1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-30 07:22:55 +01:00
Commit Graph

1243 Commits

Author SHA1 Message Date
Chris Lattner
8d8b8b05bd There is no need to call MachineInstr::print directly, just send the MI& to an ostream.
llvm-svn: 16613
2004-09-30 16:10:45 +00:00
Chris Lattner
bb425800f5 * Wrap some comments to 80 cols
* Add const_iterator stuff
* Add a print method, which means that I can now call dump() from the
  debugger.

llvm-svn: 16612
2004-09-30 15:59:17 +00:00
Chris Lattner
168a4380d9 Simplify the logic in the simple spiller and capitalize some variables
llvm-svn: 16609
2004-09-30 02:59:33 +00:00
Chris Lattner
1f61bfb971 Switch from defaulting to the 'local' spiller to the 'simple' spiller. The
two spillers produce perfectly identical code (at least on povray and eon),
but the simple spiller is substantially faster than the local spiller. Once
the local spiller is improved, we can switch back.

Switching cuts 5.2% off of the llc time for povray (about 1.3s).

llvm-svn: 16608
2004-09-30 02:40:06 +00:00
Chris Lattner
d716b5739c Don't use a densemap for keeping track of which vregs are already loaded, just
use a simple vector.  This speeds up -spiller=simple from taking 22s to taking
.1s on povray (debug build).  This change does not modify the generated code.

llvm-svn: 16607
2004-09-30 02:33:48 +00:00
Chris Lattner
d27ff8035e Use longer and more explicit names for instance vars (particularly important
data structures).  Fix the print method to send to the right ostream, not
always cerr.  Delete typedefs that are only used once.

llvm-svn: 16606
2004-09-30 02:15:18 +00:00
Chris Lattner
962d398430 Free the VirtRegMap at the end of MachineFunction processing instead of at
the beginning of processing the next one.

llvm-svn: 16605
2004-09-30 02:02:33 +00:00
Chris Lattner
418207045d Reindent code, improve comments, move huge nested methods out of classes,
prune #includes, add print/dump methods, etc.  No functionality changes.

llvm-svn: 16604
2004-09-30 01:54:45 +00:00
Alkis Evlogimenos
4f5920aaef Add includes and use std:: for standard library calls to make code
compile on windows. This patch was contributed by Paolo Invernizzi.

llvm-svn: 16539
2004-09-28 14:42:44 +00:00
Alkis Evlogimenos
95cc7f115a Fix includes. Patch contributed by Paolo Invernizzi!
llvm-svn: 16533
2004-09-28 02:38:58 +00:00
Alkis Evlogimenos
4117e2a7a9 Grow the map on entry so that we don't crash if joinIntervals never
runs (if coalescing is disabled for example).

llvm-svn: 16259
2004-09-09 19:24:38 +00:00
Alkis Evlogimenos
a5edc58139 Use a DenseMap for mapping reg->reg. This improves the LiveInterval
analysis running time from 2.7869secs to 2.5226secs on 176.gcc.

llvm-svn: 16244
2004-09-08 03:01:50 +00:00
Alkis Evlogimenos
992cf4f638 Indent to 2 spaces and cleanup excess whitespace.
llvm-svn: 16188
2004-09-05 18:41:35 +00:00
Alkis Evlogimenos
d1388a3eae Indent to 2 spaces.
llvm-svn: 16187
2004-09-05 18:39:20 +00:00
Misha Brukman
73837314b6 Order #includes alphabetically, local .h files first.
llvm-svn: 16153
2004-09-03 18:25:53 +00:00
Alkis Evlogimenos
0c50e0f211 Fixes to make LLVM compile with vc7.1.
Patch contributed by Paolo Invernizzi!

llvm-svn: 16152
2004-09-03 18:19:51 +00:00
Alkis Evlogimenos
2676000067 Change the way we choose a free register: instead of picking the first
free allocatable register, we prefer the a free one with the most uses
of inactive intervals.

llvm-svn: 16148
2004-09-02 21:24:33 +00:00
Alkis Evlogimenos
999a10318f Change the way we choose a free register: instead of picking the first
free allocatable register, we prefer the a free one with the most uses
of inactive intervals. This causes less spills and performes a bit
better compared to gcc:

Program                 | GCC/LLC (Before)| GCC/LLC (After)
164.gzip/164.gzip       | 0.59            | 0.60
175.vpr/175.vpr         | 0.57            | 0.58
176.gcc/176.gcc         | 0.59            | 0.61
181.mcf/181.mcf         | 0.94            | 0.95
186.crafty/186.crafty   | 0.62            | 0.62
197.parser/197.parser   | 0.89            | 0.88
252.eon/252.eon         | 0.61            | 0.66
253.perlbmk/253.perlbmk | 0.79            | 0.84
254.gap/254.gap         | 0.81            | 0.81
255.vortex/255.vortex   | 0.92            | 0.93
256.bzip2/256.bzip2     | 0.69            | 0.69
300.twolf/300.twolf     | 0.91            | 0.90

llvm-svn: 16147
2004-09-02 21:23:32 +00:00
Alkis Evlogimenos
7a287da6cd We don't need to sort the added vector as unhandled intervals are
stored in a binary heap.

llvm-svn: 16143
2004-09-02 18:00:38 +00:00
Reid Spencer
c4abcbefb1 Changes For Bug 352
Move include/Config and include/Support into include/llvm/Config,
include/llvm/ADT and include/llvm/Support. From here on out, all LLVM
public header files must be under include/llvm/.

llvm-svn: 16137
2004-09-01 22:55:40 +00:00
Alkis Evlogimenos
e7e63c8da4 Be a bit more efficient when processing the active and inactive
lists. Instead of scanning the vector backwards, scan it forward and
swap each element we want to erase. Then at the end erase all removed
intervals at once. This doesn't save much: 0.08s out of 4s when
compiling 176.gcc.

llvm-svn: 16136
2004-09-01 22:52:29 +00:00
Alkis Evlogimenos
ad6f0f57fa Give a better assertion if we see a use before a def.
llvm-svn: 16135
2004-09-01 22:34:52 +00:00
Alkis Evlogimenos
0513a4d8e1 Minor code clarity changes.
llvm-svn: 16123
2004-08-31 17:39:15 +00:00
Nate Begeman
efbdaad4b7 Put this change back in after testing from Reid proved its innocence. getSpillSize now returns value in bits
llvm-svn: 16102
2004-08-29 22:00:24 +00:00
Alkis Evlogimenos
4f0c0a1adc Remove dead code.
llvm-svn: 16077
2004-08-28 22:43:31 +00:00
Alkis Evlogimenos
8ae92113d6 Now that LiveIntervals::addIntervalsForSpills is fixed, do not require
LiveVariables.

llvm-svn: 16076
2004-08-27 19:00:29 +00:00
Alkis Evlogimenos
eb7256be64 Only update LiveVariables if it is available. addIntervalsForSpills
runs after the initial run of the live interval analysis.

llvm-svn: 16075
2004-08-27 18:59:22 +00:00
Alkis Evlogimenos
7c06c0aeee Back out this change as it broke the build last night. This should be
investicated further as the linearscan variants don't really need
LiveVariables...

llvm-svn: 16074
2004-08-27 18:01:21 +00:00
Nate Begeman
ad24578e7d Back out change to divide getSpillSize by 8 until I figure out why it breaks x86, which has register sizes in bits.
llvm-svn: 16073
2004-08-27 16:48:24 +00:00
Alkis Evlogimenos
f18ee1fc99 The linear scan variants do not require the LiveVariables analysis.
llvm-svn: 16071
2004-08-27 04:51:13 +00:00
Nate Begeman
c0e42be976 Register sizes are in bits, not bytes
llvm-svn: 16070
2004-08-27 04:28:10 +00:00
Alkis Evlogimenos
03913413da Use newly added API in MRegisterInfo and don't expose the allocatable
register set anymore. Its users now use the MRegisterInfo API.

llvm-svn: 16061
2004-08-26 22:23:32 +00:00
Alkis Evlogimenos
3bfc6999f2 Use newly added API in MRegisterInfo.
llvm-svn: 16060
2004-08-26 22:22:38 +00:00
Chris Lattner
66d61f3a72 Fix a bug in a previous checkin of mine, correcting
Regression.CodeGen.Generic.2004-04-09-SameValueCoalescing.llx and the
code size problem.

