Commit Graph

1609 Commits

Author SHA1 Message Date
Chris Lattner
670c889ac9 Implement SRA for global variables. This allows the other global variable
optimizations to trigger much more often.  This allows the elimination of
several dozen more global variables in Programs/External.  Note that we only
do this for non-constant globals: constant globals will already be optimized
out if the accesses to them permit it.

This implements Transforms/GlobalOpt/globalsra.llx


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16842 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-08 17:32:09 +00:00
Chris Lattner
b99d6b1cac Instcombine (X & FF00) + xx00 -> (X+xx00) & FF00, implementing and.ll:test27
This comes up when doing adds to bitfield elements.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16836 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-08 05:07:56 +00:00
Chris Lattner
1f7e160f77 Little patch to turn (shl (add X, 123), 4) -> (add (shl X, 4), 123 << 4)
This triggers in cases of bitfield additions, opening opportunities for
future improvements.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16834 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-08 03:46:20 +00:00
Chris Lattner
cf4d2a5699 Improve comments, no functionality changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16814 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-07 21:30:30 +00:00
Chris Lattner
3602549c3d Fix a bug in the safety analysis routine
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16804 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-07 06:01:25 +00:00
Chris Lattner
bc53e5e628 Comment cleanups
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16803 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-07 06:00:24 +00:00
Chris Lattner
7a90b68e5c * Rename pass to globalopt, since we do more than just constify
* Instead of handling dead functions specially, just nuke them.
* Be more aggressive about cleaning up after constification, in
  particular, handle getelementptr instructions and constantexprs.
* Be a little bit more structured about how we process globals.

*** Delete globals that are only stored to, and never read.  These are
    clearly not useful, so they should go.  This implements deadglobal.llx

This last one triggers quite a few times.  In particular, 2208 in the
external tests, 1865 of which are in 252.eon.  This shrinks eon from
1995094 to 1732341 bytes of bytecode.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16802 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-07 04:16:33 +00:00
Chris Lattner
e47ba74b07 Implement GlobalConstifier/trivialstore.llx, and also do some
simplifications of the resultant program to avoid making later passes
do it all.

This allows us to constify globals that just have the same constant that
they are initialized stored into them.

Suprisingly this comes up ALL of the freaking time, dozens of times in
SPEC, 30 times in vortex alone.

For example, on 256.bzip2, it allows us to constify these two globals:

%smallMode = internal global ubyte 0             ; <ubyte*> [#uses=8]
%verbosity = internal global int 0               ; <int*> [#uses=49]

Which (with later optimizations) results in the bytecode file shrinking
from 82286 to 69686 bytes!  Lets hear it for IPO :)

For the record, it's nuking lots of "if (verbosity > 2) { do lots of stuff }"
code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16793 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-06 20:57:02 +00:00
Chris Lattner
91ccc15b45 Instcombine: -(X sdiv C) -> (X sdiv -C), tested by sub.ll:test16
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16769 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-06 15:08:25 +00:00
Chris Lattner
c3e903fe65 Reduce code growth implied by the tail duplication pass by not duplicating
an instruction if it can be hoisted to a common dominator of the block.
This implements: test/Regression/Transforms/TailDup/MergeTest.ll


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16758 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-06 03:27:37 +00:00
Brian Gaeke
f098ced848 Add accessor function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16622 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-30 20:14:29 +00:00
Brian Gaeke
b3b1e33632 Correct type of accessor functions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16621 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-30 20:14:18 +00:00
Brian Gaeke
ccb87cdf07 Namespacify. Add accessor function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16620 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-30 20:14:07 +00:00
Chris Lattner
23367a779c Disable the 'WARNING: Found global types that are not compatible' warning
that always prints when linking programs to libstdc++ :(


