[cfe-dev] compilation regression, "fatal error: error in backend: Cannot select..."

Stuart Hastings stuart at apple.com
Thu Jun 2 08:33:35 PDT 2011


Yes, this is almost certainly my fault.

Sean, could you advise me how to extract a test case to reproduce this?

Thank you,

stuart

On Jun 2, 2011, at 8:05 AM, Evan Cheng wrote:

> Looks like it's caused by Stuart's recent change. 
> 
> Evan
> 
> On Jun 2, 2011, at 7:51 AM, Sean McBride <sean at rogue-research.com> wrote:
> 
>> Hello llvm folks,
>> 
>> I use clang to build a few open source projects nightly.  I update my
>> build of clang periodically, and yesterday updated from r130865 to
>> r132417.  As a result, the vxl project no longer compiles, with this error:
>> 
>> <http://www.cdash.org/CDash/viewBuildError.php?buildid=1189394>
>> 
>> fatal error: error in backend: Cannot select: 0x1036ddc10: f32 =
>> X86ISD::FSETCCss 0x10371ec10, 0x1036dd310, 0x10372c910 [ID=26] dbg:/usr/
>> include/c++/4.2.1/complex:450:7
>> 0x10371ec10: f80,ch = load 0x103698a10:1, 0x103698a10,
>> 0x10373ab10<LD10[%12](align=16)> [ORD=696] [ID=18] dbg:/usr/include/c++/
>> 4.2.1/complex:450:7
>>   0x103698a10: i64,ch,glue = CopyFromReg 0x1036ddb10, 0x1036dd810,
>> 0x1036ddb10:1 [ORD=694] [ID=16] dbg:/usr/include/c++/4.2.1/complex:450:7
>>     0x1036dd810: i64 = Register %RAX [ORD=694] [ID=6]
>>     0x1036ddb10: ch,glue = callseq_end 0x103698e10, 0x103698910,
>> 0x103698910, 0x103698e10:1 [ORD=694] [ID=15]
>>       0x103698910: i64 = TargetConstant<0> [ORD=694] [ID=3]
>>       0x103698910: i64 = TargetConstant<0> [ORD=694] [ID=3]
>>       0x103698e10: ch,glue = X86ISD::CALL 0x103699610, 0x10371ef10,
>> 0x10372c210, 0x103699610:1 [ORD=694] [ID=14] dbg:/usr/include/c++/4.2.1/
>> complex:450:7
>>         0x10371ef10: i64 = TargetGlobalAddress<x86_fp80*
>> (%"struct.std::complex"*)* @_ZNKSt7complexIeE4imagEv> 0 [ORD=694] [ID=5]
>> dbg:/usr/include/c++/4.2.1/complex:450:7
>>         0x10372c210: i64 = Register %RDI [ORD=694] [ID=4]
>>         0x103699610: ch,glue = CopyToReg 0x10373ad10, 0x10372c210,
>> 0x103699010 [ORD=694] [ID=13] dbg:/usr/include/c++/4.2.1/complex:450:7
>>           0x10372c210: i64 = Register %RDI [ORD=694] [ID=4]
>>           0x103699010: i64,ch = load 0x103b38608, 0x10372c410,
>> 0x10373ab10<LD8[%1]> [ORD=693] [ID=11] dbg:/usr/include/c++/4.2.1/
>> complex:450:7
>>             0x10372c410: i64 = FrameIndex<0> [ORD=693] [ID=1]
>>             0x10373ab10: i64 = undef [ORD=693] [ID=2]
>>   0x10373ab10: i64 = undef [ORD=693] [ID=2]
>> 0x1036dd310: f80,ch = load 0x103699110:1, 0x103699110,
>> 0x10373ab10<LD10[%15](align=16)> [ORD=700] [ID=25] dbg:/usr/include/c++/
>> 4.2.1/complex:450:7
>>   0x103699110: i64,ch,glue = CopyFromReg 0x103699510, 0x1036dd810,
>> 0x103699510:1 [ORD=698] [ID=24] dbg:/usr/include/c++/4.2.1/complex:450:7
>>     0x1036dd810: i64 = Register %RAX [ORD=694] [ID=6]
>>     0x103699510: ch,glue = callseq_end 0x10371e410, 0x103698910,
>> 0x103698910, 0x10371e410:1 [ORD=698] [ID=23]
>>       0x103698910: i64 = TargetConstant<0> [ORD=694] [ID=3]
>>       0x103698910: i64 = TargetConstant<0> [ORD=694] [ID=3]
>>       0x10371e410: ch,glue = X86ISD::CALL 0x10372bf10, 0x10371ef10,
>> 0x10372c210, 0x10372bf10:1 [ORD=698] [ID=22] dbg:/usr/include/c++/4.2.1/
>> complex:450:7
>>         0x10371ef10: i64 = TargetGlobalAddress<x86_fp80*
>> (%"struct.std::complex"*)* @_ZNKSt7complexIeE4imagEv> 0 [ORD=694] [ID=5]
>> dbg:/usr/include/c++/4.2.1/complex:450:7
>>         0x10372c210: i64 = Register %RDI [ORD=694] [ID=4]
>>         0x10372bf10: ch,glue = CopyToReg 0x10372c810, 0x10372c210,
>> 0x10373ae10 [ORD=698] [ID=21] dbg:/usr/include/c++/4.2.1/complex:450:7
>>           0x10372c210: i64 = Register %RDI [ORD=694] [ID=4]
>>           0x10373ae10: i64,ch = load 0x103698a10:1, 0x10372c710,
>> 0x10373ab10<LD8[%2]> [ORD=697] [ID=17] dbg:/usr/include/c++/4.2.1/
>> complex:450:7
>>             0x10372c710: i64 = FrameIndex<1> [ORD=697] [ID=7]
>>             0x10373ab10: i64 = undef [ORD=693] [ID=2]
>>   0x10373ab10: i64 = undef [ORD=693] [ID=2]
>> 0x10372c910: i8 = Constant<0> [ID=9]
>> 
>> Is this known?
>> 
>> Thanks,
>> 
>> -- 
>> ____________________________________________________________
>> Sean McBride, B. Eng                 sean at rogue-research.com
>> Rogue Research                        www.rogue-research.com 
>> Mac Software Developer              Montréal, Québec, Canada
>> 
>> 
>> 
>> _______________________________________________
>> cfe-dev mailing list
>> cfe-dev at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev





More information about the cfe-dev mailing list