Evan Cheng
00fcb0017e
Handling zero extension of 1 bit value.
...
llvm-svn: 24722
2005-12-15 01:02:48 +00:00
Evan Cheng
bc9344477e
Use MOV8rm to load 1 bit value.
...
llvm-svn: 24721
2005-12-15 00:59:17 +00:00
Nate Begeman
808f7a8abb
Remove a now unused statistic.
...
llvm-svn: 24720
2005-12-14 22:56:16 +00:00
Nate Begeman
e37cb604c1
Use the new predicate support that Evan Cheng added to remove some code
...
from the DAGToDAG cpp file. This adds pattern support for vector and
scalar fma, which passes test/Regression/CodeGen/PowerPC/fma.ll, and
does the right thing in the presence of -disable-excess-fp-precision.
Allows us to match:
void %foo(<4 x float> * %a) {
entry:
%tmp1 = load <4 x float> * %a;
%tmp2 = mul <4 x float> %tmp1, %tmp1
%tmp3 = add <4 x float> %tmp2, %tmp1
store <4 x float> %tmp3, <4 x float> *%a
ret void
}
As:
_foo:
li r2, 0
lvx v0, r2, r3
vmaddfp v0, v0, v0, v0
stvx v0, r2, r3
blr
Or, with llc -disable-excess-fp-precision,
_foo:
li r2, 0
lvx v0, r2, r3
vxor v1, v1, v1
vmaddfp v1, v0, v0, v1
vaddfp v0, v1, v0
stvx v0, r2, r3
blr
llvm-svn: 24719
2005-12-14 22:54:33 +00:00
Evan Cheng
023aef2f31
Fixed a typo: line 2323: MOVSX16rm8 -> MOVZX16rm8. This was the cause fo 12/14/2005 hbd failure.
...
llvm-svn: 24717
2005-12-14 22:28:18 +00:00
Evan Cheng
3db275d996
Added predicate !NoExcessFPPrecision to FMADD, FMADDS, FMSUB, and FMSUBS.
...
llvm-svn: 24716
2005-12-14 22:07:12 +00:00
Evan Cheng
d296a43f96
Added support to specify predicates.
...
llvm-svn: 24715
2005-12-14 22:02:59 +00:00
Evan Cheng
b8be9d1596
Fixed extload type profile. The 4th operand is a ValueType node with type
...
OtherVT, it cannot be compare to type of 1st operand which is an integer type.
llvm-svn: 24713
2005-12-14 19:40:54 +00:00
Chris Lattner
d39c60fcc8
When folding loads into ops, immediately replace uses of the op with the
...
load. This reduces number of worklist iterations and avoid missing optimizations
depending on folding of things into sext_inreg nodes (which aren't supported by
all targets).
Tested by Regression/CodeGen/X86/extend.ll:test2
llvm-svn: 24712
2005-12-14 19:25:30 +00:00
Chris Lattner
7dac1083da
Fix the (zext (zextload)) case to trigger, similarly for sign extends.
...
Allow (zext (truncate)) to apply after legalize if the target supports
AND (which all do).
This compiles
short %foo() {
%tmp.0 = load ubyte* %X ; <ubyte> [#uses=1]
%tmp.3 = cast ubyte %tmp.0 to short ; <short> [#uses=1]
ret short %tmp.3
}
to:
_foo:
movzbl _X, %eax
ret
instead of:
_foo:
movzbl _X, %eax
movzbl %al, %eax
ret
thanks to Evan for pointing this out.
llvm-svn: 24709
2005-12-14 19:05:06 +00:00
Chris Lattner
8c9e14620f
Fix Transforms/ScalarRepl/2005-12-14-UnionPromoteCrash.ll, a crash on undefined
...
behavior in 126.gcc on big-endian systems.
llvm-svn: 24708
2005-12-14 17:23:59 +00:00
Chris Lattner
f753d1a574
Fix a miscompilation in crafty due to a recent patch
...
llvm-svn: 24706
2005-12-14 07:58:38 +00:00
Evan Cheng
c273900dd8
Added sext and zext patterns.
...
llvm-svn: 24705
2005-12-14 02:22:27 +00:00
Evan Cheng
683d8515b1
Added sextld + zextld DAG nodes.
...
llvm-svn: 24703
2005-12-14 02:21:01 +00:00
Evan Cheng
bce7c47306
Fold (zext (load x) to (zextload x).
...
llvm-svn: 24702
2005-12-14 02:19:23 +00:00
Nate Begeman
40f081d8e0
Add support for fmul node of type v4f32.
...
void %foo(<4 x float> * %a) {
entry:
%tmp1 = load <4 x float> * %a;
%tmp2 = mul <4 x float> %tmp1, %tmp1
store <4 x float> %tmp2, <4 x float> *%a
ret void
}
Is selected to:
_foo:
li r2, 0
lvx v0, r2, r3
vxor v1, v1, v1
vmaddfp v0, v0, v0, v1
stvx v0, r2, r3
blr
llvm-svn: 24701
2005-12-14 00:34:09 +00:00
Nate Begeman
69caef2b78
Prepare support for AltiVec multiply, divide, and sqrt.
...
llvm-svn: 24700
2005-12-13 22:55:22 +00:00
Reid Spencer
6796b48e51
Adjust the constructor to the Linker class to take an argument that names
...
the module being constructed. This is used to correctly name the module.
Previously the name of the linker tool was used which produces confusing
output when the module identifier is used in an error message.
llvm-svn: 24699
2005-12-13 20:00:37 +00:00
Reid Spencer
175613adf6
Improve ResolveFunctions to:
...
a) use better local variable names (OldMT -> OldFT) where "M" is used to
mean "Function" (perhaps it was previously "Method"?)
b) print out the module identifier in a warning message so that it is
possible to track down in which module the error occurred.
llvm-svn: 24698
2005-12-13 19:56:51 +00:00
Chris Lattner
5d4e61dd87
Don't lump the filename and working dir together
...
llvm-svn: 24697
2005-12-13 17:40:33 +00:00
Evan Cheng
229f0ee6d7
Add load + store folding srl and sra patterns.
...
llvm-svn: 24696
2005-12-13 07:24:22 +00:00
Chris Lattner
87079884d1
Use the shared asmprinter code for printing special llvm globals
...
llvm-svn: 24695
2005-12-13 06:32:50 +00:00
Chris Lattner
f0e9aef954
Add a couple more fields, move ctor init list to .cpp file, add support
...
for emitting the ctor/dtor list for common targets.
llvm-svn: 24694
2005-12-13 06:32:10 +00:00
Chris Lattner
4d80f6e52e
Add ELF and darwin support for static ctors and dtors
...
llvm-svn: 24693
2005-12-13 04:53:51 +00:00
Chris Lattner
54a11df95d
reindent a loop, unswitch a loop. No functionality changes
...
llvm-svn: 24692
2005-12-13 04:33:58 +00:00
Nate Begeman
956aef45c9
Lowering constant pool entries on ppc exposed a bug in the recently added
...
ConstantVec legalizing code, which would return constantpool nodes that
were not of the target's pointer type.
llvm-svn: 24691
2005-12-13 03:03:23 +00:00
Evan Cheng
acec857b1a
Beautify a few patterns.
...
llvm-svn: 24690
2005-12-13 02:40:18 +00:00
Evan Cheng
89c6db4baf
Some shl patterns which do load + store folding.
...
llvm-svn: 24689
2005-12-13 02:34:51 +00:00
Evan Cheng
108beceb0f
A few helper fragments for loads. e.g. (i8 (load addr:$src)) -> (loadi8 addr:$src). Only to improve readibility.
...
llvm-svn: 24688
2005-12-13 01:57:51 +00:00
Evan Cheng
ddd5ae5a22
Add and, or, and xor patterns which fold load + stores.
...
llvm-svn: 24687
2005-12-13 01:41:36 +00:00
Evan Cheng
e5a94a03e2
Add inc + dec patterns which fold load + stores.
...
llvm-svn: 24686
2005-12-13 01:02:47 +00:00
Evan Cheng
bde9e6fca6
Add neg and not patterns which fold load + stores.
...
llvm-svn: 24685
2005-12-13 00:54:44 +00:00
Evan Cheng
c414d563f0
Missed a couple redundant explicit type casts.
...
llvm-svn: 24684
2005-12-13 00:25:07 +00:00
Evan Cheng
62e6808aa5
Fix some bad choice of names: i16SExt8 ->i16immSExt8, etc.
...
llvm-svn: 24683
2005-12-13 00:14:11 +00:00
Evan Cheng
86b2cf22d2
* Split immSExt8 to i16SExt8 and i32SExt8 for i16 and i32 immediate operands.
...
This enables the removal of some explicit type casts.