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16603 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-30 00:12:29 +00:00
Chris Lattner
a71989f11b Hrm, debugging printouts do not need to be in here
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16598 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 21:21:14 +00:00
Chris Lattner
a96879a26d * Pull range optimization code out into new InsertRangeTest function.
* SubOne/AddOne functions always return ConstantInt, declare them as such
* Pull code for handling setcc X, cst, where cst is at the end of the range,
  or cc is LE or GE up earlier in visitSetCondInst.  This reduces #iterations
  in some cases.
* Fold: (div X, C1) op C2 -> range check, implementing div.ll:test6 - test9.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16588 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 17:40:11 +00:00
Chris Lattner
0ed7f42c1b Do not insert trivially dead select instructions, which allows us to
potentially fold more in one pass.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16583 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 05:43:32 +00:00
Chris Lattner
4e998b2fea Fold binary expressions and casts into PHI nodes that have all constant inputs.
This takes something like this:

%A = phi int [ 3, %cond_false.0 ], [ 2, %endif.0.i ], [ 2, %endif.1.i ]
%B = div int %tmp.243, 4

and turns it into:

%A = phi int [ 3/4, %cond_false.0 ], [ 2/4, %endif.0.i ], [ 2/4, %endif.1.i ]

which is later simplified (in this case) into %A = 0.

This triggers thousands of times in spec, for example, 269 times in 176.gcc.

This is tested by InstCombine/add.ll:test23 and set.ll:test18.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16582 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 05:07:12 +00:00
Chris Lattner
dd763f401e Hrm, really, all tests passed without this, but it is scary to think how...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16568 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 03:16:24 +00:00
Chris Lattner
f6d1d7d2ba Remove debugging printout
Instcombine (setcc (truncate X), C1).

This occurs THOUSANDS of times in many benchmarks.  Particularlly common
seem to be things like (seteq (cast bool X to int), int 0)

This turns it into (seteq bool %X, false), which then becomes (not %X).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16567 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-29 03:09:18 +00:00
Chris Lattner
b4f40d2059 Fold (X setcc C1) | (X setcc C2)
This implements or.ll:test1[89]


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16561 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 22:33:08 +00:00
Chris Lattner
955f331a77 Fold (and (setcc X, C1), (setcc X, C2))
This is important for several reasons:

1. Benchmarks have lots of code that looks like this (perlbmk in particular):

  %tmp.2.i = setne int %tmp.0.i, 128              ; <bool> [#uses=1]
  %tmp.6343 = seteq int %tmp.0.i, 1               ; <bool> [#uses=1]
  %tmp.63 = and bool %tmp.2.i, %tmp.6343          ; <bool> [#uses=1]

   we now fold away the setne, a clear improvement.

2. In the more important cases, such as (X >= 10) & (X < 20), we now produce
   smaller code: (X-10) < 10.

3. Perhaps the nicest effect of this patch is that it really helps out the
   code generators.  In particular, for a 'range test' like the above,
   instead of generating this on X86 (the difference on PPC is even more
   pronounced):

        cmp %EAX, 50
        setge %CL
        cmp %EAX, 100
        setl %AL
        and %CL, %AL
        cmp %CL, 0

   we now generate this:

        add %EAX, -50
        cmp %EAX, 50

   Furthermore, this causes setcc's to be folded into branches more often.

These combinations trigger dozens of times in the spec benchmarks, particularly
in 176.gcc, 186.crafty, 253.perlbmk, 254.gap, & 099.go.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16559 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 21:48:02 +00:00
Chris Lattner
18d19ca6df Implement X / C1 / C2 folding
Implement (setcc (shl X, C1), C2) folding.

The second one occurs several dozen times in spec.  The first was added
just in case.  :)

These are tested by shift.ll:test2[12], and div.ll:test5


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16549 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 18:22:15 +00:00
Chris Lattner
0cba71b0d6 shl is always zero extending, so always use a zero extending shift right.
This latent bug was exposed by recent changes, and is tested as:
llvm/test/Regression/Transforms/InstCombine/2004-09-28-BadShiftAndSetCC.llx


