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

8669 Commits

Author SHA1 Message Date
Reid Spencer
24a84793ba * Use low-level unix I/O interface since we're on Unix.
* Don't use variable length arrays (replaced with alloca)

llvm-svn: 17901
2004-11-16 17:14:08 +00:00
Chris Lattner
b40f8214ea Simplify the remapper by only needing one map, since the body of the functions
being linked do not need to be remapped any longer.

llvm-svn: 17900
2004-11-16 17:12:38 +00:00
Chris Lattner
92e712b00f Do not delete dead invoke instructions!
llvm-svn: 17897
2004-11-16 16:32:28 +00:00
Misha Brukman
3885f4ec6d Erase non-applicable Unix comment, this is Win32
llvm-svn: 17896
2004-11-16 07:35:32 +00:00
Brian Gaeke
c68a1fb6b9 Update list of failing benchmarks
llvm-svn: 17895
2004-11-16 07:32:58 +00:00
Brian Gaeke
effc63fac5 Correct the implicit-defs information for indirect and direct calls.
You can't have implicit defs that overlap explicit defs, or implicit
defs that alias one another.

llvm-svn: 17894
2004-11-16 07:32:09 +00:00
Chris Lattner
0493754bcf Take advantage of the fact that we are allowed to clobber the input module
by splicing function bodies from the src module to the destination module.

This speeds up linking quite a bit, e.g. gccld time on 176.gcc from 26s -> 20s
when forming the .rbc file, with a profile build.  One of the really strange
but cool effects of this patch is that it speeds up the optimizers as well,
from 12s -> 10.7s, presumably because of better locality???

In any case, this is just a first step.  We can trivially get rid of the
LocalMap now and do other simplifications.

llvm-svn: 17893
2004-11-16 07:31:51 +00:00
Reid Spencer
35a3e7eacc Fix typeo in comment.
llvm-svn: 17892
2004-11-16 07:05:16 +00:00
Reid Spencer
026d2183e7 Remove useless #include.
Patch contributed by Jeff Cohen.

llvm-svn: 17891
2004-11-16 07:00:23 +00:00
Reid Spencer
0d16952663 Simplify code.
Patch contributed by Jeff Cohen.

llvm-svn: 17890
2004-11-16 06:59:53 +00:00
Reid Spencer
cdc0b6d962 Standardize on 'class' instead of 'struct'. Gets rid of warnings in VC++
Patch contributed by Jeff Cohen.

llvm-svn: 17889
2004-11-16 06:58:55 +00:00
Brian Gaeke
23b56332bc Give a better message for a common assertion failure.
llvm-svn: 17887
2004-11-16 06:52:35 +00:00
Reid Spencer
e64ca0563a Add a comment to some code that at first glance just doesn't look right.
llvm-svn: 17886
2004-11-16 06:50:36 +00:00
Reid Spencer
ddfafa643a Per code review:\
* Adjust indentation\
* Ensure memory do not leak if exceptions happen (std::auto_ptr use)

llvm-svn: 17885
2004-11-16 06:47:41 +00:00
Reid Spencer
2d66bbb7c3 Per code review:\
* Make the numVbrBytes function more efficient and better documented \
* Fix a bug in name truncation \
* Add comments before functions \
* Get rid of functions that are now inlined into the header \
* Do not have Archive doing symbol table printing \
* Put assert comments into the assert so they print out \
* Make sure foreign symbol tables are written

llvm-svn: 17884
2004-11-16 06:47:30 +00:00
Reid Spencer
b6d946d19d Per code review:\
* Make sure we write out the foreign symbol table if we read one \
* Make the padding calculation more efficiently and avoid Solaris warnings

llvm-svn: 17883
2004-11-16 06:47:19 +00:00
Reid Spencer
e39cdbd7f7 Per code review:\
* Make sure all members are initialized upon construction

llvm-svn: 17882
2004-11-16 06:47:07 +00:00
Chris Lattner
be4c91f75f The second arg may be clobbered by this function
llvm-svn: 17880
2004-11-16 06:41:36 +00:00
Reid Spencer
2fe884dad8 Per code review:\
* Get rid of memory leaks on exception \
* Provide better comments of how the memory handling works

llvm-svn: 17876
2004-11-16 06:41:05 +00:00
Chris Lattner
e816a3abc2 use an autoptr
llvm-svn: 17875
2004-11-16 06:40:54 +00:00
Reid Spencer
bcef3fef1d Per code review:
* get rid of (void) construct in function declarations
* make toString a const member
* add a default implementation of toString for Win32

llvm-svn: 17873
2004-11-16 06:22:17 +00:00
Reid Spencer
4031ff44bb Per code review:
* Clean up the StatusInfo constructor to construct all members and give
  them reasonable values.
* Get rid of the Vector typedef and make the interface to
  getDirectoryContent use a std::set instead of a std::vector so the dir
  content is sorted.
* Make the getStatusInfo method const and not return a useless boolean.

llvm-svn: 17872
2004-11-16 06:15:19 +00:00
Reid Spencer
fdf8ac8d29 Per code review:
*Implement/Document the cl::extrahelp feature instead of the MoreHelp ptr.

llvm-svn: 17871
2004-11-16 06:11:52 +00:00
Chris Lattner
c13149e03e Simplify namespaces
llvm-svn: 17870
2004-11-16 04:47:33 +00:00
Chris Lattner
41d31d7461 Remove a dead function, which died when we got GAS emission working (phwew,
hold your nose!)

llvm-svn: 17869
2004-11-16 04:34:29 +00:00
Chris Lattner
b378786c97 Implement a simple FIXME: if we are emitting a basic block address that has
already been emitted, we don't have to remember it and deal with it later,
just emit it directly.

llvm-svn: 17868
2004-11-16 04:30:51 +00:00
Chris Lattner
3f73c77ace * Merge some win32 ifdefs together
* Get rid of "emitMaybePCRelativeValue", either we want to emit a PC relative
  value or not: drop the maybe BS.  As it turns out, the only places where
  the bool was a variable coming in, the bool was a dynamic constant.

llvm-svn: 17867
2004-11-16 04:21:18 +00:00
Chris Lattner
89536fedee Now that we have ghost linkage, we can force resolution of external symbols
immediately instead of lazily.

In this program, for example:

int main() {
  printf("hello world\n");
  printf("hello world\n");
  printf("hello world\n");
  printf("hello world\n");
}

We used to have to go through compilation callback 4 times (once for each
call to printf), now we don't go to it at all.

Thanks to Misha for noticing this, and for adding the initial ghost linkage
patches.

llvm-svn: 17864
2004-11-15 23:20:04 +00:00
Chris Lattner
583ea4764c There is no reason to try to materialize the function from bytecode if it
already has been.  This may be a small speedup.

llvm-svn: 17863
2004-11-15 23:18:09 +00:00
Chris Lattner
3ed3e8669f Add debug-only=jit printout, so we see when lazily resolved symbols are
set up.

llvm-svn: 17862
2004-11-15 23:16:55 +00:00
Chris Lattner
9ef34d44e1 Simplify and rearrange long shift code
llvm-svn: 17861
2004-11-15 23:16:34 +00:00
Chris Lattner
10b5a5c0da Do not emit FunctionBlock blocks for external functions. This shrinks
bytecode files by about 8 bytes per external function

llvm-svn: 17859
2004-11-15 22:39:49 +00:00
Chris Lattner
558ce75905 Use a per-function flag bit to indicate whether or not there is a function
body for the function.

llvm-svn: 17858
2004-11-15 22:38:52 +00:00
Chris Lattner
c199d448e9 Disable this change, it was premature
llvm-svn: 17857
2004-11-15 21:56:33 +00:00
Chris Lattner
704cad0375 Make *SURE* to null out the pointer before throwing an exception, otherwise
the dtor for the BytecodeReader class will try to free it again!

llvm-svn: 17856
2004-11-15 21:55:33 +00:00
Chris Lattner
4a2ae87435 Cleanups. Null out pointer after freeing it for paranoia
llvm-svn: 17855
2004-11-15 21:55:06 +00:00
Chris Lattner
571b490404 If a function is external, do not output a FunctionBlock for the function
AT ALL.  This saves 11 bytes per external function from the bytecode file,
and is also required to make GhostLinkage work.

llvm-svn: 17854
2004-11-15 21:46:40 +00:00
Chris Lattner
d1e621acc6 Make functions default to having external linkage if they have no
FunctionBlock.

llvm-svn: 17853
2004-11-15 21:43:03 +00:00
Misha Brukman
757502af07 Add BCTR and LWZU instruction opcodes
llvm-svn: 17851
2004-11-15 21:20:09 +00:00
Misha Brukman
a3b5e5f569 Remove extra space char
llvm-svn: 17834
2004-11-15 19:30:05 +00:00
Reid Spencer
e986ef23d7 Remove unused variable for compilation by VC++.
Patch contributed by Morten Ofstad.

llvm-svn: 17830
2004-11-15 17:29:41 +00:00
Reid Spencer
247d690431 Provide the ThrowErrno utility.
Patch contributed by Morten Ofstad

llvm-svn: 17827
2004-11-15 17:21:57 +00:00
Reid Spencer
238e966215 Adjust implementation to match new interface.
Patch provided by Morten Ofstad

llvm-svn: 17826
2004-11-15 17:20:28 +00:00
Chris Lattner
f95f7e05a5 Minor cleanups. There is no reason for SCCP to derive from instvisitor anymore.
llvm-svn: 17825
2004-11-15 07:15:04 +00:00
Chris Lattner
4aa7dc02bf Count more accurately
llvm-svn: 17824
2004-11-15 07:02:42 +00:00
Brian Gaeke
56da74bd11 Update list of failing Benchmarks.
llvm-svn: 17823
2004-11-15 05:57:26 +00:00
Brian Gaeke
075be51e61 Expand Defs to encompass all the possibly-call-clobbered regs.
llvm-svn: 17822
2004-11-15 05:56:53 +00:00
Chris Lattner
20a9efa189 Quiet warnings on the persephone tester
llvm-svn: 17821
2004-11-15 05:54:07 +00:00
Chris Lattner
e87a1360b3 Two minor improvements:
1. Speedup getValueState by having it not consider Arguments.  It's better
    to just add them before we start SCCP'ing.
 2. SCCP can delete the contents of dead blocks.  No really, it's ok!  This
    reduces the size of the IR for subsequent passes, even though
    simplifycfg would do the same job.  In practice, simplifycfg does not
    run until much later than sccp in gccas

llvm-svn: 17820
2004-11-15 05:45:33 +00:00
Chris Lattner
4ad574191b rename InstValue to LatticeValue, as it holds for more than instructions.
llvm-svn: 17818
2004-11-15 05:03:30 +00:00
Reid Spencer
5fc587b13e Actually get the #include correct so it compiles .. duh.
llvm-svn: 17817
2004-11-15 04:47:22 +00:00
Chris Lattner
bde8da9e43 Substantially refactor the SCCP class into an SCCP pass and an SCCPSolver
class.  The only changes are minor:

 * Do not try to SCCP instructions that return void in the rewrite loop.
   This is silly and fool hardy, wasting a map lookup and adding an entry
   to the map which is never used.
 * If we decide something has an undefined value, rewrite it to undef,
   potentially leading to further simplications.

llvm-svn: 17816
2004-11-15 04:44:20 +00:00
Reid Spencer
8d0de0688c Make it actually compile on Solaris.
llvm-svn: 17815
2004-11-15 04:42:44 +00:00
Reid Spencer
2638dac5e9 Consolidate the implementation of TimeValue::now() for Unix to use the
seemingly ubiquitous gettimeofday(3) call.

llvm-svn: 17813
2004-11-15 04:36:35 +00:00
Reid Spencer
52da956231 * Make sure the string table gets read even if there isn't a foreign
symbol table.
* Make sure we update the file pointer for each member when rebuilding the
  symbol table.

llvm-svn: 17812
2004-11-15 01:40:20 +00:00
Reid Spencer
bfc924424d Changes necessary to enable linking of archives without LLVM symbol tables.
llvm-svn: 17811
2004-11-15 01:20:11 +00:00
Reid Spencer
29eb55846f Remove a forgotten debug output line.
llvm-svn: 17810
2004-11-14 23:49:55 +00:00
Reid Spencer
2012d091b2 Implement functionality suggested from code review: getStatusInfo should
returnn false if the file doesn't exist rather than throw ane exception.

llvm-svn: 17809
2004-11-14 23:30:38 +00:00
Reid Spencer
7850852268 Linker.h moved to include/llvm from include/llvm/Support.
llvm-svn: 17807
2004-11-14 23:27:04 +00:00
Reid Spencer
7de5984452 Make sure IdentifyFileType is in the sys namespace.
llvm-svn: 17806
2004-11-14 23:26:18 +00:00
Reid Spencer
7a004d2749 Linker.h has a new home.
llvm-svn: 17801
2004-11-14 23:00:08 +00:00
Reid Spencer
85698d2435 Moved to lib/Bytecode/Archive.
llvm-svn: 17800
2004-11-14 22:38:32 +00:00
Alkis Evlogimenos
a29bf44bde Add missing include.
llvm-svn: 17799
2004-11-14 22:37:42 +00:00
Reid Spencer
e51435be09 Moved to lib/Linker
llvm-svn: 17786
2004-11-14 22:15:31 +00:00
Reid Spencer
542b5216bf Forget strerror_r, it causes problems. Fix later when threading matters
llvm-svn: 17783
2004-11-14 22:10:54 +00:00
Reid Spencer
bd1d413e4e Implement the toString method
llvm-svn: 17782
2004-11-14 22:10:08 +00:00
Reid Spencer
3dfb002523 Update for prototype changes
llvm-svn: 17781
2004-11-14 22:09:22 +00:00
Reid Spencer
b73d75e4ff * Implement getDirectoryContents * Implement getStatusInfo * Implement setStatusInfo * Implement renameFile
llvm-svn: 17780
2004-11-14 22:08:36 +00:00
Reid Spencer
59872e60ef Fix bugs in class invariant
llvm-svn: 17779
2004-11-14 22:07:50 +00:00
Reid Spencer
a6b803539a Include the correct implementation file
llvm-svn: 17778
2004-11-14 22:07:04 +00:00
Reid Spencer
85b6b57c5e Don't exceed 80 columns.
llvm-svn: 17777
2004-11-14 22:06:18 +00:00
Reid Spencer
381c80ec88 Implement IdentifyFileType function
llvm-svn: 17776
2004-11-14 22:05:32 +00:00
Reid Spencer
f3923a1cf7 Implement the high level interface to make (de)compression easier.
llvm-svn: 17775
2004-11-14 22:04:46 +00:00
Reid Spencer
8bf3328ca7 Implement the MoreHelp utility that calls a function to printmore help information if the MoreHelp global is not null.
llvm-svn: 17774
2004-11-14 22:04:00 +00:00
Reid Spencer
292678447a Don't bother with a re-linked library, ensure archive library is built.
llvm-svn: 17773
2004-11-14 22:03:14 +00:00
Reid Spencer
8a2a29e286 *Adjust prototypes for public interface. *Rewrite LinkInArchive to use symbol tables.
llvm-svn: 17772
2004-11-14 22:02:27 +00:00
Reid Spencer
5c112bc910 Simplify compression code by using the high level interface to the Compressor
llvm-svn: 17771
2004-11-14 22:01:41 +00:00
Reid Spencer
cc354b445d Add wrappers to get defined symbols from bytecode
llvm-svn: 17770
2004-11-14 22:00:48 +00:00
Reid Spencer
b9baef5ca8 Simplify handling of decompression
llvm-svn: 17769
2004-11-14 22:00:09 +00:00
Reid Spencer
115f252d2a Simplify decompression code by using the high level interface to the Compressor
llvm-svn: 17768
2004-11-14 21:59:21 +00:00
Reid Spencer
2917639a35 Completely rewritten to allow reading of archives and symbol table lookup in a more efficient manner.
llvm-svn: 17767
2004-11-14 21:58:33 +00:00
Reid Spencer
0a3e3a41c3 Implementation declarations for Archive
llvm-svn: 17766
2004-11-14 21:57:46 +00:00
Reid Spencer
131f129398 First working version
llvm-svn: 17765
2004-11-14 21:56:59 +00:00
Reid Spencer
0a3ec41881 Add the Archive library
llvm-svn: 17764
2004-11-14 21:55:27 +00:00
Reid Spencer
08a949d512 Add the Linker library
llvm-svn: 17763
2004-11-14 21:54:41 +00:00
Misha Brukman
8d4e1d62dd GhostLinkage not allowed in LLVM AsmWriter, either
llvm-svn: 17751
2004-11-14 21:04:34 +00:00
Misha Brukman
8c1b4a5b9d GhostLinkage should not reach asm printing stage
llvm-svn: 17750
2004-11-14 21:03:49 +00:00
Misha Brukman
0d900050be Handle GhostLinkage (should not ever reach the assembly printing stage!)
llvm-svn: 17749
2004-11-14 21:03:30 +00:00
Misha Brukman
6da7a2e054 Mark an unmaterialized function as having GhostLinkage
llvm-svn: 17748
2004-11-14 21:02:55 +00:00
Chris Lattner
e4a97f4bee If a global is just loaded and restored, realize that it is not changing
value.  This allows us to turn more globals into constants and eliminate them.
This patch implements GlobalOpt/load-store-global.llx.

