LLVM 2.9 RC1 Pre-release Tarballs

There are LLVM 2.9 RC1 pre-release tarballs source available. You can find them here:

  http://llvm.org/pre-releases/2.9/

Please download them, build them, and compile things to your heart's content. And most importantly file a bunch of bug reports. :slight_smile:

Share and enjoy!
-bw

Your clang-2.9rc1.src.tar.gz is bad, it should named to clang-2.9rc1.src.tar.gz.gz !!!

oracle@vc:~/tmp$ gzip -cd clang-2.9rc1.src.tar.gz > clang-2.9rc1.src.tar
oracle@vc:~/tmp$ file clang-2.9rc1.src.tar
clang-2.9rc1.src.tar: gzip compressed data, from Unix, last modified: Wed Mar 9 09:38:38 2011
oracle@vc:~/tmp$ gzip -cd clang-2.9rc1.src.tar > clang-2.9rc1.src.tar2
oracle@vc:~/tmp$ tar xf clang-2.9rc1.src.tar2

Your clang-2.9rc1.src.tar.gz is bad, it should named to clang-2.9rc1.src.tar.gz.gz !!!

oracle@vc:~/tmp$ gzip -cd clang-2.9rc1.src.tar.gz > clang-2.9rc1.src.tar
oracle@vc:~/tmp$ file clang-2.9rc1.src.tar
clang-2.9rc1.src.tar: gzip compressed data, from Unix, last modified: Wed Mar 9 09:38:38 2011
oracle@vc:~/tmp$ gzip -cd clang-2.9rc1.src.tar > clang-2.9rc1.src.tar2
oracle@vc:~/tmp$ tar xf clang-2.9rc1.src.tar2

Test on mingw64-trunk (i686-w64-mingw32, 4.5.3 20110308):

Expected Passes : 5114
Expected Failures : 48
Unsupported Tests : 547
Unexpected Failures: 130

The following failures is very strange:

C:\var\pool\llvm-2.9rc1\unittests\Support\raw_ostream_test.cpp:100: Failure
Value of: printToStringUnbuffered(1.1)
Actual: “1.100000e+000”
Expected: “1.100000e+00”
[ FAILED ] raw_ostreamTest.Types_Unbuffered (0 ms)

C:\var\pool\llvm-2.9rc1\unittests\Support\raw_ostream_test.cpp:69: Failure
Value of: printToString(1.1)
Actual: “1.100000e+000”
Expected: “1.100000e+00”
[ FAILED ] raw_ostreamTest.Types_Buffered (0 ms)
[----------] 1 test from raw_ostreamTest (0 ms total)

See Bug 6745 - .ll output is different on windows.
http://llvm.org/bugs/show_bug.cgi?id=6745
On mingw32, PRINTF_EXPONENT_DIGITS affects printf and tests would run
with PRINTF_EXPONENT_DIGITS=2.

An addition, it seems mingw-w64's printf tends to drop minus sign of "-0.0".
(is it same on i686? is it fixed in trunk?)

...Takumi

Yes and Not.

llvm 2.9rc1 test on Dualcore ARM running Ubuntu Natty

gunzip llvm-2.9rc1.src.tar.gz
tar zxvf llvm-2.9rc1.src.tar
cd llvm-2.9-build
../llvm-2.9rc1/configure --enable-shared
time make clean ;time make CXXFLAGS=-marm CFLAGS=-marm