* Rename immZExt8 to i16ZExt8 as well.
llvm-svn: 24682
2005-12-13 00:01:09 +00:00
Evan Cheng
3e52756928
Add some integer mul patterns.
...
llvm-svn: 24681
2005-12-12 23:47:46 +00:00
Chris Lattner
9e8b633ec1
Accept and ignore prefetches for now
...
llvm-svn: 24678
2005-12-12 22:51:16 +00:00
Chris Lattner
b42ce7ca63
Fix CodeGen/Generic/2005-12-12-ExpandSextInreg.ll
...
llvm-svn: 24677
2005-12-12 22:27:43 +00:00
Evan Cheng
af3fe8217a
Add some sub patterns.
...
llvm-svn: 24675
2005-12-12 21:54:05 +00:00
Evan Cheng
67ed58e22b
When SelectLEAAddr() fails, it shouldn't cause the side effect of having the
...
base or index operands being selected.
llvm-svn: 24674
2005-12-12 21:49:40 +00:00
Evan Cheng
bfd259a2b7
For ISD::RET, if # of operands >= 2, try selection the real data dep. operand
...
first before the chain.
e.g.
int X;
int foo(int x)
{
x += X + 37;
return x;
}
If chain operand is selected first, we would generate:
movl X, %eax
movl 4(%esp), %ecx
leal 37(%ecx,%eax), %eax
rather than
movl $37, %eax
addl 4(%esp), %eax
addl X, %eax
which does not require %ecx. (Due to ADD32rm not matching.)
llvm-svn: 24673
2005-12-12 20:32:18 +00:00
Andrew Lenharth
cd54254af3
fix FP selects
...
llvm-svn: 24672
2005-12-12 20:30:09 +00:00
Chris Lattner
d6b17765e4
remove some never-completed and now-obsolete code.
...
llvm-svn: 24671
2005-12-12 20:12:20 +00:00
Evan Cheng
e80248b378
Add a few more add / store patterns. e.g. ADD32mi8.
...
llvm-svn: 24670
2005-12-12 19:45:23 +00:00
Andrew Lenharth
b8296181e0
restore a more restricted select
...
llvm-svn: 24668
2005-12-12 17:43:52 +00:00
Chris Lattner
3b0a62d8a5
Implement a little hack for parity with GCC on crafty. This speeds up
...
186.crafty by about 16% (from 15.109s to 13.045s) on my system.
This turns allocas with unions/casts into scalars. For example crafty has
something like this:
union doub {
unsigned short i[4];
long long d;
};
int f(long long a) {
return ((union doub){.d=a}).i[1];
}
Instead of generating loads and stores to an alloca, we now promote the
whole thing to a scalar long value.
This implements: Transforms/ScalarRepl/AggregatePromote.ll
llvm-svn: 24667
2005-12-12 07:19:13 +00:00
Chris Lattner
a4c6cc5af4
Fix typo :(
...
llvm-svn: 24664
2005-12-11 18:43:13 +00:00
Chris Lattner
f1a54c0d14
Minor tweak to get isel opt
...
llvm-svn: 24663
2005-12-11 09:05:13 +00:00
Chris Lattner
254e0a842f
add selectcc
...
llvm-svn: 24662
2005-12-11 08:35:54 +00:00
Chris Lattner
090eed0483
Remove type casts that are no longer needed
...
llvm-svn: 24661
2005-12-11 07:45:47 +00:00
Chris Lattner
e6f2c82073
Realize the constant pool & global addrs must always be ptr type
...
llvm-svn: 24660
2005-12-11 07:45:04 +00:00
Chris Lattner
3d9559fedc
Fix the JIT failures from last night.
...
llvm-svn: 24659
2005-12-11 07:37:41 +00:00
Andrew Lenharth
20d0b81c04
FP select improvements (and likely breakage), oh and crazy people might want to *return* floating point values. Don't see why myself
...
llvm-svn: 24658
2005-12-11 03:54:31 +00:00
Nate Begeman
4e56db674c
Add support for TargetConstantPool nodes to the dag isel emitter, and use
...
them in the PPC backend, to simplify some logic out of Select and
SelectAddr.
llvm-svn: 24657
2005-12-10 02:36:00 +00:00
Evan Cheng
3c5198336c
Use SDTCisPtrTy type property for store address.
...
llvm-svn: 24656
2005-12-10 01:59:36 +00:00
Evan Cheng
0d6cfee704
* Added X86 store patterns.
...
* Added X86 dec patterns.
llvm-svn: 24654
2005-12-10 00:48:20 +00:00
Evan Cheng
dadc1057ac
Added new getNode and getTargetNode variants for X86 stores.
...
llvm-svn: 24653
2005-12-10 00:37:58 +00:00
Nate Begeman
ade6f9a255
Add support patterns to many load and store instructions which will
...
hopefully use patterns in the near future.
llvm-svn: 24651
2005-12-09 23:54:18 +00:00
Chris Lattner
27656ac89c
Add SDTCisPtrTy and use it for loads, to indicate that the operand of a load
...
must be a pointer. This removes a type check out of the code generated by
tblgen for load matching.
llvm-svn: 24650
2005-12-09 22:58:42 +00:00
Evan Cheng
275a3ed80c
Added patterns for ADD8rm, etc. These fold load operands. e.g. addb 4(%esp), %al
...
llvm-svn: 24648
2005-12-09 22:48:48 +00:00
Chris Lattner
a6f835f5a0
Avoid emitting two tabs when switching to a named section
...
llvm-svn: 24646
2005-12-09 19:28:49 +00:00
Chris Lattner
e0f5f8e43c
Teach the PPC backend about the ctor and dtor list when not using __main and
...
linking the entire program into one bc file.
llvm-svn: 24645
2005-12-09 18:24:29 +00:00
Chris Lattner
268d457b69
Teach legalize how to promote sext_inreg to fix a problem Andrew pointed
...
out to me.
llvm-svn: 24644
2005-12-09 17:32:47 +00:00
Andrew Lenharth
87bf2234b5
it helps if your conditionals are not reversed
...
llvm-svn: 24641
2005-12-09 00:45:42 +00:00
Chris Lattner
be73d6eece
improve code insertion in two ways:
...
1. Only forward subst offsets into loads and stores, not into arbitrary
things, where it will likely become a load.
2. If the source is a cast from pointer, forward subst the cast as well,
allowing us to fold the cast away (improving cases when the cast is
from an alloca or global).
This hasn't been fully tested, but does appear to further reduce register
pressure and improve code. Lets let the testers grind on it a bit. :)
llvm-svn: 24640
2005-12-08 08:00:12 +00:00
Chris Lattner
29e6c3dbf9
Add another important case we miss
...
llvm-svn: 24639
2005-12-08 07:13:28 +00:00
Evan Cheng
790af6d18f
Added support for ComplexPattern.
...
llvm-svn: 24638
2005-12-08 04:28:48 +00:00
Evan Cheng
f039648614
Added explicit type field to ComplexPattern.
...
llvm-svn: 24637
2005-12-08 02:15:07 +00:00
Evan Cheng
c9fab31098
* Added intelligence to X86 LEA addressing mode matching routine so it returns
...
false if the match is not profitable. e.g. leal 1(%eax), %eax.
* Added patterns for X86 integer loads and LEA32.
llvm-svn: 24635
2005-12-08 02:01:35 +00:00
Nate Begeman
ae89d862f5
Fix a crash where ConstantVec nodes were being generated with the wrong
...
type when the target did not support them. Also teach Legalize how to
expand ConstantVecs.
This allows us to generate
_test:
lwz r2, 12(r3)
lwz r4, 8(r3)
lwz r5, 4(r3)
lwz r6, 0(r3)
addi r2, r2, 4
addi r4, r4, 3
addi r5, r5, 2
addi r6, r6, 1
stw r2, 12(r3)
stw r4, 8(r3)
stw r5, 4(r3)
stw r6, 0(r3)
blr
For:
void %test(%v4i *%P) {
%T = load %v4i* %P
%S = add %v4i %T, <int 1, int 2, int 3, int 4>
store %v4i %S, %v4i * %P
ret void
}
On PowerPC.
llvm-svn: 24633
2005-12-07 19:48:11 +00:00
Chris Lattner
57c882edf8
Only transform (sext (truncate x)) -> (sextinreg x) if before legalize or
...
if the target supports the resultant sextinreg
llvm-svn: 24632
2005-12-07 18:02:05 +00:00
Chris Lattner
3225733e65
X86 doesn't support sextinreg for 8-bit things either.