Note that this patch speeds up 255.vortex from:

Output/255.vortex.out-cbe.time:program 7.640000
Output/255.vortex.out-llc.time:program 9.810000

to:

Output/255.vortex.out-cbe.time:program 7.250000
Output/255.vortex.out-llc.time:program 9.490000

Which isn't bad at all!

llvm-svn: 17746
2004-11-14 20:50:30 +00:00
Misha Brukman
c98cd22aae Fix build on Linux/PowerPC64 using SuSE GCC (#undef PPC)
llvm-svn: 17744
2004-11-14 20:34:01 +00:00
Reid Spencer
9534caa3bb Moved to lib/Bytecode/Archive in preparation for re-write.
llvm-svn: 17742
2004-11-14 19:59:40 +00:00
Chris Lattner
3d61b688a9 This optimization makes MANY phi nodes that all have the same incoming value.
If this happens, detect it early instead of relying on instcombine to notice
it later.  This can be a big speedup, because PHI nodes can have many
incoming values.

llvm-svn: 17741
2004-11-14 19:29:34 +00:00
Chris Lattner
1e4cad9176 Implement instcombine/phi.ll:test6 - pulling operations through PHI nodes.
This exposes subsequent optimization possiblities and reduces code size.
This triggers 1423 times in spec.

llvm-svn: 17740
2004-11-14 19:13:23 +00:00
Chris Lattner
fdd41995d8 Transform this:
%X = alloca ...
  %Y = alloca ...
    X == Y

into false.  This allows us to simplify some stuff in eon (and probably
many other C++ programs) where operator= was checking for self assignment.
Folding this allows us to SROA several additional structs.

llvm-svn: 17735
2004-11-14 07:33:16 +00:00
Chris Lattner
d6d9a04344 Remove note to self
llvm-svn: 17734
2004-11-14 06:57:47 +00:00
Brian Gaeke
f9ec4c8ee2 Fix problem with insertion point for ADJCALLSTACKDOWN.
llvm-svn: 17733
2004-11-14 06:32:08 +00:00
Brian Gaeke
0e77dab4e8 Update lists of failing unit tests.
Exclude bigfib, so that we effectively exclude all C++ benchmarks.
Update to-do list: mention va_start.

llvm-svn: 17732
2004-11-14 06:32:07 +00:00
Chris Lattner
363e78c357 If a function always returns a constant, replace all calls sites with that
constant value.  This makes the return value dead and allows for
simplification in the caller.

This implements IPConstantProp/return-constant.ll

This triggers several dozen times throughout SPEC.

llvm-svn: 17730
2004-11-14 06:10:11 +00:00
Brian Gaeke
ab60855423 Fix NotTest - round up extraStack to the nearest doubleword, if it is
not zero.

llvm-svn: 17728
2004-11-14 05:19:00 +00:00
Chris Lattner
d0a0af0818 Teach SROA how to promote an array index that is variable, if the dimension
of the array is just two.  This occurs 8 times in gcc, 6 times in crafty, and
12 times in 099.go.

This implements ScalarRepl/sroa_two.ll

llvm-svn: 17727
2004-11-14 05:00:19 +00:00
Brian Gaeke
79f5f96a60 Update failing Benchmarks; point out that I'm skipping Shootout-C++.
llvm-svn: 17725
2004-11-14 04:43:12 +00:00
Chris Lattner
bc35272f86 Rearrange some code, no functionality changes.
llvm-svn: 17724
2004-11-14 04:24:28 +00:00
Brian Gaeke
505a9ee73d Update expected UnitTests failures.
llvm-svn: 17723
2004-11-14 03:22:08 +00:00
Brian Gaeke
1b3d701769 Rewrite outgoing arg handling to handle more weird corner cases.
llvm-svn: 17722
2004-11-14 03:22:07 +00:00
Brian Gaeke
294af88a7a Support UndefValue emission.
llvm-svn: 17721
2004-11-14 03:22:05 +00:00
Chris Lattner
9b2f0f93e3 Remove debugging code
llvm-svn: 17719
2004-11-13 23:32:53 +00:00
Chris Lattner
9749bf21f8 Argument promotion transforms functions to unconditionally load their
argument pointers.  This is only valid to do if the function already
unconditionally loaded an argument or if the pointer passed in is known
to be valid.  Make sure to do the required checks.

This fixed ArgumentPromotion/control-flow.ll and the Burg program.

llvm-svn: 17718
2004-11-13 23:31:34 +00:00
Chris Lattner
09b7f968e0 Don't print unneeded labels
llvm-svn: 17714
2004-11-13 23:27:11 +00:00
Chris Lattner
3684294c71 Hack around stupidity in GCC, fixing Burg with the CBE and
CBackend/2004-11-13-FunctionPointerCast.llx

llvm-svn: 17710
2004-11-13 22:21:56 +00:00
Chris Lattner
1cde11aa95 shld is a very high latency operation. Instead of emitting it for shifts of
two or three, open code the equivalent operation which is faster on athlon
and P4 (by a substantial margin).

For example, instead of compiling this:

long long X2(long long Y) { return Y << 2; }

to:

X3_2:
        movl 4(%esp), %eax
        movl 8(%esp), %edx
        shldl $2, %eax, %edx
        shll $2, %eax
        ret

Compile it to:

X2:
        movl 4(%esp), %eax
        movl 8(%esp), %ecx
        movl %eax, %edx
        shrl $30, %edx
        leal (%edx,%ecx,4), %edx
        shll $2, %eax
        ret

Likewise, for << 3, compile to:

X3:
        movl 4(%esp), %eax
        movl 8(%esp), %ecx
        movl %eax, %edx
        shrl $29, %edx
        leal (%edx,%ecx,8), %edx
        shll $3, %eax
        ret

This matches icc, except that icc open codes the shifts as adds on the P4.

llvm-svn: 17707
2004-11-13 20:48:57 +00:00
Chris Lattner
c531e090db Add missing check
llvm-svn: 17706
2004-11-13 20:04:38 +00:00
Chris Lattner
d1381380ae Compile:
long long X3_2(long long Y) { return Y+Y; }
int X(int Y) { return Y+Y; }

into:

X3_2:
        movl 4(%esp), %eax
        movl 8(%esp), %edx
        addl %eax, %eax
        adcl %edx, %edx
        ret
X:
        movl 4(%esp), %eax
        addl %eax, %eax
        ret

instead of:

X3_2:
        movl 4(%esp), %eax
        movl 8(%esp), %edx
        shldl $1, %eax, %edx
        shll $1, %eax
        ret

X:
        movl 4(%esp), %eax
        shll $1, %eax
        ret

llvm-svn: 17705
2004-11-13 20:03:48 +00:00
Chris Lattner
70e351fb1c Simplify handling of shifts to be the same as we do for adds. Add support
for (X * C1) + (X * C2) (where * can be mul or shl), allowing us to fold:

   Y+Y+Y+Y+Y+Y+Y+Y

into
         %tmp.8 = shl long %Y, ubyte 3           ; <long> [#uses=1]

instead of

        %tmp.4 = shl long %Y, ubyte 2           ; <long> [#uses=1]
        %tmp.12 = shl long %Y, ubyte 2          ; <long> [#uses=1]
        %tmp.8 = add long %tmp.4, %tmp.12               ; <long> [#uses=1]

This implements add.ll:test25

Also add support for (X*C1)-(X*C2) -> X*(C1-C2), implementing sub.ll:test18

llvm-svn: 17704
2004-11-13 19:50:12 +00:00
Chris Lattner
7a8d26a581 Fold:
(X + (X << C2)) --> X * ((1 << C2) + 1)
   ((X << C2) + X) --> X * ((1 << C2) + 1)

This means that we now canonicalize "Y+Y+Y" into:

        %tmp.2 = mul long %Y, 3         ; <long> [#uses=1]

instead of:

        %tmp.10 = shl long %Y, ubyte 1          ; <long> [#uses=1]
        %tmp.6 = add long %Y, %tmp.10               ; <long> [#uses=1]

llvm-svn: 17701
2004-11-13 19:31:40 +00:00
Chris Lattner
d348f5b9fb Lazily create the abort message, so only translation units that use unwind
will actually get it.

llvm-svn: 17700
2004-11-13 19:07:32 +00:00
Chris Lattner
ab917141d0 Fix: CodeExtractor/2004-11-12-InvokeExtract.ll
llvm-svn: 17699
2004-11-13 00:06:45 +00:00
Chris Lattner
c6b1d7a081 Fix a bug where the code extractor would get a bit confused handling invoke
instructions, setting DefBlock to a block it did not have dom info for.

llvm-svn: 17697
2004-11-12 23:50:44 +00:00
Chris Lattner
922a1b4601 Simplify handling of constant initializers
llvm-svn: 17696
2004-11-12 22:42:57 +00:00
Reid Spencer
f6caf3b1c2 Makefile for lib/Linker
llvm-svn: 17695
2004-11-12 20:38:45 +00:00
Reid Spencer
f755c169b3 This file originated in lib/VMCore/Linker.cpp but now lives in
lib/Linker/LinkModules.cpp

llvm-svn: 17694
2004-11-12 20:37:43 +00:00
Reid Spencer
13a6af95ad This file originated in tools/gccld/Linker.cpp but now lives in
lib/Linker/LinkArchives.cpp

llvm-svn: 17693
2004-11-12 20:34:32 +00:00
Chris Lattner
1a469385bd Actually, leave the check in. This prevents us from counting dead arguments
as IPCP opportunities.

llvm-svn: 17680
2004-11-11 07:47:54 +00:00
Chris Lattner
dba9c2b0ef Fix bug: IPConstantProp/deadarg.ll
llvm-svn: 17679
2004-11-11 07:46:29 +00:00
Chris Lattner
d920b5b770 Make IP Constant prop more aggressive about handling self recursive calls.
This implements IPConstantProp/recursion.ll

llvm-svn: 17666
2004-11-10 19:43:59 +00:00
John Criswell
402e338f11 Correct the name of stosd for the AT&T syntax:
It's stosl (l for long == 32 bit).

llvm-svn: 17658
2004-11-10 04:48:15 +00:00
Chris Lattner
b214a04a16 Do not let dead constant expressions hanging off of functions prevent IPCP.
This allows to elimination of a bunch of global pool descriptor args from
programs being pool allocated (and is also generally useful!)

llvm-svn: 17657
2004-11-09 20:47:30 +00:00
Reid Spencer
d17e35893c Fix isBytecodeFile to correctly recognized compressed bytecode too.
llvm-svn: 17655
2004-11-09 20:27:23 +00:00
Reid Spencer
28ceb24005 * Implement getStatusInfo for getting stat(2) like information
* Implement createTemporaryFile for mkstemp(3) functionality
* Fix isBytecodeFile to accept llvc magic # (compressed) as bytecode.

llvm-svn: 17654
2004-11-09 20:26:31 +00:00
John Criswell
5be81b837b Recognize compressed LLVM bytecode files.
This should fix the problem of not being able to link compressed LLVM
bytecode files from LLVM libraries.

llvm-svn: 17648
2004-11-09 19:37:07 +00:00
Reid Spencer
d9387fee02 Tune compression:
bzip2: block size 9 -> 5, reduces memory by 400Kbytes, doesn't affect speed
       or compression ratio on all but the largest bytecode files (>1MB)
zip:   level 9 -> 6, this speeds up compression time by ~30% but only
       degrades the compressed size by a few bytes per megabyte. Those few
       bytes aren't worth the effort.

llvm-svn: 17647
2004-11-09 17:58:09 +00:00
Chris Lattner
9f035c9fdb Change this back so that I get stable numbers to reflect the change from the
nightly testers

llvm-svn: 17646
2004-11-09 08:05:23 +00:00
Chris Lattner
b924e8be62 Fix bug: 2004-11-08-FreeUseCrash.ll
llvm-svn: 17642
2004-11-09 05:10:56 +00:00
Misha Brukman
5718caa3ab * Convert tabs to spaces
* Order #includes according to style guide
* Remove extraneous blank lines

llvm-svn: 17639
2004-11-09 04:27:19 +00:00
Nate Begeman
398bd2b9f6 Allow hbd to be bugpointable on darwin by fixing common and linkonce codegen
llvm-svn: 17637
2004-11-09 04:01:18 +00:00
Misha Brukman
f28a01454d Handle headers for compressed bytecode files
llvm-svn: 17634
2004-11-08 22:03:32 +00:00
Chris Lattner
0efef015fa Fix a bug that was preventing povray and namd from pool allocating correctly.
llvm-svn: 17632
2004-11-08 21:08:46 +00:00
Chris Lattner
dcee21898f Handle assert_fail special
llvm-svn: 17631
2004-11-08 21:08:28 +00:00
Reid Spencer
af24480f3e Fix a typo: isvalid -> isValid
llvm-svn: 17621
2004-11-08 08:27:51 +00:00
Nate Begeman
a0c15f3ffd Put int the getReg cast optimization from x86 so that we generate fewer
move instructions for the register allocator to coalesce.