This bug prevented us from doing most register coallesces.

llvm-svn: 16031
2004-08-24 17:48:29 +00:00
Chris Lattner
05d5a44c02 Do not use .xword and friends to emit zeros on V9. Apparently there are issues
with emitting .xwords when not on an 8-byte boundary (.xword 0 is not the
same as 8 .byte 0's).  Because we do not know when or when we are not aligned,
just emit bytes like the old V9 asmprinter did.

llvm-svn: 16006
2004-08-24 00:26:11 +00:00
Chris Lattner
0df8bcf2fa Register info alignment is in bits, frame object alignment is (currently) in
bytes.

llvm-svn: 15970
2004-08-21 20:04:59 +00:00
Chris Lattner
7ead36e4a4 Now that we have per-register spill size/alignment info, remove more uses
of getRegClass

llvm-svn: 15967
2004-08-21 19:45:10 +00:00
Brian Gaeke
7364ea9c20 Instead of using isDummyPhiInstr, we just compare the opcode with V9::PHI.
llvm-svn: 15906
2004-08-18 20:04:24 +00:00
Brian Gaeke
bec1f54a8e Instead of using isDummyPhiInstr, we just compare the opcode with V9::PHI.
Also, squash a use of TargetInstrInfo::isNop().

llvm-svn: 15905
2004-08-18 20:04:21 +00:00
Chris Lattner
8c5096d223 Rename var
llvm-svn: 15897
2004-08-18 02:22:55 +00:00
Chris Lattner
d8c7f4bf3c Add support for targets without a .zero directive
llvm-svn: 15894
2004-08-17 21:38:40 +00:00
Chris Lattner
659175521b Add support for alignment
llvm-svn: 15888
2004-08-17 19:14:29 +00:00
Chris Lattner
b8703980b4 Use a designated comment character when printing comments.
llvm-svn: 15880
2004-08-17 16:27:05 +00:00
Chris Lattner
bf03260147 Add support for targets that have .quad, drop extra tab inserted
llvm-svn: 15870
2004-08-17 06:48:16 +00:00
Chris Lattner
e47c31e25e Implement emitGlobalConstant
llvm-svn: 15868
2004-08-17 06:36:49 +00:00
Chris Lattner
1362ba130b We now allow targets to use any prefix they want for global symbols. Lets
hear it for ".".

llvm-svn: 15863
2004-08-17 06:06:19 +00:00
Chris Lattner
460d80e77d Some asmwriters want an _ prefix
llvm-svn: 15845
2004-08-17 02:28:26 +00:00
Chris Lattner
aac51aa3d8 Initial implementation of the asmprinter base class
llvm-svn: 15838
2004-08-16 23:15:22 +00:00
Chris Lattner
f3450010d5 Move this code to lib/Target/SparcV9/MachineFunctionInfo.cpp
llvm-svn: 15834
2004-08-16 22:36:34 +00:00
Chris Lattner
ea590216d9 This file is moving to lib/Target/SparcV9
llvm-svn: 15833
2004-08-16 22:36:10 +00:00
Chris Lattner
6d8d39e17b Move MachineCodeForInstruction.h and MachineFunctionInfo.h into lib/Target/SparcV9
llvm-svn: 15830
2004-08-16 21:55:02 +00:00
Chris Lattner
2eaaf7660b Nuke this file
llvm-svn: 15829
2004-08-16 21:36:40 +00:00
Chris Lattner
625e4e8f34 Fold MachineInstrAnnot.cpp into this file
llvm-svn: 15828
2004-08-16 21:36:31 +00:00
Brian Gaeke
8ca16af75f Add a note that people shouldn't use MachineFunctionInfo.
llvm-svn: 15823
2004-08-16 18:27:24 +00:00
Chris Lattner
0e350557bf Reduce usage of MRegisterInfo::getRegClass
llvm-svn: 15784
2004-08-15 22:23:09 +00:00
Chris Lattner
a27bbd33e6 The insertion method returns void now
llvm-svn: 15779
2004-08-15 22:14:31 +00:00
Chris Lattner
450d93b2ff Nuke ifdef'd out code
llvm-svn: 15777
2004-08-15 22:03:57 +00:00
Chris Lattner
463d46f17a Stop using CreateStackObject(RegClass*)
llvm-svn: 15775
2004-08-15 22:02:22 +00:00
Chris Lattner
e58190f5f6 These methods no longer take a TargetRegisterClass* operand.
llvm-svn: 15774
2004-08-15 21:56:44 +00:00
Alkis Evlogimenos
dbe432aee7 Make this compile on gc 3.4.1 (static_cast to non-const type was not
allowed).