...
llvm-svn: 24631
2005-12-07 17:59:14 +00:00
Chris Lattner
cbd3d01a43
Teach the dag combiner to turn a truncate/sign_extend pair into a sextinreg
...
when the types match up. This allows the X86 backend to compile:
sbyte %toggle_value(sbyte* %tmp.1) {
%tmp.2 = load sbyte* %tmp.1
ret sbyte %tmp.2
}
to this:
_toggle_value:
mov %EAX, DWORD PTR [%ESP + 4]
movsx %EAX, BYTE PTR [%EAX]
ret
instead of this:
_toggle_value:
mov %EAX, DWORD PTR [%ESP + 4]
movsx %EAX, BYTE PTR [%EAX]
movsx %EAX, %AL
ret
noticed in Shootout/objinst.
-Chris
llvm-svn: 24630
2005-12-07 07:11:03 +00:00
Andrew Lenharth
26473b6b58
fix divide and remainder
...
llvm-svn: 24628
2005-12-06 23:27:39 +00:00
Chris Lattner
de085f0165
Silence another annoying GCC warning
...
llvm-svn: 24627
2005-12-06 20:56:18 +00:00
Andrew Lenharth
973cd1c845
more decent branches for FP. I might have to make some intermediate nodes to actually be able to use the DAG for FPcmp
...
llvm-svn: 24625
2005-12-06 20:43:30 +00:00
Andrew Lenharth
29b7ef0065
OK, this does wonders for broken stuff
...
llvm-svn: 24624
2005-12-06 20:40:34 +00:00
Sumant Kowshik
a69fcbdeb8
Collapsing node if variable length struct with final field of length zero
...
llvm-svn: 24621
2005-12-06 18:04:30 +00:00
Chris Lattner
d490d4f023
Fix test/Regression/Linker/2005-12-06-AppendingZeroLengthArrays.ll and
...
PR662. Thanks to Markus for providing me with a ton of files to
reproduce the problem!
llvm-svn: 24619
2005-12-06 17:30:58 +00:00
Nate Begeman
41b1cdc771
Teach the SelectionDAG ISel how to turn ConstantPacked values into
...
constant nodes with vector types. Also teach the asm printer how to print
ConstantPacked constant pool entries. This allows us to generate altivec
code such as the following, which adds a vector constantto a packed float.
LCPI1_0: <4 x float> < float 0.0e+0, float 0.0e+0, float 0.0e+0, float 1.0e+0 >
.space 4
.space 4
.space 4
.long 1065353216 ; float 1
.text
.align 4
.globl _foo
_foo:
lis r2, ha16(LCPI1_0)
la r2, lo16(LCPI1_0)(r2)
li r4, 0
lvx v0, r4, r2
lvx v1, r4, r3
vaddfp v0, v1, v0
stvx v0, r4, r3
blr
For the llvm code:
void %foo(<4 x float> * %a) {
entry:
%tmp1 = load <4 x float> * %a;
%tmp2 = add <4 x float> %tmp1, < float 0.0, float 0.0, float 0.0, float 1.0 >
store <4 x float> %tmp2, <4 x float> *%a
ret void
}
llvm-svn: 24616
2005-12-06 06:18:55 +00:00
Chris Lattner
fea33f7e64
Use new PPC-specific nodes to represent shifts which require the 6-bit
...
amount handling that PPC provides. These are generated by the lowering code
and prevents the dag combiner from assuming (rightfully) that the shifts
don't only look at 5 bits. This fixes a miscompilation of crafty with
the new front-end.
llvm-svn: 24615
2005-12-06 02:10:38 +00:00
Andrew Lenharth
e788bbf6ef
added instructions with inverted immediates
...
llvm-svn: 24614
2005-12-06 00:33:53 +00:00
Andrew Lenharth
08c4a775e6
yea, it helps to have your path set right when testing
...
llvm-svn: 24613
2005-12-05 23:41:45 +00:00
Andrew Lenharth
3c7c4d7508
These never trigger, but whatever
...
llvm-svn: 24612
2005-12-05 23:19:44 +00:00
Evan Cheng
c0c190239d
Remove unnecessary let hasCtrlDep=1 now it can be inferred.
...
llvm-svn: 24611
2005-12-05 23:09:43 +00:00
Andrew Lenharth
5bfcd1e63a
move this over to the dag
...
llvm-svn: 24609
2005-12-05 20:50:53 +00:00
Chris Lattner
077200737c
getRawValue zero extens for unsigned values, use getsextvalue so that we
...
know that small negative values fit into the immediate field of addressing
modes.
llvm-svn: 24608
2005-12-05 18:23:57 +00:00
Andrew Lenharth
9410433966
fix constant pool loads
...
llvm-svn: 24607
2005-12-05 17:51:02 +00:00
Chris Lattner
3539778883
Fix the #1 code quality problem that I have seen on X86 (and it also affects
...
PPC and other targets). In a particular, consider code like this:
struct Vector3 { double x, y, z; };
struct Matrix3 { Vector3 a, b, c; };
double dot(Vector3 &a, Vector3 &b) {
return a.x * b.x + a.y * b.y + a.z * b.z;
}
Vector3 mul(Vector3 &a, Matrix3 &b) {
Vector3 r;
r.x = dot( a, b.a );
r.y = dot( a, b.b );
r.z = dot( a, b.c );
return r;
}
void transform(Matrix3 &m, Vector3 *x, int n) {
for (int i = 0; i < n; i++)
x[i] = mul( x[i], m );
}
we compile transform to a loop with all of the GEP instructions for indexing
into 'm' pulled out of the loop (9 of them). Because isel occurs a bb at a time
we are unable to fold the constant index into the loads in the loop, leading to
PPC code that looks like this:
LBB3_1: ; no_exit.preheader
li r2, 0
addi r6, r3, 64 ;; 9 values live across the loop body!
addi r7, r3, 56
addi r8, r3, 48
addi r9, r3, 40
addi r10, r3, 32
addi r11, r3, 24
addi r12, r3, 16
addi r30, r3, 8
LBB3_2: ; no_exit
lfd f0, 0(r30)
lfd f1, 8(r4)
fmul f0, f1, f0
lfd f2, 0(r3) ;; no constant indices folded into the loads!
lfd f3, 0(r4)
lfd f4, 0(r10)
lfd f5, 0(r6)
lfd f6, 0(r7)
lfd f7, 0(r8)
lfd f8, 0(r9)
lfd f9, 0(r11)
lfd f10, 0(r12)
lfd f11, 16(r4)
fmadd f0, f3, f2, f0
fmul f2, f1, f4
fmadd f0, f11, f10, f0
fmadd f2, f3, f9, f2
fmul f1, f1, f6
stfd f0, 0(r4)
fmadd f0, f11, f8, f2
fmadd f1, f3, f7, f1
stfd f0, 8(r4)
fmadd f0, f11, f5, f1
addi r29, r4, 24
stfd f0, 16(r4)
addi r2, r2, 1
cmpw cr0, r2, r5
or r4, r29, r29
bne cr0, LBB3_2 ; no_exit
uh, yuck. With this patch, we now sink the constant offsets into the loop, producing
this code:
LBB3_1: ; no_exit.preheader
li r2, 0
LBB3_2: ; no_exit
lfd f0, 8(r3)
lfd f1, 8(r4)
fmul f0, f1, f0
lfd f2, 0(r3)
lfd f3, 0(r4)
lfd f4, 32(r3) ;; much nicer.
lfd f5, 64(r3)
lfd f6, 56(r3)
lfd f7, 48(r3)
lfd f8, 40(r3)
lfd f9, 24(r3)
lfd f10, 16(r3)
lfd f11, 16(r4)
fmadd f0, f3, f2, f0
fmul f2, f1, f4
fmadd f0, f11, f10, f0
fmadd f2, f3, f9, f2
fmul f1, f1, f6
stfd f0, 0(r4)
fmadd f0, f11, f8, f2
fmadd f1, f3, f7, f1
stfd f0, 8(r4)
fmadd f0, f11, f5, f1
addi r6, r4, 24
stfd f0, 16(r4)
addi r2, r2, 1
cmpw cr0, r2, r5
or r4, r6, r6
bne cr0, LBB3_2 ; no_exit
This is much nicer as it reduces register pressure in the loop a lot. On X86,
this takes the function from having 9 spilled registers to 2. This should help
some spec programs on X86 (gzip?)
This is currently only enabled with -enable-gep-isel-opt to allow perf testing
tonight.
llvm-svn: 24606
2005-12-05 07:10:48 +00:00
Chris Lattner
7d4d93c52c
Add a flag to Module::getGlobalVariable to allow it to return vars with
...
internal linkage.
Patch provided by Evan Jones, thanks!
llvm-svn: 24604
2005-12-05 05:30:21 +00:00
Chris Lattner
165998207e
Wrap a long line, never internalize llvm.used.
...
llvm-svn: 24602
2005-12-05 05:07:38 +00:00
Chris Lattner
3c0b8f577d
Several things:
...
1. Remove redundant type casts now that PR673 is implemented.
2. Implement the OUT*ir instructions correctly. The port number really
*is* a 16-bit value, but the patterns should only match if the number
is 0-255. Update the patterns so they now match.