llvm-svn: 17608
2004-11-08 02:25:40 +00:00
Misha Brukman
8c397821a5 Implement ExecutionEngine::freeMachineCodeForFunction()
llvm-svn: 17601
2004-11-07 23:58:46 +00:00
Chris Lattner
a9e56ddaa3 VERY large functions that are only called from one place are not really
exciting to inline.  Only inline medium or small sized functions with a
single call site.

llvm-svn: 17588
2004-11-07 21:46:47 +00:00
Nate Begeman
a7541b19fc Disable bogus cast elimination when the cast is used by a setcc instruction.
llvm-svn: 17583
2004-11-07 20:23:42 +00:00
Reid Spencer
c9c7c7fbee * Comments & cleanup per CL code review.
* Accept 'llvc' as signature for compressed bytecode

llvm-svn: 17579
2004-11-07 18:20:55 +00:00
Reid Spencer
152f895e9b Add comments per CL code review.
llvm-svn: 17578
2004-11-07 18:19:00 +00:00
Reid Spencer
1446cd065f * Add comments and cleanup per CL code review
* Make signature for compressed bytecode llvc instead of unreadable
* Make the CompressionContext have a constructor and destructor.

llvm-svn: 17576
2004-11-07 18:17:38 +00:00
Misha Brukman
ce8c2c2aee Replace uses of llvm.org with llvm.cs.uiuc.edu
llvm-svn: 17549
2004-11-07 00:58:38 +00:00
Misha Brukman
820aa44589 Consistently use llvm.cs.uiuc.edu instead of llvm.org
llvm-svn: 17548
2004-11-07 00:58:20 +00:00
Chris Lattner
f5c5cb033e Decompose* is V9 specific, make it internal
llvm-svn: 17547
2004-11-07 00:43:24 +00:00
Chris Lattner
a060b55dea This is V9 specific, move it there.
llvm-svn: 17545
2004-11-07 00:39:26 +00:00
Chris Lattner
f11d034489 Move this file from lib/Transforms/Scalar
llvm-svn: 17544
2004-11-07 00:39:09 +00:00
Reid Spencer
a81994464f Add support for compressed bytecode
llvm-svn: 17535
2004-11-06 23:17:23 +00:00
Reid Spencer
bf14271ed6 Implement the GetBytecodeSymbols interface function to extract just the
externally visible defined symbols from a bytecode file.

llvm-svn: 17503
2004-11-06 08:56:40 +00:00
Reid Spencer
c3a4c8a346 Makefile for the LLVMArchive library.
llvm-svn: 17500
2004-11-06 08:52:36 +00:00
Reid Spencer
8907e56650 First kinda/sorta working version of the Archive library. Reading is not
yet supported but writing works. Way too early to review this. More to come

llvm-svn: 17499
2004-11-06 08:51:45 +00:00
Reid Spencer
7090e6a875 Stop propagating method names that violate the coding standard
llvm-svn: 17498
2004-11-05 22:15:36 +00:00
John Criswell
97da76178c Fix compilation problem; make the cast and the LHS be the same type.
llvm-svn: 17488
2004-11-05 16:17:06 +00:00
Chris Lattner
9f23af6d68 Remove dead var
llvm-svn: 17485
2004-11-05 04:52:08 +00:00
Chris Lattner
499e1b16a7 Quiet VC++ warnings
llvm-svn: 17484
2004-11-05 04:50:59 +00:00
Chris Lattner
bde92f3c03 Do not make i have bigger scope that we need
llvm-svn: 17483
2004-11-05 04:47:37 +00:00
Chris Lattner
b5ff07e46e Remove dead vars
llvm-svn: 17482
2004-11-05 04:46:22 +00:00
Chris Lattner
064071600d Fix some warnings on VC++
llvm-svn: 17481
2004-11-05 04:45:43 +00:00
Nate Begeman
bc8bc24d28 Thanks to sabre for pointing out that we were incorrectly codegen'ing
int test(int x) { return 32768 - x; }

Fixed by teaching the function that checks a constant's validity to be used
as an immediate argument about subtract-from instructions.

llvm-svn: 17476
2004-11-04 19:43:18 +00:00
Brian Gaeke
85e7df8097 Handle "call" operands of type long/ulong passed in registers.
llvm-svn: 17464
2004-11-04 00:27:04 +00:00
Chris Lattner
3b6d1045d3 Don't call Constant::getNullValue when the argument could be VoidTy
llvm-svn: 17457
2004-11-03 18:51:26 +00:00
Chris Lattner
f2976f16ce Fix this function to not say that longs have 8-byte alignment on X86/PPC.
This method is really a gross hack, but at least we can make it work on
the targets we support right now.

This bug fix stops a crash in a testcase reduced from 176.gcc

llvm-svn: 17443
2004-11-02 22:18:18 +00:00
Tanya Lattner
bd82bbff7c Added gross hacks such as creating my own def-use map, and picking on Instruction that I can add all my TmpInstructions to its MCFI.
llvm-svn: 17441
2004-11-02 21:04:56 +00:00
Chris Lattner
d9696aa7b8 Fix a warning
llvm-svn: 17431
2004-11-02 15:27:57 +00:00
Chris Lattner
33ed2c8984 * Rearrange code slightly
* Disable broken transforms for simplifying (setcc (cast X to larger), CI)
  where CC is not != or ==

llvm-svn: 17422
2004-11-02 03:50:32 +00:00
Chris Lattner
10de12fd46 Add placeholder variable to make Win32 work, applied for Morten Ofstad
llvm-svn: 17406
2004-11-01 20:10:20 +00:00
Chris Lattner
b81520727e Speed up the tail duplication pass on the testcase below from 68.2s to 1.23s:
#define CL0(a) case a: f(); goto c;
 #define CL1(a) CL0(a##0) CL0(a##1) CL0(a##2) CL0(a##3) CL0(a##4) CL0(a##5) \
 CL0(a##6) CL0(a##7) CL0(a##8) CL0(a##9)
 #define CL2(a) CL1(a##0) CL1(a##1) CL1(a##2) CL1(a##3) CL1(a##4) CL1(a##5) \
 CL1(a##6) CL1(a##7) CL1(a##8) CL1(a##9)
 #define CL3(a) CL2(a##0) CL2(a##1) CL2(a##2) CL2(a##3) CL2(a##4) CL2(a##5) \
 CL2(a##6) CL2(a##7) CL2(a##8) CL2(a##9)
 #define CL4(a) CL3(a##0) CL3(a##1) CL3(a##2) CL3(a##3) CL3(a##4) CL3(a##5) \
 CL3(a##6) CL3(a##7) CL3(a##8) CL3(a##9)

 void f();

 void a() {
     int b;
  c: switch (b) {
         CL4(1)
     }
 }

This comes from GCC PR 15524

llvm-svn: 17390
2004-11-01 07:05:07 +00:00
Chris Lattner
3083f2a424 Do not compute the predecessor list for a block unless we need it.
This speeds up simplifycfg on this program, from 44.87s to 0.29s (with
a profiled build):

 #define CL0(a) case a: goto c;
 #define CL1(a) CL0(a##0) CL0(a##1) CL0(a##2) CL0(a##3) CL0(a##4) CL0(a##5) \
 CL0(a##6) CL0(a##7) CL0(a##8) CL0(a##9)
 #define CL2(a) CL1(a##0) CL1(a##1) CL1(a##2) CL1(a##3) CL1(a##4) CL1(a##5) \
 CL1(a##6) CL1(a##7) CL1(a##8) CL1(a##9)
 #define CL3(a) CL2(a##0) CL2(a##1) CL2(a##2) CL2(a##3) CL2(a##4) CL2(a##5) \
 CL2(a##6) CL2(a##7) CL2(a##8) CL2(a##9)
 #define CL4(a) CL3(a##0) CL3(a##1) CL3(a##2) CL3(a##3) CL3(a##4) CL3(a##5) \
 CL3(a##6) CL3(a##7) CL3(a##8) CL3(a##9)

 void f();

 void a() {
     int b;
  c: switch (b) {
         CL4(1)
     }
 }

This testcase is contrived to expose N^2 behavior, but this patch should speedup
simplifycfg on any programs that use large switch statements.  This testcase
comes from GCC PR17895.

llvm-svn: 17389
2004-11-01 06:53:58 +00:00
Chris Lattner
8572d582cf Fix comment
llvm-svn: 17377
2004-10-31 21:54:51 +00:00
Chris Lattner
d72b6b1e77 Improve comment
llvm-svn: 17375
2004-10-31 19:57:43 +00:00
Chris Lattner
14d85c9b09 Add more paranoid assertions :)
llvm-svn: 17367
2004-10-31 17:45:40 +00:00
Reid Spencer
acb1305e99 Internalize variable names to prevent recursive assignment. Cleanup docs.
llvm-svn: 17359
2004-10-30 09:19:36 +00:00
Chris Lattner
29b6259de3 Fix some more problems where we called getOffset before getNode()
llvm-svn: 17358
2004-10-30 07:21:19 +00:00
Chris Lattner
4b741eab69 Fix three bugs:
1. Calls to external global VARIABLES should not be treated as a call to an
    external function
 2. Efficiently deleting an element from a vector by using std::swap with
    the back, then pop_back is NOT a good way to keep the vector sorted.
 3. Our hope of having stuff get deleted by making them redundant just won't
    work.  In particular, if we have three calls in sequence that should be
    merged: A, B, C   first we unify B into A.  To be sure that they appeared
    identical (so B would be erased) we set B = A.  On the next step, we
    unified C into A and set C = A.  Unfortunately, this is no guarantee that
    C = B, so we would fail to delete the dead call.  Switch to a more
    explicit scheme.

llvm-svn: 17357
2004-10-30 05:41:23 +00:00
Chris Lattner
14dfc01ad5 Fix more undefined behavior
llvm-svn: 17356
2004-10-30 04:22:45 +00:00
Chris Lattner
c17cd85776 * Add a method
* change some uses of NH.getNode() in a bool context to use !NH.isNull()
* Fix a bunch of places where we depended on the (undefined) order of
  evaluation of arguments to function calls to ensure that getNode() was
  called before getOffset().  In practice, this was NOT happening.

llvm-svn: 17354
2004-10-30 04:05:01 +00:00
Tanya Lattner
8b938744dc Fixed bug with infinite epilogues.
Fixed issue with generating the partial order. It now adds the nodes not in recurrences in sets for each connected component.

llvm-svn: 17351
2004-10-30 00:39:07 +00:00
Misha Brukman
4e78536fa3 * Fix compilation on AIX: GCC's fixincludes eliminates isinf() declaration
* Move file comment to the top of the header where it belongs

llvm-svn: 17349
2004-10-29 23:17:45 +00:00
Brian Gaeke
c66a1d540b Change name of target lib to conform to new naming scheme.
llvm-svn: 17347
2004-10-29 21:57:16 +00:00
Brian Gaeke
db07ac46f0 Remove dependency on MRegisterInfo::getRegClass
llvm-svn: 17346
2004-10-29 21:42:27 +00:00
Misha Brukman
dba13ac3cb The Alpha (tm) intrinsics have never been used anywhere
llvm-svn: 17340
2004-10-29 18:43:17 +00:00
Brian Gaeke
a757bae04f When emitting debug msgs for function stubs, don't truncate the
printed pointer value if sizeof(unsigned) != pointer size.  Instead,
use uintptr_t.

llvm-svn: 17338
2004-10-29 18:22:45 +00:00
Alkis Evlogimenos
e88c2aaed7 Gep indices must be of int, uint, long or ulong type.
llvm-svn: 17313
2004-10-28 06:43:38 +00:00
Reid Spencer
1be896397a Fix library name.
llvm-svn: 17307
2004-10-28 05:37:24 +00:00
Alkis Evlogimenos
1162280a0d Fix library name.
llvm-svn: 17306
2004-10-28 05:36:48 +00:00
Reid Spencer
36b5b2c07e Fix library name.
llvm-svn: 17305
2004-10-28 05:32:01 +00:00
Reid Spencer
ccdfdd7346 Fix name of library
llvm-svn: 17304
2004-10-28 05:30:54 +00:00
Reid Spencer
a909f95ebd Make sure that the yacc and lex output are specified as BUILT_SOURCES.
Correct the dependency of the Lexer.o file on the constructed
llvmAsmParser.h header file. It is not the Lexer.cpp file that depends on
the header, its the output of compiling Lexer.cpp, Lexer.o

llvm-svn: 17289
2004-10-28 00:43:24 +00:00
Reid Spencer
d3f7233495 Change Library Names Not To Conflict With Others When Installed
llvm-svn: 17286
2004-10-27 23:18:45 +00:00
Reid Spencer
1a344dfd95 Changes to support rand48 tests
llvm-svn: 17284
2004-10-27 23:03:44 +00:00
Chris Lattner
dd0094e4ed Convert 'struct' to 'class' in various places to adhere to the coding standards
and work better with VC++.  Patch contributed by Morten Ofstad!

llvm-svn: 17281
2004-10-27 16:14:51 +00:00
Nate Begeman
2a64d462d9 Move destructor out of line to avoid vtable emission in every file that includes the header. Thanks to sabre.
llvm-svn: 17278
2004-10-27 06:00:53 +00:00
Chris Lattner
000424b69e Hrm, this code was severely botched. As it turns out, this patch:
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20041018/019708.html

exposed ANOTHER latent bug in this xform, which caused Prolangs-C/bison to fill
the zion nightly tester disk up and make the tester barf.

