RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 19:52:54 +01:00
Code Issues Projects Releases Wiki Activity
21,533 Commits 4 Branches 5 Tags 704 MiB
bd79f871a2
Commit Graph

21533 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Evan Cheng
bd79f871a2 * Support for hasInFlag and hasOutFlag (on instructions). Remove nameless FLAG 
support which is fragile.
* Fixed a number of bugs.

llvm-svn: 24996
 2005-12-23 22:11:47 +00:00
Jim Laskey
d8cc1062ed Remove redundant debug locations. 
llvm-svn: 24995
 2005-12-23 20:08:28 +00:00
Chris Lattner
d84f4721b0 this test fails 
llvm-svn: 24994
 2005-12-23 16:14:29 +00:00
Chris Lattner
70b2f56c5a this test is malformed 
llvm-svn: 24993
 2005-12-23 16:13:52 +00:00
Chris Lattner
c97655052a unbreak the build :-/ 
llvm-svn: 24992
 2005-12-23 16:12:20 +00:00
Chris Lattner
738cec7a6e not a good idea 
llvm-svn: 24991
 2005-12-23 07:37:47 +00:00
Evan Cheng
995503fc91 More X86 floating point patterns. 
llvm-svn: 24990
 2005-12-23 07:31:11 +00:00
Evan Cheng
74cf3e8823 Operand 1 of TRUNCSTORE can be any of integer and floating point types. 
llvm-svn: 24989
 2005-12-23 07:30:30 +00:00
Evan Cheng
bcfef42cd4 Allow custom lowering of LOAD, EXTLOAD, ZEXTLOAD, STORE, and TRUNCSTORE. Not 
currently used.

llvm-svn: 24988
 2005-12-23 07:29:34 +00:00
Chris Lattner
2e386f9d46 fix something-o 
llvm-svn: 24987
 2005-12-23 07:08:39 +00:00
Chris Lattner
a4a2d4c3fe implement vaarg. Varargs now should work. 
llvm-svn: 24986
 2005-12-23 06:37:38 +00:00
Chris Lattner
4e8124bd9b implement vastart. The dag isel compiles this: 
void test3(va_list Y);
void test2(int F, ...) {
  va_list X;
  va_start(X, F);
  test3(X);
}

into this:

test2:
        save -104, %o6, %o6
        st %i5, [%i6+88]
        st %i4, [%i6+84]
        st %i3, [%i6+80]
        st %i2, [%i6+76]
        st %i1, [%i6+72]
        add %i6, 72, %o0
        st %o0, [%i6+-4]
        call test3
        nop
        restore %g0, %g0, %g0
        retl
        nop

The simple isel emits:

test2:
        save -96, %o6, %o6
        st %i0, [%i6+68]
        st %i1, [%i6+72]
        st %i2, [%i6+76]
        st %i3, [%i6+80]
        st %i4, [%i6+84]
        st %i5, [%i6+88]
        or %g0, 1, %l0
        or %g0, 4, %l1
        umul %l0, %l1, %l0
        add %l0, 7, %l0
        and %l0, -8, %l0
        sub %o6, %l0, %o6
        add %o6, 96, %l0
        add %i6, 72, %l1
        st %l1, [%l0]
        ld [%l0], %o0
        call test3
        nop
        restore %g0, %g0, %g0
        retl
        nop

llvm-svn: 24985
 2005-12-23 06:24:04 +00:00
Chris Lattner
73db2dc9fc remove benchmark list, remove issues addressed by the dag-dag isel 
llvm-svn: 24984
 2005-12-23 06:09:30 +00:00
Chris Lattner
0637d38ec2 Simplify store(bitconv(x)) to store(x). This allows us to compile this: 
void bar(double Y, double *X) {
  *X = Y;
}

to this:

bar:
        save -96, %o6, %o6
        st %i1, [%i2+4]
        st %i0, [%i2]
        restore %g0, %g0, %g0
        retl
        nop

instead of this:

bar:
        save -104, %o6, %o6
        st %i1, [%i6+-4]
        st %i0, [%i6+-8]
        ldd [%i6+-8], %f0
        std  %f0, [%i2]
        restore %g0, %g0, %g0
        retl
        nop

on sparcv8.

llvm-svn: 24983
 2005-12-23 05:48:07 +00:00
Chris Lattner
20e6338732 fold (conv (load x)) -> (load (conv*)x). 
This allows us to compile this:
void foo(double);
void bar(double *X) { foo(*X); }