3. Fix patterns for shifts to reflect that the shift amount is always an
i8, not an i16 as they were believed to be before. This previous fib
stopped working when we started knowing that CL has type i8.
4. Change use of i16i8imm in SH*ri patterns to all be imm.
llvm-svn: 24599
2005-12-05 02:40:25 +00:00
Chris Lattner
282c2af4d8
On some targets (e.g. X86), shift amounts are not the same as the value
...
being shifted. Don't assume they are.
llvm-svn: 24598
2005-12-05 02:37:26 +00:00
Chris Lattner
c54cddd25b
Add some explicit type casts so that tblgen knows the type of the shiftamount, which is not necessarily the same as the type being shifted.
...
llvm-svn: 24595
2005-12-05 02:34:29 +00:00
Chris Lattner
f3322af5c6
Add some explicit type casts so that tblgen knows the type of the shift
...
amount, which is not necessarily the same as the type being shifted.
llvm-svn: 24594
2005-12-05 02:34:05 +00:00
Chris Lattner
efc86f5f7a
The basic fneg cases are already autogen'd
...
llvm-svn: 24592
2005-12-04 19:04:38 +00:00
Chris Lattner
f979794717
Autogen matching code for ADJCALLSTACK[UP|DOWN], thanks to Evan's tblgen
...
improvements.
llvm-svn: 24591
2005-12-04 19:01:59 +00:00
Chris Lattner
fd857daa0d
Finish moving uncond br over to .td file, remove from .cpp file.
...
llvm-svn: 24590
2005-12-04 18:48:01 +00:00
Chris Lattner
d9d18aff6a
Define BR in the .td file now that Evan made tblgen smarter.
...
llvm-svn: 24589
2005-12-04 18:42:54 +00:00
Evan Cheng
95cb763818
Added isel patterns for RET, JMP, and WRITEPORT.
...
llvm-svn: 24588
2005-12-04 08:19:43 +00:00
Evan Cheng
e8531381e2
* Added instruction property hasCtrlDep for those which r/w control-flow
...
chains.
* Added DAG node property SDNPHasChain for nodes which r/w control-flow
chains.
* Renamed SDTVT to SDTOther.
* Added several new SDTypeProfiles for BR, BRCOND, RET, and WRITEPORT.
* Added SDNode definitions for BR, BRCOND, RET, and WRITEPORT.
llvm-svn: 24586
2005-12-04 08:13:17 +00:00
Chris Lattner
7e79292fef
Fix PR672 another way which should be more robust
...
llvm-svn: 24585
2005-12-04 06:03:50 +00:00
Chris Lattner
8782b782cd
dbg.stoppoint returns a value, don't forget to init it
...
llvm-svn: 24583
2005-12-03 18:50:48 +00:00
Chris Lattner
2820b8c855
Fix SimplifyCFG/2005-12-03-IncorrectPHIFold.ll
...
llvm-svn: 24581
2005-12-03 18:25:58 +00:00
Chris Lattner
ecfc7e56c5
Fix test/Regression/ExecutionEngine/2005-12-02-TailCallBug.ll and PR672.
...
This also fixes 177.mesa, the only program that fails with --enable-x86-fastcc
turned on. Given a clean nightly tester run, we should be able to turn it
on by default!
llvm-svn: 24578
2005-12-03 07:15:55 +00:00
Andrew Lenharth
f9b27d7011
bah, must generate all results
...
llvm-svn: 24574
2005-12-02 06:08:08 +00:00
Andrew Lenharth
73420b3795
cycle counter fix
...
llvm-svn: 24573
2005-12-02 04:56:24 +00:00
Chris Lattner
986cb40953
add a note
...
llvm-svn: 24572
2005-12-02 00:11:20 +00:00
Chris Lattner
0142afd6c1
Don't remove two operand, two result nodes from the binary ops map. These
...
should come from the arbitrary ops map.
This fixes Regression/CodeGen/PowerPC/2005-12-01-Crash.ll
llvm-svn: 24571
2005-12-01 23:14:50 +00:00
Chris Lattner
05b0b4575b
Promote line and column number information for our friendly 64-bit targets.
...
llvm-svn: 24568
2005-12-01 18:21:35 +00:00
Chris Lattner
93feffb4ed
IA64 doesn't support the LOCATION node, and for some reason the ISelPattern
...
stuff isn't using ISelLowering.cpp
llvm-svn: 24567
2005-12-01 18:19:53 +00:00
Chris Lattner
df9287836e
Make sure these get added into the codegenmap when appropriate
...
llvm-svn: 24566
2005-12-01 18:09:22 +00:00
Chris Lattner
9d0d715e83
This is a bugfix for SelectNodeTo. In certain situations, we could be
...
selecting a node and use a mix of getTargetNode() and SelectNodeTo. Because
SelectNodeTo didn't check the CSE maps for a preexisting node and didn't insert
its result into the CSE maps, we would sometimes miss a CSE opportunity.
This is extremely rare, but worth fixing for completeness.
llvm-svn: 24565
2005-12-01 18:00:57 +00:00
Andrew Lenharth
7bb09912c8
major think-o
...
llvm-svn: 24564
2005-12-01 17:48:51 +00:00
Nate Begeman
006bb04f3a
Support multiple ValueTypes per RegisterClass, needed for upcoming vector
...
work. This change has no effect on generated code.
llvm-svn: 24563
2005-12-01 04:51:06 +00:00
Nate Begeman
aa5f8f2a26
Cosmetic change, better reflects actual values
...
llvm-svn: 24562
2005-12-01 04:48:26 +00:00
Chris Lattner
bd099102f0
Fix a regression caused by a patch earlier today
...
llvm-svn: 24561
2005-12-01 03:50:19 +00:00
Andrew Lenharth
ce68ef8073
Flags where I think I need them, quick, before the nightly tester starts
...
llvm-svn: 24560
2005-12-01 01:53:10 +00:00
Evan Cheng
4b02426130
Proper support for shifts with register shift value.
...
llvm-svn: 24559
2005-12-01 00:43:55 +00:00
Evan Cheng
d94aa71e1a
Use a getCopyToReg() variant to generate a flaggy CopyToReg node.
...
llvm-svn: 24558
2005-12-01 00:41:50 +00:00
Chris Lattner
4b25924d2a
SelectNodeTo now returns its result, we must pay attention to it.
...
llvm-svn: 24552
2005-11-30 23:04:38 +00:00
Chris Lattner
a75694aa16
Pay attn to the node returned by SelectNodeTo
...
llvm-svn: 24551
2005-11-30 23:02:08 +00:00
Chris Lattner
af2e0373dd
SelectNodeTo now returns its result, we must pay attention to it.
...
llvm-svn: 24550
2005-11-30 22:59:19 +00:00
Chris Lattner
e318977940
SelectNodeTo now returns N. Use it instead of return N directly.
...
llvm-svn: 24549
2005-11-30 22:53:06 +00:00
Chris Lattner
be5dd5da19
Make SelectNodeTo return N
...
llvm-svn: 24548
2005-11-30 22:45:14 +00:00
Chris Lattner
3713e6b49c
Fix Regression/CodeGen/PowerPC/2005-11-30-vastart-crash.ll
...
llvm-svn: 24547
2005-11-30 20:40:54 +00:00
Chris Lattner
dc4ffef633
Fix a bug where we didn't realize that vaarg reads memory. This fixes
...
Transforms/DeadStoreElimination/2005-11-30-vaarg.ll
llvm-svn: 24545
2005-11-30 19:38:22 +00:00
Nate Begeman
11695c0537
Fix a typo in my latest change
...
llvm-svn: 24542
2005-11-30 18:57:39 +00:00
Nate Begeman
6f8c1ace6e
No longer track value types for asm printer operands, and remove them as
...
an argument to every operand printing function. Requires some slight
tweaks to x86, the only user.
llvm-svn: 24541
2005-11-30 18:54:35 +00:00
Chris Lattner
c174048430
CALLSEQ_START/END nodes don't get memoized, do not add them in when
...
replaceAllUses'ing.
llvm-svn: 24539
2005-11-30 18:20:52 +00:00
Andrew Lenharth
7ffe3affda
remove redundant code
...
llvm-svn: 24538
2005-11-30 17:14:11 +00:00
Andrew Lenharth
6ee8566cae
At long last, you can say that f32 isn't supported for setcc
...
llvm-svn: 24537
2005-11-30 17:12:26 +00:00
Andrew Lenharth
ede966e8ee
Make typesafe that which isn't: FCMOVxx
...
llvm-svn: 24536
2005-11-30 17:11:20 +00:00
Andrew Lenharth
873ed82a36
FPSelect and more custom lowering
...
llvm-svn: 24535
2005-11-30 16:10:29 +00:00
Nate Begeman
1064d6ec43
First chunk of actually generating vector code for packed types. These
...
changes allow us to generate the following code:
_foo:
li r2, 0
lvx v0, r2, r3
vaddfp v0, v0, v0
stvx v0, r2, r3
blr
for this llvm:
void %foo(<4 x float>* %a) {
entry:
%tmp1 = load <4 x float>* %a
%tmp2 = add <4 x float> %tmp1, %tmp1
store <4 x float> %tmp2, <4 x float>* %a
ret void
}
llvm-svn: 24534
2005-11-30 08:22:07 +00:00
Andrew Lenharth
6db615df14
All sorts of stuff.