This is obviously not a good thing, so lets fix this bug shall we? :)

llvm-svn: 17276
2004-10-27 05:57:15 +00:00
Nate Begeman
a6fafb1f64 Fix the build by eliminating some more dead code. That'll learn me not to listen to Reid
llvm-svn: 17275
2004-10-27 05:44:23 +00:00
Chris Lattner
abfae1c72e Initialize with the correct constant type
llvm-svn: 17270
2004-10-27 03:55:24 +00:00
Chris Lattner
d8ac6d4a25 Plug a memory leak in the asmparser. It turns out that we were leaking
the strings for basic block labels in some cases.  This amounted to about
120K of memory for namd, a medium sized program.

llvm-svn: 17262
2004-10-26 18:26:14 +00:00
Chris Lattner
3a79669263 add support for UndefValue
llvm-svn: 17260
2004-10-26 16:23:03 +00:00
Chris Lattner
2c73917686 Move method bodies that depend on <algorithm> from MBB.h to MBB.cpp
llvm-svn: 17253
2004-10-26 15:43:42 +00:00
Chris Lattner
a361504a90 Clean up the MachineBasicBlock.h file, percolating #includes into this file.
Patch contributed by Morten Ofstad

llvm-svn: 17251
2004-10-26 15:35:58 +00:00
Nate Begeman
64ddf151fa Remove include of MRegisterInfo.h, since it is already included by
SkeletonGenRegisterInfo.h.inc

llvm-svn: 17245
2004-10-26 06:04:23 +00:00
Nate Begeman
46c3a8875f Remove file that is no longer used, and move include of MRegisterInfo.h
from PowerPCFrameInfo.h to PowerPCAsmPrinter.cpp where it is actually
needed.

llvm-svn: 17244
2004-10-26 06:02:38 +00:00
Nate Begeman
7c3c97af10 Eliminate usage of MRegisterInfo::getRegClass(physreg)
llvm-svn: 17240
2004-10-26 05:40:45 +00:00
Chris Lattner
0dbd792854 Fix the interpreter crash that Michael McCracken found
llvm-svn: 17239
2004-10-26 05:35:14 +00:00
Chris Lattner
6e775d56cf Reduce usage of MRegisterInfo::getRegClass
llvm-svn: 17238
2004-10-26 05:29:18 +00:00
Nate Begeman
ae98298003 Update to-do list
llvm-svn: 17235
2004-10-26 04:10:53 +00:00
Nate Begeman
113f516f6b Fix treecc. Also fix a latent bug in emitBinaryConstOperation that would
allow and const, 0 to be incorrectly codegen'd into a rlwinm instruction.

llvm-svn: 17234
2004-10-26 03:48:25 +00:00
Reid Spencer
2f74ad1e36 Add EXTRA_DIST for additional files to be distributed.
llvm-svn: 17233
2004-10-26 03:12:11 +00:00
Chris Lattner
9c356da87d Disable the JIT until it can sorta kinda work.
llvm-svn: 17230
2004-10-25 20:53:41 +00:00
Chris Lattner
c178fc879a Remove dead assert
llvm-svn: 17221
2004-10-25 19:04:01 +00:00
Chris Lattner
e8d5408a45 Patch to support MSVC, contributed by Morten Ofstad
llvm-svn: 17220
2004-10-25 18:47:10 +00:00
Chris Lattner
e5ccecc27e Patch to support MSVC, contributed by Morten Ofstad
llvm-svn: 17219
2004-10-25 18:46:05 +00:00
Chris Lattner
895f16ebfc Fix compatibility with MSVC, patch by Morten Ofstad
llvm-svn: 17218
2004-10-25 18:45:16 +00:00
Chris Lattner
84b07af401 Do not use variable sized arrays in C++, they are non-portable. Patch
contributed by Morten Ofstad

llvm-svn: 17217
2004-10-25 18:44:14 +00:00
John Criswell
3c14d0815c Removed dead method, printPHICopiesForSuccessors().
llvm-svn: 17216
2004-10-25 18:41:50 +00:00
Chris Lattner
ec942219ad Patch to support MSVC better, contributed by Morten Ofstad
llvm-svn: 17215
2004-10-25 18:40:47 +00:00
Chris Lattner
1cef20bac1 Patch to support MSVC, contributed by Morten Ofstad
llvm-svn: 17214
2004-10-25 18:40:08 +00:00
John Criswell
259f362897 Modified switch generation so that only the phi values associated with the
destination basic block are copied.

llvm-svn: 17212
2004-10-25 18:30:09 +00:00
Nate Begeman
4b5ed899fd Implement more complete and correct codegen for bitfield inserts, as tested
by the recently committed rlwimi.ll test file.  Also commit initial code
for bitfield extract, although it is turned off until fully debugged.

llvm-svn: 17207
2004-10-24 10:33:30 +00:00
Chris Lattner
75772beb3b Fix the previous bug the correct way. This fixes ptrdist/bc
llvm-svn: 17201
2004-10-24 04:27:59 +00:00
Alkis Evlogimenos
4c30220f3a Make this actually work.
llvm-svn: 17199
2004-10-24 03:02:16 +00:00
Alkis Evlogimenos
a893aad8fb Add ConstantExpr::getSizeOf(Type*).
llvm-svn: 17196
2004-10-24 01:41:10 +00:00
Misha Brukman
f8bd6fc901 * Correctly handle the MovePCtoLR pseudo-instr with a bl to next instr
* Stop the confusion of using rv and Addr for global addresses: just use rv

llvm-svn: 17195
2004-10-23 23:47:34 +00:00
Misha Brukman
58d9e43fa1 Add BA, BL, and BLA opcodes
llvm-svn: 17193
2004-10-23 20:29:24 +00:00
Misha Brukman
0342392149 * Do not emit IMPLICIT_DEF pseudo-instructions
* Convert register numbers from their opcode value to the real value, e.g.
  PPC::R1 => 1 and PPC::F1 => 1
* Add correct handling of loading of global values which are PC-relative --
  implement ha16() and lo16()

llvm-svn: 17190
2004-10-23 18:28:01 +00:00
Misha Brukman
bd6a01c3ef DForm_1, particularly used by store instructions, needs the immediate operand to
be listed second as that is how the instructions are usually created (and is the
correct asm syntax) so that it's assembled correctly from its constituents

llvm-svn: 17183
2004-10-23 06:08:38 +00:00
Misha Brukman
e4ae05e099 Fix the SPR field for MTLR, MFLR, MTCTR, and MFCTR instructions.
The decimal value given in the manual (8 or 9) really needs to be multiplied by
a factor of 32 because of the group of 5 zero bits after the register code.

llvm-svn: 17182
2004-10-23 06:05:49 +00:00
Misha Brukman
5cea06807d The value of the XO field for MFLR and MFCTR is 339, not 399
llvm-svn: 17181
2004-10-23 05:38:55 +00:00
Misha Brukman
39d23c81d4 Remove extraneous blank line
llvm-svn: 17180
2004-10-23 04:59:22 +00:00
Misha Brukman
c660c6c514 Align function arguments in function headers
llvm-svn: 17178
2004-10-23 04:58:32 +00:00
Chris Lattner
9595df9f4e Fix a bug that Brian brought to my attention. This corrects:
Assembler/2004-10-22-BCWriterUndefBug.llx

llvm-svn: 17177
2004-10-23 03:10:23 +00:00
Nate Begeman
91ef127999 Kill casts from integer types to unsigned byte, when the cast was only used
as the shift amount operand to a shift instruction.  This was causing us to
emit unnecessary clear operations for code such as:
int foo(int x) { return 1 << x; }

llvm-svn: 17175
2004-10-23 00:50:23 +00:00
Misha Brukman
cfd8a7eb53 Use cleaner quoting and eliminate blank space
llvm-svn: 17174
2004-10-22 23:35:57 +00:00
Reid Spencer
4e31ccbe2e Clean up the output from this makefile so its not verbose.
llvm-svn: 17173
2004-10-22 23:24:39 +00:00
Misha Brukman
2493cb2a2a Adjust rules for building .inc files due to Reid's changes of Makefile.rules
llvm-svn: 17169
2004-10-22 22:16:24 +00:00
Reid Spencer
019621a1ea Adjust to changes in Makefile.rules
llvm-svn: 17167
2004-10-22 21:02:08 +00:00
Reid Spencer
9f6f128630 Eliminate compilation warning on uninitialized variable.
llvm-svn: 17163
2004-10-22 16:10:39 +00:00
Chris Lattner
f0e2ce2279 *** empty log message ***
llvm-svn: 17161
2004-10-22 06:43:28 +00:00
Chris Lattner
6321852143 Fix a bug Nate noticed, where we miscompiled a simple testcase
llvm-svn: 17157
2004-10-22 04:53:16 +00:00
Reid Spencer
e48ba34fd4 We won't use automake
llvm-svn: 17155
2004-10-22 03:35:04 +00:00
Misha Brukman
66a092f3df Remove debug code emitter from the JIT
llvm-svn: 17151
2004-10-21 03:07:38 +00:00
Alkis Evlogimenos
67449982b8 Make this compile.
llvm-svn: 17150
2004-10-21 02:44:16 +00:00
Misha Brukman
9514a75e36 * Added basic support for JITing functions, basic blocks, instruction encoding,
including registers, constants, and partial support for global addresses
* The JIT is disabled by default to allow building llvm-gcc, which wants to test
  running programs during configure

llvm-svn: 17149
2004-10-21 01:42:02 +00:00
Nate Begeman
d7cbf1d28e Don't clear or sign extend bool->int. This fires a few dozen times on the test suite
llvm-svn: 17147
2004-10-20 21:55:41 +00:00
Brian Gaeke
71ff3efdd5 Explain what this pass does.
llvm-svn: 17146
2004-10-20 19:38:58 +00:00
John Criswell
0ba9295e3c Small performance improvement in generated C code:
Instead of unconditionally copying all phi node values into temporaries for
all successor blocks, generate code that will determine what successor
block will be called and then copy only those phi node values needed by
the successor block.

This seems to cut down namd execution time from being 8% higher than GCC to
4% higher than GCC.

llvm-svn: 17144
2004-10-20 14:38:39 +00:00
Misha Brukman
9113553178 * Add baseline structural JIT code, but disable the JIT to allow llvm-gcc builds
- Support added for functions, basic blocks, constant pool, constants,
    registers, and some basic support for globals, all untested
* Turn assert()s into abort()s so that unimplemented functions fail in release

llvm-svn: 17143
2004-10-19 19:49:42 +00:00
Chris Lattner
e555b5c5cb Hrm, some people complain when the compiler cheerfully tells them what it's
doing... I guess they're right.

llvm-svn: 17142
2004-10-19 06:33:16 +00:00
Misha Brukman
4cd5cde095 Fix some grammar
llvm-svn: 17141
2004-10-19 05:55:54 +00:00
Brian Gaeke
62ecd75aeb Simplify mapping info generation. In particular, the LLVM-to-MachineInstr map
is no longer emitted, and we do not reference any MachineCodeForInstruction
information.

llvm-svn: 17138
2004-10-19 05:15:21 +00:00
Reid Spencer
ce514b1c2c Initial automake generated Makefile template
llvm-svn: 17136
2004-10-18 23:55:41 +00:00
Nate Begeman
49e38c2185 Initial implementation of the strength reduction for GEP instructions in
loops.  This optimization is not turned on by default yet, but may be run
with the opt tool's -loop-reduce flag.  There are many FIXMEs listed in the
code that will make it far more applicable to a wide range of code, but you
have to start somewhere :)

This limited version currently triggers on the following tests in the
MultiSource directory:
pcompress2: 7 times
cfrac: 5 times
anagram: 2 times
ks: 6 times
yacr2: 2 times

llvm-svn: 17134
2004-10-18 21:08:22 +00:00
Misha Brukman
ed985abc65 AIX does not have mkdtemp() so emulate its behavior using mktemp() and mkdir()
llvm-svn: 17131
2004-10-18 17:39:45 +00:00
Chris Lattner
cac643c78f Improve compatibility with VC++, patch contributed by Morten Ofstad!
llvm-svn: 17126
2004-10-18 15:54:17 +00:00
Chris Lattner
9e2649a7be Get this file compiling with VC++, patch contributed by Morten Ofstad. Thanks Morten!
llvm-svn: 17125
2004-10-18 15:43:46 +00:00
Reid Spencer
2b5cfe8be8 Correction to allow compilation with Visual C++.
Patch contributed by Morten Ofstad. Thanks Morten!

llvm-svn: 17123
2004-10-18 14:38:48 +00:00
Chris Lattner
b1827a765a Simplify code by deleting instructions that preceed unreachable instructions.
Simplify code by simplifying terminators that branch to blocks that start
with an unreachable instruction.

llvm-svn: 17116
2004-10-18 04:07:22 +00:00
Chris Lattner
3c1be63b41 Turn store -> null/undef into the LLVM unreachable instruction! This simple
change hacks off 10K of bytecode from perlbmk (.5%) even though the front-end
is not generating them yet and we are not optimizing the resultant code.
This isn't too bad.

llvm-svn: 17111
2004-10-18 03:00:50 +00:00
Chris Lattner
18333c64de Turn things with obviously undefined semantics into 'store -> null'
llvm-svn: 17110
2004-10-18 02:59:09 +00:00
Chris Lattner
dbfa9a5e9c My friend the invoke instruction does not dominate all basic blocks if it
occurs in the entry node of a function

llvm-svn: 17109
2004-10-18 01:48:31 +00:00
Chris Lattner
f7bf129037 Fix a bug that occurs when the constant value is the result of an invoke. In
particular, invoke ret values are only live in the normal dest of the invoke
not in the unwind dest.

llvm-svn: 17108
2004-10-18 01:21:17 +00:00
Chris Lattner
5ae94ef614 Print a semicolon for the unreacahble instruction. This fixes problems
where C requires semicolons in some cases to indicate null statements.

llvm-svn: 17107
2004-10-17 23:49:11 +00:00
Chris Lattner
44a7cf480d Getting ADCE to interact well with unreachable instructions seems like a nontrivial
exercise that I'm not interested in tackling right now.  Just punt and treat them
like unwind's.

This 'fixes' test/Regression/Transforms/ADCE/unreachable-function.ll

llvm-svn: 17106
2004-10-17 23:45:06 +00:00
Chris Lattner
9e7ddf8596 Fix Regression/Transforms/Inline/2004-10-17-InlineFunctionWithoutReturn.ll
If a function had no return instruction in it, and the result of the inlined
call instruction was used, we would crash.

llvm-svn: 17104
2004-10-17 23:21:07 +00:00
Nate Begeman
927c4cc3dc Pass -single_module option to gcc when linking dynamic libraries for use with bugpoint, so that we can bugpoint multiple .cp files
llvm-svn: 17102
2004-10-17 23:03:32 +00:00
Nate Begeman
1c408f7319 Generate correct stubs for weak-linked symbols
llvm-svn: 17101
2004-10-17 23:01:34 +00:00
Chris Lattner
c088680a5f fold gep undef, ... -> undef
This comes up many times in perlbmk and probably others.

llvm-svn: 17100
2004-10-17 21:54:55 +00:00
Chris Lattner
795ae57cde Remove printout, realize that instructions in the entry block dominate all
other blocks.

llvm-svn: 17099
2004-10-17 21:31:34 +00:00
Chris Lattner
ac92e5fcd0 When inserting PHI nodes, don't insert any phi nodes that are obviously
unneccesary.  This allows us to delete several hundred phi nodes of the
form PHI(x,x,x,undef) from 253.perlbmk and probably other programs as well.

This implements Mem2Reg/UndefValuesMerge.ll

llvm-svn: 17098
2004-10-17 21:25:56 +00:00
Chris Lattner
306dd8a44a Enhance hasConstantValue to ignore undef values in phi nodes. This allows it
to think that PHI[4, undef] == 4.

llvm-svn: 17096
2004-10-17 21:23:26 +00:00
Chris Lattner
ef0888e493 hasConstantValue will soon return instructions that don't dominate the PHI node,
so prepare for this.

llvm-svn: 17095
2004-10-17 21:22:38 +00:00
Chris Lattner
caf0d76a8a The first hunk corrects a bug when printing undef null values. We would print
0->field, which is illegal.  Now we print ((foo*)0)->field.

