maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
12,490 Commits 23 Branches 0 Tags 2.2 GiB
Commit Graph

12490 Commits

Author SHA1 Message Date
Chris Lattner ee59d4bf04 Fix a bug in my checkin from last night that caused miscompilations of 
186.crafty, fhourstones and 132.ijpeg.

Bugpoint makes really nasty miscompilations embarassingly easy to find.  It
narrowed it down to the instcombiner and this testcase (from fhourstones):

bool %l7153_l4706_htstat_loopentry_2E_4_no_exit_2E_4(int* %i, [32 x int]* %works, int* %tmp.98.out) {
newFuncRoot:
        %tmp.96 = load int* %i          ; <int> [#uses=1]
        %tmp.97 = getelementptr [32 x int]* %works, long 0, int %tmp.96         ; <int*> [#uses=1]
        %tmp.98 = load int* %tmp.97             ; <int> [#uses=2]
        %tmp.99 = load int* %i          ; <int> [#uses=1]
        %tmp.100 = and int %tmp.99, 7           ; <int> [#uses=1]
        %tmp.101 = seteq int %tmp.100, 7                ; <bool> [#uses=2]
        %tmp.102 = cast bool %tmp.101 to int            ; <int> [#uses=0]
        br bool %tmp.101, label %codeRepl4.exitStub, label %codeRepl3.exitStub

codeRepl4.exitStub:             ; preds = %newFuncRoot
        store int %tmp.98, int* %tmp.98.out
        ret bool true

codeRepl3.exitStub:             ; preds = %newFuncRoot
        store int %tmp.98, int* %tmp.98.out
        ret bool false
}

... which only has one combination performed on it:

$ llvm-as < t.ll | opt -instcombine -debug | llvm-dis
IC: Old =       %tmp.101 = seteq int %tmp.100, 7                ; <bool> [#uses=1]
    New =       setne int %tmp.100, 0           ; <bool>:<badref> [#uses=0]
IC: MOD =       br bool %tmp.101, label %codeRepl3.exitStub, label %codeRepl4.exitStub
IC: MOD =       %tmp.97 = getelementptr [32 x int]* %works, uint 0, int %tmp.96         ; <int*> [#uses=1]

It doesn't get much better than this.  :)

llvm-svn: 14109
 2004-06-10 02:33:20 +00:00
Chris Lattner c8e7e298c1 More minor cleanups 
llvm-svn: 14108
 2004-06-10 02:12:35 +00:00
Chris Lattner df20a4d589 Eliminate many occurrances of Instruction:: 
llvm-svn: 14107
 2004-06-10 02:07:29 +00:00
Chris Lattner 88ba175c1e Share some code 
llvm-svn: 14106
 2004-06-10 01:57:38 +00:00
Chris Lattner 867653ad6a Add new BinaryOperator::createAdd/Sub/... methods to avoid having to type 
llvm::Instruction:: all of the time.

llvm-svn: 14105
 2004-06-10 01:43:29 +00:00
Chris Lattner 68a038e6a1 Tolerate more errors 
llvm-svn: 14104
 2004-06-09 22:22:10 +00:00
Brian Gaeke 5b03a0628f Encode %fsr correctly; don't fail an assertion. 
llvm-svn: 14103
 2004-06-09 21:54:59 +00:00
Brian Gaeke 87d8878f6d Fix encoding of ST*FSR instructions. 
llvm-svn: 14102
 2004-06-09 21:54:58 +00:00
Brian Gaeke 87c75632ee Fix assertion failure message to have the right method name. 
llvm-svn: 14101
 2004-06-09 20:44:42 +00:00
Chris Lattner 757ee0b533 Make the asmwriter much more tolerant of errors (which are common when working 
on new front-ends and stuff).  Also get rid of some tabs that snuck in.

