Commits · 05e6267466bb5b869189f51800293be104abd06e · Roger Ferrer / llvm-epi-0.8

Apr 24, 2008
- Drop dead includes · 1ae135c8
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50192
```
  1ae135c8
- Adjust option names for C++ backend · 9dcc3e97
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50190
```
  9dcc3e97
- First step of implementing PR1538: move llvm2cpp logic to new 'target' · 78695035
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50189
```
  78695035
Apr 23, 2008
- Initial CBE support for multiple return values. · d871fa5c
  Dan Gohman authored Apr 23, 2008
```
llvm-svn: 50187
```
  d871fa5c
- tighten up verifier checks which missed cases where · 1e31bf5b
  Chris Lattner authored Apr 23, 2008
```
return instrs operands didn't match up with function results.

llvm-svn: 50182
```
  1e31bf5b
- Use isa instead of dyn_cast. · e9e3891c
  Dan Gohman authored Apr 23, 2008
```
llvm-svn: 50181
```
  e9e3891c
- Add support to codegen for getresult instructions with undef operands. · b418aafa
  Dan Gohman authored Apr 23, 2008
```
llvm-svn: 50180
```
  b418aafa
- Check type instead of no. of operands. · 8f83081f
  Devang Patel authored Apr 23, 2008
```
llvm-svn: 50179
```
  8f83081f
- Allow llvm-as to parse a getresult with an undef operand. · 6e918509
  Dan Gohman authored Apr 23, 2008
```
llvm-svn: 50175
```
  6e918509
- Rewrite previous patch to suit Chris's preference. · f6e15a47
  Dale Johannesen authored Apr 23, 2008
```
llvm-svn: 50174
```
  f6e15a47
- Unbreak JIT · 0516b6f2
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50173
```
  0516b6f2
- Fix typo · 0d6df367
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50169
```
  0d6df367
- Only allow increase of max alignment value · 965babda
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50168
```
  965babda
- Be over-conservative: scan for all used virtual registers and calculate... · c1534dca
  Anton Korobeynikov authored Apr 23, 2008
```
Be over-conservative: scan for all used virtual registers and calculate maximal stack alignment in assumption, that there will be spill of vector register.

llvm-svn: 50167
```
  c1534dca
- Add X86 Maximal Stack Alignment Calculator Pass before RA · 2659011b
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50166
```
  2659011b
- Add facility for pre-RA passes · 7e859dd7
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50165
```
  7e859dd7
- Use precomputed value, if any · 41334635
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50164
```
  41334635
- Do proper book-keeping of offsets and prologue/epilogue code for stack realignment · 156550ae
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50163
```
  156550ae
- If stack realignment is used - incoming args will use EBP as base register and locals - ESP · 89a0a017
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50162
```
  89a0a017
- Eastimate required stack alignment early, so we can decide, whether we will... · ba512907
  Anton Korobeynikov authored Apr 23, 2008
```
Eastimate required stack alignment early, so we can decide, whether we will need frame pointer or not

llvm-svn: 50161
```
  ba512907
- Cleanup · f49bc9f8
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50160
```
  f49bc9f8
- Cleanup · c756b460
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50159
```
  c756b460
- Simplify · a8aac3db
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50158
```
  a8aac3db
- Make stack alignment options global for all targets · cb195f51
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50157
```
  cb195f51
- Provide option for enabling-disabling stack realignment · 9328fbc4
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50156
```
  9328fbc4
- Disable stack realignment for functions with dynamic-sized alloca's · ca150edd
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50155
```
  ca150edd
- Provide ABI-correct stack alignment · a7495260
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50154
```
  a7495260
- Provide convenient helpers for some operations · 8843487e
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50153
```
  8843487e
- Whitespace cleanup · 2ccafa47
  Anton Korobeynikov authored Apr 23, 2008
```
llvm-svn: 50152
```
  2ccafa47
- simplify code for propagation of constant arguments into · a82d691c
  Chris Lattner authored Apr 23, 2008
```
callees.

llvm-svn: 50142
```
  a82d691c
- Fix a number of bugs in ipconstantprop, simplify the code, fit in 80 cols, · 5f1802cf
  Chris Lattner authored Apr 23, 2008
```
fix read after free bug (PR2238).

llvm-svn: 50141
```
  5f1802cf
- Rewrite multiple return value handling in SCCP. Before, the -sccp pass · 5a58a4dc
  Chris Lattner authored Apr 23, 2008
```
would turn every getresult instruction into undef.  This helps with
rdar://5778210

llvm-svn: 50140
```
  5a58a4dc
- regenerate · 6e0284bb
  Chris Lattner authored Apr 23, 2008
```
llvm-svn: 50139
```
  6e0284bb
- Validate that the result of a function type is valid using shared · ff9089eb
  Chris Lattner authored Apr 23, 2008
```
logic with vmcore.

llvm-svn: 50138
```
  ff9089eb
- Enforce that multiple return values have to have at least one result. · b6917d2a
  Chris Lattner authored Apr 23, 2008
```
llvm-svn: 50137
```
  b6917d2a
- Verify that the operand of a getresult instruction is a · 31abd54e
  Chris Lattner authored Apr 23, 2008
```
call/invoke or undef.

llvm-svn: 50129
```
  31abd54e
- Do not change the type of a ByVal argument to a · 493527d8
  Dale Johannesen authored Apr 23, 2008
```
type of a different size.

llvm-svn: 50121
```
  493527d8
- Don't do: "(X & 4) >> 1 == 2 --> (X & 4) == 4" if there are more than one... · 1c89ca72
  Evan Cheng authored Apr 23, 2008
```
Don't do: "(X & 4) >> 1 == 2  --> (X & 4) == 4" if there are more than one uses of the shift result.

llvm-svn: 50118
```
  1c89ca72
Apr 22, 2008

Start doing the significantly useful part of jump threading: handle cases · 37e9c187

Chris Lattner authored Apr 22, 2008

where a comparison has a phi input and that phi is a constant.  For example,
stuff like:

  Threading edge through bool from 'bb2149' to 'bb2231' with cost: 1, across block:
bb2237:		; preds = %bb2231, %bb2149
	%tmp2328.rle = phi i32 [ %tmp2232, %bb2231 ], [ %tmp2232439, %bb2149 ]		; <i32> [#uses=2]
	%done.0 = phi i32 [ %done.2, %bb2231 ], [ 0, %bb2149 ]		; <i32> [#uses=1]
	%tmp2239 = icmp eq i32 %done.0, 0		; <i1> [#uses=1]
	br i1 %tmp2239, label %bb2231, label %bb2327

or

bb38.i298:		; preds = %bb33.i295, %bb1693
	%tmp39.i296.rle = phi %struct.ibox* [ null, %bb1693 ], [ %tmp39.i296.rle1109, %bb33.i295 ]		; <%struct.ibox*> [#uses=2]
	%minspan.1.i291.reg2mem.1 = phi i32 [ 32000, %bb1693 ], [ %minspan.0.i288, %bb33.i295 ]		; <i32> [#uses=1]
	%tmp40.i297 = icmp eq %struct.ibox* %tmp39.i296.rle, null		; <i1> [#uses=1]
	br i1 %tmp40.i297, label %implfeeds.exit311, label %bb43.i301

This triggers thousands of times in spec.

llvm-svn: 50110

37e9c187

Dig through multiple levels of AND to thread jumps if needed. · d5425e8f
Chris Lattner authored Apr 22, 2008
```
llvm-svn: 50106
```
d5425e8f