The second hunk is an optimization to not print undefined phi values.

llvm-svn: 17094
2004-10-17 17:48:59 +00:00
Chris Lattner
f96fb0c946 Don't print stuff out from the code generator. This broke the JIT horribly
last night. :)  bork!

llvm-svn: 17093
2004-10-17 17:40:50 +00:00
Reid Spencer
210d95cffb Make the library name SparcV9 specific
llvm-svn: 17089
2004-10-17 15:01:12 +00:00
Reid Spencer
7ece7ff509 Consolidate the definitions
llvm-svn: 17088
2004-10-17 15:00:26 +00:00
Reid Spencer
9a97056275 PPC32GenCodeEmitter instead of PowerPCGenCodeEmitter
llvm-svn: 17087
2004-10-17 14:59:38 +00:00
Chris Lattner
63e6bdd207 Rewrite support for cast uint -> FP. In particular, we used to compile this:
double %test(uint %X) {
        %tmp.1 = cast uint %X to double         ; <double> [#uses=1]
        ret double %tmp.1
}

into:

test:
        sub %ESP, 8
        mov %EAX, DWORD PTR [%ESP + 12]
        mov %ECX, 0
        mov DWORD PTR [%ESP], %EAX
        mov DWORD PTR [%ESP + 4], %ECX
        fild QWORD PTR [%ESP]
        add %ESP, 8
        ret

... which basically zero extends to 8 bytes, then does an fild for an
8-byte signed int.

Now we generate this:


test:
        sub %ESP, 4
        mov %EAX, DWORD PTR [%ESP + 8]
        mov DWORD PTR [%ESP], %EAX
        fild DWORD PTR [%ESP]
        shr %EAX, 31
        fadd DWORD PTR [.CPItest_0 + 4*%EAX]
        add %ESP, 4
        ret

        .section .rodata
        .align  4
.CPItest_0:
        .quad   5728578726015270912

This does a 32-bit signed integer load, then adds in an offset if the sign
bit of the integer was set.

It turns out that this is substantially faster than the preceeding sequence.
Consider this testcase:

unsigned a[2]={1,2};
volatile double G;

void main() {
    int i;
    for (i=0; i<100000000; ++i )
        G += a[i&1];
}

On zion (a P4 Xeon, 3Ghz), this patch speeds up the testcase from 2.140s
to 0.94s.

On apoc, an athlon MP 2100+, this patch speeds up the testcase from 1.72s
to 1.34s.

Note that the program takes 2.5s/1.97s on zion/apoc with GCC 3.3 -O3
-fomit-frame-pointer.

llvm-svn: 17083
2004-10-17 08:01:28 +00:00
Chris Lattner
bf114f32c0 Unify handling of constant pool indexes with the other code paths, allowing
us to use index registers for CPI's

llvm-svn: 17082
2004-10-17 07:49:45 +00:00
Chris Lattner
892b15538d Give the asmprinter the ability to print memrefs with a constant pool index,
index reg and scale

llvm-svn: 17081
2004-10-17 07:16:32 +00:00
Chris Lattner
2fdca0bc02 fold:
%X = and Y, constantint
  %Z = setcc %X, 0

instead of emitting:

        and %EAX, 3
        test %EAX, %EAX
        je .LBBfoo2_2   # UnifiedReturnBlock

We now emit:

        test %EAX, 3
        je .LBBfoo2_2   # UnifiedReturnBlock

This triggers 581 times on 176.gcc for example.

llvm-svn: 17080
2004-10-17 06:10:40 +00:00
Chris Lattner
3f095f3c33 All of these labels are off by one now that the unreachable instruction exists
llvm-svn: 17079
2004-10-17 05:37:47 +00:00
Nate Begeman
f9aac7846c Implement bitfield insert by recognizing the following pattern:
1. optional shift left
2. and x, immX
3. and y, immY
4. or z, x, y
==> rlwimi z, x, y, shift, mask begin, mask end

where immX == ~immY and immX is a run of set bits. This transformation
fires 32 times on voronoi, once on espresso, and probably several
dozen times on external benchmarks such as gcc.

To put this in terms of actual code generated for
struct B { unsigned a : 3; unsigned b : 2; };
void storeA (struct B *b, int v) { b->a = v;}
void storeB (struct B *b, int v) { b->b = v;}

Old:
_storeA:
        rlwinm r2, r4, 0, 29, 31
        lwz r4, 0(r3)
        rlwinm r4, r4, 0, 0, 28
        or r2, r4, r2
        stw r2, 0(r3)
        blr

_storeB:
        rlwinm r2, r4, 3, 0, 28
        rlwinm r2, r2, 0, 27, 28
        lwz r4, 0(r3)
        rlwinm r4, r4, 0, 29, 26
        or r2, r2, r4
        stw r2, 0(r3)
        blr

New:
_storeA:
        lwz r2, 0(r3)
        rlwimi r2, r4, 0, 29, 31
        stw r2, 0(r3)
        blr

_storeB:
        lwz r2, 0(r3)
        rlwimi r2, r4, 3, 27, 28
        stw r2, 0(r3)
        blr

llvm-svn: 17078
2004-10-17 05:19:20 +00:00
Chris Lattner
e5aa085c1d Fix constant folding relational operators with undef operands.
llvm-svn: 17077
2004-10-17 04:01:51 +00:00
Chris Lattner
dc55caa720 I forgot that sparc no longer uses the shared asmwriter. Give it support
for undef.

llvm-svn: 17075
2004-10-17 02:44:45 +00:00
Chris Lattner
bb5b3f0b2f Add support for unreachable and undef
llvm-svn: 17074
2004-10-17 02:42:42 +00:00
Chris Lattner
8c86882a99 Implement constant folding of undef values.
llvm-svn: 17070
2004-10-16 23:31:32 +00:00
Chris Lattner
68f14bc09c Fix a type violation
llvm-svn: 17069
2004-10-16 23:28:04 +00:00
Nate Begeman
d4c970aa3d Finally fix one of the oldest FIXMEs in the PowerPC backend: correctly
flag rotate left word immediate then mask insert (rlwimi) as a two-address
instruction, and update the ISel usage of the instruction accordingly.

This will allow us to properly schedule rlwimi, and use it to efficiently
codegen bitfield operations.

llvm-svn: 17068
2004-10-16 20:43:38 +00:00
Chris Lattner
d12442c206 Kill the bogon that slipped into my buffer before I committed.
llvm-svn: 17067
2004-10-16 19:46:33 +00:00
Chris Lattner
b55574181d Implement InstCombine/getelementptr.ll:test9, which is the source of many
ugly and giant constnat exprs in some programs.

llvm-svn: 17066
2004-10-16 19:44:59 +00:00
Chris Lattner
a0019a2104 Do not erroneously accept revision 6 bytecode files when the format hasn't
been defined yet!

llvm-svn: 17063
2004-10-16 18:56:02 +00:00
Chris Lattner
3a1215ce83 Fix fix fix
llvm-svn: 17057
2004-10-16 18:21:50 +00:00
Chris Lattner
2fae8a1ef9 Add support for unreachable
llvm-svn: 17056
2004-10-16 18:21:33 +00:00
Chris Lattner
8d479b62ad Add support for undef
llvm-svn: 17055
2004-10-16 18:19:26 +00:00
Chris Lattner
8336590b1f Add support for undef, unreachable, and function flags
llvm-svn: 17054
2004-10-16 18:18:16 +00:00
Chris Lattner
eb973c8226 Parse undef and unreachable
llvm-svn: 17053
2004-10-16 18:17:13 +00:00
Chris Lattner
cbdf19fed2 Add support
llvm-svn: 17052
2004-10-16 18:16:19 +00:00
Chris Lattner
5fac2c8212 Add support for undef and unreachable
llvm-svn: 17051
2004-10-16 18:14:10 +00:00
Chris Lattner
3662abfd5a ADd support for undef and unreachable
llvm-svn: 17050
2004-10-16 18:13:47 +00:00
Chris Lattner
ae2e5f4de1 Teach the X86 backend about unreachable and undef. Among other things, we
now compile:

'foo() {}' into "ret" instead of "mov EAX, 0; ret"

llvm-svn: 17049
2004-10-16 18:13:05 +00:00
Chris Lattner
08ad95ec1f Add support for unreachable and undef
llvm-svn: 17048
2004-10-16 18:12:13 +00:00
Chris Lattner
3ebca6fb19 Optimize instructions involving undef values. For example X+undef == undef.
llvm-svn: 17047
2004-10-16 18:11:37 +00:00
Chris Lattner
4fca8caaee Add support for UndefValue
llvm-svn: 17046
2004-10-16 18:10:31 +00:00
Chris Lattner
ca01f160ee When promoting mem2reg, make uninitialized values become undef isntead of 0.
llvm-svn: 17045
2004-10-16 18:10:06 +00:00
Chris Lattner
80f963c30b Handle undef values as undefined on the constant lattice
ignore unreachable instructions

llvm-svn: 17044
2004-10-16 18:09:41 +00:00
Chris Lattner
c630ba08cf Add note
llvm-svn: 17043
2004-10-16 18:09:25 +00:00
Chris Lattner
4a37579191 Add support for the undef value. Implement a new optimization based on globals
that are initialized with undef.  When promoting malloc to a global, start out
initialized to undef

llvm-svn: 17042
2004-10-16 18:09:00 +00:00
Chris Lattner
93bf5a8066 Add support for undef and unreachable
llvm-svn: 17041
2004-10-16 18:08:06 +00:00
Chris Lattner
8824a5ee1c Implement UndefValue class
llvm-svn: 17040
2004-10-16 18:07:16 +00:00
Chris Lattner
72d8078a36 Add a missing dependency
llvm-svn: 17031
2004-10-16 17:12:55 +00:00
Chris Lattner
b1e427f563 Fix file header
llvm-svn: 17030
2004-10-16 16:37:42 +00:00
Chris Lattner
b3a86dc93f Be more careful about looking for constants when we really want constantint's.
llvm-svn: 17029
2004-10-16 16:07:10 +00:00
Chris Lattner
d3844cc216 Move the implementation of the instructions clone methods to this file so
that the vtables for these classes are only instantiated in this translation
unit, not in every xlation unit they are used.

llvm-svn: 17026
2004-10-15 23:52:53 +00:00
Chris Lattner
33dd5f87b8 There is no reason not to build these in parallel
llvm-svn: 17023
2004-10-15 23:22:15 +00:00
Misha Brukman
209f5ea32b Add a space between the type and name of value when printing error message
llvm-svn: 17022
2004-10-15 23:08:50 +00:00
Chris Lattner
b346be57a2 Don't print a bunch of metrics that are meaningless for external functions
llvm-svn: 17017
2004-10-15 19:40:31 +00:00
Chris Lattner
25b5777485 Instruction select globals with offsets better. For example, on this test
case:

int C[100];
int foo() {
  return C[4];
}

We now codegen:

foo:
        mov %EAX, DWORD PTR [C + 16]
        ret

instead of:

foo:
        mov %EAX, OFFSET C
        mov %EAX, DWORD PTR [%EAX + 16]
        ret

Other impressive features may be coming later.

This patch is contributed by Jeff Cohen!

llvm-svn: 17011
2004-10-15 05:05:29 +00:00
Chris Lattner
38de76365d Give the X86 JIT the ability to encode global+disp constants. Patch
contributed by Jeff Cohen!

llvm-svn: 17010
2004-10-15 04:53:13 +00:00
Chris Lattner
812d56631a Give the X86 asm printer the ability to print out addressing modes that have
constant displacements from global variables.  Patch by Jeff Cohen!

llvm-svn: 17009
2004-10-15 04:44:53 +00:00
Chris Lattner
1b9a284e54 Allow X86 addressing modes to represent globals with offsets. Patch contributed
by Jeff Cohen!

llvm-svn: 17008
2004-10-15 04:43:20 +00:00
Chris Lattner
65976f4178 Allow machine operands to represent global variables with offsets. This is
useful when you have a reference like:

int A[100];

void foo() { A[10] = 1; }

In this case, &A[10] is a single constant and should be treated as such.

Only MO_GlobalAddress and MO_ExternalSymbol are allowed to use this field, no
other operand type is.

This is another fine patch contributed by Jeff Cohen!!

llvm-svn: 17007
2004-10-15 04:38:41 +00:00
Chris Lattner
34acee9dbd This patch fixes the nasty bug that caused 175.vpr to fail for X86 last night.
The problem occurred when trying to reload this instruction:

MOV32mr %reg2326, 8, %reg2297, 4, %reg2295

The value of reg2326 was available in EBX, so it was reused from there, instead
of reloading it into EDX.

The value of reg2297 was available in EDX, so it was reused from there, instead
of reloading it into EDI.

The value of reg2295 was not available, so we tried reloading it into EBX, its
assigned register.  However, we checked and saw that we already reloaded
something into EBX, so we chose what reg2326 was assigned to (EDX) and reloaded
into that register instead.

Unfortunately EDX had already been used by reg2297, so reloading into EDX
clobbered the value used by the reg2326 operand, breaking the program.

The fix for this is to check that the newly picked register is ok.  In this
case we now find that EDX is already used and try using EDI, which succeeds.

llvm-svn: 17006
2004-10-15 03:19:31 +00:00
Chris Lattner
2c87b68231 This patch adds and improves debugging output. No functionality changes.
llvm-svn: 17005
2004-10-15 03:16:29 +00:00
Nate Begeman
d8183bd297 Better codegen of binary integer ops with 32 bit immediate operands.
This transformation fires a few dozen times across the testsuite.

