Commit Graph

20154 Commits

Author SHA1 Message Date
Chris Lattner
f31f09e67c allow for a target to ask for a dag isel
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23237 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-03 01:15:25 +00:00
Chris Lattner
4a24c645c8 Add an option and stuff implementation of a dag isel emitter
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23236 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-03 01:14:03 +00:00
Chris Lattner
6a8a21ced4 Fix a checking failure in gs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23235 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-03 01:04:40 +00:00
Chris Lattner
25dae727f3 Change the isel to not break out of the big giant switch. Instead, the
switch should never be exited, so its bottom is now unreachable.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23234 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-03 00:53:47 +00:00
Chris Lattner
7cd09cf942 rearrange logical ops to group them together more consistently.
Define the PatFrag class which can be used to define subpatterns to match
things with.  Define 'not', and use it to define the patterns for andc,
nand, etc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23233 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-03 00:21:51 +00:00
Chris Lattner
6159fb20c2 Add AND/OR/XOR
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23232 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 22:35:53 +00:00
Nate Begeman
646d7e2727 Next round of DAG Combiner changes. Just need to support multiple return
values, and then we should be able to hook it up.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23231 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 21:18:40 +00:00
Chris Lattner
218a15d02c Add some initial patterns to simple binary instructions, though they
currently don't do anything.  This elides patterns for binary operators
that ping on the carry flag, since we don't model it yet.

This patch also removes PPC::SUB, because it is dead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23230 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 21:18:00 +00:00
Chris Lattner
6fdcb250d5 Clean up some code from the last checkin
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23229 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 20:32:45 +00:00
Chris Lattner
99c25b86aa Fix a bug in legalize where it would emit two calls to libcalls that return
i64 values on targets that need that expanded to 32-bit registers.  This fixes
PowerPC/2005-09-02-LegalizeDuplicatesCalls.ll and speeds up 189.lucas from
taking 122.72s to 81.96s on my desktop.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23228 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 20:26:58 +00:00
Chris Lattner
c789701360 Test that converting from double to int64 results in one libcall, not one
and a dead one.  This is a legalize bug


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23227 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 20:24:10 +00:00
Chris Lattner
73e37c35be turn on dag isel by default
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23226 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 19:53:54 +00:00
Chris Lattner
70b9b1098a Make sure to auto-cse nullary ops
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23224 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 19:36:17 +00:00
Chris Lattner
ab2eb66f07 add a map for nullary ops
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23223 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 19:35:42 +00:00
Jim Laskey
839615a510 Add help support for -mcpu and -mattr.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23222 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 19:27:43 +00:00
Chris Lattner
6621e3b963 Fix some buggy logic where we would try to remove nodes with two operands
from the binary ops map, even if they had multiple results.  This latent bug
caused a few failures with the dag isel last night.

To prevent stuff like this from happening in the future, add some really
strict checking to make sure that the CSE maps always match up with reality!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23221 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 19:15:44 +00:00
Andrew Lenharth
aa38ce436b Pull out Lowering in preperation for multiple ISels. Oh, and get rid of some stuff
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23220 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 18:46:02 +00:00
Chris Lattner
fd88f64934 Don't create zero sized stack objects even for array allocas with a zero
number of elements.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23219 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 18:41:28 +00:00
Chris Lattner
1e9de3ed2d Decouple fsqrt from gpul optimizations, implementing fsqrt.ll.
Remove the -enable-gpopt option which is subsumed by feature flags.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23218 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 18:33:05 +00:00
Chris Lattner
446a4eaee5 new testcase to ensure fsqrt is generated for correct subtargets only, and
that the fsqrt feature works.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23217 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 18:32:22 +00:00
Chris Lattner
0271077eba Move a bunch of non-deprecated methods above the "deprecated line"
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23216 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 18:16:20 +00:00
Chris Lattner
a639a43602 Fix the release build, noticed by Eric van Riet Paap
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23215 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 07:09:28 +00:00
Chris Lattner
7b4ad94282 Fix a problem that Dan Berlin noticed, where reassociation would not succeed
in building maximal expressions before simplifying them.  In particular, i
cases like this:

X-(A+B+X)

the code would consider A+B+X to be a maximal expression (not understanding
that the single use '-' would be turned into a + later), simplify it (a noop)
then later get simplified again.

