Commit Graph

18558 Commits

Author SHA1 Message Date
Chris Lattner
51d2ed976f rename getPPCOpcodeForSetCCNumber -> getPPCOpcodeForSetCCOpode to be more
correct.  Remove the EmitComparison retvalue, as it is always the first arg.

Fix a place where we incorrectly passed in the setcc opcode instead of the
setcc number, causing us to miscompile crafty.  Crafty now works!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21195 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-10 01:03:31 +00:00
Nate Begeman
706471e291 fix ISD::BRCONDTWOWAY codegen to not deference the end() iterator
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21193 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 23:35:05 +00:00
Chris Lattner
9184bfbbf4 Fix CodeGen/Generic/2005-05-09-GlobalInPHI.ll, which was reduced from 254.gap.
This caused the "use before a def" assertion on some programs.

With this patch, 254.gap now passes with the PPC backend.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21191 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 22:05:17 +00:00
Chris Lattner
d0abb7a2ec new testcase that used to crash the ppc fe. It could effect any simpleisel
that is not careful, so I'm checking it into the generic tests.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21190 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 22:03:10 +00:00
Chris Lattner
a2daa8c78d add a little peephole optimization. This allows us to codegen:
int a(short i) {
        return i & 1;
}

as

_a:
        andi. r3, r3, 1
        blr

instead of:

_a:
        rlwinm r2, r3, 0, 16, 31
        andi. r3, r2, 1
        blr

on ppc.  It should also help the other risc targets.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21189 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 21:43:54 +00:00
Chris Lattner
91277ea45c do not set the root to null if an argument is dead
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21188 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 21:23:24 +00:00
Nate Begeman
cd08e4cb7e Add rlwnm instruction for variable rotate
Generate rotate left/right immediate
Generate code for brcondtwoway
Use new livein/liveout functionality


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21187 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 20:09:12 +00:00
Chris Lattner
27ee3a332d Fix a crash on 173.applu by asking for a constant bigger than 32-bits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21185 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 19:47:21 +00:00
Chris Lattner
f429a3e0f6 Switch this instruction selector over to using liveins and liveouts, eliminating
implicit defs on entry to the function.  yaay :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21184 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 16:32:30 +00:00
Chris Lattner
8e7d87b228 there is no need to remove this instruction, linscan does it already as it
removes noop moves.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21183 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 16:24:20 +00:00
Chris Lattner
799a919dbc Adjust live intervals to support a livein set
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21182 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 16:17:50 +00:00
Chris Lattner
4c52f0ec8c Use live out sets for return values instead of imp_defs, which is cleaner and faster.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21181 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 15:23:56 +00:00
Chris Lattner
d493b34d31 Consider the livein/out set for a function, allowing targets to not have to
use ugly imp_def/imp_uses for arguments and return values.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21180 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 15:23:25 +00:00
Chris Lattner
dde3a9abc3 add routines to track the livein/out set for a function
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21179 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 15:22:53 +00:00
Duraid Madina
e00e5ded51 ok, the "ia64 has a boatload of registers" joke stopped being funny today ;)
* fix overallocation of integer (stacked) registers: we can't allocate
  registers for local use if they are required as output registers

this fixes 'toast' in the test suite, and all sorts of larger programs
like bzip2 etc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21178 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 11:53:00 +00:00
Nate Begeman
af4ab1b103 Optimize FSEL a bit for fneg arguments. This fixes the recently added test
case so that we emit

_test_fneg_sel:
.LBB_test_fneg_sel_0:   ;
        fsel f1, f1, f3, f2
        blr

instead of:

_test_fneg_sel:
.LBB_test_fneg_sel_0:   ;
        fneg f0, f1
        fneg f0, f0
        fsel f1, f0, f3, f2
        blr


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21177 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 09:33:07 +00:00
Nate Begeman
9a98a77446 Add a testcase to make sure that we don't emit two fneg instructions back
to back for certain fsel instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21176 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 09:30:09 +00:00
Nate Begeman
e9c0f43a1b Add cases to cover the rest of the patterns we should be matching
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21175 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 08:29:59 +00:00
Chris Lattner
67ac5f459e Fix CodeGen/SparcV9/2005-05-09-GEP-Crash.ll a crash on some specfp program
lets hope this doesn't break other programs with induced entropy


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21174 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 06:27:14 +00:00
Chris Lattner
2f65f8e724 New testcase that the sparc backend crashes on
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21173 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 06:26:27 +00:00
Chris Lattner
59723e9698 recognize some patterns as fabs operations, so that fabs at the source level
is deconstructed then reconstructed here.  This catches 19 fabs's in 177.mesa
9 in 168.wupwise, 5 in 171.swim, 3 in 172.mgrid, and 14 in 173.applu out of
specfp2000.

This allows the X86 code generator to make MUCH better code than before for
each of these and saves one instr on ppc.