llvm-svn: 15766
2004-08-15 09:18:55 +00:00
Nate Begeman
cbc21460f2 Elminiate MachineFunction& argument from eliminateFrameIndex
llvm-svn: 15736
2004-08-14 22:00:10 +00:00
Chris Lattner
1a0fa932e1 Split saveCallerSavedRegisters into two methods for clarity, and add comments.
Add support for targets that must spill certain physregs at certain locations.

Patch contributed by Nate Begeman, slightly hacked by me.

llvm-svn: 15701
2004-08-12 19:01:14 +00:00
Chris Lattner
5ba5260225 Forward substitute some constants into their users
llvm-svn: 15693
2004-08-12 18:10:18 +00:00
Chris Lattner
8d142c6f87 The only target that uses this code (v9) always has argsOnStackHaveFixedSize
set to true (obviously)

llvm-svn: 15692
2004-08-12 18:06:35 +00:00
Reid Spencer
554bdc52b2 Get rid of a warning when compiling optimized. Uninitialized variable has
been initialized.

llvm-svn: 15565
2004-08-07 15:19:31 +00:00
Chris Lattner
c5f185c269 Ok get rid of the REST of the tabs
llvm-svn: 15564
2004-08-07 07:18:41 +00:00
Chris Lattner
560e2f0751 Death to tabs
llvm-svn: 15563
2004-08-07 07:07:57 +00:00
Alkis Evlogimenos
4084c50fad Clean up whitespace.
llvm-svn: 15490
2004-08-04 09:46:56 +00:00
Alkis Evlogimenos
db3b503b6f Convert indentation to 2 spaces.
llvm-svn: 15489
2004-08-04 09:46:26 +00:00
Brian Gaeke
a3937dba33 Include SparcV9TmpInstr.h to pick up the def. of TmpInstruction,
instead of InstrSelection.h, which is dead.

llvm-svn: 15476
2004-08-04 07:34:57 +00:00
Brian Gaeke
10dd34f5d8 Tighten up some whitespace. Include SparcV9TmpInstr.h to pick up
the def. of TmpInstruction, instead of InstrSelection.h, which is
dead.

llvm-svn: 15475
2004-08-04 07:34:44 +00:00
Chris Lattner
cccb1332ed Squelch warnings in release mode
llvm-svn: 15460
2004-08-04 03:51:55 +00:00
Misha Brukman
245315547e Add #include <cstdlib> and abort() to silence a warning
llvm-svn: 15413
2004-08-02 14:02:21 +00:00
Misha Brukman
1fd8909108 * ceil() requires #include <cmath> for compilation
* Alphabetize #includes
* Fix some lines to fit within 80 cols

llvm-svn: 15412
2004-08-02 13:59:10 +00:00
Tanya Lattner
414e014c5f Adding ModuloScheduling so that it compiles for everyone.
llvm-svn: 15408
2004-08-01 19:00:17 +00:00
Chris Lattner
786c788d12 Dereferencing end() is bad.
llvm-svn: 15402
2004-08-01 09:51:42 +00:00
Alkis Evlogimenos
1dd3ea5515 Make OptimizeBlock take a MachineFunction::iterator instead of a
MachineBasicBlock* as a parameter so that nxext() and prior() helper
functions can work naturally on it.

llvm-svn: 15376
2004-07-31 19:24:41 +00:00
Chris Lattner
ecca71ce92 Next on a pointer increments the pointer, not an iterator
llvm-svn: 15375
2004-07-31 18:40:36 +00:00
Alkis Evlogimenos
b28bc78d0a Use next() helper to make code more readable. Use
MachineFunction::iterator instead of MachineBasicBlock* to avoid
dereferencing end iterators.

llvm-svn: 15373
2004-07-31 15:14:29 +00:00
Alkis Evlogimenos
143aeb646c Use MachineFunction::iterator instead of a MachineBasicBlock* because
FallThrough maybe == to MF.end().

