RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-28 14:32:51 +01:00
Code Issues Projects Releases Wiki Activity
2,168 Commits 4 Branches 5 Tags 704 MiB
8f81cd7da6
Commit Graph

1262 Commits

Author SHA1 Message Date
Chris Lattner
ff6ec9b552 Enable better debug output. When debugging the type system, print out the 
type tables after reading a bytecode file to make sure they are ok

llvm-svn: 2126
 2002-04-07 06:11:22 +00:00
Chris Lattner
35c1a4347b * Reenable apparently neccesary code, that breaks testcases when enabled, 
but causes MANY FUNDAMENTAL PROBLEMS when not enabled. :(
* Add debugging function

llvm-svn: 2125
 2002-04-06 00:21:11 +00:00
Chris Lattner
83b6eecc84 * FIX: test/Regression/Assembler/2002-02-19-TypeParsing.ll 
That was due to the StructType object not getting called multiple times
   to remove itself from the ATU list for an object, because it stopped
   getting queried due to this wrong test

* Improve debug output when DEBUG_MERGE_TYPES is on

llvm-svn: 2124
 2002-04-05 23:02:46 +00:00
Chris Lattner
68857bb2bf * Move DerivedType::addAbstractTypeUser from DerivedType.h 
* Refactor a bunch of code.  Types now only remove one subelement at a time
  when they are told that they have changed
* Improve debugging output, add more assertions...
* FIX Bugs:
  * test/Regression/Assembler/2002-04-04-PureVirtMethCall.ll
  * test/Regression/Assembler/2002-04-04-PureVirtMethCall2.ll
  * The change to fix these bugs was the to ValTypeBase::refineAbstractType
    method.  Basically we #if 0'd out the chunk of code there and make the
    table get reorganized EVEN IF the type has the same pointer.  Merging
    opportunities were being missed!

llvm-svn: 2121
 2002-04-05 22:25:26 +00:00
Chris Lattner
3b80438423 Clean up debugging code 
llvm-svn: 2119
 2002-04-05 19:53:06 +00:00
Chris Lattner
e105f97672 * Refactor loop in removeAbstractTypeUser to be more clear 
* Fix a bug in DerivedType::refineAbstractTypeTo that caused
  test/Regression/Assembler/2002-04-05-TypeParsing.ll to fail.
  The added comment describes the fix

llvm-svn: 2118
 2002-04-05 19:44:07 +00:00
Chris Lattner
8afff0c29a s/MethodType/FunctionType 
llvm-svn: 2115
 2002-04-04 22:19:18 +00:00
Chris Lattner
39d7d981ed Make build work in release mode 
llvm-svn: 2113
 2002-04-04 19:35:24 +00:00
Chris Lattner
ffc9ef8be9 Make the release build work 
llvm-svn: 2109
 2002-04-04 19:26:26 +00:00
Chris Lattner
87cabdd757 * Add debug support for PATypeHolder 
* PATypeHolder is now not a template

llvm-svn: 2108
 2002-04-04 19:26:02 +00:00
Chris Lattner
a0a6c8d4a2 Make the release build compile 
llvm-svn: 2107
 2002-04-04 19:24:48 +00:00
Chris Lattner
1ee6bf5249 PATypeHolder is now not a template 
llvm-svn: 2106
 2002-04-04 19:24:34 +00:00
Chris Lattner
798be9a43a Add debug support 
llvm-svn: 2105
 2002-04-04 19:24:11 +00:00
Chris Lattner
78b786813b * Make PATypeHolder not take a type argument 
* Eliminate by inlining the old newTH, newTH, and TypeDone functions
* OPAQUE is now just a token that gets returned by the lexer, not a type
  Parser now creates type, not lexer

llvm-svn: 2104
 2002-04-04 19:23:55 +00:00
Chris Lattner
feeac9c7ea Lexer doesn't create typehandle gross stuff now, parser does. 
llvm-svn: 2103
 2002-04-04 19:22:17 +00:00
Chris Lattner
5cbfde5744 Call nodes are never equivalent 
Shadow nodes are never critical.