For example, int test2(int X) { return X ^ 0x0FF00FF0; }
Old:
_test2:
        lis r2, 4080
        ori r2, r2, 4080
        xor r3, r3, r2
        blr

New:
_test2:
        xoris r3, r3, 4080
        xori r3, r3, 4080
        blr

llvm-svn: 17004
2004-10-15 00:50:19 +00:00
Misha Brukman
e05e5f0655 The field is called imm22', not simply imm'
llvm-svn: 17003
2004-10-14 22:33:32 +00:00
Misha Brukman
4932cb0e94 Synthetic instructions RET and RETL need to have all 3 parameters specified
llvm-svn: 17002
2004-10-14 22:32:49 +00:00
Misha Brukman
7b29a7bc80 Class F2_1 already inherits the imm22 field from class F2
llvm-svn: 17001
2004-10-14 22:32:24 +00:00
Misha Brukman
4e691d2adb Generate the SparcV8 code emitter from .td files
llvm-svn: 17000
2004-10-14 21:57:19 +00:00
Misha Brukman
5bc4009783 * In the F3_1 class, default asi to 0 because it's not currently used
* In the F3_3 class, remove mention of asi because it's not part of the format

llvm-svn: 16999
2004-10-14 21:53:39 +00:00
Chris Lattner
7319d9289f Fix a bug John tracked down in libstdc++ where we were incorrectly deleting
weak functions.  Thanks for finding this John!

llvm-svn: 16997
2004-10-14 19:53:50 +00:00
Brian Gaeke
ce6dcb0713 Add FSTOI, FDTOI (fp to integer cast) instructions.
llvm-svn: 16996
2004-10-14 19:39:35 +00:00
Brian Gaeke
ba65401a4d Rewrite emitCastOperation, refactoring parts of it into emitIntegerCast, and
adding emitFPToIntegerCast.

llvm-svn: 16995
2004-10-14 19:39:34 +00:00
Brian Gaeke
31fa12ece7 Add list of libc procedures we'll use, at some point.
Update list of currently failing tests.
ADJCALLSTACK* support is done.

llvm-svn: 16994
2004-10-14 19:39:33 +00:00
Chris Lattner
44e34a155c Make sure any client of Dominators.h links in Dominators.cpp
Patch by Morten Ofstad

llvm-svn: 16987
2004-10-14 15:47:16 +00:00
Chris Lattner
3fc5e3778f Do not use the same variable name for two different variables in the
same scope.  This confused VC++ (and probably people too!).  Patch by
Morten Ofstad!

llvm-svn: 16985
2004-10-14 14:59:16 +00:00
Misha Brukman
66261f021d * Claim to support machine code emission - return false from
addPassesToEmitMachineCode()
* Add support for registers and constants in getMachineOpValue()

This enables running "int main() { ret 0 }" via the PowerPC JIT.

llvm-svn: 16983
2004-10-14 06:39:56 +00:00
Misha Brukman
cb4130c28c * Include the real (generated) version of getBinaryCodeForInstr()
* Add implementation of getMachineOpValue() for generated code emitter
* Convert assert()s in unimplemented functions to abort()s so that non-debug
  builds fail predictably
* Add file header comments

llvm-svn: 16981
2004-10-14 06:07:25 +00:00
Misha Brukman
11d1764f74 * Make a PPC32-specific code emitter because we have separate classes for 32-
and 64-bit code emitters that cannot share code unless we use virtual
  functions
* Identify components being built by tablegen with more detail by assigning them
  to PowerPC, PPC32, or PPC64 more specifically; also avoids seeing 'building
  PowerPC XYZ' messages twice, where one is for PPC32 and one for PPC64

llvm-svn: 16980
2004-10-14 06:04:56 +00:00
Tanya Lattner
8a4773acb3 Checking in code that works on my simple test case. However, there is still a bug with branches that I need to fix.
llvm-svn: 16979
2004-10-14 06:04:28 +00:00
Misha Brukman
5e8bfd0675 There is only one field in an instruction, and that is `Inst', the final view of
the instruction binary format, all others are simply operands and should not
have the `field' label

llvm-svn: 16978
2004-10-14 05:55:37 +00:00
Misha Brukman
47c2236ae9 PowerPC instruction definitions use LittleEndian-style encoding [0..31]
llvm-svn: 16977
2004-10-14 05:54:38 +00:00
Misha Brukman
628b2ba095 Add isLittleEndianEncoding to InstrInfo class, defaults to `off'
llvm-svn: 16976
2004-10-14 05:53:40 +00:00
Chris Lattner
88ad718c75 When converting phi nodes into select instructions, we shouldn't promote PHI
nodes unless we KNOW that we are able to promote all of them.

This fixes: test/Regression/Transforms/SimplifyCFG/PhiNoEliminate.ll

llvm-svn: 16973
2004-10-14 05:13:36 +00:00
Reid Spencer
7689de0e49 Allow this file to compile on Darwin.
llvm-svn: 16971
2004-10-14 03:33:25 +00:00
Reid Spencer
d3c912a7b0 Use __MINGW instead of __MING. Patch contributed by Henrik Bach.
llvm-svn: 16970
2004-10-14 03:09:02 +00:00
Reid Spencer
86b05297be Get proper BSD #includes for MappedFile implementation.
llvm-svn: 16969
2004-10-14 03:06:59 +00:00
Reid Spencer
27ae06c1db Implementation of MappedFile for Win32. Patch provided by Jeff Cohen.
llvm-svn: 16968
2004-10-14 03:05:59 +00:00
Chris Lattner
ad57c9f54c Today is not my day. Fix broken #
llvm-svn: 16967
2004-10-14 02:31:35 +00:00
Chris Lattner
d2a5066c6a unbreak previous checkin :(
llvm-svn: 16966
2004-10-14 02:06:48 +00:00
Chris Lattner
93bfff6da2 Add back a missing paren
llvm-svn: 16965
2004-10-14 01:57:28 +00:00
Chris Lattner
aef5a2d78d Fit to 80 cols
llvm-svn: 16964
2004-10-14 01:49:34 +00:00
Chris Lattner
2ed4986e00 Wrap to 80 cols, delete some seriously old debugging printouts
llvm-svn: 16963
2004-10-14 01:46:07 +00:00
Chris Lattner
3f3d5aad47 Fit in 80 columns
llvm-svn: 16962
2004-10-14 01:39:18 +00:00
Chris Lattner
bdb8db1a40 Minor cleanups
llvm-svn: 16961
2004-10-14 01:35:17 +00:00
Reid Spencer
e6418ec30f Update to reflect changes in Makefile rules.
llvm-svn: 16950
2004-10-13 11:46:52 +00:00
Chris Lattner
63f15158f4 Transform memmove -> memcpy when the source is obviously constant memory.
llvm-svn: 16932
2004-10-12 04:52:52 +00:00
Chris Lattner
58435df175 Minor tweaks
llvm-svn: 16929
2004-10-12 04:32:37 +00:00
Chris Lattner
8428411c46 Implement a new method
llvm-svn: 16927
2004-10-12 04:20:25 +00:00
Chris Lattner
6b6e5a5815 This nutty patch has been in my tree since before 1.3 went out, and it needs
to go in.  This patch allows us to compute the trip count of loops controlled
by values loaded from constant arrays.  The cannonnical example of this is
strlen when passed a constant argument:

for (int i = 0; "constantstring"[i]; ++i) ;
return i;

In this case, it will compute that the loop executes 14 times, which means
that the exit value of i is 14.  Because of this, the loop gets DCE'd and
we are happy.  This also applies to anything that does similar things, e.g.
loops like this:

  const float Array[] = { 0.1, 2.1, 3.2, 23.21 };
  for (int i = 0; Array[i] < 20; ++i)

and is actually fairly general.

The problem with this is that it almost never triggers.  The reason is that
we run indvars and the loop optimizer only at compile time, which is before
things like strlen and strcpy have been inlined into the program from libc.
Because of this, it almost never is used (it triggers twice in specint2k).

I'm committing it because it DOES work, may be useful in the future, and
doesn't slow us down at all.  If/when we start running the loop optimizer
at link-time (-O4?) this will be very nice indeed :)

llvm-svn: 16926
2004-10-12 01:49:27 +00:00
Chris Lattner
b605c860fc Fix a REALLY obscure bug in my previous checkin, which was splicing the END
marker from one ilist into the middle of another basic block!

llvm-svn: 16925
2004-10-12 01:02:29 +00:00
Chris Lattner
a6d0e41b72 Handle a common case more carefully. In particular, instead of transforming
pointer recurrences into expressions from this:

  %P_addr.0.i.0 = phi sbyte* [ getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), %entry ], [ %inc.0.i, %no_exit.i ]
  %inc.0.i = getelementptr sbyte* %P_addr.0.i.0, int 1            ; <sbyte*> [#uses=2]

into this:

  %inc.0.i = getelementptr sbyte* getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), int %inc.0.i.rec

Actually create something nice, like this:

  %inc.0.i = getelementptr [8 x sbyte]* %.str_1, int 0, int %inc.0.i.rec

llvm-svn: 16924
2004-10-11 23:06:50 +00:00
Chris Lattner
63ba63bb56 Allow creation of GEP constantexprs with a vector of value* operands as
well as a vector of constant*'s.  It turns out that this is more efficient
and all of the clients want to do that, so we should cater to them.

llvm-svn: 16923
2004-10-11 22:52:25 +00:00
Chris Lattner
6496308d18 Implement remove/eraseFromParent methods
llvm-svn: 16922
2004-10-11 22:21:39 +00:00
Chris Lattner
f9e79568c2 Reenable the transform, turning X/-10 < 1 into X > -10
llvm-svn: 16918
2004-10-11 19:40:04 +00:00
Chris Lattner
2bc979427d Fix a warning that is emitted on the suns
llvm-svn: 16917
2004-10-11 15:50:40 +00:00
Chris Lattner
9d662982e0 This patch implements two things (sorry).
First, it allows SRA of globals that have embedded arrays, implementing
GlobalOpt/globalsra-partial.llx.  This comes up infrequently, but does allow,
for example, deleting several stores to dead parts of globals in dhrystone.

Second, this implements GlobalOpt/malloc-promote-*.llx, which is the
following nifty transformation:

Basically if a global pointer is initialized with malloc, and we can tell
that the program won't notice, we transform this:

struct foo *FooPtr;
...
  FooPtr = malloc(sizeof(struct foo));
...
  FooPtr->A   FooPtr->B

Into:

struct foo FooPtrBody;
...
   FooPtrBody.A  FooPtrBody.B

This comes up occasionally, for example, the 'disp' global in 183.equake (where
the xform speeds the CBE version of the program up from 56.16s to 52.40s (7%)
on apoc), and the 'desired_accept', 'fixLRBT', 'macroArray', & 'key_queue'
globals in 300.twolf (speeding it up from 22.29s to 21.55s (3.4%)).

The nice thing about this xform is that it exposes the resulting global to
global variable optimization and makes alias analysis easier in addition to
eliminating a few loads.

llvm-svn: 16916
2004-10-11 05:54:41 +00:00
Chris Lattner
f8dab00902 Fix SingleSource/Benchmarks/McGill/chomp
llvm-svn: 16912
2004-10-11 04:07:27 +00:00
Chris Lattner
f8380b953a If we are trying to create a ConstantExpr cast that is really a GEP to the
first element of an array, return a GEP instead of a cast.  This allows us
to transparently fold this:

int* getelementptr (int* cast ([100 x int]* %Gbody to int*), int 40)

into this:

int* getelementptr ([100 x int]* %Gbody, int 0, int 40)

llvm-svn: 16911
2004-10-11 03:57:30 +00:00
Misha Brukman
5e9e7cada2 ModuloScheduling has moved to lib/Target/SparcV9
llvm-svn: 16906
2004-10-10 23:37:40 +00:00
Misha Brukman
995af8d2b1 Add ModuloScheduling to the recursive build tree
llvm-svn: 16905
2004-10-10 23:36:09 +00:00
Misha Brukman
95cbabd1b5 Adjust header file inclusion due to move
llvm-svn: 16904
2004-10-10 23:34:50 +00:00
Misha Brukman
0f102f7fc9 Adjust comment header and paths to refect move
llvm-svn: 16903
2004-10-10 23:34:36 +00:00
Misha Brukman
020c3ab94c ModuloScheduling moved to lib/Target/SparcV9 as it is SparcV9-specific
llvm-svn: 16902
2004-10-10 23:33:20 +00:00
Chris Lattner
21dcce4e33 Just because we cannot completely eliminate all uses of a global, we can
still optimize away all of the indirect calls and loads, etc from it.
This turns code like this:

  if (G != 0)
    G();

into
   if (G != 0)
     ActualCallee();

This triggers a couple of times in gcc and libstdc++.

llvm-svn: 16901
2004-10-10 23:14:11 +00:00
Reid Spencer
85d2758f11 Initial version of automake Makefile.am file.
llvm-svn: 16898
2004-10-10 22:52:14 +00:00
Reid Spencer
04cb07f3ce Add the new InstrSched directory.
llvm-svn: 16897
2004-10-10 22:51:03 +00:00
Tanya Lattner
be65f6cd02 Added debug information. Fixed several bugs in the reconstruct loop function.
llvm-svn: 16895
2004-10-10 22:44:35 +00:00
Reid Spencer
1b7459b29d Initial version of automake Makefile.am file.
llvm-svn: 16893
2004-10-10 22:20:40 +00:00
Reid Spencer
bc135b067a Build both archive and relinked objects
llvm-svn: 16892
2004-10-10 22:17:39 +00:00
Reid Spencer
b317fd443c Remove the InstrSched directory (moved to SparcV9)
llvm-svn: 16887
2004-10-10 21:19:41 +00:00
Reid Spencer
064bebfba3 Directory no long exists (moved to Targets/SparcV9).
llvm-svn: 16886
2004-10-10 21:18:31 +00:00
Reid Spencer
7d9cba7a0f Initial version of automake Makefile.am file.
llvm-svn: 16885
2004-10-10 20:43:57 +00:00
Brian Gaeke
26b353ebd6 Fix assertion failure when calling or returning from a function which
returns 'bool' type.

llvm-svn: 16884
2004-10-10 20:34:17 +00:00
Brian Gaeke
3af0547680 Implement eliminateCallFramePseudoInstr().
Wrap a long comment line.

llvm-svn: 16883
2004-10-10 19:57:21 +00:00
Brian Gaeke
b6239cd5ed Model calls as *both* using *and* killing O0..O5, because callees use the
argument values passed in (so they're not dead until *after* the call),
and callees are free to modify those registers.

llvm-svn: 16882
2004-10-10 19:57:20 +00:00
Brian Gaeke
245a073aa6 Fix whitespace and wrap some long lines.
Deal with allocating stack space for outgoing args and copying them into the
correct stack slots (at least, we can copy <=32-bit int args).
We now correctly generate ADJCALLSTACK* instructions.

llvm-svn: 16881
2004-10-10 19:57:18 +00:00
Chris Lattner
d10f9ed455 Fix 2004-10-10-CastStoreOnce.llx, by adjusting types back if we strip off a cast
llvm-svn: 16878
2004-10-10 17:07:12 +00:00
Chris Lattner
cd4a7a5293 Implement GlobalOpt/deadglobal-2.llx, deletion of globals that are only
stored to, but are stored at variable indexes.  This occurs at least in
176.gcc, but probably others, and we should handle it for completeness.

llvm-svn: 16876
2004-10-10 16:47:33 +00:00
Chris Lattner
b0d3511aa6 Avoid calling use_size() which could (in theory) be expensive if the global
has a large number of users.  Instead, just keep track of whether we're
making changes as we do so.

This patch has no functionlity changes.

llvm-svn: 16874
2004-10-10 16:43:46 +00:00
Chris Lattner
6ff0fd4837 bling bling!
llvm-svn: 16873
2004-10-10 16:26:13 +00:00
Chris Lattner
6b0cd15f9c Eliminate global pointers that are only stored a single value and null if
we know that all uses of the global will trap if the pointer contained is
null.  In this case, we forward substitute the stored value to any uses.