Each of these simplify steps is where the cost of reassociation comes from,
so this patch should speed up the already fast pass a bit.

Thanks to Dan for noticing this!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23214 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 07:07:58 +00:00
Chris Lattner
2cd85da3ed Avoid creating garbage instructions, just move the old add instruction
to where we need it when converting -(A+B+C) -> -A + -B + -C.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23213 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 06:38:04 +00:00
Chris Lattner
d6c69e9bba new testcase for recent bugfix
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23212 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 06:12:12 +00:00
Chris Lattner
368a3aabb2 add some assertions and fix problems where reassociate could access the
Ops vector out of range


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23211 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 05:23:22 +00:00
Jeff Cohen
7cd57f4c45 Fix VC++ build errors
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23210 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 02:51:42 +00:00
Chris Lattner
f505949580 Restore this patch now that the latent bug has been fixed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23209 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 01:24:55 +00:00
Chris Lattner
08951a32fb Make sure to legalize assert[zs]ext's operand correctly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23208 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 01:15:01 +00:00
Chris Lattner
5b3224fe7e Revert the previous patch which causes a mysterious regression in toast.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23207 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:47:05 +00:00
Chris Lattner
5ab6f5fe66 Teach live intervals to not crash on dead livein regs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23206 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:20:32 +00:00
Chris Lattner
fab08875b7 For values that are live across basic blocks and need promotion, use ANY_EXTEND
instead of ZERO_EXTEND to eliminate extraneous extensions.  This eliminates
dead zero extensions on formal arguments and other cases on PPC, implementing
the newly tightened up test/Regression/CodeGen/PowerPC/small-arguments.ll test.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23205 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:19:37 +00:00
Chris Lattner
13c78e2e5f legalize ANY_EXTEND appropriately
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23204 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:18:10 +00:00
Chris Lattner
4ed11b4b79 Add support for ANY_EXTEND and add a few minor folds for it
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23203 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:17:32 +00:00
Chris Lattner
fb8b8a75ff Handle any_extend like zext
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23202 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:16:09 +00:00
Chris Lattner
26e04bbf77 Handle ANY_EXTEND like ZERO_EXTEND. Simplify the extend/truncate code on
the observation that it only has to handle i1 -> i64 and i64 -> i1.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23201 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:15:30 +00:00
Chris Lattner
7e122db776 Add a new ANY_EXTEND node, which operates like an extension but has undefined
top bits.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23200 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:14:40 +00:00
Chris Lattner
7dd8b05fd7 This should permit NOT and's, not even dead ones.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23199 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-02 00:13:56 +00:00
Chris Lattner
2a00daac58 Implement small-arguments.ll:test3 by teaching the DAG optimizer that
the results of calls to functions returning small values are properly
sign/zero extended.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23198 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 23:44:32 +00:00
Chris Lattner
9832ce0f0a Another case that shouldn't have sign extends: functions returning short
are known to return properly sign extended values, no need for an explicit
extension.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23197 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 23:43:58 +00:00
Nate Begeman
39f60a2302 Fix some code in the current node combining code, spotted when it was moved
over to DAGCombiner.cpp

1. Don't assume that SetCC returns i1 when folding (xor (setcc) constant)
2. Don't duplicate code in folding AND with AssertZext that is handled by
   MaskedValueIsZero


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23196 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 23:25:49 +00:00
Nate Begeman
4ebd805c6a Implement first round of feedback from chris (there's still a couple things
left to do).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23195 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 23:24:04 +00:00
Chris Lattner
2b8ad8e55a Align functions to 16-byte boundaries, to eliminate noise in performance measurements. This improves the performance of 'treeadd' by about 20% with the dag
isel, restoring it to the pattern-isel level (which happens to get the alignment right).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23194 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 23:08:50 +00:00
Chris Lattner
a82f7b2be0 Local labels on darwin apparently start with just 'L', not .L like other
platforms.  This reduces executable size and makes shark realize the actual
bounds of functions instead of showing each MBB as a function :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23193 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 21:48:35 +00:00
Jim Laskey
b1e1180ca0 1. Use SubtargetFeatures in llc/lli.
2. Propagate feature "string" to all targets.