xranby@panda:/media/dh0/llvm-2.9-build$ time make check

Failing Tests (64):
    LLVM :: CodeGen/CellSPU/2009-01-01-BrCond.ll
    LLVM :: CodeGen/CellSPU/call_indirect.ll
    LLVM :: CodeGen/CellSPU/extract_elt.ll
    LLVM :: CodeGen/CellSPU/fcmp64.ll
    LLVM :: CodeGen/CellSPU/fneg-fabs.ll
    LLVM :: CodeGen/CellSPU/i64ops.ll
    LLVM :: CodeGen/CellSPU/immed64.ll
    LLVM :: CodeGen/CellSPU/private.ll
    LLVM :: CodeGen/CellSPU/rotate_ops.ll
    LLVM :: CodeGen/CellSPU/sext128.ll
    LLVM :: CodeGen/CellSPU/shuffles.ll
    LLVM :: CodeGen/CellSPU/stores.ll
    LLVM :: CodeGen/CellSPU/struct_1.ll
    LLVM :: CodeGen/CellSPU/trunc.ll
    LLVM :: CodeGen/CellSPU/v2f32.ll
    LLVM :: CodeGen/CellSPU/v2i32.ll
    LLVM :: CodeGen/CellSPU/vec_const.ll
    LLVM :: CodeGen/CellSPU/vecinsert.ll
    LLVM :: CodeGen/MSP430/2009-05-10-CyclicDAG.ll
    LLVM :: CodeGen/MSP430/2009-05-17-Rot.ll
    LLVM :: CodeGen/MSP430/2009-05-17-Shift.ll
    LLVM :: CodeGen/MSP430/2009-05-19-DoubleSplit.ll
    LLVM :: CodeGen/MSP430/2009-08-25-DynamicStackAlloc.ll
    LLVM :: CodeGen/MSP430/2009-09-18-AbsoluteAddr.ll
    LLVM :: CodeGen/MSP430/2009-10-10-OrImpDef.ll
    LLVM :: CodeGen/MSP430/2009-11-08-InvalidResNo.ll
    LLVM :: CodeGen/MSP430/2009-11-20-NewNode.ll
    LLVM :: CodeGen/MSP430/2009-12-21-FrameAddr.ll
    LLVM :: CodeGen/MSP430/AddrMode-bis-rx.ll
    LLVM :: CodeGen/MSP430/AddrMode-bis-xr.ll
    LLVM :: CodeGen/MSP430/AddrMode-mov-rx.ll
    LLVM :: CodeGen/MSP430/AddrMode-mov-xr.ll
    LLVM :: CodeGen/MSP430/Inst16mi.ll
    LLVM :: CodeGen/MSP430/Inst16mm.ll
    LLVM :: CodeGen/MSP430/Inst16mr.ll
    LLVM :: CodeGen/MSP430/Inst16rm.ll
    LLVM :: CodeGen/MSP430/Inst8mi.ll
    LLVM :: CodeGen/MSP430/Inst8mm.ll
    LLVM :: CodeGen/MSP430/Inst8mr.ll
    LLVM :: CodeGen/MSP430/Inst8rm.ll
    LLVM :: CodeGen/MSP430/bit.ll
    LLVM :: CodeGen/MSP430/indirectbr.ll
    LLVM :: CodeGen/MSP430/indirectbr2.ll
    LLVM :: CodeGen/MSP430/inline-asm.ll
    LLVM :: CodeGen/MSP430/mult-alt-generic-msp430.ll
    LLVM :: CodeGen/Mips/2008-07-03-SRet.ll
    LLVM :: CodeGen/Mips/2008-07-05-ByVal.ll
    LLVM :: CodeGen/Mips/2008-07-15-SmallSection.ll
    LLVM :: CodeGen/Mips/2008-08-03-ReturnDouble.ll
    LLVM :: CodeGen/Mips/2008-10-13-LegalizerBug.ll
    LLVM :: CodeGen/Mips/2008-11-10-xint_to_fp.ll
    LLVM :: CodeGen/Mips/2010-07-20-Switch.ll
    LLVM :: CodeGen/Mips/blockaddr.ll
    LLVM :: CodeGen/Mips/cmov.ll
    LLVM :: CodeGen/Mips/divrem.ll
    LLVM :: CodeGen/Mips/private.ll
    LLVM :: CodeGen/SPARC/2007-07-05-LiveIntervalAssert.ll
    LLVM :: CodeGen/SPARC/2009-08-28-PIC.ll
    LLVM :: CodeGen/SPARC/2011-01-11-CC.ll
    LLVM :: CodeGen/SPARC/2011-01-19-DelaySlot.ll
    LLVM :: CodeGen/SPARC/2011-01-22-SRet.ll
    LLVM :: CodeGen/SPARC/mult-alt-generic-sparc.ll
    LLVM :: CodeGen/Thumb/select.ll
    LLVM :: CodeGen/X86/fold-pcmpeqd-0.ll

  Expected Passes : 5191
  Expected Failures : 50
  Unsupported Tests : 543
  Unexpected Failures: 64

llvm 2.9rc1 test on Dualcore ARM running Ubuntu Natty.txt (97.2 KB)

Hello Xerxes,

llvm 2.9rc1 test on Dualcore ARM running Ubuntu Natty

What is the gcc used for the compilation? Can you try to do the -O0
build and see whether this changed the stuff?

xranby@panda:/media/dh0/llvm-2.9-build-O0$ gcc --version
gcc (Ubuntu/Linaro 4.5.2-5ubuntu1) 4.5.2
Copyright (C) 2010 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

-O0 did change quite a lot see below:

xranby@panda:/media/dh0/llvm-2.9-build-O0$ time make check
llvm[0]: Running test suite
make[1]: Entering directory `/media/dh0/llvm-2.9-build-O0/test'
Making a new site.exp file...
Making LLVM 'lit.site.cfg' file...
Making LLVM unittest 'lit.site.cfg' file...
( ulimit -t 600 ; ulimit -d 512000 ; ulimit -m 512000 ; ulimit -v 1024000 ; \
      /media/dh0/llvm-2.9rc1/utils/lit/lit.py -s -v . )
FAIL: LLVM :: CodeGen/Thumb/select.ll (1595 of 5848)
******************** TEST 'LLVM :: CodeGen/Thumb/select.ll' FAILED ********************
Script:

Hello Xerxes,

xranby@panda:/media/dh0/llvm-2.9-build-O0$ gcc --version
gcc (Ubuntu/Linaro 4.5.2-5ubuntu1) 4.5.2

Ok. Looks like we'll need to mark this gcc as "known bad" for ARM.

Hmm...This might be missing a flag? But Eric was the last to touch it (because of his ILP stuff). Eric, do you have any comments?

-bw

Should be fixed I think on the branch.

-eric