llvm-svn: 14100
 2004-06-09 19:41:19 +00:00
Reid Spencer f8d610ca73 Columnized the compilation statistics for easier reading. 
llvm-svn: 14099
 2004-06-09 18:29:15 +00:00
Chris Lattner 4408630c67 Fix the really bizarre stuff that happened last night in the tester 
due to non-numeric diff failures that caused fpcmp to go into infinite loops

llvm-svn: 14098
 2004-06-09 18:28:53 +00:00
Reid Spencer 8beac69819 Made it possible for the printInfoComment method to invoke getSlot in 
such a way that if the Value being printed is standalone that we don't
assert and abort but just print ":??" for the slot number instead.

llvm-svn: 14097
 2004-06-09 15:26:53 +00:00
John Criswell 9095c64147 Fix for PR#366. We use getClassB() so that we can handle cast instructions 
that cast to bool.

llvm-svn: 14096
 2004-06-09 15:18:51 +00:00
Chris Lattner 35167c3087 Implement InstCombine/select.ll:test15* 
llvm-svn: 14095
 2004-06-09 07:59:58 +00:00
Chris Lattner ce527c8e6d More instcombine testcases 
llvm-svn: 14094
 2004-06-09 07:59:40 +00:00
Reid Spencer 2acee7e4d6 Cleanup alignment of output. 
llvm-svn: 14093
 2004-06-09 06:22:00 +00:00
Reid Spencer 1e47e066ad Add some new fields for bytecode analysis. 
llvm-svn: 14092
 2004-06-09 06:18:53 +00:00
Reid Spencer cc4e35a404 Implement analysis output. Don't dump function details unless requested. 
llvm-svn: 14091
 2004-06-09 06:17:58 +00:00
Reid Spencer 839fa681f6 Implement some rudimentary analysis. 
llvm-svn: 14090
 2004-06-09 06:16:43 +00:00
Reid Spencer f1598b0efa Clean up indentation of file body output. 
Don't dump functions unless "detailedResults" is requested.

llvm-svn: 14089
 2004-06-09 06:16:19 +00:00
Reid Spencer fc6f5508aa Move parsing details to Parser.cpp. 
llvm-svn: 14088
 2004-06-09 06:15:21 +00:00
Reid Spencer 670c26d3af Bring some things out of header files that belong only in this file. 
llvm-svn: 14087
 2004-06-09 06:14:52 +00:00
Chris Lattner 396dbfe327 Be more careful about the order we put stuff onto the worklist. This allow us to 
collapse this:
bool %le(int %A, int %B) {
        %c1 = setgt int %A, %B
        %tmp = select bool %c1, int 1, int 0
        %c2 = setlt int %A, %B
        %result = select bool %c2, int -1, int %tmp
        %c3 = setle int %result, 0
        ret bool %c3
}

into:

bool %le(int %A, int %B) {
        %c3 = setle int %A, %B          ; <bool> [#uses=1]
        ret bool %c3
}

which is handy, because the Java FE makes these sequences all over the place.

This is tested as: test/Regression/Transforms/InstCombine/JavaCompare.ll

llvm-svn: 14086
 2004-06-09 05:08:07 +00:00
Chris Lattner 7ed7d82512 New testcase that is important for the Java FE 
llvm-svn: 14085
 2004-06-09 05:07:01 +00:00
Reid Spencer e73febc64a Cleaned up a dead header file to prevent duplicate definition warnings 
in doxygen.

llvm-svn: 14084
 2004-06-09 04:38:34 +00:00
Chris Lattner 2dd017402b Implement select.ll:test14* 
llvm-svn: 14083
 2004-06-09 04:24:29 +00:00
Chris Lattner 52d7430fc5 New testcase for the instruction combiner that happen often in the Java FE 
llvm-svn: 14082
 2004-06-09 04:23:57 +00:00
Chris Lattner 0c09852909 Minor tweaks 
llvm-svn: 14081
 2004-06-09 03:59:05 +00:00
Reid Spencer 4d598edc59 Regularize title of Regression Test section. Clean up some blank space. 
llvm-svn: 14080
 2004-06-09 02:07:25 +00:00
Chris Lattner 4769612be1 Bug fixed 
llvm-svn: 14079
 2004-06-09 01:09:16 +00:00
Chris Lattner 74248512eb Workaround or a VS miscompilation bug 
llvm-svn: 14078
 2004-06-08 23:21:39 +00:00
Chris Lattner 8b52a68356 I checked and no clients expect this to return null for unconditional branches 
Simplify code and make it more uniform.