This has the effect of devirtualizing trivial globals in trivial cases.  For
example, 164.gzip contains this:

gzip.h:extern   int (*read_buf) OF((char *buf, unsigned size));
bits.c: read_buf  = file_read;
deflate.c:    lookahead = read_buf((char*)window,
deflate.c:        n = read_buf((char*)window+strstart+lookahead, more);

Since read_buf has to point to file_read at every use, we just replace
the calls through read_buf with a direct call to file_read.

This occurs in several benchmarks, including 176.gcc and 164.gzip.  Direct
calls are good and stuff.

llvm-svn: 16871
2004-10-09 21:48:45 +00:00
Chris Lattner
955a220ad2 Instead of silently breaking, print notification of why this doesn't work.
llvm-svn: 16870
2004-10-09 21:13:51 +00:00
Chris Lattner
f8a6e4402e Use DEBUG instead of DebugFlag directly, as DebugFlag does not respect
-debug-only!

llvm-svn: 16868
2004-10-09 19:30:36 +00:00
Brian Gaeke
ba13791a01 update according to tonight's info
llvm-svn: 16866
2004-10-09 05:58:27 +00:00
Brian Gaeke
3abdd11420 Implement getModuleMatchQuality and getJITMatchQuality so that v8 will be the
default 32/BE target on sparc hosts, and ppc will continue to be the default
on other hosts.

llvm-svn: 16865
2004-10-09 05:57:01 +00:00
Chris Lattner
51a762c414 Fix infinite loop due to iteration
llvm-svn: 16864
2004-10-09 03:32:52 +00:00
Chris Lattner
c5085d212c Implement sub.ll:test17, -X/C -> X/-C
llvm-svn: 16863
2004-10-09 02:50:40 +00:00
Chris Lattner
e2607964ae Add a check to avoid an assertion on malformed input
llvm-svn: 16861
2004-10-09 02:18:58 +00:00
Chris Lattner
2419e1d27e The person who was planning to add SSE support isn't anymore, so disable
the -sse* options (to avoid misleading people).

Also, the stack alignment of the target doesn't depend on whether SSE is
eventually implemented, so remove a comment.

llvm-svn: 16860
2004-10-08 22:41:46 +00:00
Chris Lattner
1291307d27 Fix a major regression from the bugfix for 2004-10-08-SelectSetCCFold.llx,
which prevented setcc's from being folded into branches.  It appears that
conditional branchinst's CC operand is actually operand(2), not operand(0)
as we might expect. :(

llvm-svn: 16859
2004-10-08 22:24:31 +00:00
Chris Lattner
4f504b0751 If we found a dead global, we should at least delete it...
llvm-svn: 16858
2004-10-08 22:05:31 +00:00
Chris Lattner
7fc483bf28 * Pull out the meat of runOnModule into another function for clarity.
* Do not lead dangling dead constants prevent optimization
* Iterate global optimization while we're making progress.

These changes allow us to be more aggressive, handling cases like
GlobalOpt/iterate.llx without a problem (turning it into 'ret int 0').

llvm-svn: 16857
2004-10-08 20:59:28 +00:00
Chris Lattner
d777571d0c We might as well delete the known-dead global sooner rather than later since
we know it is dead.

llvm-svn: 16855
2004-10-08 20:25:55 +00:00
Misha Brukman
af84f00600 Hyphenate target-(in)dependent for more tasty grammar goodness (tm)
llvm-svn: 16854
2004-10-08 19:43:31 +00:00
Chris Lattner
816a8a5e1e Temporarily disable a buggy transformation until it can be fixed. This fixes
254.gap.

llvm-svn: 16853
2004-10-08 19:15:44 +00:00
Misha Brukman
d858079005 Adjust paths due to moving InstrSched to lib/Target/SparcV9
llvm-svn: 16852
2004-10-08 18:30:22 +00:00
Misha Brukman
a01e5cd2ea InstrSched has been moved to lib/Target/SparcV9
llvm-svn: 16850
2004-10-08 18:12:53 +00:00
Misha Brukman
049f559995 InstrSched is SparcV9-specific and so has been moved to lib/Target/SparcV9/
llvm-svn: 16849
2004-10-08 18:12:14 +00:00
Misha Brukman
b1851ac70c Single-space instead of double-spacing in the Makefile
llvm-svn: 16848
2004-10-08 18:11:14 +00:00
Misha Brukman
cfe86b257b Build InstrSched as well, and all three subdirs can be built independently
llvm-svn: 16847
2004-10-08 18:10:48 +00:00
Misha Brukman
5ea7613a13 Single-space instead of double-spacing in the Makefile
llvm-svn: 16845
2004-10-08 18:05:25 +00:00
Chris Lattner
71aecc5006 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

llvm-svn: 16842
2004-10-08 17:32:09 +00:00
Chris Lattner
1ac1e54bf9 Fix bug: 2004-10-08-SelectSetCCFold.llx. Normally this is hidden by the
instcombine xform, which is why we didn't notice it before.

llvm-svn: 16840
2004-10-08 16:34:13 +00:00
Chris Lattner
5839d93b51 Instcombine (X & FF00) + xx00 -> (X+xx00) & FF00, implementing and.ll:test27
This comes up when doing adds to bitfield elements.

llvm-svn: 16836
2004-10-08 05:07:56 +00:00
Chris Lattner
87259c3ce9 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.

llvm-svn: 16834
2004-10-08 03:46:20 +00:00
Nate Begeman
dfefd2f3fc Implement logical and with an immediate that consists of a contiguous block
of one or more 1 bits (may wrap from least significant bit to most
significant bit) as the rlwinm rather than andi., andis., or some longer
instructons sequence.

int andn4(int z) { return z & -4; }
int clearhi(int z) { return z & 0x0000FFFF; }
int clearlo(int z) { return z & 0xFFFF0000; }
int clearmid(int z) { return z & 0x00FFFF00; }
int clearwrap(int z) { return z & 0xFF0000FF; }

_andn4:
        rlwinm r3, r3, 0, 0, 29
        blr

_clearhi:
        rlwinm r3, r3, 0, 16, 31
        blr

_clearlo:
        rlwinm r3, r3, 0, 0, 15
        blr

_clearmid:
        rlwinm r3, r3, 0, 8, 23
        blr

_clearwrap:
        rlwinm r3, r3, 0, 24, 7
        blr

llvm-svn: 16832
2004-10-08 02:49:24 +00:00
Nate Begeman
370b1b7a9a Several fixes and enhancements to the PPC32 backend.
1. Fix an illegal argument to getClassB when deciding whether or not to
   sign extend a byte load.

2. Initial addition of isLoad and isStore flags to the instruction .td file
   for eventual use in a scheduler.

3. Rewrite of how constants are handled in emitSimpleBinaryOperation so
   that we can emit the PowerPC shifted immediate instructions far more
   often.  This allows us to emit the following code:

int foo(int x) { return x | 0x00F0000; }

_foo:
.LBB_foo_0:     ; entry
        ; IMPLICIT_DEF
        oris r3, r3, 15
        blr

llvm-svn: 16826
2004-10-07 22:30:03 +00:00
Nate Begeman
76d2a77998 Add ori reg, reg, 0 as a move instruction. This can be generated from
loading a 32bit constant into a register whose low halfword is all zeroes.

We now omit the ori after the lis for the following C code:

int bar(int y) { return y * 0x00F0000; }

_bar:
.LBB_bar_0:     ; entry
        ; IMPLICIT_DEF
        lis r2, 15
        mullw r3, r3, r2
        blr

llvm-svn: 16825
2004-10-07 22:26:12 +00:00
Nate Begeman
f60feea650 Remove unnecessary header include
llvm-svn: 16824
2004-10-07 22:24:32 +00:00
Chris Lattner
7882b54197 Improve comments, no functionality changes
llvm-svn: 16814
2004-10-07 21:30:30 +00:00
Chris Lattner
d15e144241 Fix a nasty dangling pointer problem, due to a free'd pointer being left in
a map.  This caused problems if a later object happened to be allocated at
the free'd object's address.

llvm-svn: 16813
2004-10-07 20:01:31 +00:00
Chris Lattner
50e55bcdb0 Unfortunately the fix for the previous bug introduced the previous
exponential behavior (bork!).  This patch processes stuff with an
explicit SCC finder, allowing the algorithm to be more clear,
efficient, and also (as a bonus) correct!  This gets us back to taking
0.6s to disassemble my horrible .bc file that previously took something
> 30 mins.

llvm-svn: 16811
2004-10-07 19:20:48 +00:00
Chris Lattner
dfdbd62d37 Fix a bug in my previous change. Unfortunately this reverts most of the
speedup, but has the advantage of not breaking a bunch of programs!

llvm-svn: 16806
2004-10-07 16:19:40 +00:00
Chris Lattner
e1d5d599bd Fix a bug in the safety analysis routine
llvm-svn: 16804
2004-10-07 06:01:25 +00:00
Chris Lattner
e7ec24c63e Comment cleanups
llvm-svn: 16803
2004-10-07 06:00:24 +00:00
Chris Lattner
ad9fe72e72 * 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.

llvm-svn: 16802
2004-10-07 04:16:33 +00:00
Chris Lattner
4a19983f2d 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.

llvm-svn: 16793
2004-10-06 20:57:02 +00:00
Chris Lattner
e412d10cc0 Dont' let null nodes sneak past cast instructions
llvm-svn: 16779
2004-10-06 19:29:13 +00:00
Chris Lattner
c2563bf614 Change Type::isAbstract to have better comments, a more correct name
(PromoteAbstractToConcrete), and to use a set to avoid recomputation.
In particular, this set eliminates the potentially exponential cases
from this little recursive algorithm.

On a particularly nasty testcase, llvm-dis on the .bc file went from 34
minutes (which is when I killed it, it still hadn't finished) to 0.57s.
Remember kids, exponential algorithms are bad.

llvm-svn: 16772
2004-10-06 16:36:46 +00:00
Chris Lattner
38fbf09104 Correct some typeos
llvm-svn: 16770
2004-10-06 16:28:24 +00:00
Chris Lattner
ff8cbd01e7 Instcombine: -(X sdiv C) -> (X sdiv -C), tested by sub.ll:test16
llvm-svn: 16769
2004-10-06 15:08:25 +00:00
Chris Lattner
82aa8544a5 Remove debugging code, fix encoding problem. This fixes the problems
the JIT had last night.

llvm-svn: 16766
2004-10-06 14:31:50 +00:00
Nate Begeman
79d42a185e Turning on fsel code gen now that we can do so would be good.
llvm-svn: 16765
2004-10-06 11:03:30 +00:00
Nate Begeman
7b4fe83ba8 Implement floating point select for lt, gt, le, ge using the powerpc fsel
instruction.

Now, rather than emitting the following loop out of bisect:
.LBB_main_19:	; no_exit.0.i
	rlwinm r3, r2, 3, 0, 28
	lfdx f1, r3, r27
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f2, lo16(.CPI_main_1-"L00000$pb")(r3)
	fsub f2, f2, f1
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f4, lo16(.CPI_main_1-"L00000$pb")(r3)
	fcmpu cr0, f1, f4
	bge .LBB_main_64	; no_exit.0.i
.LBB_main_63:	; no_exit.0.i
	b .LBB_main_65	; no_exit.0.i
.LBB_main_64:	; no_exit.0.i
	fmr f2, f1
.LBB_main_65:	; no_exit.0.i
	addi r3, r2, 1
	rlwinm r3, r3, 3, 0, 28
	lfdx f1, r3, r27
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f4, lo16(.CPI_main_1-"L00000$pb")(r3)
	fsub f4, f4, f1
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f5, lo16(.CPI_main_1-"L00000$pb")(r3)
	fcmpu cr0, f1, f5
	bge .LBB_main_67	; no_exit.0.i
.LBB_main_66:	; no_exit.0.i
	b .LBB_main_68	; no_exit.0.i
.LBB_main_67:	; no_exit.0.i
	fmr f4, f1
.LBB_main_68:	; no_exit.0.i
	fadd f1, f2, f4
	addis r3, r30, ha16(.CPI_main_2-"L00000$pb")
	lfd f2, lo16(.CPI_main_2-"L00000$pb")(r3)
	fmul f1, f1, f2
	rlwinm r3, r2, 3, 0, 28
	lfdx f2, r3, r28
	fadd f4, f2, f1
	fcmpu cr0, f4, f0
	bgt .LBB_main_70	; no_exit.0.i
.LBB_main_69:	; no_exit.0.i
	b .LBB_main_71	; no_exit.0.i
.LBB_main_70:	; no_exit.0.i
	fmr f0, f4
.LBB_main_71:	; no_exit.0.i
	fsub f1, f2, f1
	addi r2, r2, -1
	fcmpu cr0, f1, f3
	blt .LBB_main_73	; no_exit.0.i
.LBB_main_72:	; no_exit.0.i
	b .LBB_main_74	; no_exit.0.i
.LBB_main_73:	; no_exit.0.i
	fmr f3, f1
.LBB_main_74:	; no_exit.0.i
	cmpwi cr0, r2, -1
	fmr f16, f0
	fmr f17, f3
	bgt .LBB_main_19	; no_exit.0.i

We emit this instead:
.LBB_main_19:	; no_exit.0.i
	rlwinm r3, r2, 3, 0, 28
	lfdx f1, r3, r27
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f2, lo16(.CPI_main_1-"L00000$pb")(r3)
	fsub f2, f2, f1
	fsel f1, f1, f1, f2
	addi r3, r2, 1
	rlwinm r3, r3, 3, 0, 28
	lfdx f2, r3, r27
	addis r3, r30, ha16(.CPI_main_1-"L00000$pb")
	lfd f4, lo16(.CPI_main_1-"L00000$pb")(r3)
	fsub f4, f4, f2
	fsel f2, f2, f2, f4
	fadd f1, f1, f2
	addis r3, r30, ha16(.CPI_main_2-"L00000$pb")
	lfd f2, lo16(.CPI_main_2-"L00000$pb")(r3)
	fmul f1, f1, f2
	rlwinm r3, r2, 3, 0, 28
	lfdx f2, r3, r28
	fadd f4, f2, f1
	fsub f5, f0, f4
	fsel f0, f5, f0, f4
	fsub f1, f2, f1
	addi r2, r2, -1
	fsub f2, f1, f3
	fsel f3, f2, f3, f1
	cmpwi cr0, r2, -1
	fmr f16, f0
	fmr f17, f3
	bgt .LBB_main_19	; no_exit.0.i

llvm-svn: 16764
2004-10-06 09:53:04 +00:00
Chris Lattner
b0e465f0cb Codegen signed mod by 2 or -2 more efficiently. Instead of generating:
t:
        mov %EDX, DWORD PTR [%ESP + 4]
        mov %ECX, 2
        mov %EAX, %EDX
        sar %EDX, 31
        idiv %ECX
        mov %EAX, %EDX
        ret

Generate:
t:
        mov %ECX, DWORD PTR [%ESP + 4]
***     mov %EAX, %ECX
        cdq
        and %ECX, 1
        xor %ECX, %EDX
        sub %ECX, %EDX
***     mov %EAX, %ECX
        ret

Note that the two marked moves are redundant, and should be eliminated by the
register allocator, but aren't.