llvm-svn: 2102
 2002-04-04 19:21:51 +00:00
Chris Lattner
d4c7ce8d78 Add hack to get timing of analysis 
llvm-svn: 2101
 2002-04-04 19:21:27 +00:00
Chris Lattner
3b604e846a * Eliminate commented out code 
* Do not demand a shadow node when resolving
* Raise arbitrary inline limit

llvm-svn: 2100
 2002-04-04 19:21:06 +00:00
Chris Lattner
19689af3bb Support resolving function arguments/return values to pointers that index 
into other objects.

llvm-svn: 2094
 2002-04-01 22:20:48 +00:00
Vikram S. Adve
0035fcbea1 Case numbers were not updated when Sparc.burg.in was changed 
to add separate label for bitwise NOT operator.

llvm-svn: 2091
 2002-04-01 20:28:48 +00:00
Chris Lattner
0f0829f28d Add a new replaceWith method useful for replacing instructions 
llvm-svn: 2088
 2002-04-01 17:52:16 +00:00
Chris Lattner
2e3a4dece6 Allow scalars that point to multiple nodes when building the scalar map. 
llvm-svn: 2087
 2002-04-01 01:26:53 +00:00
Chris Lattner
99da111daf Add extra case here to avoid getting spurious output 
llvm-svn: 2086
 2002-04-01 00:45:33 +00:00
Chris Lattner
62756e4208 Avoid incorrectly adding null values to the scalar map! 
llvm-svn: 2085
 2002-04-01 00:45:09 +00:00
Chris Lattner
e66e651b68 Critical shadow nodes no do not know their parent explictly. 
llvm-svn: 2084
 2002-04-01 00:15:30 +00:00
Chris Lattner
61f6065cef Minor cleanups (use dyn_cast instead of testing manually) 
Shadow nodes now don't explicitly know their parent.

llvm-svn: 2083
 2002-04-01 00:14:41 +00:00
Chris Lattner
a68c92a3d4 Allow merging of identical call nodes. Make the shadow node pointed to 
by the call node noncritical before the call is destroyed.

llvm-svn: 2082
 2002-04-01 00:13:56 +00:00
Chris Lattner
f9097b54be Increase limit for perimeter 
llvm-svn: 2081
 2002-04-01 00:12:58 +00:00
Chris Lattner
42145d8a25 Unify the destruction code used for node pairs vs normal nodes. This was 
causing a problem before because global values with incoming edges didn't
copy the incoming edges to the node they were being merged from, causing
the poolalloc pass to die.

llvm-svn: 2079
 2002-03-31 19:57:44 +00:00
Vikram S. Adve
e665a992a9 1. Fix a leftover bug in generating memory instructions. 
2. Fix type used for TmpInstruction holding the return address.

llvm-svn: 2075
 2002-03-31 19:07:35 +00:00
Vikram S. Adve
257e4d18da Added function getCallInstIndirectAddrVal() to set call interference 
for that value.

llvm-svn: 2074
 2002-03-31 19:04:50 +00:00
Vikram S. Adve
8efa40ac63 Minor cleanup in printing constants. I think this included a bug 
fix related to putting a read-write variable in a read-only section,
but I'm not sure now.

llvm-svn: 2073
 2002-03-31 19:03:58 +00:00
Vikram S. Adve
44ee4184e4 Add method clearCallInterference(). 
llvm-svn: 2071
 2002-03-31 18:58:14 +00:00
Vikram S. Adve
c2c33ba3fd Include temp. values when computing max. size of stack frame! 
llvm-svn: 2070
 2002-03-31 18:57:49 +00:00
Vikram S. Adve
1e6ab6f476 Fix a bug in previous bug fix. 
llvm-svn: 2069
 2002-03-31 18:56:51 +00:00
Vikram S. Adve
fbdd533059 Bug fix: address used by indirect call instruction should also 
be marked as having a Call Interference, even though it may not
be live after the call.