...
Getting in on the custom lowering thing, yay
evilness with fp setcc, yuck
trivial int select, hmmm
in memory args for functions, yay
DIV and REM, always handy. They should be custom lowered though.
Lots more stuff compiles now (go go single source!). Of course, none of it
probably works, but that is what the nightly tester can find out :)
llvm-svn: 24533
2005-11-30 07:19:56 +00:00
Andrew Lenharth
8d17c70171
add support for custom lowering SINT_TO_FP
...
llvm-svn: 24531
2005-11-30 06:43:03 +00:00
Reid Spencer
3fd1b4c9bf
Fix a problem with llvm-ranlib that (on some platforms) caused the archive
...
file to become corrupted due to interactions between mmap'd memory segments
and file descriptors closing. The problem is completely avoiding by using
a third temporary file.
Patch provided by Evan Jones
llvm-svn: 24527
2005-11-30 05:21:10 +00:00
Chris Lattner
9c7af08bc9
Fix a bug in a recent patch that broke shifts
...
llvm-svn: 24526
2005-11-30 05:11:18 +00:00
Evan Cheng
4eb7af9bc9
Added support to STORE and shifts to DAG to DAG isel.
...
llvm-svn: 24525
2005-11-30 02:51:20 +00:00
Evan Cheng
11d61613af
Fixed a bug introduced by my last commit: TargetGlobalValues should key on
...
GlobalValue * and index pair. Update getGlobalAddress() for symmetry.
llvm-svn: 24524
2005-11-30 02:49:21 +00:00
Evan Cheng
0e0de2f3f0
Added an index field to GlobalAddressSDNode so it can represent X+12, etc.
...
llvm-svn: 24523
2005-11-30 02:04:11 +00:00
Evan Cheng
d2cb70513d
Fixed a minor bug: - -offset != offset iff offset == MININT
...
llvm-svn: 24522
2005-11-30 01:59:00 +00:00
Nate Begeman
048b26387b
Represent the encoding of the SPR instructions as they actually are, so
...
that we can use the correct SPR numbers in the InstrInfo.td file. This is
necessary to support VRsave.
llvm-svn: 24521
2005-11-29 22:42:50 +00:00
Evan Cheng
72ab335858
Add more X86 ISel patterns.
...
llvm-svn: 24520
2005-11-29 19:38:52 +00:00
Nate Begeman
3e7db9c6d5
Hook up one type, v4f32, to the VR RegisterClass for now.
...
llvm-svn: 24517
2005-11-29 08:17:20 +00:00
Nate Begeman
c138118cdb
Add the remainder of the AltiVec 4 x float instructions. Further
...
enhancements will be necessary to teach the code generator that since
there is no fmul, it will have to do vmaddfp, adding +0.0.
llvm-svn: 24516
2005-11-29 08:04:45 +00:00
Chris Lattner
435b402e1f
Add support for a new STRING and LOCATION node for line number support, patch
...
contributed by Daniel Berlin, with a few cleanups here and there by me.
llvm-svn: 24515
2005-11-29 06:21:05 +00:00
Chris Lattner
9c415364cf
No targets support line number info yet.
...
llvm-svn: 24513
2005-11-29 06:16:21 +00:00
Nate Begeman
89b049af90
Add the majority of the vector machien value types we expect to support,
...
and make a few changes to the legalization machinery to support more than
16 types.
llvm-svn: 24511
2005-11-29 05:45:29 +00:00
Evan Cheng
1d4af53444
Fixed a comment bug:
...
createPPCPatternInstructionSelector -> createPPCISelPattern
llvm-svn: 24510
2005-11-29 04:59:46 +00:00
Chris Lattner
c418b5d938
refix typo
...
llvm-svn: 24505
2005-11-29 00:42:30 +00:00
Chris Lattner
7a18a25d33
don't say this is i128, because it isn't yet. Hopefully nate will change
...
this to be something sane, but in the mean time it is unused, so safe to
make something bogus.
llvm-svn: 24504
2005-11-29 00:41:40 +00:00
Chris Lattner
d3bc8cb26d
revert my change for the time being, which broke the build
...
llvm-svn: 24503
2005-11-29 00:24:08 +00:00
Chris Lattner
50986909dc
fix a typo :)
...
llvm-svn: 24501
2005-11-28 22:42:15 +00:00
Andrew Lenharth
d251192910
a few more comments on the interfaces and functions
...
llvm-svn: 24500
2005-11-28 18:10:59 +00:00
Andrew Lenharth
517caef495
Added documented rsprofiler interface. Also remove new profiler passes, the
...
old ones have been updated to implement the interface.
llvm-svn: 24499
2005-11-28 18:00:38 +00:00
Jeff Cohen
7ff44ec372
Fix VC++ warning.
...
llvm-svn: 24496
2005-11-28 06:45:57 +00:00
Chris Lattner
820c94e467
Add a missed optimization
...
llvm-svn: 24495
2005-11-28 04:52:39 +00:00
Andrew Lenharth
93e59f6032
Random sampling (aka Arnold and Ryder) profiling. This is still preliminary, but it works on spec on x86 and alpha. The idea is to allow profiling passes to remember what profiling they inserted, then a random sampling framework is inserted which consists of duplicated basic blocks (without profiling), such that at each backedge in the program and entry into every function, the framework chooses whether to use the instrumented code or the instrumentation free code. The goal of such a framework is to make it reasonably cheap to do random sampling of very expensive profiling products (such as load-value profiling).
...
The code is organized into 3 parts (2 passes)
1) a linked set of profiling passes, which implement an analysis group (linked, like alias analysis are). These insert profiling into the program, and remember what they inserted, so that at a later time they can be queried about any instruction.
2) a pass that handles inserting the random sampling framework. This also has options to control how random samples are choosen. Currently implemented are Global counters, register allocated global counters, and read cycle counter (see? there was a reason for it).
The profiling passes are almost identical to the existing ones (block, function, and null profiling is supported right now), and they are valid passes without the sampling framework (hence the existing passes can be unified with the new ones, not done yet).
Some things are a bit ugly still, but that should be fixed up soon enough.
Other todo? making the counter values not "magic 2^16 -1" values, but dynamically choosable.
llvm-svn: 24493
2005-11-28 00:58:09 +00:00
Nate Begeman
11fd6b22b1
Small tweaks noticed while on the plane.
...
llvm-svn: 24492
2005-11-26 22:39:34 +00:00
Andrew Lenharth
5fc3794e71
since reg2mem requires it, might as well mention that it preserves it
...
llvm-svn: 24491
2005-11-25 16:04:54 +00:00
Duraid Madina
0c88f150c6
add support for dynamic_stackalloc to the dag isel (thanks andrew ;)
...
next up: support argument passing in memory, not just registers
llvm-svn: 24490
2005-11-25 07:49:25 +00:00
Nate Begeman
8492fd30ab
Some first bits of AltiVec stuff: Instruction Formats, Encodings, and
...
Registers. Apologies to Jim if the scheduling info so far isn't accurate.
There's a few more things like VRsave support that need to be finished up
in my local tree before I can commit code that Does The Right Thing for
turning 4 x float into the various altivec packed float instructions.
llvm-svn: 24489
2005-11-23 05:29:52 +00:00
Andrew Lenharth
061029dee2
Reg2Mem is something a pass may depend on, so allow that
...
llvm-svn: 24488
2005-11-22 22:14:23 +00:00
Andrew Lenharth
71b09bbb07
turns out, demotion and invokes and critical edges don't mix
...
llvm-svn: 24487
2005-11-22 21:45:19 +00:00
Andrew Lenharth
d6a0308470
Fix warning, the better way. Really, this is what this instruction is for, so use it
...
llvm-svn: 24486
2005-11-22 20:59:00 +00:00
Andrew Lenharth
03390557fa
Fix warning
...
llvm-svn: 24485
2005-11-22 20:56:05 +00:00
Nate Begeman
d37c13154a
Check in code to scalarize arbitrarily wide packed types for some simple
...
vector operations (load, add, sub, mul).