3. Implement use of SubtargetFeatures in PowerPCTargetSubtarget.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23192 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 21:38:21 +00:00
Jim Laskey
b3302db18a This new class provides support for platform specific "features". The intent
is to manage processor specific attributes from the command line.  See examples
of use in llc/lli and PowerPCTargetSubtarget.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23191 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 21:36:18 +00:00
Chris Lattner
75592e4137 Implement dynamic allocas correctly. In particular, because we were copying
directly out of R1 (without using a CopyFromReg, which uses a chain), multiple
allocas were getting CSE'd together, producing bogus code.  For this:

int %foo(bool %X, int %A, int %B) {
        br bool %X, label %T, label %F
F:
        %G = alloca int
        %H = alloca int
        store int %A, int* %G
        store int %B, int* %H
        %R = load int* %G
        ret int %R
T:
        ret int 0
}

We were generating:

_foo:
        stwu r1, -16(r1)
        stw r31, 4(r1)
        or r31, r1, r1
        stw r1, 12(r31)
        cmpwi cr0, r3, 0
        bne cr0, .LBB_foo_2     ; T
.LBB_foo_1:     ; F
        li r2, 16
        subf r2, r2, r1   ;; One alloca
        or r1, r2, r2
        or r3, r1, r1
        or r1, r2, r2
        or r2, r1, r1
        stw r4, 0(r3)
        stw r5, 0(r2)
        lwz r3, 0(r3)
        lwz r1, 12(r31)
        lwz r31, 4(r31)
        lwz r1, 0(r1)
        blr
.LBB_foo_2:     ; T
        li r3, 0
        lwz r1, 12(r31)
        lwz r31, 4(r31)
        lwz r1, 0(r1)
        blr

Now we generate:

_foo:
        stwu r1, -16(r1)
        stw r31, 4(r1)
        or r31, r1, r1
        stw r1, 12(r31)
        cmpwi cr0, r3, 0
        bne cr0, .LBB_foo_2     ; T
.LBB_foo_1:     ; F
        or r2, r1, r1
        li r3, 16
        subf r2, r3, r2  ;; Alloca 1
        or r1, r2, r2
        or r2, r1, r1
        or r6, r1, r1
        subf r3, r3, r6  ;; Alloca 2
        or r1, r3, r3
        or r3, r1, r1
        stw r4, 0(r2)
        stw r5, 0(r3)
        lwz r3, 0(r2)
        lwz r1, 12(r31)
        lwz r31, 4(r31)
        lwz r1, 0(r1)
        blr
.LBB_foo_2:     ; T
        li r3, 0
        lwz r1, 12(r31)
        lwz r31, 4(r31)
        lwz r1, 0(r1)
        blr

This fixes Povray and SPASS with the dag isel, the last two failing cases.
Tommorow we will hopefully turn it on by default! :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23190 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 21:31:30 +00:00
Chris Lattner
393ecd6d2d Fix a bug where we were useing HA to get the high part, which seems like it
could cause a miscompile.  Fixing this didn't fix the two programs that fail
though.  :(

This also changes the implementation to follow the pattern selector more
closely, causing us to select 0 to li instead of lis.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23189 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 19:38:28 +00:00
Chris Lattner
50ff55c2c7 Do not select the operands being passed into SelectCC. IT does this itself
and selecting early prevents folding immediates into the cmpw* instructions


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23188 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 19:20:44 +00:00
Chris Lattner
da8abb0239 It is NDEBUG not _NDEBUG
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23186 91177308-0d34-0410-b5e6-96231b3b80d8
2005-09-01 18:44:10 +00:00