llvm-svn: 15372
2004-07-31 15:03:52 +00:00
Chris Lattner
7b12b8a13e Implement a simple target-independent CFG cleanup pass
llvm-svn: 15368
2004-07-31 10:01:27 +00:00
Tanya Lattner
e757e5a231 Updated ModuloScheduling. It makes it all the wya through register allocation on the new code!!
llvm-svn: 15351
2004-07-30 23:36:10 +00:00
Brian Gaeke
9abd4de830 Convert a few assertions with side-effects into regular old runtime checks.
These side-effects seem to make a difference when using llc -march=sparcv9
in Release mode (i.e., with -DNDEBUG); when they are left out, lots of
instructions just get dropped on the floor, because they never end up
in the schedule.

llvm-svn: 15339
2004-07-29 21:31:20 +00:00
Misha Brukman
58104df77b Fix #includes of i*.h => Instructions.h as per PR403.
llvm-svn: 15334
2004-07-29 17:30:56 +00:00
Chris Lattner
42b2748566 Fix #includes of i*.h => Instructions.h as per PR403:
http://llvm.cs.uiuc.edu/PR403 .

llvm-svn: 15333
2004-07-29 17:23:00 +00:00
Chris Lattner
d5247fc917 Fix #includes of i*.h => Instructions.h as per PR403:
http://llvm.cs.uiuc.edu/PR403 .

llvm-svn: 15331
2004-07-29 17:11:37 +00:00
Brian Gaeke
99f18cc97f TargetInstrInfo::hasOperandInterlock() is always true, because it is
never overridden by any target.

llvm-svn: 15308
2004-07-28 19:24:48 +00:00
Alkis Evlogimenos
3e3fe9ff5a Add some comments to the backtracking code.
llvm-svn: 15200
2004-07-25 08:10:33 +00:00
Chris Lattner
fd9f044bcf Fix the sense of joinable
llvm-svn: 15196
2004-07-25 07:47:25 +00:00
Chris Lattner
7630df9aa0 This patch makes use of the infrastructure implemented before to safely and
aggressively coallesce live ranges even if they overlap.  Consider this LLVM
code for example:

int %test(int %X) {
        %Y = mul int %X, 1      ;; Codegens to Y = X
        %Z = add int %X, %Y
        ret int %Z
}

The mul is just there to get a copy into the code stream.  This produces
this machine code:

 (0x869e5a8, LLVM BB @0x869b9a0):
        %reg1024 = mov <fi#-2>, 1, %NOREG, 0    ;; "X"
        %reg1025 = mov %reg1024                 ;; "Y"  (subsumed by X)
        %reg1026 = add %reg1024, %reg1025
        %EAX = mov %reg1026
        ret

Note that the life times of reg1024 and reg1025 overlap, even though they
contain the same value.  This results in this machine code:

test:
        mov %EAX, DWORD PTR [%ESP + 4]
        mov %ECX, %EAX
        add %EAX, %ECX
        ret

Another, worse case involves loops and PHI nodes.  Consider this trivial loop:
testcase:

int %test2(int %X) {
entry:
        br label %Loop
Loop:
        %Y = phi int [%X, %entry], [%Z, %Loop]
        %Z = add int %Y, 1
        %cond = seteq int %Z, 100
        br bool %cond, label %Out, label %Loop
Out:
        ret int %Z
}

Because of interactions between the PHI elimination pass and the register
allocator, this got compiled to this code:

test2:
        mov %ECX, DWORD PTR [%ESP + 4]
.LBBtest2_1:
***     mov %EAX, %ECX
        inc %EAX
        cmp %EAX, 100
***     mov %ECX, %EAX
        jne .LBBtest2_1

        ret

Or on powerpc, this code:

_test2:
        mflr r0
        stw r0, 8(r1)
        stwu r1, -60(r1)
.LBB_test2_1:
        addi r2, r3, 1
        cmpwi cr0, r2, 100
***     or r3, r2, r2
        bne cr0, .LBB_test2_1

***     or r3, r2, r2
        lwz r0, 68(r1)
        mtlr r0
        addi r1, r1, 60
        blr 0



With this improvement in place, we now generate this code for these two
testcases, which is what we want:


test:
        mov %EAX, DWORD PTR [%ESP + 4]
        add %EAX, %EAX
        ret

test2:
        mov %EAX, DWORD PTR [%ESP + 4]
.LBBtest2_1:
        inc %EAX
        cmp %EAX, 100
        jne .LBBtest2_1 # Loop
        ret