llvm-svn: 2068
 2002-03-31 18:54:37 +00:00
Chris Lattner
c620b825b6 Add debug output 
llvm-svn: 2066
 2002-03-31 07:17:46 +00:00
Chris Lattner
164b766137 * Fix nondeleted type handle which could cause type pool corruption (and 
a memory leak)
* Fix memory leak of Argument nodes on function prototypes

llvm-svn: 2065
 2002-03-31 07:16:49 +00:00
Chris Lattner
8528806b33 * Move the isEquivalentTo implementations here. They can probably be put 
someplace nicer in the file though.
* Add new dump method for debugging

llvm-svn: 2064
 2002-03-31 07:16:08 +00:00
Chris Lattner
d2c39a02f7 Print out the instruction instead of just the address! 
llvm-svn: 2063
 2002-03-31 07:15:20 +00:00
Chris Lattner
aba1ef6fc9 * Move isEquivalentTo implementations to NodeImpl 
* Implement a new form of node folding to catch cases missed in Addtree
* Add removeIndistinguishableNodePairs to merge calls (todo) and globals

llvm-svn: 2062
 2002-03-31 07:14:52 +00:00
Chris Lattner
f1a2a9c168 * Convert CallMap to be a vector, because the keys can change, and the map 
doesn't resort!
* Be more generous with the cached matches that we allow now.
* Recursive calls should all work now!

llvm-svn: 2061
 2002-03-31 07:13:27 +00:00
Vikram S. Adve
f057ff78fb Fix CreateUIntSetInstruction() to handle sign extensions correctly. 
llvm-svn: 2059
 2002-03-31 00:13:12 +00:00
Chris Lattner
4c4a273f43 * Catch and ignore (for now) return instructions in tranformed functions 
* Add more debugging output
* Fix problems refering to wrong versions of various graphs
* Build the pool descriptor map!
* Clear the nodemapping map after building pool descriptor to avoid assert
* Transform the NEw function body, not the old one...
* Matrix.ll now works!

llvm-svn: 2058
 2002-03-30 20:53:14 +00:00
Chris Lattner
410c6394d3 * Clean up data structures [AllocDSNode -> DSNode] 
* TransformFunctionInfo now has call field form field mapping.  May be
  removed in the future.
* Moved the computation of "Scalars" into transformFunctionBody so
  transformFunction didn't have to recompute it.
* Implement the node mapping calculation in preparation to calculate
  PoolDescriptors to pass to transformFunctionBody
* Print out the node mapping [it looks right!]
* Other minor changes

llvm-svn: 2056
 2002-03-30 09:12:35 +00:00
Chris Lattner
36d50d01ab Maintain enough information so that the pools for all of the nodes of 
the graph can be passed around.

llvm-svn: 2053
 2002-03-30 04:02:31 +00:00
Chris Lattner
a4b9a9f573 Implement the first batch of transformations to the methods. So far it: 
* Converts malloc instructions to poolalloc
 * Converts free instructions to poolfree
 * Convert calls to call the new cloned hacked up versions

Note that this does not modify hacked up stuff yet, just the top level
function

llvm-svn: 2052
 2002-03-29 21:25:19 +00:00
Chris Lattner
c7df07c3f8 * s/Method/Function 
* Implement a check to make sure a function is not internal and external
  at the same time

llvm-svn: 2050
 2002-03-29 19:06:18 +00:00
Chris Lattner
9166ac0b3f Correctly clone the function with the extra argument types. Now we need 
to modify the function next.

This also properly recycles functions so that we don't get exponential
code blowup in the common case.

llvm-svn: 2049
 2002-03-29 19:05:48 +00:00
Chris Lattner
b6b7c3fe60 s/Method/Function 
llvm-svn: 2048
 2002-03-29 19:04:45 +00:00
Chris Lattner
57dd5ed245 New clone function routine 
llvm-svn: 2045
 2002-03-29 19:03:54 +00:00
Chris Lattner
6b1ce137c2 Find out which calls in the function we need to transform and how. 
Next step is to start hacking functions up.

llvm-svn: 2044
 2002-03-29 17:13:46 +00:00
Chris Lattner
5df9c977e1 s/Method/Function 
llvm-svn: 2043
 2002-03-29 17:08:29 +00:00
Chris Lattner
3a9f1dd66a Modularize code a bit 
llvm-svn: 2040
 2002-03-29 06:21:38 +00:00
Chris Lattner
7c2529076a Implement the first step of pool allocation - Creating, initialization, and 
destruction of the pools.