To this:

bar:
        save -96, %o6, %o6
        ld [%i0+4], %o1
        ld [%i0], %o0
        call foo
        nop
        restore %g0, %g0, %g0
        retl
        nop

instead of this:

bar:
        save -104, %o6, %o6
        ldd [%i0], %f0
        std %f0, [%i6+-8]
        ld [%i6+-4], %o1
        ld [%i6+-8], %o0
        call foo
        nop
        restore %g0, %g0, %g0
        retl
        nop

on SparcV8.

llvm-svn: 24982
 2005-12-23 05:44:41 +00:00
Chris Lattner
28887b3ca6 Fold bitconv(bitconv(x)) -> x. We now compile this: 
void foo(double);
void bar(double X) { foo(X); }

to this:

bar:
        save -96, %o6, %o6
        or %g0, %i0, %o0
        or %g0, %i1, %o1
        call foo
        nop
        restore %g0, %g0, %g0
        retl
        nop

instead of this:

bar:
        save -112, %o6, %o6
        st %i1, [%i6+-4]
        st %i0, [%i6+-8]
        ldd [%i6+-8], %f0
        std %f0, [%i6+-16]
        ld [%i6+-12], %o1
        ld [%i6+-16], %o0
        call foo
        nop
        restore %g0, %g0, %g0
        retl
        nop

on V8.

llvm-svn: 24981
 2005-12-23 05:37:50 +00:00
Chris Lattner
9ee4ecfe74 constant fold bits_convert in getNode and in the dag combiner for fp<->int 
conversions.  This allows V8 to compiles this:

void %test() {
        call float %test2( float 1.000000e+00, float 2.000000e+00, double 3.000000e+00, double* null )
        ret void
}

into:

test:
        save -96, %o6, %o6
        sethi 0, %o3
        sethi 1049088, %o2
        sethi 1048576, %o1
        sethi 1040384, %o0
        or %g0, %o3, %o4
        call test2
        nop
        restore %g0, %g0, %g0
        retl
        nop

instead of:

test:
        save -112, %o6, %o6
        sethi 0, %o4
        sethi 1049088, %l0
        st %o4, [%i6+-12]
        st %l0, [%i6+-16]
        ld [%i6+-12], %o3
        ld [%i6+-16], %o2
        sethi 1048576, %o1
        sethi 1040384, %o0
        call test2
        nop
        restore %g0, %g0, %g0
        retl
        nop

llvm-svn: 24980
 2005-12-23 05:30:37 +00:00
Chris Lattner
8e80a247ff make sure bit_convert's are expanded 
llvm-svn: 24979
 2005-12-23 05:15:23 +00:00
Chris Lattner
8b2bd265f4 make sure bit_converts are expanded 
llvm-svn: 24978
 2005-12-23 05:13:35 +00:00
Jeff Cohen
fcce4dc8b6 Get bugpoint building with VC++ again. 
llvm-svn: 24977
 2005-12-23 05:00:38 +00:00
Chris Lattner
e47f28d044 fix the int<->fp instructions, which apparently take a single float register 
to represent the int part (because it's always 32-bits)

llvm-svn: 24976
 2005-12-23 05:00:16 +00:00
Chris Lattner
aaeb6774c7 Use BIT_CONVERT to simplify this code 
llvm-svn: 24975
 2005-12-23 02:31:39 +00:00
Chris Lattner
3c4a3dd86e Simplify some code by using BIT_CONVERT 
llvm-svn: 24974
 2005-12-23 00:59:59 +00:00
Chris Lattner
4bcbe2d378 Fix a pasto 
llvm-svn: 24973
 2005-12-23 00:52:30 +00:00
Chris Lattner
4a929edf04 fix a thinko in the bit_convert handling code 
llvm-svn: 24972
 2005-12-23 00:50:25 +00:00
Chris Lattner
1d16f29ca7 improve comment: mention vectors 
llvm-svn: 24971
 2005-12-23 00:46:10 +00:00
Chris Lattner
a59cc5ebbb add very simple support for the BIT_CONVERT node 
llvm-svn: 24970
 2005-12-23 00:16:34 +00:00
Chris Lattner
77ebc833c1 add a new node 
llvm-svn: 24969
 2005-12-23 00:15:59 +00:00
Reid Spencer
63a1f99627 Revert previous patch. Additional tests fail. 
llvm-svn: 24968
 2005-12-22 21:46:37 +00:00
Chris Lattner
1cbff1f3cd clean up .td file by using evan's new FLAG thing 
llvm-svn: 24967
 2005-12-22 21:18:39 +00:00
Chris Lattner
0b240343df remove a dead node 
llvm-svn: 24966
 2005-12-22 21:16:35 +00:00
Chris Lattner
73f38507d9 remove dead code 
llvm-svn: 24965
 2005-12-22 21:16:08 +00:00
Chris Lattner
8c2622a14e fix handling of weak linkage 
llvm-svn: 24964
 2005-12-22 21:15:17 +00:00
Reid Spencer
56825ced86 Fix PR409: 
Implement the suggested check to ensure that out-of-range float constants
don't get accepted by LLVM accidentally. Adjust the supporting test cases
as well.