Compare this to GCC, which generates:

t:
        mov     %eax, DWORD PTR [%esp+4]
        mov     %edx, %eax
        shr     %edx, 31
        lea     %ecx, [%edx+%eax]
        and     %ecx, -2
        sub     %eax, %ecx
        ret

or ICC 8.0, which generates:

t:
        movl      4(%esp), %ecx                                 #3.5
        movl      $-2147483647, %eax                            #3.25
        imull     %ecx                                          #3.25
        movl      %ecx, %eax                                    #3.25
        sarl      $31, %eax                                     #3.25
        addl      %ecx, %edx                                    #3.25
        subl      %edx, %eax                                    #3.25
        addl      %eax, %eax                                    #3.25
        negl      %eax                                          #3.25
        subl      %eax, %ecx                                    #3.25
        movl      %ecx, %eax                                    #3.25
        ret                                                     #3.25

We would be in great shape if not for the moves.

llvm-svn: 16763
2004-10-06 05:01:07 +00:00
Chris Lattner
c959314701 Really fix FreeBSD, which apparently doesn't tolerate the extern.
Thanks to Jeff Cohen for pointing out my goof.

llvm-svn: 16762
2004-10-06 04:21:52 +00:00
Chris Lattner
09b6b3f514 Fix a scary bug with signed division by a power of two. We used to generate:
s:   ;; X / 4
        mov %EAX, DWORD PTR [%ESP + 4]
        mov %ECX, %EAX
        sar %ECX, 1
        shr %ECX, 30
        mov %EDX, %EAX
        add %EDX, %ECX
        sar %EAX, 2
        ret

When we really meant:

s:
        mov %EAX, DWORD PTR [%ESP + 4]
        mov %ECX, %EAX
        sar %ECX, 1
        shr %ECX, 30
        add %EAX, %ECX
        sar %EAX, 2
        ret

Hey, this also reduces register pressure too :)

llvm-svn: 16761
2004-10-06 04:19:43 +00:00
Chris Lattner
9258948b08 Codegen signed divides by 2 and -2 more efficiently. In particular
instead of:

s:   ;; X / 2
        movl 4(%esp), %eax
        movl %eax, %ecx
        shrl $31, %ecx
        movl %eax, %edx
        addl %ecx, %edx
        sarl $1, %eax
        ret

t:   ;; X / -2
        movl 4(%esp), %eax
        movl %eax, %ecx
        shrl $31, %ecx
        movl %eax, %edx
        addl %ecx, %edx
        sarl $1, %eax
        negl %eax
        ret

Emit:

s:
        movl 4(%esp), %eax
        cmpl $-2147483648, %eax
        sbbl $-1, %eax
        sarl $1, %eax
        ret

t:
        movl 4(%esp), %eax
        cmpl $-2147483648, %eax
        sbbl $-1, %eax
        sarl $1, %eax
        negl %eax
        ret

llvm-svn: 16760
2004-10-06 04:02:39 +00:00
Chris Lattner
acd213fba3 Add some new instructions. Fix the asm string for sbb32rr
llvm-svn: 16759
2004-10-06 04:01:02 +00:00
Chris Lattner
5f0c904ec0 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

llvm-svn: 16758
2004-10-06 03:27:37 +00:00
Chris Lattner
b2e8fdc431 FreeBSD uses GCC. Patch contributed by Jeff Cohen!
llvm-svn: 16756
2004-10-06 03:15:44 +00:00
Brian Gaeke
38641114a3 Must include sys/stat.h before declaring a 'struct stat'
llvm-svn: 16728
2004-10-05 18:46:59 +00:00
Chris Lattner
6023408a6e Make sure the const bit gets inherited correctly when linking declarations
of disagreeing constness.  This fixes
test/Regression/Linker/ConstantGlobals[123].ll

llvm-svn: 16692
2004-10-05 02:28:11 +00:00
Reid Spencer
bba26329ab Adjust sys/stat.h inclusion so its only for SunOS.
llvm-svn: 16686
2004-10-05 00:56:46 +00:00
Tanya Lattner
7198953962 Added a couple of includes to get this to compile on Sparc.
llvm-svn: 16685
2004-10-05 00:51:26 +00:00
Chris Lattner
6547451f32 Solaris doesn't have MAP_FILE.
llvm-svn: 16682
2004-10-05 00:46:21 +00:00
Reid Spencer
079b225788 Excise the ill-advised RLCOMP compression algorithm and simply leave the
previously temporary NULLCOMP implementation that merely copies the data
verbatim without compression. Also, don't warn if there's no compression
library as that is taken care of during configuration time.

llvm-svn: 16654
2004-10-04 17:45:44 +00:00
Reid Spencer
49089d64c2 Add a context for the callback so different compression scenarios can be
distinguished. Tidy up documentation.  Thanks, Chris.

llvm-svn: 16652
2004-10-04 17:29:25 +00:00
Chris Lattner
9f6c72d660 Fix build if not HAVE_BZIP2
llvm-svn: 16650
2004-10-04 16:33:25 +00:00
Reid Spencer
da2e8b9943 First version of the MappedFile abstraction for operating system idependent
mapping of files. This first version uses mmap where its available. The
class needs to implement an alternate mechanism based on malloc'd memory
and file reading/writing for platforms without virtual memory.

llvm-svn: 16649
2004-10-04 11:08:32 +00:00
Reid Spencer
d2bedc512d First version of a support utility to provide generalized compression in
LLVM that handles availability and unavailability of bzip2 and zlib.

llvm-svn: 16648
2004-10-04 10:49:41 +00:00
Chris Lattner
0228f228df * Prune #includes
* Update comments
* Rearrange code a bit
* Finally ELIMINATE the GAS workaround emitter for Intel mode.  woot!

llvm-svn: 16647
2004-10-04 07:31:08 +00:00
Chris Lattner
581948c8f6 Add support for emitting AT&T style .s files, and make it the default. Users
may now choose their output format with the -x86-asm-syntax={intel|att} flag.

llvm-svn: 16646
2004-10-04 07:24:48 +00:00
Chris Lattner
5959f4a108 Convert some missed patterns to support AT&T style
llvm-svn: 16645
2004-10-04 07:23:07 +00:00
Chris Lattner
a05d9f53bb Apparently the GNU assembler has a HUGE hack to be compatible with really
old and broken AT&T syntax assemblers.  The problem with this hack is that
*SOME* forms of the fdiv and fsub instructions have the 'r' bit inverted.
This was a real pain to figure out, but is trivially easy to support: thus
we are now bug compatible with gas and gcc.

llvm-svn: 16644
2004-10-04 07:08:46 +00:00
Chris Lattner
08098895db Fix incorrect suffix
llvm-svn: 16642
2004-10-04 05:20:16 +00:00
Chris Lattner
c2fc9597bd Fix some more missed suffixes and swapped operands
llvm-svn: 16641
2004-10-04 01:38:10 +00:00
Chris Lattner
7b15a84728 Add missing suffixes to FP instructions for AT&T mode
llvm-svn: 16640
2004-10-04 00:43:31 +00:00
Chris Lattner
8d44dcca97 Add support for the -x86-asm-syntax flag, which can be used to choose between
Intel and AT&T style assembly language.  The ultimate goal of this is to
eliminate the GasBugWorkaroundEmitter class, but for now AT&T style emission
is not fully operational.

llvm-svn: 16639
2004-10-03 20:36:57 +00:00
Chris Lattner
94780713a8 Add support to the instruction patterns for AT&T style output, which will
hopefully lead to the death of the 'GasBugWorkaroundEmitter'.  This also
includes changes to wrap the whole file to 80 columns! Woot! :)

Note that the AT&T style output has not been tested at all.

llvm-svn: 16638
2004-10-03 20:35:00 +00:00
Chris Lattner
30b5b79aa0 Add initial support for variants
llvm-svn: 16635
2004-10-03 19:34:18 +00:00
Chris Lattner
815b635639 Do not repeat the map lookup
llvm-svn: 16633
2004-10-01 23:16:43 +00:00
Chris Lattner
c521544a32 When a virtual register is folded into an instruction, keep track of whether
it was a use, def, or both.  This allows us to be less pessimistic in our
analysis of them.  In practice, this doesn't make a big difference, but it
doesn't hurt either.

llvm-svn: 16632
2004-10-01 23:15:36 +00:00
Chris Lattner
38467b8a66 Add a simple little improvement to the local spiller to keep track of stores
and delete them if they turn out to be dead.  This is a useful little hack
that even speeds up some programs.  For example, it speeds up Ptrdist/ks
from 17.53s to 15.59s, and 188.ammp from 149s to 146s.

This also speeds up llc :)

llvm-svn: 16630
2004-10-01 19:47:12 +00:00
Chris Lattner
8a5b40154f Substantially revamp the local spiller, causing it to actually improve the
generated code over the simple spiller.  The new local spiller generates
substantially better code than the simple one in some cases, by reusing
values that are loaded out of stack slots and kept available in registers.

This primarily helps programs that are spilling a lot, and there is still
stuff that can be done to improve it.  This patch makes the local spiller
the default, as it's only a tiny bit slower than the simple spiller (it
increases the runtime of llc by < 1%).

Here are some numbers with speedups.

Program    #reuse  old(s)    new(s)  Speedup

Povray:     3452,  16.87 ->  15.93   (5.5%)
177.mesa:   2176,   2.77 ->   2.76   (0%)
179.art:      35,  28.43 ->  28.01   (1.5%)
183.equake:   55,  61.44 ->  61.41   (0%)
188.ammp:    869, 174    -> 149      (15%)

164.gzip:     43,  40.73 ->  40.71   (0%)
175.vpr:     351,  18.54 ->  17.34   (6.5%)
176.gcc:    2471,   5.01 ->   4.92   (1.8%)
181.mcf       42,  79.30 ->  75.20   (5.2%)
186.crafty:  484,  29.73 ->  30.04   (-1%)
197.parser:  251,  10.47 ->  10.67   (-1%)
252.eon:    1501,   1.98 ->   1.75   (12%)
253.perlbm: 1183,  14.83 ->  14.42   (2.8%)
254.gap:     825,   7.46 ->   7.29   (2.3%)
255.vortex:  285,  10.51 ->  10.27   (2.3%)
256.bzip2:    63,  55.70 ->  55.20   (0.9%)
300.twolf:   830,  21.63 ->  22.00   (-1%)

PtrDist/ks    14,  32.75 -> 17.53    (46.5%)
Olden/tsp     46,   8.71 ->  8.24    (5.4%)
Free/distray  70,   1.09 ->  0.99    (9.2%)

llvm-svn: 16629
2004-10-01 19:04:51 +00:00
Chris Lattner
2ddd74fa8b Pretty print a bit nicer :)
llvm-svn: 16628
2004-10-01 19:01:39 +00:00
Alkis Evlogimenos
8e4ed30dcb Document this class a bit :-)
llvm-svn: 16626
2004-10-01 00:35:07 +00:00
Brian Gaeke
ee0dfdd1e1 Make EmitMappingInfo into an "external location" option, so that it can be set
or cleared externally.

llvm-svn: 16623
2004-09-30 20:20:01 +00:00
Brian Gaeke
ab7dd80200 Add accessor function.
llvm-svn: 16622
2004-09-30 20:14:29 +00:00
Brian Gaeke
65540b3e58 Correct type of accessor functions.
llvm-svn: 16621
2004-09-30 20:14:18 +00:00
Brian Gaeke
90a286872c Namespacify. Add accessor function.
llvm-svn: 16620
2004-09-30 20:14:07 +00:00
Brian Gaeke
c667e351ed I think this will handle double args.
llvm-svn: 16618
2004-09-30 19:44:32 +00:00
Misha Brukman
8a43086705 That should actually be __MING, not __MINGW.
llvm-svn: 16617
2004-09-30 18:28:07 +00:00
Misha Brukman
549b3533be Check for __MINGW define instead of __MINGW_H, patch by Henrik Bach.
llvm-svn: 16615
2004-09-30 18:24:58 +00:00
Chris Lattner
db2a0987cc Use more efficient map operations. Fix a bug that would affect hypothetical
targets that supported multiple memory operands.

llvm-svn: 16614
2004-09-30 16:35:08 +00:00
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
Brian Gaeke
0d4d060dbd Mark the instructions that have delay slots with the hasDelaySlot flag.
Add some comments.

llvm-svn: 16611
2004-09-30 04:04:48 +00:00
Brian Gaeke
738005408e Use TargetMachine::hasDelaySlot() instead of our old switch statement
to find instrs that have delay slots.

llvm-svn: 16610
2004-09-30 04:04:47 +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
Chris Lattner
af68e9a012 Disable the 'WARNING: Found global types that are not compatible' warning
that always prints when linking programs to libstdc++ :(

llvm-svn: 16603
2004-09-30 00:12:29 +00:00
Misha Brukman
7c87c6b215 Change the #ifdefs to allow compilation with a V8 compiler, but the JIT still
won't work if not compiled in V9 mode, currently by GCC only, because Sun's
system compiler does not tell us if it's a V8 or V9 system.

llvm-svn: 16602
2004-09-29 23:01:17 +00:00
Chris Lattner
8341306cba Hrm, debugging printouts do not need to be in here
llvm-svn: 16598
2004-09-29 21:21:14 +00:00
Chris Lattner
7af052f5c8 Make sure to check select instructions for generic instruction properties
llvm-svn: 16597
2004-09-29 21:19:28 +00:00
Brian Gaeke
ee2dab29f9 Update list of shootout programs that should be working.
llvm-svn: 16595
2004-09-29 20:45:06 +00:00
Brian Gaeke
d4c38d8c4b Tell the target description that calls clobber registers O0...O5.
llvm-svn: 16594
2004-09-29 20:45:05 +00:00
Chris Lattner
da5c635143 Don't use DominatorSet::dominates for intra-block instruction dom checks.
This method is linear time in the size of the basic block, which is very
bad for large basic blocks.  On the Assembler/2004-09-29-VerifierIsReallySlow.llx
testcase, the verifier changes from taking 50s to 0.23s with this patch.

llvm-svn: 16593
2004-09-29 20:07:45 +00:00
Brian Gaeke
5bca3b8e5f FITOD is spelled "fitod", not "fitos". Ouch.
llvm-svn: 16591
2004-09-29 19:59:07 +00:00
Brian Gaeke
e9aabee09d Don't use .quad to output double constants. The assembler must have a bug or
something, because the wrong bit patterns get output.

llvm-svn: 16590
2004-09-29 19:59:06 +00:00
Chris Lattner
79ceb6ba53 * 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.

llvm-svn: 16588
2004-09-29 17:40:11 +00:00
Brian Gaeke
b8237dfbb9 Recognize FpMOVD as a move.
llvm-svn: 16586
2004-09-29 16:45:47 +00:00
Chris Lattner
778a49acfd Do not insert trivially dead select instructions, which allows us to
potentially fold more in one pass.

llvm-svn: 16583
2004-09-29 05:43:32 +00:00