llvm-svn: 2039
 2002-03-29 05:50:20 +00:00
Chris Lattner
18cc6a9277 Add an insert method to VAlueHolder to allow batch insertion 
llvm-svn: 2038
 2002-03-29 05:49:37 +00:00
Chris Lattner
474495e742 Add new addTypeName method to Module class 
llvm-svn: 2037
 2002-03-29 04:48:40 +00:00
Chris Lattner
dc0566e721 s/Method/Function 
llvm-svn: 2036
 2002-03-29 03:57:15 +00:00
Chris Lattner
8c98acc07f s/Method/Function 
llvm-svn: 2035
 2002-03-29 03:51:11 +00:00
Chris Lattner
084601e6f4 s/Method/Function 
llvm-svn: 2034
 2002-03-29 03:44:36 +00:00
Chris Lattner
c32e9901b1 Implement new getFunction and getOrInsertFunction methods 
llvm-svn: 2033
 2002-03-29 03:44:18 +00:00
Chris Lattner
a2d3c99dfc Simplify code a bit by using Module::getOrInsertFunction 
llvm-svn: 2032
 2002-03-29 03:43:24 +00:00
Chris Lattner
815e2f4daa Flesh out a bunch more code, print allocations that are poolable. 
llvm-svn: 2031
 2002-03-29 03:40:59 +00:00
Chris Lattner
e46e35806a s/method/function 
llvm-svn: 2030
 2002-03-29 03:40:25 +00:00
Chris Lattner
4ca6fd00f2 s/Method/Function/g 
llvm-svn: 2029
 2002-03-29 03:39:36 +00:00
Chris Lattner
b7f4043df7 Simplify code a lot by using the Module::getFunction & getOrInsertFunction 
llvm-svn: 2028
 2002-03-29 03:38:05 +00:00
Chris Lattner
ab9cc61ceb Remove the reduceApply functions they are obsolete things from the days before 
we had a reasonable pass system

llvm-svn: 2022
 2002-03-28 22:49:49 +00:00
Chris Lattner
cab0ce8244 Implement getEscapingAllocations & getNonEscapingAllocations 
llvm-svn: 2021
 2002-03-28 19:33:00 +00:00
Chris Lattner
513ecd7256 * Rename UnlinkUndistinguishableShadowNodes & RemoveUnreachableShadowNodes 
to reflect that they can eliminate arbitrary nodes.
* Rename the ShadowNodeEliminate.cpp file to EliminateNodes.cpp for the
  same reason

llvm-svn: 2020
 2002-03-28 19:16:48 +00:00
Chris Lattner
fa67ad938c Ooops, I did such a great job pruning nodes, that I accidentally deleted 
ALL allocation nodes... hrm... bad.

llvm-svn: 2018
 2002-03-28 18:38:38 +00:00
Chris Lattner
0336a78488 Initial checkin of Noop pass that will be the pool allocator 
llvm-svn: 2014
 2002-03-28 18:08:31 +00:00
Chris Lattner
55186e7abf Many changes 
* Simplify a lot of the inlining stuff.  There are still problems, but not
  many
* Break up the Function representation to have a vector for every different
  node type so it is fast to find nodes of a particular flavor.
* Do more intelligent merging of call values
* Allow elimination of unreachable shadow and allocation nodes
* Generalize indistinguishability testing to allow merging of identical calls.
* Increase shadow node merging power

llvm-svn: 2010
 2002-03-28 17:56:03 +00:00
Cameron Buschardt
49c09d5b24 * Move classes around 
* rename constructor function
* Move stuff into anonymous namespaces

llvm-svn: 2007
 2002-03-27 23:28:40 +00:00
Cameron Buschardt
bcff69b0f4 Implemented promote mem->reg pass. 
llvm-svn: 2005
 2002-03-27 23:17:37 +00:00
Chris Lattner
d771725ed1 * Add #define to enabled debug messages 
* Move removeEdgesTo to be a member of DSNode
* Implement (but #ifdef out) the new, spiffier, method of determining
  shadow node equivalence.  This cannot be enabled until more is
  implemented.

