[cfe-dev] choosing the gcc toolchain
Reed Kotler
rkotler at mips.com
Sun Mar 2 18:33:43 PST 2014
With this latest patch I'm seeing:
rkotler at mipsswbrd006-le:~/caviumnativellvm/build$
./Release+Asserts/bin/clang++ -v
clang version 3.5 (gitosis at dmz-portal.mips.com:clang.git
f4481e0e412e3f817e65d5ce1c835381dc7a7773)
(gitosis at dmz-portal.mips.com:llvm.git
18fe44cb057e4b11533bd3f814b96fdca724baf0)
Target: mips64el-unknown-linux-gnu
Thread model: posix
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.4
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.4.6
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.4.7
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.6
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.6.3
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.7
Found candidate GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.7.2
Selected GCC installation: /usr/lib/gcc/mipsel-linux-gnu/4.6
Candidate multilib: .;@m32
Candidate multilib: 64;@m64
Candidate multilib: n32;@mabi=n32
Selected multilib: 64;@m64
rkotler at mipsswbrd006-le:~/caviumnativellvm/build$
The day before I was getting 4.7
On 03/02/2014 06:20 PM, Reed Kotler wrote:
> I've been testing this part on little endian and there are currently
> some additional "make check" issues on little endian no present on big
> endian.
>
> The little endian native compiler without your patch gets:
>
> ********************
> Testing Time: 143.31s
> ********************
> Failing Tests (7):
> LLVM :: CodeGen/Generic/2009-03-17-LSR-APInt.ll
> LLVM :: CodeGen/Generic/2011-07-07-ScheduleDAGCrash.ll
> LLVM :: CodeGen/Generic/add-with-overflow-128.ll
> LLVM :: CodeGen/Generic/i128-addsub.ll
> LLVM :: DebugInfo/2010-10-01-crash.ll
> LLVM-Unit ::
> ADT/Release+Asserts/ADTTests/HashingTest.HashCombineBasicTest
> LLVM-Unit :: ADT/Release+Asserts/ADTTests/HashingTest.HashValueStdPair
>
> The main regression we have been seeing are these two LLVM-Unit tests,
> since switching to gcc 4.7 for the build but those tests did not fail
> with the clang/llvm based build when we used your first patch.
>
>
> On 03/02/2014 03:36 PM, reed kotler wrote:
>> On 03/02/2014 01:37 AM, Simon Atanasyan wrote:
>>> Here is the fix:
>>> http://llvm-reviews.chandlerc.com/D2911
>>>
>>> On Sat, Mar 1, 2014 at 1:57 AM, reed kotler
>>> <rkotler-8NJIiSa5LzA-XMD5yJDbdMReXY1tMh2IBg at public.gmane.org> wrote:
>>>> So what do you propose?
>> I will try this patch tonight.
>>
>> I will try the new patch tonight.
>>
>> The one the night before had some new errors but it was still not
>> chosing the right toolchain as we know.
>>
>> The good news is that with clang/llvm we eliminated the errors:
>> LLVM-Unit ::
>> ADT/Release+Asserts/ADTTests/HashingTest.HashCombineBasicTest
>> LLVM-Unit ::
>> ADT/Release+Asserts/ADTTests/HashingTest.HashValueStdPair
>>
>>
>> In particular, check mips-cs.cpp and mips-fs.cpp.
>>
>> When built with gcc 4.7:
>>
>> Failing Tests (11):
>> Clang :: Analysis/crash-trace.c
>> Clang :: Driver/mips-cs.cpp
>> Clang :: Driver/mips-fsf.cpp
>> Clang :: Parser/crash-report.c
>> LLVM :: CodeGen/Generic/2009-03-17-LSR-APInt.ll
>> LLVM :: CodeGen/Generic/2011-07-07-ScheduleDAGCrash.ll
>> LLVM :: CodeGen/Generic/add-with-overflow-128.ll
>> LLVM :: CodeGen/Generic/i128-addsub.ll
>> LLVM :: DebugInfo/2010-10-01-crash.ll
>> LLVM-Unit ::
>> ADT/Release+Asserts/ADTTests/HashingTest.HashCombineBasicTest
>> LLVM-Unit ::
>> ADT/Release+Asserts/ADTTests/HashingTest.HashValueStdPair
>>
>> Expected Passes : 16716
>> Expected Failures : 90
>> Unsupported Tests : 41
>> Unexpected Failures: 11
>> make[1]: *** [check-local-all] Error 1
>> make[1]: Leaving directory `/home/rkotler/caviumnativellvm/build/test'
>> make: *** [check-all] Error 2
>> rkotler at mipsswbrd006-le:~/caviumnativellvm/build$
>>
>> When built with recent clang/llvm:
>>
>> Failing Tests (9):
>> Clang :: Analysis/crash-trace.c
>> Clang :: Driver/mips-cs.cpp
>> Clang :: Driver/mips-fsf.cpp
>> Clang :: Parser/crash-report.c
>> LLVM :: CodeGen/Generic/2009-03-17-LSR-APInt.ll
>> LLVM :: CodeGen/Generic/2011-07-07-ScheduleDAGCrash.ll
>> LLVM :: CodeGen/Generic/add-with-overflow-128.ll
>> LLVM :: CodeGen/Generic/i128-addsub.ll
>> LLVM :: DebugInfo/2010-10-01-crash.ll
More information about the cfe-dev
mailing list