llvm-svn: 14077
 2004-06-08 22:03:05 +00:00
Chris Lattner 4c7a24af44 Don't grab the condition of unconditional branches! 
This fixes PR363

llvm-svn: 14076
 2004-06-08 21:50:30 +00:00
Brian Gaeke c365e7cc99 Fix a minor bug in the map - since this pass adds a global symbol, it must be 
accounted for in the map (at least, in its current format).

llvm-svn: 14075
 2004-06-08 20:08:30 +00:00
Brian Gaeke 78d03b523f Add a forwarding method pop_front() that allows you to delete instructions 
from the beginning of a MBB.

llvm-svn: 14074
 2004-06-08 18:52:47 +00:00
Brian Gaeke 76b6426a8c Add a TmpInstruction ctor that doesn't take a MCFI. 
llvm-svn: 14073
 2004-06-08 18:52:46 +00:00
Brian Gaeke 223b345279 Explicitly specify libtool tag "CXX" so that if you setenv CXX to something 
libtool can't parse, e.g., "/path/to/g++ -some-funny-options",
then it will still be able to compile and link.

llvm-svn: 14072
 2004-06-08 18:52:45 +00:00
Chris Lattner a20f4aab7f Fix a link error using VS8.0 
llvm-svn: 14071
 2004-06-08 17:53:24 +00:00
Chris Lattner 4409487746 Apparently a particular vendor compiler uses the struct/class tag to MANGLE 
symbols with.  Therefore, if you do not use struct/class consistently, you can
get LINK ERRORS.  grr.

This fixes the link errors for libsupport and vmcore.

-Chris

llvm-svn: 14070
 2004-06-08 17:44:21 +00:00
Misha Brukman 23c2f67fda Fix case of doxygen directive `\p': it's lowercase. 
llvm-svn: 14068
 2004-06-08 13:49:17 +00:00
Reid Spencer 7d4b814e86 * Fix indentation of Regression Test Results by making sure the 
GetQMTestResults function closes all its open tags.
* Cause XFAIL results to not be reported at all except in the stats.

llvm-svn: 14067
 2004-06-08 08:01:33 +00:00
Reid Spencer b14a0a69fc Put in a place holder for describing the differences in bytecode format 
between LLVM versions. This is just a reminder so I don't forget to
document it.

llvm-svn: 14066
 2004-06-08 07:41:41 +00:00
Reid Spencer 850f85f81f Add a note about llvm-abcd, the Analysis of ByteCode Dumper 
llvm-svn: 14065
 2004-06-08 07:30:31 +00:00
Chris Lattner bd948cfe38 Ooh, that fixed the annoying warning!! 
llvm-svn: 14064
 2004-06-08 07:19:29 +00:00
Chris Lattner 25b3a03e17 Bug fixed 
llvm-svn: 14063
 2004-06-08 07:15:52 +00:00
Chris Lattner 79dd360555 New testcase for PR355 
llvm-svn: 14062
 2004-06-08 07:11:14 +00:00
Chris Lattner 12eb60f213 Testcase for PR355 
llvm-svn: 14061
 2004-06-08 07:10:11 +00:00
Chris Lattner 522ef67fb2 Add documentation to the TargetFrameInfo class, contributed by Vladimir Prus 
llvm-svn: 14060
 2004-06-08 06:23:17 +00:00
Reid Spencer ddc6fb125e Clean up the documentation. 
Simplify the file handling. It now only writes to std::cout.

llvm-svn: 14059
 2004-06-08 05:56:58 +00:00