llvm-svn: 2004
 2002-03-27 19:48:03 +00:00
Chris Lattner
4c6e9cda3f * Implement DSNode::removeAllIncomingEdges 
* Implement Critical Shadow node handling
* Implement routines to determine whether an allocation node is a malloc
  or alloca

llvm-svn: 2003
 2002-03-27 19:46:05 +00:00
Chris Lattner
a16c85b43d * Add critical node support 
* Optimize graph after building it.  This should be unneccesary in the future

llvm-svn: 2002
 2002-03-27 19:45:12 +00:00
Chris Lattner
9aff5a8047 * Destroy alloca nodes when a graph gets inlined 
* Add links to all subtrees when a shadow node gets resolved
* Add critical node handling

llvm-svn: 2001
 2002-03-27 19:44:33 +00:00
Chris Lattner
2a3d6c83ff * Optimizers return true on change 
* Implement indistinguishable shadow node elimination

llvm-svn: 1999
 2002-03-27 00:55:13 +00:00
Chris Lattner
ae0192e2f6 Fix long line 
llvm-svn: 1998
 2002-03-27 00:54:31 +00:00
Chris Lattner
8a44f81684 * Because of optimization, the shadow nodes between arguments might get 
removed.  Check to see if they are there.
* Repeat optimizations while changing

llvm-svn: 1997
 2002-03-27 00:53:57 +00:00
Chris Lattner
d5b115e687 Initial checkin of Datastructure analysis. 
Has bugs, but shouldn't crash in theory.

llvm-svn: 1994
 2002-03-26 22:39:06 +00:00
Chris Lattner
18073550b8 Change from Method to Function 
llvm-svn: 1992
 2002-03-26 18:02:30 +00:00
Chris Lattner
849c762b02 Change references from Method to Function 
change references from MethodARgument to FunctionArgument

llvm-svn: 1991
 2002-03-26 18:01:55 +00:00
Chris Lattner
8972179a78 Convert to use new style casts instead of direct checking 
llvm-svn: 1990
 2002-03-26 17:58:12 +00:00
Chris Lattner
57efd6ffd1 change refs to Method to Function 
Change references to MEthodArgument to FunctionArgument

llvm-svn: 1989
 2002-03-26 17:55:33 +00:00
Vikram S. Adve
bd1e4e8bd8 Moved to directory Scalar/ and renamed. 
llvm-svn: 1977
 2002-03-24 03:55:06 +00:00
Vikram S. Adve
3dfd5b888b Destroy MethodLiveVarInfo after register allocation. 
llvm-svn: 1976
 2002-03-24 03:54:03 +00:00
Vikram S. Adve
fa2ed7e6b1 Use deterministic iterator for SchedGraphs. This is actually not 
useful right now when we only do local scheduling.

llvm-svn: 1975
 2002-03-24 03:53:03 +00:00
Vikram S. Adve
058ec11ea3 Bug fix: nextToTry was not being initialized in one case. 
llvm-svn: 1974
 2002-03-24 03:46:15 +00:00
Vikram S. Adve
37d44e7de9 Add option to disable scheduling. 
llvm-svn: 1973
 2002-03-24 03:45:35 +00:00
Vikram S. Adve
0cdcf40121 Add option to disable scheduling. 
Destroy live-variable information after scheduling so it is
recomputed before later phases (e.g., reg. allocation).
Use deterministic iterator to enumerate sched graphs.

llvm-svn: 1972
 2002-03-24 03:44:55 +00:00
Vikram S. Adve
d18a1cc9fb Minor changes. 
llvm-svn: 1971
 2002-03-24 03:40:59 +00:00
Vikram S. Adve
e06cac463f Re-fix bug: Put back MachineCodeForInstruction::dropAllReferences. 
llvm-svn: 1970
 2002-03-24 03:40:11 +00:00
Vikram S. Adve
b347d2796c Fix padding for variables allocated on stack. 
llvm-svn: 1969
 2002-03-24 03:39:26 +00:00
Vikram S. Adve
34306e0e1c Major overhaul to FoldGetElemPtr to handle mixed array and struct refs. 
llvm-svn: 1968
 2002-03-24 03:37:53 +00:00
Vikram S. Adve
8493ae74b2 Use deterministic iterator for treeRoots. 
llvm-svn: 1967
 2002-03-24 03:36:52 +00:00
Vikram S. Adve
8ada4c031e Bug re-fix: put back MachineCodeForInstruction::get(*I).dropAllReferences(). 
Also re-enable instr. scheduling pass.