This allows us to codegen:
void %foo(<4 x float> * %a) {
entry:
%tmp1 = load <4 x float> * %a;
%tmp2 = add <4 x float> %tmp1, %tmp1
store <4 x float> %tmp2, <4 x float> *%a
ret void
}
on ppc as:
_foo:
lfs f0, 12(r3)
lfs f1, 8(r3)
lfs f2, 4(r3)
lfs f3, 0(r3)
fadds f0, f0, f0
fadds f1, f1, f1
fadds f2, f2, f2
fadds f3, f3, f3
stfs f0, 12(r3)
stfs f1, 8(r3)
stfs f2, 4(r3)
stfs f3, 0(r3)
blr
llvm-svn: 24484
2005-11-22 18:16:00 +00:00
Andrew Lenharth
0294e33ea4
massive DAGISel patch. lots and lots more stuff compiles now
...
llvm-svn: 24483
2005-11-22 04:20:06 +00:00
Nate Begeman
07890bbec4
Rather than attempting to legalize 1 x float, make sure the SD ISel never
...
generates it. Make MVT::Vector expand-only, and remove the code in
Legalize that attempts to legalize it.
The plan for supporting N x Type is to continually epxand it in ExpandOp
until it gets down to 2 x Type, where it will be scalarized into a pair of
scalars.
llvm-svn: 24482
2005-11-22 01:29:36 +00:00
Chris Lattner
ac6cb46429
Use HasDotTypeDotSizeDirective instead of forELF
...
llvm-svn: 24481
2005-11-21 23:06:54 +00:00
Chris Lattner
78161dbc84
Remove a level of indentation by using a continue.
...
llvm-svn: 24479
2005-11-21 22:48:18 +00:00
Chris Lattner
40f8c8450d
Simplify the subtarget info, allow the asmwriter to do some target sensing
...
based on TargetType.
llvm-svn: 24478
2005-11-21 22:43:58 +00:00
Chris Lattner
99be8f766f
Use subtarget information computed by X86Subtarget instead of rolling our own.
...
llvm-svn: 24477
2005-11-21 22:39:40 +00:00
Chris Lattner
3eb876117a
Make the X86 subtarget compute the basic target type: ELF, Cygwin, Darwin,
...
or native Win32
llvm-svn: 24476
2005-11-21 22:31:58 +00:00
Chris Lattner
ebc39f5a9c
Add a forELF flag, allowing the removal of forCygwin and simplification of
...
conditionals.
llvm-svn: 24475
2005-11-21 22:19:48 +00:00
Chris Lattner
7df25ab429
simplify and genericize this code
...
llvm-svn: 24473
2005-11-21 19:50:31 +00:00
Duraid Madina
8edbf805d7
add support for div/rem to the dag->dag isel. yay.
...
llvm-svn: 24472
2005-11-21 14:14:54 +00:00
Duraid Madina
f28b3bd8b4
I think I know what you meant here, but just to be safe I'll let you
...
do it. :)
<_sabre_> excuses excuses
llvm-svn: 24471
2005-11-21 14:09:40 +00:00
Chris Lattner
b9db67a045
Eliminate unneeded intermediate class. Move doFinalizeMethod to bottom of
...
file.
llvm-svn: 24470
2005-11-21 08:40:17 +00:00
Chris Lattner
b55de47595
Start using shared asmprinter Constant Pool emitter, use shorter cpi names.
...
llvm-svn: 24469
2005-11-21 08:38:26 +00:00
Chris Lattner
4a7eb5132b
prune #include
...
llvm-svn: 24468
2005-11-21 08:33:17 +00:00
Chris Lattner
8a5f3c1b68
Switch to using the shared constant pool printer, along with using shorter
...
CPI ids
llvm-svn: 24467
2005-11-21 08:32:23 +00:00
Chris Lattner
882b9fa977
Switch to using the generic constant pool emitter impl, use shorter
...
CPI names
llvm-svn: 24466
2005-11-21 08:29:17 +00:00
Chris Lattner
ef83ebd45d
Use generic constant pool emission code in the AsmPrinter class.
...
llvm-svn: 24465
2005-11-21 08:26:15 +00:00
Chris Lattner
f2991cee1f
Allow target to customize directive used to switch to arbitrary section in SwitchSection,
...
add generic constant pool emitter
llvm-svn: 24464
2005-11-21 08:25:09 +00:00
Chris Lattner
ffbfa71866
Use the FunctionNumber provided by the AsmPrinter class
...
llvm-svn: 24462
2005-11-21 08:14:07 +00:00
Chris Lattner
08adbd13ff
increment the function number in SetupMachineFunction
...
llvm-svn: 24461
2005-11-21 08:13:27 +00:00
Chris Lattner
dd3bf8e4a2
Use CommentString where possible, fix a bug where aix mode wouldn't assemble
...
due to basic blocks being misnamed.
llvm-svn: 24459
2005-11-21 08:02:41 +00:00
Chris Lattner
a0222a1698
unify the darwin and aix constant pool printers
...
llvm-svn: 24458
2005-11-21 07:57:37 +00:00
Chris Lattner
bb644e39c0
Adjust to capitalized asmprinter method names
...
llvm-svn: 24457
2005-11-21 07:51:36 +00:00
Chris Lattner
99946fb63f
Adjust to capitalized AsmPrinter method names
...
llvm-svn: 24456
2005-11-21 07:51:23 +00:00
Chris Lattner
d365627d3e
Use PrivateGlobalPrefix for basic block labels. This allows the x86 darwin
...
port to properly use L for the bb prefix instead of .
llvm-svn: 24454
2005-11-21 07:43:59 +00:00
Chris Lattner
c2bc19af57
use PrivateGlobalPrefix for basic blocks
...
llvm-svn: 24453
2005-11-21 07:41:05 +00:00
Chris Lattner
2b53ea99b0
Use PrivateGlobalPrefix for basic block labels
...
llvm-svn: 24452
2005-11-21 07:39:22 +00:00
Chris Lattner
9885c97088
Use PrivateGlobalPrefix for basic blocks
...
llvm-svn: 24451
2005-11-21 07:38:08 +00:00
Chris Lattner
a43b832f7f
Switch to the new shared SwitchSection
...
llvm-svn: 24450
2005-11-21 07:30:28 +00:00
Chris Lattner
2bccd73dbb
Start using SwitchSection, allowing globals and functions to be emitted
...
to specific sections. Delete some dead functions copied from the X86 backend.
llvm-svn: 24449
2005-11-21 07:26:04 +00:00
Chris Lattner
050bf2faf8
convert the rest of this over to use SwitchSection
...
llvm-svn: 24448
2005-11-21 07:16:34 +00:00
Chris Lattner
024e32e118
Start using the AsmPrinter shared SwitchSection code. This allows the X86
...
backend to implement global variables in sections.
llvm-svn: 24447
2005-11-21 07:11:11 +00:00
Chris Lattner
b650241f8b
This is now implemented in common codegen code
...
llvm-svn: 24446
2005-11-21 07:06:58 +00:00
Chris Lattner
2ea5c99eca
Add section switching to common code generator code. Add a couple of
...
asserts.
llvm-svn: 24445
2005-11-21 07:06:27 +00:00
Chris Lattner
2c0b435ba6
Rename SwitchSection -> switchSection to avoid conflicting with a future
...
change.
llvm-svn: 24443
2005-11-21 06:55:27 +00:00
Chris Lattner
8a4995e42a
Start using PrivateGlobalPrefix correctly
...
llvm-svn: 24442
2005-11-21 06:51:52 +00:00
Chris Lattner
41cb115afb
set PrivateGlobalPrefix on darwin, use it when printing out CP references
...
llvm-svn: 24441
2005-11-21 06:47:58 +00:00
Chris Lattner
618981fd03
Naturally align doubles in the constant pool, set PrivateGlobalPrefix on
...
darwin, use it when printing the constant pool indices so the labels are
appropriately private, emit cp entries to .const instead of .data on darwin
and only emit a single .section for the constant pool, not one for each
entry.
llvm-svn: 24440
2005-11-21 06:46:22 +00:00
Chris Lattner
6c1ca888d4
Lower READCYCLECOUNTER correctly, preserving the chain result
...
llvm-svn: 24438
2005-11-20 22:57:19 +00:00
Chris Lattner
44c28c22b7
Legalize MERGE_VALUES, expand READCYCLECOUNTER correctly, so it doesn't
...
break control dependence.
llvm-svn: 24437
2005-11-20 22:56:56 +00:00
Chris Lattner
d1061ac8d1
encode rdtsc correctly
...
llvm-svn: 24435
2005-11-20 22:13:18 +00:00
Chris Lattner
6df9e11989
use chain operands to ensure the copies don't wander from the rdtsc instruction.
...
llvm-svn: 24434
2005-11-20 22:01:40 +00:00
Andrew Lenharth
0bf68ae434
The second patch of X86 support for read cycle counter.
