similar to: [LLVMdev] DAG Debugging

Hi I am running into a legalize dag issue after custom lowering a load. If someone can give me pointers that would be great. I am using llvm 2.6 When custom lowering a load, the following code gets executed. Tmp1 = TLI.LowerOperation(Tmp3, DAG); if (Tmp1.getNode()) { Tmp3 = LegalizeOp(Tmp1); Tmp4 = LegalizeOp(Tmp1.getValue(1)); <----------------- what

thanks for the help ..I do add the chain and the result. My code is like this ... SDValue Ops[] = { load->getChain(), load->getOperand(1), load->getBasePtr(), des }; DAG.getNode(CustomOpc, NodeTys, Ops, 4); thanks again! shrey On Thu, Nov 12, 2009 at 4:41 PM, Anton Korobeynikov <anton at korobeynikov.info> wrote: >> My problem is that the second call asserts inside

thanks. One question - Isnt it needed for other arithmetic operations such as add etc ? shrey On Fri, Aug 13, 2010 at 10:36 AM, Anton Korobeynikov <anton at korobeynikov.info> wrote: > Hello > >> if needed conversions are generated. So if  I needed to generate >> different instructions for signed/unsigned, is that possible ? > Yes, see e.g.

> My problem is that the second call asserts inside legalize ops at > ResultVals[Op.getResNo()]; b'cos ResultVals has only 1 element and > Op.resno is 0. Looks like you lowered the load improperly. It should return 2 values: the value loaded and a chain. -- With best regards, Anton Korobeynikov Faculty of Mathematics and Mechanics, Saint Petersburg State University

Hi I was wondering if within the TargetDescription string it was possible to specify different alignment for variables on stack versus globals. If not, is there any other way to specify that ? thanks shrey

Hi I have a question abt type information such as unsigned/signed. I find that the information is not retained in the intermediate form and if needed conversions are generated. So if I needed to generate different instructions for signed/unsigned, is that possible ? thanks shrey

On Aug 14, 2012, at 2:05 PM, Steve Montgomery <stephen.montgomery3 at btinternet.com> wrote: > Further to my earlier question, I'm perhaps a bit confused about memory serialisation. The following example, compiled using clang for the MSP430: > > target datalayout = "e-p:16:16:16-i8:8:8-i16:16:16-i32:16:32-n8:16" > target triple = "msp430-??-??" >

> No, a chain is supposed to mean "later than". It sounds like MSP430 is bending > the rules here. The instruction selector for ADD16mm is autogenerated, so, this is not MSP430 bug alone :) This is just the single target in the tree which has mem-mem instructions. -- With best regards, Anton Korobeynikov Faculty of Mathematics and Mechanics, Saint Petersburg State University

Yes I am on darwin. On Mon, Oct 19, 2009 at 12:17 AM, Anton Korobeynikov < anton at korobeynikov.info> wrote: > > warning: Could not find object file > > "/var/folders/h6/h6He5G34EF8g9zpg3SNFF++++TI/-Tmp-//ccoTexQw.o" - no > debug > > information available for "ReflectionTest.tart". > It seems you're on darwin, right? Then you need another

> I can't find any release_32 branch at http://llvm.org/git/llvm.git or http://llvm.org/git/clang.git. Unfortunately, this requires manual grafting, since git-svn does really bad job here. I'm going to work on this tonight. -- With best regards, Anton Korobeynikov Faculty of Mathematics and Mechanics, Saint Petersburg State University

On Oct 19, 2010, at 8:00 PM, Jakob Stoklund Olesen wrote: > > > One problem is that calling conventions are handled while building the selection DAG, and the DAG doesn't really know to represent clobbered registers. > > Perhaps X86TargetLowering::LowerCall() could decorate the X86ISD::CALL node with the calling convention somehow? > > Dan, do you have any thoughts on

Anton, please add release_32 also in; clang-tools-extra compiler-rt dragonegg libcxxabi lldb They have release_32 in svn. I don't know they might be released, though. And, could you suppress generating refs/heads/svn-tags and prune them for now? I am sure that orphan branches will stress the llvm.org server to begin git-pack-ing whole tree. ...Takumi 2012/11/15 Anton Korobeynikov

I've been struggling with this for a while. I've gone back and reviewed all the messages in this topic on the subject, and I still can't figure it out. -- I'm disabling frame pointer elimination via "-disable-fp-elim" -- I see debugging information in my .s file. -- My command line for compiling the LLVM-generated looks like this: gcc -g -O0 -x assembler

thanks for the help ..I could build it with cmake generated project files Being a newbie with cmake one question, is it equivalent to configure in windows environment and so would need to be run everytime I change the source location ? thanks shrey On Wed, Oct 28, 2009 at 4:31 PM, shreyas krishnan <shreyas76 at gmail.com> wrote: > done > >

done http://llvm.org/bugs/show_bug.cgi?id=5331 shrey On Wed, Oct 28, 2009 at 3:29 PM, Óscar Fuentes <ofv at wanadoo.es> wrote: > shreyas krishnan <shreyas76 at gmail.com> writes: > >>     I am running into bunch of windows build issues. Can someone >> please provide help on what might be going wrong >> >> >>

I recently filed this bug: http://llvm.org/bugs/show_bug.cgi?id=8323 It's a dodgy one because you have to patch LLVM to demonstrate it. I suspect that the cause of the problem in that "bug" is that the peephole optimisation in PerformDAGCombine results in a Flag output from one node being used as input by two other nodes in the DAG, and the scheduler then can't cope with that.

Should be there. Please let me know if there are any problems with them On Wed, Nov 14, 2012 at 9:19 PM, Anton Korobeynikov <anton at korobeynikov.info> wrote: >> I can't find any release_32 branch at http://llvm.org/git/llvm.git or http://llvm.org/git/clang.git. > Unfortunately, this requires manual grafting, since git-svn does > really bad job here. > > I'm going

I'm trying to change the Endian of X86 from little to big in the TargetData. I only care about the way in which this effects the LLVM IR, not the actual backend code. I've changed the "e-" to "E-" in X86TargetMachine.cpp (where it sets the DataLayout) with no luck. Are there any other obvious places that I need to change this? The TargetData docs were somewhat helpful

Anton Korobeynikov: > > Is it, or should it be legal for a Flag output to be used as input by > > more than one other node? > It's illegal. Multiple uses of the flag output do not make any sense, > this breaks the semantics of flag operands. All right. Then what should the Selection DAG look like in a case where the flag value generated by one instruction is to be used as

Hi I am running into bunch of windows build issues. Can someone please provide help on what might be going wrong llvm\win32\Configure\..\llvm\ADT\hash_set.h 1>The system cannot find the file specified. Similarly 2>c1xx : fatal error C1083: Cannot open source file: '..\..\lib\Support\Annotation.cpp': No such file or directory thanks in advance shrey