llvm-svn: 1966
 2002-03-24 03:35:16 +00:00
Vikram S. Adve
7b84ec22d8 Change latencies for Load, Store and Branch instructions. 
llvm-svn: 1965
 2002-03-24 03:33:53 +00:00
Vikram S. Adve
57d735db7a Major enhancements to how array and structure indices are handled. 
Improve checking for constants in Multiply.
Simpler method to keep track of when a node is folded into its parent.
Several other bug fixes.

llvm-svn: 1964
 2002-03-24 03:33:02 +00:00
Vikram S. Adve
a141a7190b Change treeRoots data structure to make enumeration deterministic. 
llvm-svn: 1962
 2002-03-24 03:24:00 +00:00
Vikram S. Adve
1f5e8f93cc Major overhaul to support arbitrary mixed array and structure indices. 
Also moved to the Scalar/ directory and renamed to "MultiDim".

llvm-svn: 1961
 2002-03-24 03:21:18 +00:00
Chris Lattner
c481e92886 Rename Method to Function 
llvm-svn: 1957
 2002-03-23 22:51:58 +00:00
Vikram S. Adve
3294262573 New pass to decompose multi-dimensional array references into 
a sequence of 1-D references, using a sequence of getElementPtrs.

llvm-svn: 1948
 2002-03-23 20:43:39 +00:00
Chris Lattner
0b9734b293 Add an assertion that would hopefully catch a problem that was tripping 
cameron up

llvm-svn: 1944
 2002-03-22 18:01:58 +00:00
Chris Lattner
2ed4beefea Fix bug test/Regression/Transforms/LevelRaise/2002-03-21-MissedRaise3.ll 
llvm-svn: 1943
 2002-03-21 23:02:37 +00:00
Chris Lattner
ad241cc3ea Fix test/Regression/Transforms/LevelRaise/2002-03-21-MissedRaise2.ll 
llvm-svn: 1941
 2002-03-21 22:39:59 +00:00
Chris Lattner
8051e7c953 * AllocationInst ctor moved here from iMemory.h 
* AllocationInst now always has an array size operand

llvm-svn: 1939
 2002-03-21 22:37:48 +00:00
Chris Lattner
a80b65c208 Fix test/Regression/Other/2002-03-21-LevelRaiseMiss.ll 
llvm-svn: 1933
 2002-03-21 21:21:13 +00:00
Chris Lattner
c2b5cb5619 Make ConvertableToGEP handle cases where the user is indexing into the 
first element of a structure type.  Before this would not be handled because
getStructOffset would either stop immediately (because StopEarly was true
and Offset = 0), or blast past the level we wanted.

Now ConvertableToGEP steps down through the type one level at a time, checking
the Offset and Scale conditions at each step

llvm-svn: 1931
 2002-03-21 06:27:20 +00:00
Chris Lattner
559c71f6f4 Implement the new InsertInstBeforeInst function 
llvm-svn: 1930
 2002-03-21 06:24:00 +00:00
Chris Lattner
14fe793707 Fix the: test/Regression/Other/2002-03-20-LevelRaiseCrash.ll bug. It turned 
out that almost always the result of the add was the same as the GEP that
would be generated.  In this case, however, it wasn't, so bad things happened.