...
llvm-svn: 24430
2005-11-20 21:41:10 +00:00
Andrew Lenharth
627cbd49b1
The first patch of X86 support for read cycle counter
...
llvm-svn: 24429
2005-11-20 21:32:07 +00:00
Chris Lattner
a8d37d748f
more progress towards bug 291 being finished. Patch by Owen Anderson,
...
HAVE_GV case fixed up by me.
llvm-svn: 24428
2005-11-20 03:45:52 +00:00
Chris Lattner
19baba67b5
Unbreak codegen of bools. This should fix the llc/jit/llc-beta failures
...
from last night.
llvm-svn: 24427
2005-11-19 18:40:42 +00:00
Chris Lattner
377bdbff91
Improve Selection DAG printer portability. Patch by Owen Anderson!
...
llvm-svn: 24425
2005-11-19 07:44:09 +00:00
Chris Lattner
d7102c4980
Teach the x86 backend about the register constraints of its addressing mode.
...
Patch by Evan Cheng
llvm-svn: 24423
2005-11-19 07:01:30 +00:00
Chris Lattner
252d88c68e
Capture more operand info, patch by Evan Cheng
...
llvm-svn: 24422
2005-11-19 07:00:10 +00:00
Chris Lattner
a22eae0163
Teach the graph viewer to handle register operands that are zero.
...
llvm-svn: 24421
2005-11-19 06:58:46 +00:00
Chris Lattner
301015a703
Silence a bogus warning
...
llvm-svn: 24420
2005-11-19 05:51:46 +00:00
Chris Lattner
3f0f71b92b
Add load and other support to the dag-dag isel. Patch contributed by Evan
...
Cheng!
llvm-svn: 24419
2005-11-19 02:11:08 +00:00
Chris Lattner
f090f7eb0e
Add some method variants, patch by Evan Cheng
...
llvm-svn: 24418
2005-11-19 01:44:53 +00:00
Nate Begeman
b2e089c31b
Teach LLVM how to scalarize packed types. Currently, this only works on
...
packed types with an element count of 1, although more generic support is
coming. This allows LLVM to turn the following code:
void %foo(<1 x float> * %a) {
entry:
%tmp1 = load <1 x float> * %a;
%tmp2 = add <1 x float> %tmp1, %tmp1
store <1 x float> %tmp2, <1 x float> *%a
ret void
}
Into:
_foo:
lfs f0, 0(r3)
fadds f0, f0, f0
stfs f0, 0(r3)
blr
llvm-svn: 24416
2005-11-19 00:36:38 +00:00
Chris Lattner
9c37f23645
Fix a crash building 176.gcc due to my recent patch, which only fixed
...
half the problem.
llvm-svn: 24414
2005-11-18 18:30:47 +00:00
Andrew Lenharth
6bc51c6f34
prevent latent switch creation
...
llvm-svn: 24413
2005-11-18 13:57:03 +00:00
Nate Begeman
127321b14c
Split out the shift code from visitBinary.
...
llvm-svn: 24412
2005-11-18 07:42:56 +00:00
Chris Lattner
3e9e8bd25c
Implement a refinement to the mem2reg algorithm for cases where an alloca
...
has a single def. In this case, look for uses that are dominated by the def
and attempt to rewrite them to directly use the stored value.
This speeds up mem2reg on these values and reduces the number of phi nodes
inserted. This should address PR665.
llvm-svn: 24411
2005-11-18 07:31:42 +00:00
Chris Lattner
31dc3827d3
This needs proper dominance
...
llvm-svn: 24410
2005-11-18 07:29:44 +00:00
Chris Lattner
b14c41378d
post-dom-frontiers requires proper post-dominance
...
llvm-svn: 24409
2005-11-18 07:28:26 +00:00
Chris Lattner
f519d1c6ef
This requires proper dominance
...
llvm-svn: 24408
2005-11-18 07:27:53 +00:00
Chris Lattner
57ce97862d
add more patterns, patch by Evan Cheng.
...
llvm-svn: 24406
2005-11-18 01:04:42 +00:00
Nate Begeman
227f1bdd2f
Also add the new vector value type here, for completeness.
...
llvm-svn: 24405
2005-11-18 00:53:32 +00:00
Chris Lattner
cdde9990b7
only use dyld stubs if not in ppc-static mode. This completes support for
...
non-static codegen.
llvm-svn: 24403
2005-11-17 19:40:30 +00:00
Chris Lattner
bca0be812d
This was checking the wrong GEP expression. Fixing this fixes a gccas crash
...
compiling mysql reported by Ted Kremenek.
llvm-svn: 24402
2005-11-17 19:35:42 +00:00
Chris Lattner
6ab87fa360
refactor call operand handling to eliminate special cases from printOp.
...
llvm-svn: 24401
2005-11-17 19:25:59 +00:00
Chris Lattner
bd9efdb64c
disentangle call operands from branch operands a bit
...
llvm-svn: 24400
2005-11-17 19:16:08 +00:00
Chris Lattner
3570cf456b
add an option to generate completely non-pic code, corresponding to what
...
gcc -static produces on PPC. This is used for building kexts and other things.
With this, materializing the address of a global looks like:
lis r2, ha16(L_H$non_lazy_ptr)
la r3, lo16(L_H$non_lazy_ptr)(r2)
we're still emitting stubs for functions, which is wrong. That is next.
llvm-svn: 24399
2005-11-17 18:55:48 +00:00
Chris Lattner
8f8ed28a64
Fix a bug that resistor on IRC hit where we tried to create token factor
...
nodes of load results, not of their chain results.
llvm-svn: 24398
2005-11-17 18:30:17 +00:00
Chris Lattner
5aba6ae3b3
Enable global address legalization, fixing a todo and allowing the removal
...
of some code. This exposes the implicit load from the stubs to the DAG, allowing
them to be optimized by the dag combiner. It also moves darwin specific stuff
out of the isel into the legalizer, and allows more to be moved to the .td file.
llvm-svn: 24397
2005-11-17 18:26:56 +00:00
Chris Lattner
0fe88e3f32
Teach the selector to fold lo(g) into load instruction immediate fields
...
llvm-svn: 24396
2005-11-17 18:02:16 +00:00
Chris Lattner
4b11fa284d
Generate LA and ADDIS when possible.
...
llvm-svn: 24395
2005-11-17 17:52:01 +00:00
Chris Lattner
3648c20472
Use the right accessor to create this node
...
llvm-svn: 24394
2005-11-17 17:51:38 +00:00
Chris Lattner
595088aa0f
Add an initial hack at legalizing GlobalAddress into the appropriate nodes
...
on Darwin to remove smarts from the isel. This is currently disabled by
default (uncomment setOperationAction(ISD::GlobalAddress to enable it).
tblgen needs to become smarter about tglobaladdr nodes and bigger patterns
needed to be added to the .td file. However, we can currently emit stuff like
this: :)
li r2, lo16(L_x$non_lazy_ptr)
lis r3, ha16(L_x$non_lazy_ptr)
lwzx r2, r3, r2
The obvious improvements will follow.
llvm-svn: 24390
2005-11-17 07:30:41 +00:00
Chris Lattner
840458276c
Add globaladdress and targetglobaladdress nodes for dag patterns
...
llvm-svn: 24389
2005-11-17 07:20:15 +00:00
Chris Lattner
63ed749ce0
LI could theoretically be used for the lo-part of a global address, just like
...
lis can be used for the high part.
llvm-svn: 24388
2005-11-17 07:04:43 +00:00
Chris Lattner
45ca1c0194
Allow targets to custom legalize leaf nodes like GlobalAddress.
...
llvm-svn: 24387
2005-11-17 06:41:44 +00:00
Chris Lattner
b7025749e1
When lowering direct calls, lower them to use a targetglobaladress directly
...
instead of a globaladdress. This has no effect on the generated code at all.
llvm-svn: 24386
2005-11-17 05:56:14 +00:00
Chris Lattner
4ff65ec745
Teach legalize about targetglobaladdress
...
llvm-svn: 24385
2005-11-17 05:52:24 +00:00
Chris Lattner
2bf458af92
Add patterns for some 16-bit immediate instructions, patch contributed by
...
Evan Cheng.
llvm-svn: 24384
2005-11-17 02:01:55 +00:00
Chris Lattner
5930d3df3d
Add patterns for several simple instructions that take i32 immediates.
...
Patch contributed by Evan Cheng!
llvm-svn: 24382
2005-11-16 22:59:19 +00:00
Andrew Lenharth
59eefd4787
who would have thought you would want to write into globals too
...
llvm-svn: 24381
2005-11-16 21:15:53 +00:00
Chris Lattner
f2b62f317c
when debugging lower dbg intrinsics to calls
...
llvm-svn: 24377
2005-11-16 07:22:30 +00:00
Chris Lattner
168fe79229
tell selectiondag when we're debugging
...
llvm-svn: 24376
2005-11-16 07:21:47 +00:00
Chris Lattner
fdc8b19ad6
indicate when a tool is a debug build.