This depends on the previous CFE patch to expose these correctly.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21171 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 05:15:53 +00:00
Chris Lattner
2e6027ccab make this test more interesting
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21170 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 04:55:14 +00:00
Chris Lattner
45485234ee add a test for fnabs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21169 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 04:03:16 +00:00
Chris Lattner
e23836cf77 add a partial test for the fma operations that ppc supports. I'm sure I'm
missing some and not all of these match yet, but I'm sure that Nate will
clean up my mess :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21168 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 04:01:32 +00:00
Chris Lattner
e7ccd4acd2 Emit BRCONDTWOWAY when possible.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21167 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:30:29 +00:00
Chris Lattner
411e888c1b Legalize BRCONDTWOWAY into a BRCOND/BR pair if a target doesn't support it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21166 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:30:19 +00:00
Chris Lattner
ef847df087 print and fold BRCONDTWOWAY correctly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21165 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:27:28 +00:00
Chris Lattner
da4d4694a8 This target does not support/want ISD::BRCONDTWOWAY
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21164 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:22:37 +00:00
Chris Lattner
644db4ec5f This target does not yet support ISD::BRCONDTWOWAY
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21163 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:22:30 +00:00
Chris Lattner
1df633821b Add a new node
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21162 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:21:50 +00:00
Nate Begeman
e88aa5b4d1 64b: Expand S/UREM
32b: No longer pattern match fneg(fsub(fmul)) as fnmsub
     Pattern match fsub a, mul(b, c) as fnmsub
     Pattern match fadd a, mul(b, c) as fmadd
Those changes speed up hydro2d by 2.5%, distray by 6%, and scimark by 8%


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21161 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:05:51 +00:00
Chris Lattner
485df9b84b canonicalize a bunch of operations involving fneg
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21160 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-09 03:02:46 +00:00
Nate Begeman
a9532d5dde Fix 64b shifts
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21159 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 23:45:01 +00:00
Chris Lattner
2bffad3eeb fix this method for 64-bit constants
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21158 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 21:31:29 +00:00
Nate Begeman
f3f2d6d378 Match Mac OS X 64 bit calling conventions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21157 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 21:26:05 +00:00
Andrew Lenharth
a32b9e31c6 collect a few statistics, factor constants (constant loading and mult), fix logic operation pattern matchs, supress FP div when int dividing by a constant
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21156 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 17:28:49 +00:00
Andrew Lenharth
a844b017a6 oops
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21155 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 16:55:15 +00:00
Andrew Lenharth
b8a5d63b18 added some tests to check stupid pattern matching mistakes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21154 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 16:46:44 +00:00
Duraid Madina
6dcceb5ecb fix bogus division-by-power-of-2 (was wrong for negative input, adds extr insn)
fix hack in division (clean up frcpa instruction)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21153 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 10:01:48 +00:00
Chris Lattner
bf70b838e7 Fix bug: InstCombine/2005-05-07-UDivSelectCrash.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21152 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 04:03:26 +00:00
Chris Lattner
b37246d65e new testcase that crashes the instcombiner.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21151 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-08 03:58:21 +00:00
Nate Begeman
7e7fadd2ea Optimized code sequences for setcc reg, 0
Optimized code sequence for (a < 0) ? b : 0


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21150 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 20:30:01 +00:00
Andrew Lenharth
d3355e22a7 Alpha zero extends setcc results
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21149 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 20:11:32 +00:00
Chris Lattner
5a6bace3ab If a target zero or sign extends the result of its setcc, allow folding of
this into sign/zero extension instructions later.

On PPC, for example, this testcase:

%G = external global sbyte
implementation
void %test(int %X, int %Y) {
  %C = setlt int %X, %Y
  %D = cast bool %C to sbyte
  store sbyte %D, sbyte* %G
  ret void
}

Now codegens to:

        cmpw cr0, r3, r4
        li r3, 1
        li r4, 0
        blt .LBB_test_2 ;
.LBB_test_1:    ;
        or r3, r4, r4
.LBB_test_2:    ;
        addis r2, r2, ha16(L_G$non_lazy_ptr-"L00000$pb")
        lwz r2, lo16(L_G$non_lazy_ptr-"L00000$pb")(r2)
        stb r3, 0(r2)

instead of:

        cmpw cr0, r3, r4
        li r3, 1
        li r4, 0
        blt .LBB_test_2 ;
.LBB_test_1:    ;
        or r3, r4, r4
.LBB_test_2:    ;
***     rlwinm r3, r3, 0, 31, 31
        addis r2, r2, ha16(L_G$non_lazy_ptr-"L00000$pb")
        lwz r2, lo16(L_G$non_lazy_ptr-"L00000$pb")(r2)
        stb r3, 0(r2)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21148 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 19:43:53 +00:00
Chris Lattner
cbd06fc3d7 PowerPC zero extends setcc results
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21147 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 19:41:49 +00:00
Chris Lattner
6659bd7592 X86 zero extends setcc results
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21146 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 19:41:46 +00:00
Chris Lattner
7b5987d56e Allow targets which produce setcc results in non-MVT::i1 registers to describe
what the contents of the top bits of these registers are, in the common cases
of targets that sign and zero extend the results.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21145 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 19:41:18 +00:00
Chris Lattner
2467392c59 Remove somethign I had for testing
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21144 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 18:58:54 +00:00
Andrew Lenharth
2966e84a8d fix a small optimization opertunity and make gcc happy
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21143 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 18:15:28 +00:00
Chris Lattner
67255a1d17 This patch does two things. First, it canonicalizes 'X >= C' -> 'X > C-1'
(likewise for <= >=u >=u).

Second, it implements a special case hack to turn 'X gtu SINTMAX' -> 'X lt 0'

On powerpc, for example, this changes this:

        lis r2, 32767
        ori r2, r2, 65535
        cmplw cr0, r3, r2
        bgt .LBB_test_2

into:

        cmpwi cr0, r3, 0
        blt .LBB_test_2


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21142 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-07 18:14:58 +00:00