llvm-svn: 24963
 2005-12-22 21:07:29 +00:00
Chris Lattner
d752db6276 silence some bogus gcc warnings 
llvm-svn: 24962
 2005-12-22 20:37:36 +00:00
Reid Spencer
36b05575fe For PR351: 
Generally, remove use of fork/exec from bugpoint in favor of the portable
sys::Program::ExecuteAndWait method. This change requires two new options
to bugpoint to tell it that it is running in "child" mode. In this mode,
it reads its input and runs the passes. The result code signals to the
parent instance of bugpoint what happened (success, fail, crash).

This change should make bugpoint usable on Win32 systems.

llvm-svn: 24961
 2005-12-22 20:02:55 +00:00
Reid Spencer
5086a114a5 For PR351: 
* Allow the ExecuteAndWait to return negative values if a signal is
  detected as the reason for the child termination. This is needed to
  support bugpoint detecting bad things in its child processes.

llvm-svn: 24960
 2005-12-22 20:00:16 +00:00
Chris Lattner
06a349fb05 Fix test/Regression/Other/2002-01-31-CallGraph.ll after the recent callgraph 
rework.

llvm-svn: 24959
 2005-12-22 19:26:06 +00:00
Chris Lattner
f7ed832fcf move some random notes out of my email into someplace useful 
llvm-svn: 24956
 2005-12-22 17:19:28 +00:00
Duraid Madina
e009e40008 this is a hack, which may or may not hang around. In short: 
whimper out of doing things the Right Way, and hack up a generic
'BRCALL' instruction, that gets generated when calls are lowered.
This gets selected by hand in the DAG isel, where it gets turned
into real (i.e. in tablegen) br.call instructions.

BUG: this dies on void calls, but seems to work otherwise?
llvm-svn: 24952
 2005-12-22 13:29:14 +00:00
Duraid Madina
325c2cc07b we can't do this directly in lowering, so we need this case 
llvm-svn: 24951
 2005-12-22 07:14:45 +00:00
Duraid Madina
bf2a5de0d5 oops, back this out 
llvm-svn: 24950
 2005-12-22 07:13:51 +00:00
Duraid Madina
a83ab290ec i need to do this to take over the earth 
when I take over the earth, sabre will be happy because i'll stop asking
him silly questions!!!!!!

llvm-svn: 24949
 2005-12-22 07:02:51 +00:00
Duraid Madina
d87738a097 we can't all have brains now, can we 
llvm-svn: 24948
 2005-12-22 06:41:39 +00:00
Duraid Madina
b43dc0ff5b this should take care of calls to varadic functions, but it doesn.,t 
BUG: calling printf(string, float) will load the float into the wrong
register, completely forget about loading the string, etce

llvm-svn: 24947
 2005-12-22 06:39:57 +00:00
Duraid Madina
d1c9d5ae4d we need to emit the getf.d instruction in lowering, so add it 
to IA64ISD

llvm-svn: 24946
 2005-12-22 06:38:38 +00:00
Chris Lattner
3d84ef7903 credit where credit is due 
llvm-svn: 24945
 2005-12-22 06:09:08 +00:00
Chris Lattner
7f45e655a8 Separate the call graph implementation from its interface. This implements 
the rough idea sketched out in http://nondot.org/sabre/LLVMNotes/CallGraphClass.txt,
allowing new spiffy implementations of the callgraph interface to be built.

Many thanks to Saem Ghani for contributing this!

llvm-svn: 24944
 2005-12-22 06:07:52 +00:00
Chris Lattner
6f708e886f The 81st column doesn't like code in it. 
llvm-svn: 24943
 2005-12-22 05:23:45 +00:00
Duraid Madina
47cd4eeab0 I shoulda done this a *long* time ago (tm): implement calls properly, 
i.e. r1/r12/rp are saved/restored regardless of scheduling/luck

TODO: calls to external symbols, indirect (function descriptor) calls,
      performance (we're being paranoid right now)

BUG: the code for handling calls to vararg functions breaks if FP
args are passed (this will make printf() go haywire so a bunch of
tests will fail)

BUG: this seems to trigger some legalize nastiness
llvm-svn: 24942
 2005-12-22 04:07:40 +00:00