...
llvm-svn: 24374
2005-11-16 06:36:47 +00:00
Chris Lattner
6b41edb30f
* Fix DerivedType::dropAllTypeUses to not change the number of types in a
...
type when it gets refined. This allows us to hash on this crucial value.
* Fix several issues in TypeMap::RefineAbstractType that prevent it from
handling hash values that change correctly.
* Define hashTypeStructure to not always return 0. :)
This last part (which depends on the first two) speeds up gccld time on eon
from 3.78s to 2.75s with a release build (a 28% speedup!). This resolves
PR474.
llvm-svn: 24372
2005-11-16 06:09:47 +00:00
Chris Lattner
655e7dfd0d
initial step at adding a dag-to-dag isel for X86 backend. Patch contributed
...
by Evan Cheng!
llvm-svn: 24371
2005-11-16 01:54:32 +00:00
Nate Begeman
a171f6b20c
Patch to clean up function call pseudos and support the BLA instruction,
...
which branches to an absolute address. This is required to support objc
direct dispatch.
llvm-svn: 24370
2005-11-16 00:48:01 +00:00
Chris Lattner
63985e2892
Make sure to use SwitchSection to switch sections so that we don't accidentally emit
...
functions into the .const section. Whoops.
llvm-svn: 24363
2005-11-15 01:45:01 +00:00
Chris Lattner
7c727b2736
Fix handling of multiple unnamed globals with the same type
...
llvm-svn: 24362
2005-11-15 01:32:03 +00:00
Chris Lattner
76ac068568
Separate X86ISelLowering stuff out from the X86ISelPattern.cpp file. Patch
...
contributed by Evan Cheng.
llvm-svn: 24358
2005-11-15 00:40:23 +00:00
Chris Lattner
bba9c372c1
Remove extraneous parents around constants when using a constant expr cast.
...
llvm-svn: 24357
2005-11-15 00:03:16 +00:00
Chris Lattner
1a4adc7aee
Handle globals with explicit alignment requests
...
llvm-svn: 24355
2005-11-14 19:00:30 +00:00
Chris Lattner
dd8eeed096
Teach emitAlignment to handle explicit alignment requests by globals.
...
llvm-svn: 24354
2005-11-14 19:00:06 +00:00
Chris Lattner
0aacd2ab9b
Teach the PPC asmwriter to honor globals with explicit section requests.
...
llvm-svn: 24353
2005-11-14 18:52:46 +00:00
Chris Lattner
16cbc6a177
instead of using mstats, use malloc_zone_statistics which returns numbers
...
that actually make sense.
llvm-svn: 24352
2005-11-14 07:27:56 +00:00
Chris Lattner
698fa760f4
Teach -track-memory to work on darwin. Looking at sbrk doesn't work because
...
the default allocator uses mmap.
llvm-svn: 24349
2005-11-14 07:00:29 +00:00
Duraid Madina
76034f95f6
add FP select. next up - divide!
...
llvm-svn: 24346
2005-11-14 01:17:30 +00:00
Chris Lattner
0333e68f72
Add some dummy method impls
...
llvm-svn: 24345
2005-11-13 03:26:33 +00:00
Chris Lattner
06b453b2dc
Move some methods around. Refactor the parts of TypeMap that do not depend
...
on its template arguments into a base class so that the code isn't duplicated
5 times.
llvm-svn: 24343
2005-11-13 03:14:09 +00:00
Chris Lattner
5c28b96304
Split the type refinement case completely out of the type-becomes-concrete
...
case, simplifying the code.
llvm-svn: 24340
2005-11-13 01:58:06 +00:00
Chris Lattner
54c8fcf303
unbreak the build
...
llvm-svn: 24339
2005-11-13 01:45:23 +00:00
Chris Lattner
708bb251d8
Separate the type-became-concrete case from the type-is-resolved case, the
...
former of which takes much less work than the later. This speeds up linking
eon from 3.749 to 3.637s with a release build (about 3%).
llvm-svn: 24338
2005-11-13 01:27:50 +00:00
Andrew Lenharth
ab72424488
enable LSR by default on alpha
...
llvm-svn: 24337
2005-11-12 19:21:08 +00:00
Andrew Lenharth
2ba45d1ee9
fix more regressions
...
llvm-svn: 24335
2005-11-12 19:06:28 +00:00
Andrew Lenharth
56526ec1a9
fix READCYCLECOUNTER
...
llvm-svn: 24334
2005-11-12 19:04:09 +00:00
Chris Lattner
026a5ef177
trivial cleanup
...
llvm-svn: 24330
2005-11-12 18:34:09 +00:00
Chris Lattner
b264c5d4dc
Fix reading LLVM 1.3 bc files.
...
llvm-svn: 24329
2005-11-12 18:31:54 +00:00
Chris Lattner
11ca4d636b
regenerate
...
llvm-svn: 24328
2005-11-12 18:22:38 +00:00
Chris Lattner
bce7ca8a73
refactor grammar to eliminate shift-reduce conflict. Move alignment checking
...
code out of all of the clients and into OptAlign/OptCAlign
llvm-svn: 24327
2005-11-12 18:21:21 +00:00
Chris Lattner
f2794fa073
Unswitch a loop, add more assertions, eliminate use of removeUserFromConcrete.
...
llvm-svn: 24324
2005-11-12 08:39:48 +00:00
Chris Lattner
11edcc19da
Tiny speedup, improve comments and code organization.
...
llvm-svn: 24323
2005-11-12 08:22:41 +00:00
Chris Lattner
8e1e6117d0
Read and write section info from/to .bc files
...
llvm-svn: 24321
2005-11-12 01:33:40 +00:00
Jeff Cohen
045f096b18
Wrap long line.
...
llvm-svn: 24320
2005-11-12 01:01:50 +00:00
Jeff Cohen
93b3d8f415
Fix VC++ warning that bools cannot be ORed.
...
llvm-svn: 24319
2005-11-12 00:59:39 +00:00
Jeff Cohen
cf1f782a2f
Fix operator precedence bug caught by VC++.
...
llvm-svn: 24318
2005-11-12 00:59:01 +00:00
Chris Lattner
a0c20b4f8b
regenerate
...
llvm-svn: 24316
2005-11-12 00:11:49 +00:00
Chris Lattner
9da804d8e9
add a token
...
llvm-svn: 24315
2005-11-12 00:11:30 +00:00
Chris Lattner
71b936cab9
Parse section info
...
llvm-svn: 24314
2005-11-12 00:11:10 +00:00
Chris Lattner
4b96c540f6
print section info
...
llvm-svn: 24313
2005-11-12 00:10:19 +00:00
Andrew Lenharth
97e8207a05
fix yet more regressions
...
llvm-svn: 24308
2005-11-11 23:08:46 +00:00
Andrew Lenharth
fab772045e
generate chain result
...
llvm-svn: 24307
2005-11-11 23:02:55 +00:00
Andrew Lenharth
de1b5d6baa
added a chain output
...
llvm-svn: 24306
2005-11-11 22:48:54 +00:00
Andrew Lenharth
5b3b9d7052
Fix a bunch more alpha regressions
...
llvm-svn: 24304
2005-11-11 19:52:25 +00:00
Andrew Lenharth
5278ca3fa2
prevent cse of readcyclecounter
...
llvm-svn: 24303
2005-11-11 19:02:54 +00:00
Andrew Lenharth
01aa56397d
continued readcyclecounter support
...
llvm-svn: 24300
2005-11-11 16:47:30 +00:00
Chris Lattner
ecc301df17
Fix the optimized code handling of user asm strings
...
llvm-svn: 24296
2005-11-10 23:24:26 +00:00
Chris Lattner
fafff9ba1d
Make BB and CPI labels use the function number, not the function name as a
...
uniquing id. This makes things happy when the function name is quoted,
preventing labels like LBB"foo"_2.
llvm-svn: 24295
2005-11-10 21:59:25 +00:00
Chris Lattner
86164e6e9e
speedup the common case where nothing needs to be quoted
...
llvm-svn: 24294
2005-11-10 21:47:01 +00:00
Chris Lattner
e1d34bac0b
Allow per-character control over what target assemblers allow in symbol
...
names. This also changes the default to allow all of "$_." in addition
to letters and numbers as symbol names. If you don't want this, use
markCharUnacceptable to remove one of these or markCharAcceptable to add
to the set. This corresponds with what GAS accepts by default.
Also, this includes some minor speedups
llvm-svn: 24293
2005-11-10 21:40:01 +00:00
Chris Lattner
4b3b9192b2
do not allow '.' in symbol names
...
llvm-svn: 24292
2005-11-10 21:39:29 +00:00