llvm-svn: 1929
 2002-03-21 06:22:23 +00:00
Chris Lattner
e9ec6058e4 Declare the new InsertInstBeforeInst function 
llvm-svn: 1927
 2002-03-21 06:15:53 +00:00
Chris Lattner
ceb4e429a2 Add an assertion to catch a misuse of replaceAllUsesWith 
llvm-svn: 1924
 2002-03-21 05:38:15 +00:00
Chris Lattner
72367834b0 1. Fix bug test/Regression/Other/2002-03-20-LevelRaiseBadCodegen.ll 
2. Allow negative scales in ConvertableToGEP, so that this loop from
   alloc_tree in the Olden health benchmark is correctly raised:

       for (i = 3; i >= 0; i--)
         fval[i] = alloc_tree(level - 1, label*4 + i + 1, new);

   Stepping through the array backwards caused a negative scale.

llvm-svn: 1922
 2002-03-21 03:04:38 +00:00
Chris Lattner
db90f5f6ea Namespace qualify debugging code 
llvm-svn: 1921
 2002-03-21 03:02:07 +00:00
Vikram S. Adve
92fb8b1f9c Fixed several problems with handling arguments to Phis. 
llvm-svn: 1910
 2002-03-18 03:47:26 +00:00
Vikram S. Adve
cd6cb1962f Big bug fix: killed uses were being inserted instead of erased! 
Also added a set of debug options.

llvm-svn: 1909
 2002-03-18 03:45:41 +00:00
Vikram S. Adve
308f672ac8 Just a comment. 
llvm-svn: 1908
 2002-03-18 03:40:25 +00:00
Vikram S. Adve
5f910def00 New pass to decompose multi-dimensional array references into 
a sequence of 1-D references, using a sequence of getElementPtrs.

llvm-svn: 1907
 2002-03-18 03:39:09 +00:00
Vikram S. Adve
8553fc5d21 Big bug fix: getUsableUniRegAtMI needed to return values in arguments 
but did not pass the arguments by reference!

llvm-svn: 1906
 2002-03-18 03:37:19 +00:00
Vikram S. Adve
855ed3d941 Added function computeOffsetforLocalVar to check how big the 
offset-from-FP will be before allocating space for a local variable.

llvm-svn: 1905
 2002-03-18 03:36:30 +00:00
Vikram S. Adve
edc1adae11 Bug fix in setting CC register. 
Also use distinct names for the three types of SetMachineOperand
to avoid painful overloading problems and errors.

llvm-svn: 1904
 2002-03-18 03:35:24 +00:00
Vikram S. Adve
b26ed73c33 Constants are now added to the constant pool only when a load 
instruction is actually generated for them.
Rename the different versions of SetMachineOperand.

llvm-svn: 1903
 2002-03-18 03:33:43 +00:00
Vikram S. Adve
3fe5a9e913 Use vector for machine instructions returned by GetInstructionsByRule. 
Fix constants in instructions generated for Phi elimination.

llvm-svn: 1902
 2002-03-18 03:31:54 +00:00
Vikram S. Adve
125cd33f99 Big bug fix: getUsableUniRegAtMI needed to return values in arguments 
but did not pass the arguments by reference!
Also added a function to get a register class by ID.

llvm-svn: 1901
 2002-03-18 03:26:48 +00:00
Vikram S. Adve
7f139ccf85 Add support for code generation for array references. 
Also, use distinct names for the three types of SetMachineOperand to
avoid painful overloading problems and errors.
Bug fix in converting-int-to-float.  Other minor stuff.

llvm-svn: 1896
 2002-03-18 03:15:35 +00:00
Vikram S. Adve
217f7b1fe1 Bug fixes for not using unified reg. numbers, and for using the 
wrong register class for saving CC registers.
Also, use distinct names for the three types of SetMachineOperand.

llvm-svn: 1895
 2002-03-18 03:12:16 +00:00
Vikram S. Adve
6a74ad7030 Use distinct names for the three types of SetMachineOperand to 
avoid painful overloading problems and errors.

llvm-svn: 1893
 2002-03-18 03:09:15 +00:00
Vikram S. Adve
f7d5ad2124 Minor fixes. 
llvm-svn: 1892
 2002-03-18 03:08:07 +00:00
Vikram S. Adve
23d4011ea1 Handling of global names was severely broken when converting to 2 passes. 
Different labels were generated for the same name in different passes.
Use an annotation to hold a fixed table of names.