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16546 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 17:54:07 +00:00
Alkis Evlogimenos
c72c617a4e Add includes and use std:: for standard library calls to make code
compile on windows. This patch was contributed by Paolo Invernizzi.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16539 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 14:42:44 +00:00
Alkis Evlogimenos
200a360ec6 Pull assignment out of for loop conditional in order for this to
compile under windows. Patch contributed by Paolo Invernizzi!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16534 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-28 02:40:37 +00:00
Chris Lattner
83c4ec0372 Fix two bugs: one where a condition was mistakenly swapped, and another
where we folded (X & 254) -> X < 1 instead of X < 2.  These problems were
latent problems exposed by the latest patch.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16528 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-27 19:29:18 +00:00
Chris Lattner
f63f647345 Fold: (setcc (shr X, ShAmt), CI), where 'cc' is eq or ne. This xform
triggers often, for example:

6x in povray, 1x in gzip, 279x in gcc, 1x in crafty, 8x in eon, 11x in perlbmk,
362x in gap, 4x in vortex, 14 in m88ksim, 211x in 126.gcc, 1x in compress,
11x in ijpeg, and 4x in 147.vortex.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16521 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-27 16:18:50 +00:00
Chris Lattner
0c96766c90 Implement shift-and combinations, implementing InstCombine/and.ll:test19-21
These combinations trigger 4 times in povray, 7x in gcc, 4x in gap, and 2x in bzip2.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16508 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-24 15:21:34 +00:00
Chris Lattner
648e3bc39c Move LHSI->hasOneUse() into the arms of the conditional, reindenting code.
No functionality changes here.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16505 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-23 21:52:49 +00:00
Chris Lattner
b20ba0a83f Implement Transforms/InstCombine/and.ll:test18, a case that occurs 20 times
in perlbmk


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16504 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-23 21:46:38 +00:00
Chris Lattner
684fe210a9 Implement select.ll:test16: fold load (select C, X, null) -> load X
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16499 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-23 15:46:00 +00:00
Chris Lattner
3d834bf2c3 Do not fold (X + C1 != C2) if there are other users of the add. Doing
this transformation used to take a loop like this:

int Array[1000];
void test(int X) {
  int i;
  for (i = 0; i < 1000; ++i)
    Array[i] += X;
}

Compiled to LLVM is:

no_exit:                ; preds = %entry, %no_exit
        %indvar = phi uint [ 0, %entry ], [ %indvar.next, %no_exit ]            ; <uint> [#uses=2]
        %tmp.4 = getelementptr [1000 x int]* %Array, int 0, uint %indvar                ; <int*> [#uses=2]
        %tmp.7 = load int* %tmp.4               ; <int> [#uses=1]
        %tmp.9 = add int %tmp.7, %X             ; <int> [#uses=1]
        store int %tmp.9, int* %tmp.4
***     %indvar.next = add uint %indvar, 1              ; <uint> [#uses=2]
***     %exitcond = seteq uint %indvar.next, 1000               ; <bool> [#uses=1]
        br bool %exitcond, label %return, label %no_exit

and turn it into a loop like this:

no_exit:                ; preds = %entry, %no_exit
        %indvar = phi uint [ 0, %entry ], [ %indvar.next, %no_exit ]            ; <uint> [#uses=3]
        %tmp.4 = getelementptr [1000 x int]* %Array, int 0, uint %indvar                ; <int*> [#uses=2]
        %tmp.7 = load int* %tmp.4               ; <int> [#uses=1]
        %tmp.9 = add int %tmp.7, %X             ; <int> [#uses=1]
        store int %tmp.9, int* %tmp.4
***     %indvar.next = add uint %indvar, 1              ; <uint> [#uses=1]
***     %exitcond = seteq uint %indvar, 999             ; <bool> [#uses=1]
        br bool %exitcond, label %return, label %no_exit

Note that indvar.next and indvar can no longer be coallesced.  In machine
code terms, this patch changes this code:

.LBBtest_1:     # no_exit
        mov %EDX, OFFSET Array
        mov %ESI, %EAX
        add %ESI, DWORD PTR [%EDX + 4*%ECX]
        mov %EDX, OFFSET Array
        mov DWORD PTR [%EDX + 4*%ECX], %ESI
        mov %EDX, %ECX
        inc %EDX
        cmp %ECX, 999
        mov %ECX, %EDX
        jne .LBBtest_1  # no_exit

into this:

.LBBtest_1:     # no_exit
        mov %EDX, OFFSET Array
        mov %ESI, %EAX
        add %ESI, DWORD PTR [%EDX + 4*%ECX]
        mov %EDX, OFFSET Array
        mov DWORD PTR [%EDX + 4*%ECX], %ESI
        inc %ECX
        cmp %ECX, 1000
        jne .LBBtest_1  # no_exit

We need better instruction selection to get this:

.LBBtest_1:     # no_exit
        add DWORD PTR [Array + 4*%ECX], EAX
        inc %ECX
        cmp %ECX, 1000
        jne .LBBtest_1  # no_exit

... but at least there is less register juggling


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16473 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-21 21:35:23 +00:00
Chris Lattner
79f0c8e4ee Fix potential miscompilations: InstCombine/2004-09-20-BadLoadCombine*.llx
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16447 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-20 10:15:10 +00:00
Alkis Evlogimenos
7b6ec600c5 Fix loop condition so that we don't decrement off the beginning of the
list.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16440 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-20 06:42:58 +00:00
Chris Lattner
b12914bfc0 'Pass' should now not be derived from by clients. Instead, they should derive
from ModulePass.  Instead of implementing Pass::run, then should implement
ModulePass::runOnModule.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16436 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-20 04:48:05 +00:00
Chris Lattner
ded6d0c9fb Prototype more accurately
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16433 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-20 04:43:57 +00:00
Chris Lattner
4b50156049 Prototype these functions more accurately
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16432 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-20 04:43:15 +00:00
Chris Lattner
8a37520f80 Make isSafeToLoadUnconditionally a bit smarter, implementing PR362 and
Regression/Transforms/InstCombine/CPP_min_max.llx


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16409 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-19 19:18:10 +00:00
Chris Lattner
a744b77e11 Remove a whole bunch of horrible hacky code that was used to promote allocas
whose addresses where used by trivial phi nodes and select instructions.  This
is now performed by the instcombine pass, which is more powerful, is much
simpler, and is faster.  This allows the deletion of a bunch of code, two
FIXME's and two gotos.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16406 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-19 18:51:51 +00:00
Chris Lattner
c10acedb2d Make instruction combining a bit more aggressive in the face of volatile
loads, and implement two new transforms: InstCombine/load.ll:test[56].


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16404 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-19 18:43:46 +00:00
Chris Lattner
f5afcabff8 Add comment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16400 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-19 01:05:16 +00:00
Chris Lattner
0c0aa711b8 Fix the inliner to always delete any edges from the external call node to
a function being deleted.  Due to optimizations done while inlining, there
can be edges from the external call node to a function node that were not
apparent any longer.

This fixes the compiler crash while compiling 175.vpr


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16399 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-18 21:37:03 +00:00
Chris Lattner
5eb6f6c829 Convert this pass to be a CallGraphSCCPass instead of a Pass, which eliminates
the worklist and makes it more efficient.  This does not change functionality
at all.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16390 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-18 00:34:13 +00:00
Chris Lattner
afbb1ccd2f Make sure to remove the Select instruction as well
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16389 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-18 00:32:40 +00:00
Chris Lattner
ebeb0cba94 Fix typo in comment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16384 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-17 03:58:39 +00:00
Chris Lattner
acf808951b Add a newline
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16369 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-15 17:53:52 +00:00
Reid Spencer
2da5c3dda6 Convert code to compile with vc7.1.
Patch contributed by Paolo Invernizzi. Thanks Paolo!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16368 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-15 17:06:42 +00:00
Chris Lattner
bdacd87957 Fix a bug in the previous checkin that broke 255.vortex
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16355 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-15 02:34:40 +00:00
Chris Lattner
fb3ee1930b Make sure to update alias analysis information as we transform the function.
This fixes PR420 and Regression/Transforms/LICM/2004-09-14-AliasAnalysisInvalidate.llx


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16348 91177308-0d34-0410-b5e6-96231b3b80d8
2004-09-15 01:04:07 +00:00