hs -22 +111
M ./compiler/cmm/ZipCfgExtras.hs -3 +6
M ./compiler/cmm/ZipDataflow.hs -81 +99
_
Thu Sep 13 10:35:12 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* tidying cmm/CmmSpillReload.hs
M ./compiler/cmm/CmmSpillReload.hs -16 +12
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Thu Sep 13 10:34:46 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* tidying cmm/CmmLiveZ.hs
M ./compiler/cmm/CmmLiveZ.hs -7 +6
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Tue Sep 11 14:45:18 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* a test to show that multiple calls correctly share a return address
to see the optimization at work, compile cps027.cmm with
-fconvert-to-zipper-and-back -frun-cps
A ./tests/ghc-regress/cps/cps027.cmm
__
Wed Sep 12 09:58:51 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* remove remaining redundancies from ZipCfgCmmRep
-- LastBranch no longer takes parameters
-- LastJump and LastReturn no longer carry CmmActuals;
instead, those are carried by a CopyOut in the same basic block
Wed Sep 12 08:38:52 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* change the zipper representation of calls
This patch combines two changes:
1. As requested by SimonPJ, the redundancy inherent in having
LastCall bear actual parameters has been removed. The actual
parameters a
Wed Sep 12 08:25:02 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* make it easier to have debugging code typechecked even when debugging is
turned off
M ./compiler/HsVersions.h +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org
Wed Sep 12 03:25:45 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* fix a typo!
M ./compiler/cmm/ZipCfgCmmRep.hs -1 +1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Wed Sep 12 03:25:26 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* cleaned up all warnings (and added many type signatures) in Outputable
M ./compiler/utils/Outputable.lhs -58 +87
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/
Wed Sep 12 02:39:20 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* overlooked ZipCfgExtras for a name change
M ./compiler/cmm/ZipCfgExtras.hs r1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Wed Sep 12 02:38:44 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* extra prettyprinting only when debugging
M ./compiler/cmm/ZipCfgCmmRep.hs -22 +31
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Tue Sep 11 15:55:42 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* renaming, reorganizing, and better doco for ZipCfg
M ./compiler/cmm/CmmCvt.hs r2
M ./compiler/cmm/CmmProcPointZ.hs r2
M ./compiler/cmm/PprCmmZ.hs r2
M ./compiler/cmm/ZipCfg.hs -110 +158 r2
M ./compiler/cmm/ZipD
Tue Sep 11 08:45:33 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* scrape some unused barnacles off of ZipCfg and put them into ZipCfgExtras
M ./compiler/cmm/ZipCfg.hs -63 +23
M ./compiler/cmm/ZipCfgCmmRep.hs -6 +1
A ./compiler/cmm/ZipCfgExtras.hs
Tue Sep 11 08:06:35 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* split the CmmGraph constructor interface from the representation
Interface MkZipCfgCmm should now be sufficient for all construction
needs, though some identifiers are re-exported from (and explained in)
MkZipCfg. ZipCfgCmmR
Tue Sep 11 07:29:14 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* correct two single-identifier bugs that stopped the Adams optimization from
working
M ./compiler/cmm/CmmProcPointZ.hs -2 +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.
Tue Sep 11 07:27:01 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* default ppr method for CmmGraph now tells more about the representation
(Previously, ppr had tried to make the zipper representation look as
much like the ListGraph representation as possible. This decision was
unhelpful for
Tue Sep 11 07:25:35 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* prettyprint 'hinted' things in a more readable way
M ./compiler/cmm/PprCmm.hs -1 +7
M ./compiler/cmm/ZipCfgCmm.hs -14 +20
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www
Tue Sep 11 07:24:10 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* check for unreachable code only with -DDEBUG
M ./compiler/cmm/MkZipCfg.hs +5
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Tue Sep 11 07:23:38 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* add a big diagnostic for failures in CmmCvt.toZgraph
M ./compiler/cmm/CmmCvt.hs -3 +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Sat Sep 8 08:51:41 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* implement the outOfLine primitive in MkZipCfg (proposed as mkBlock)
M ./compiler/cmm/MkZipCfg.hs -7 +13
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/l
Fri Sep 7 10:20:30 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* withUnique and mkBlock as requested by SLPJ (but only one is implemented)
M ./compiler/cmm/MkZipCfg.hs -3 +35
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mai
Fri Sep 7 10:08:43 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* no registers are available after a call
M ./compiler/cmm/CmmSpillReload.hs +1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Fri Sep 7 09:59:55 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* wrote an analysis to help in sinking Reload instructions
M ./compiler/cmm/CmmExpr.hs -2 +6
M ./compiler/cmm/CmmLiveZ.hs -1 +1
M ./compiler/cmm/CmmSpillReload.hs -11 +64
___
Fri Sep 7 09:12:46 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* a good deal of salutory renaming
I've renamed a number of type and data constructors within Cmm so that
the names used in the compiler may more closely reflect the C--
specification 2.1. I've done a bit of other renaming as
Fri Sep 7 06:24:42 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* refactor duplicated code in main/HscMain
I kept making mistakes because all the ZipCfg and CPS stuff
was called from two different places (compiling Haskell and
compiling Cmm). Now it is called from a single place, a
Fri Sep 7 06:24:17 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* in CmmExpr, always have (Show GlobalReg), regardless of DEBUG setting
M ./compiler/cmm/CmmExpr.hs -5 +1
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/l
Fri Sep 7 00:57:54 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* adding new files to do with new cmm functionality
A ./compiler/cmm/CmmCPSZ.hs
A ./compiler/cmm/CmmContFlowOpt.hs
A ./compiler/cmm/CmmCvt.hs
A ./compiler/cmm/CmmExpr.hs
A ./compiler/cmm/CmmLiveZ.hs
A ./compi
Thu Sep 6 09:19:48 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* massive changes to add a 'zipper' representation of C--
Changes too numerous to comment on, but here is some old history that
I saved:
Wed Aug 15 11:07:13 BST 2007 Norman Ramsey <[EMAIL PROTECTED]>
* type synon
Wed Sep 5 09:48:02 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* change of representation for GenCmm, GenCmmTop, CmmProc
The type parameter to a C-- procedure now represents a control-flow
graph, not a single instruction. The newtype ListGraph preserves the
current representation while e
Wed Sep 5 09:46:46 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* enable and slay warnings in cmm/Cmm.hs
M ./compiler/cmm/Cmm.hs -7 +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Mon Aug 20 12:54:41 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* put CmmReturnInfo into a CmmCall (and related types)
M ./compiler/cmm/Cmm.hs -1 +2 r3
M ./compiler/cmm/CmmBrokenBlock.hs -12 +13
M ./compiler/cmm/CmmCPS.hs -3 +3
M ./compiler/cmm/CmmCPSGen.hs -6 +8
M ./compiler
Mon Aug 20 09:41:05 PDT 2007 Norman Ramsey <[EMAIL PROTECTED]>
* annotate C-- calls that do not return
* The correct definition of C-- requires that a procedure not
'fall off the end'. The 'never returns' annotation tells us
if a (foreign) call is not going to return.
V
67 matches
Mail list logo