llvm-svn: 1891
 2002-03-18 03:07:26 +00:00
Vikram S. Adve
d311d0d546 Several sundry bug fixes. 
llvm-svn: 1890
 2002-03-18 03:02:42 +00:00
Chris Lattner
240209eb5b Make an assertion provide a more helpful error message 
llvm-svn: 1883
 2002-03-17 21:16:01 +00:00
Chris Lattner
88ccdcd73e Remove code designed to compensate for a bug in GCC. The bug has since 
been fixed.

llvm-svn: 1881
 2002-03-15 20:35:21 +00:00
Chris Lattner
ba358e8e65 Checkin new test for problem anand ran into 
llvm-svn: 1880
 2002-03-15 20:25:09 +00:00
Chris Lattner
55aff5c433 Cleanup error message output a bit 
llvm-svn: 1879
 2002-03-15 20:21:29 +00:00
Chris Lattner
8cc6186707 Fix test/Regression/Other/2002-03-14-ValueToLarge.ll. Note that this 
fix could be better, but we don't have the infrastructure to find out what
size types are yet.

llvm-svn: 1876
 2002-03-14 22:35:50 +00:00
Chris Lattner
1a4aac1060 Add a check to ensure that only PHI nodes are self referential. Code 
input to instruction combination was broken, which caused it to explode.

llvm-svn: 1870
 2002-03-14 16:53:48 +00:00
Chris Lattner
2ecc9a337c Fix bug: test/Regression/Other/2002-03-14-LevelRaiseHang.ll 
llvm-svn: 1869
 2002-03-14 16:37:04 +00:00
Chris Lattner
b0128957b1 Fix bug: test/Regression/Other/2002-03-11-InstCombineHang.ll 
llvm-svn: 1864
 2002-03-11 23:28:45 +00:00
Chris Lattner
ba4660ff3d update comments 
llvm-svn: 1862
 2002-03-11 22:21:04 +00:00
Chris Lattner
0f8d0257f9 * Fix bug: test/Regression/Other/2002-03-11-LevelRaiseIterInvalidate.ll 
* Use more concise form of ReplaceInstWithInst to make code simpler
* Output nicer message for dead code eliminated when debugging

llvm-svn: 1861
 2002-03-11 22:19:48 +00:00
Chris Lattner
23379a852e Implement TODO for better diagnostic output 
llvm-svn: 1859
 2002-03-11 22:12:39 +00:00
Chris Lattner
889b1e5b00 Fix bug: test/Regression/Other/2002-03-11-ConstPropCrash.ll 
llvm-svn: 1858
 2002-03-11 22:11:07 +00:00
Chris Lattner
35c1c7353b Bugfix for test/Regression/Other/2002-03-11-ExprAssertion.ll 
llvm-svn: 1856
 2002-03-11 20:50:24 +00:00
Chris Lattner
558efd4bf0 Remove runtime library in favor of users linking against real libraries. 
llvm-svn: 1853
 2002-03-11 17:57:13 +00:00
Chris Lattner
32669e6211 Do not link in runtime library anymore, assume that user manually links in 
libraries.  Code should be generalized to allow lli to take -lc -lm arguments
in the future (todo)

llvm-svn: 1852
 2002-03-11 17:52:43 +00:00
Chris Lattner
34eab0d19c Fix bug: test/Regression/Other/2002-03-11-LevelRaiseCalls.ll 
llvm-svn: 1844
 2002-03-11 17:27:34 +00:00
Chris Lattner
e71c9ba6d0 Implement a bunch of new functions to support the SPECINT mcf 
benchmark

llvm-svn: 1841
 2002-03-08 22:51:07 +00:00
Chris Lattner
b44f81b987 * Add new method localLookup 
* SymbolTable::remove(Value *N) checks to see if we are internally
  inconsistent before looking for a type plane (caused a crash)
* insertEntry now does a local lookup instead of a global lookup, which was
  causing an infinite loop in the renamer logic.
* Added assertions to make sure stuff stays happy
* Now the linker correctly links the SPECINT2000 mcf benchmark

llvm-svn: 1840
 2002-03-08 20:26:17 +00:00