llvm dev - Aug 2010 - [LLVMdev] What does this error mean: psuedo instructions should be removed before code emission?

On 08/26/2010 11:11, Bob Wilson wrote:
> A pseudo instruction is used internally without llvm, but it does not
correspond directly to a real instruction in the target architecture.  Before
emitting the final compiled code, all the pseudo instructions must be expanded
to real instructions.  If you're seeing that error, it means that something
was not expanded.
>    
So what is likely to be wrong? This error only happens on amd64 and not 
on i386, variable Opcode is 2493 when the error occurs.
It is triggered from under the call that I make: JIT::runFunction.

Yuri

On Aug 26, 2010, at 11:47 AM, Yuri wrote:
> On 08/26/2010 11:11, Bob Wilson wrote:
>> A pseudo instruction is used internally without llvm, but it does not
correspond directly to a real instruction in the target architecture.  Before
emitting the final compiled code, all the pseudo instructions must be expanded
to real instructions.  If you're seeing that error, it means that something
was not expanded.
>> 
> 
> So what is likely to be wrong? This error only happens on amd64 and not 
> on i386, variable Opcode is 2493 when the error occurs.
> It is triggered from under the call that I make: JIT::runFunction.
Could you get it to print out the instruction when it happens? (just change the
line above the error message to print it out to errs()).

It basically means that a pseudo wasn't lowered to something that the jit
can output before the jit was run.  Is this on ToT?

-eric

On 08/26/2010 11:53, Eric Christopher wrote:
> Could you get it to print out the instruction when it happens? (just change
the line above the error message to print it out to errs()).
>
> It basically means that a pseudo wasn't lowered to something that the
jit can output before the jit was run.  Is this on ToT?
>    
Insn before the error: TCRETURNri64 %RAX<kill>, 0, %RDI<kill>, 
%RAX<imp-def,dead>, %RDI<imp-def,dead>, %RSP<imp-use>, ...

Yuri

On 08/26/2010 11:53, Eric Christopher wrote:
> It basically means that a pseudo wasn't lowered to something that the
jit can output before the jit was run.  Is this on ToT?
>    
I don't know what is ToT. 42 meanings listed here 
http://acronyms.thefreedictionary.com/TOT don't seem to apply.

Yuri