Or on PPC:

_test2:
        mflr r0
        stw r0, 8(r1)
        stwu r1, -60(r1)
.LBB_test2_1:
        addi r3, r3, 1
        cmpwi cr0, r3, 100
        bne cr0, .LBB_test2_1

        lwz r0, 68(r1)
        mtlr r0
        addi r1, r1, 60
        blr 0


Static numbers for spill code loads/stores/reg-reg copies (smaller is better):

em3d:       before: 47/25/26         after: 44/22/24
164.gzip:   before: 433/245/310      after: 403/231/278
175.vpr:    before: 3721/2189/1581   after: 4144/2081/1423
176.gcc:    before: 26195/8866/9235  after: 25942/8082/8275
186.crafty: before: 4295/2587/3079   after: 4119/2519/2916
252.eon:    before: 12754/7585/5803  after: 12508/7425/5643
256.bzip2:  before: 463/226/315      after: 482:241/309


Runtime perf number samples on X86:

gzip: before: 41.09 after: 39.86
bzip2: runtime: before: 56.71s after: 57.07s
gcc: before: 6.16 after: 6.12
eon: before: 2.03s after: 2.00s
llvm-svn: 15194
2004-07-25 07:11:19 +00:00
Chris Lattner
0997e50af5 Make a method const, no functionality changes
llvm-svn: 15193
2004-07-25 06:23:01 +00:00
Chris Lattner
32cab750d4 Fix a bug where we incorrectly value numbered the first PHI definition the
same as the PHI use.  This is not correct as the PHI use value is different
depending on which branch is taken.  This fixes espresso with aggressive
coallescing, and perhaps others.

llvm-svn: 15189
2004-07-25 05:45:18 +00:00
Chris Lattner
e3da94e11f Fix a bug in the range remover
llvm-svn: 15188
2004-07-25 05:43:53 +00:00
Chris Lattner
4ed2b97d51 Add debugging output for joining assignments
llvm-svn: 15187
2004-07-25 03:24:11 +00:00
Alkis Evlogimenos
2faa6cdc75 Remove implementation of operator= and make it private so that it is
not used accidentally.

llvm-svn: 15172
2004-07-24 18:55:15 +00:00
Alkis Evlogimenos
0618429149 Change std::map<unsigned, LiveInterval*> into a std::map<unsigned,
LiveInterval>. This saves some space and removes the pointer
indirection caused by following the pointer.

llvm-svn: 15167
2004-07-24 11:44:15 +00:00
Chris Lattner
26856ab7c3 whoops, didn't mean to remove this
llvm-svn: 15157
2004-07-24 04:32:22 +00:00
Chris Lattner
de8efecef1 In the joiner, merge the small interval into the large interval. This restores
us back to taking about 10.5s on gcc, instead of taking 15.6s!  The net result
is that my big patches have hand no significant effect on compile time or code
quality.  heh.

llvm-svn: 15156
2004-07-24 03:41:50 +00:00
Chris Lattner
c71a9684f8 Completely eliminate the intervals_ list. instead, the r2iMap_ maintains
ownership of the intervals.

llvm-svn: 15155
2004-07-24 03:32:06 +00:00
Chris Lattner
b05762ade7 Big change to compute logical value numbers for each LiveRange added to an
Interval.  This generalizes the isDefinedOnce mechanism that we used before
to help us coallesce ranges that overlap.  As part of this, every logical
range with a different value is assigned a different number in the interval.
For example, for code that looks like this:

0  X = ...
4  X += ...
  ...
N    = X

We now generate a live interval that contains two ranges: [2,6:0),[6,?:1)
reflecting the fact that there are two different values in the range at
different positions in the code.

Currently we are not using this information at all, so this just slows down
liveintervals.  In the future, this will change.

Note that this change also substantially refactors the joinIntervalsInMachineBB
method to merge the cases for virt-virt and phys-virt joining into a single
case, adds comments, and makes the code a bit easier to follow.

llvm-svn: 15154
2004-07-24 02:59:07 +00:00