maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
21,585 Commits 23 Branches 0 Tags 2.2 GiB
Commit Graph

11995 Commits

Author SHA1 Message Date
Andrew Lenharth 30db2ec59f allow custom lowering to return null for legal results 
llvm-svn: 25007
 2005-12-25 01:07:37 +00:00
Andrew Lenharth 7259426d88 Support Custom lowering of a few more operations. 
Alpha needs to custom lower *DIV and *REM

llvm-svn: 25006
 2005-12-24 23:42:32 +00:00
Andrew Lenharth 0dc12c38e5 support targetexternalsym 
llvm-svn: 25005
 2005-12-24 23:36:59 +00:00
Evan Cheng 782b654e6f Let the helper functions know about X86::FR32RegClass and X86::FR64RegClass. 
llvm-svn: 25004
 2005-12-24 09:48:35 +00:00
Andrew Lenharth 5b18ed9e60 All addressing modes are now exposed. The only remaining relocated forms 
are for function prologue.

TODO: move external symbols over to using RelLit.
    : have a pattern that matches constpool|globaladdr
    : have a pattern that matches (add x imm) -> x, imm or (...) -> ..., 0
llvm-svn: 25003
 2005-12-24 08:29:32 +00:00
Andrew Lenharth b9aaea3564 Unify the patterns for loads and stores. Now offset addressing should be 
supported.  This almost completes memory operations.

llvm-svn: 25002
 2005-12-24 07:34:33 +00:00
Andrew Lenharth 4621488965 Let's see if we can break things. 
Lower GOT relative addresses to Lo and HI.
Update both ISels to select them when they can.
Saves instructions here and there.

llvm-svn: 25001
 2005-12-24 05:36:33 +00:00
Andrew Lenharth 636e1aed43 move loads and stores over. Smart addr selection comming 
llvm-svn: 25000
 2005-12-24 03:41:56 +00:00
Nate Begeman 9aea6e4691 Fix one of the things in the todo file, and get a bit closer to folding 
constant offsets from statics into the address arithmetic.

llvm-svn: 24999
 2005-12-24 01:00:15 +00:00
Evan Cheng 9ae486047e * Removed the use of FLAG. Now use hasFlagIn and hasFlagOut instead. 
* Added a pseudo instruction (for each target) that represent "return void".
  This is a workaround for lack of optional flag operand (return void is not
  lowered so it does not have a flag operand.)

llvm-svn: 24997
 2005-12-23 22:14:32 +00:00
Jim Laskey bdba3e2a46 Remove redundant debug locations. 
llvm-svn: 24995
 2005-12-23 20:08:28 +00:00
Chris Lattner c7037abc5b unbreak the build :-/ 
llvm-svn: 24992
 2005-12-23 16:12:20 +00:00
Chris Lattner 4f52796f79 not a good idea 
llvm-svn: 24991
 2005-12-23 07:37:47 +00:00
Evan Cheng 5c59d49630 More X86 floating point patterns. 
llvm-svn: 24990
 2005-12-23 07:31:11 +00:00
Evan Cheng 8be8067b19 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 31d15fa093 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 55823ae03f fix something-o 
llvm-svn: 24987
 2005-12-23 07:08:39 +00:00
Chris Lattner 5427aab5f8 implement vaarg. Varargs now should work. 
llvm-svn: 24986
 2005-12-23 06:37:38 +00:00
Chris Lattner 5ee896aad9 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 6e6d5a1fa1 remove benchmark list, remove issues addressed by the dag-dag isel 
llvm-svn: 24984
 2005-12-23 06:09:30 +00:00
Chris Lattner 26943b9691 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 54560f6887 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 efbbedbf4a 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 a187460552 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 30107e65c8 make sure bit_convert's are expanded 
llvm-svn: 24979
 2005-12-23 05:15:23 +00:00
Chris Lattner c46fc2482c make sure bit_converts are expanded 
llvm-svn: 24978
 2005-12-23 05:13:35 +00:00
Chris Lattner 5ce81edbf6 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 c9849274bd Use BIT_CONVERT to simplify this code 
llvm-svn: 24975
 2005-12-23 02:31:39 +00:00
Chris Lattner f474034432 Simplify some code by using BIT_CONVERT 
llvm-svn: 24974
 2005-12-23 00:59:59 +00:00
Chris Lattner 884eb3adc3 Fix a pasto 
llvm-svn: 24973
 2005-12-23 00:52:30 +00:00
Chris Lattner 9eae8d5d03 fix a thinko in the bit_convert handling code 
llvm-svn: 24972
 2005-12-23 00:50:25 +00:00
Chris Lattner 36e663d6e1 add very simple support for the BIT_CONVERT node 
llvm-svn: 24970
 2005-12-23 00:16:34 +00:00
Reid Spencer e4ae72a4e6 Revert previous patch. Additional tests fail. 
llvm-svn: 24968
 2005-12-22 21:46:37 +00:00
Chris Lattner 4c141d047c clean up .td file by using evan's new FLAG thing 
llvm-svn: 24967
 2005-12-22 21:18:39 +00:00
Chris Lattner 177d7af5d5 remove dead code 
llvm-svn: 24965
 2005-12-22 21:16:08 +00:00
Chris Lattner 6b0325aa26 fix handling of weak linkage 
llvm-svn: 24964
 2005-12-22 21:15:17 +00:00
Reid Spencer 8f8124edaa 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
Reid Spencer ccd5b90f58 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 ffe3542726 move some random notes out of my email into someplace useful 
llvm-svn: 24956
 2005-12-22 17:19:28 +00:00
Duraid Madina 644e7db818 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 3692fa14b8 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 3608ab87c0 oops, back this out 
llvm-svn: 24950
 2005-12-22 07:13:51 +00:00
Duraid Madina b1d57fb175 we can't all have brains now, can we 
llvm-svn: 24948
 2005-12-22 06:41:39 +00:00
Duraid Madina d0c146d59f 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 a8de8a5db4 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 be19877731 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 1408c05a8b The 81st column doesn't like code in it. 
llvm-svn: 24943
 2005-12-22 05:23:45 +00:00
Duraid Madina 5ccf76fed3 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
Duraid Madina 8f6c86fe3b kill SelectCALL() in the DAG isel, we handle this in lowering now, like 
SPARCv8. (we copy sparcv8's workaround for tablegen not being nice about
ISD::CALL/TAILCALL)

llvm-svn: 24941
 2005-12-22 03:58:17 +00:00
Duraid Madina a743e00e7a update tablegen files - nothing to see here 
llvm-svn: 24939
 2005-12-22 03:56:03 +00:00