llvm dev - Mar 2004 - [LLVMdev] Re: Can't Figure Out My Error :(

On Sun, 28 Mar 2004, Reid Spencer wrote:
> The XPL compiler produced the attached OneOfEach.bc file without error
> or warning from either LLVM or XPLC. However when I llvm-dis the file, I
> get:
>
> llvm-dis: Failed value look-up for name 'entry'
I'm guessing that 'entry' is a basic block?
> The Bytecode Reader prints this out when it can't find the associated
> value for the name.  Obviously I botched something in the XPL compiler
> but I would have thought the verified would catch my mistake before
> writing the byte code.
It should!  It probably is missing a check somewhere... once you find the
problem, it would be helpful to make sure that the verifier catches that
problem in the future...
> The patch to the Reader.cpp file that I sent you earlier added the
"for
> name 'entry'" part to the output. Unfortunately, this
doesn't help
> because my source input doesn't have a symbol named "entry".
> Any hints?
My completely speculative *guess* is that you are using a basic block as
an operand to an instruction that does not expect it.  You can't add two
basic blocks together, and I would guess the verifier doesn't check for
this.  The only place where basic blocks can appear is in PHI nodes and
terminators.

I just checked this patch into the Verifier.  Maybe it will help?

--- Verifier.cpp        14 Mar 2004 03:23:54 -0000      1.88
+++ Verifier.cpp        29 Mar 2004 00:27:29 -0000
@@ -493,6 +493,11 @@
   Assert1(I.getType() != Type::VoidTy || !I.hasName(),
           "Instruction has a name, but provides a void value!",
&I);

+  // Check that the return value of the instruction is either void or a legal
+  // value type.
+  Assert1(I.getType() == Type::VoidTy || I.getType()->isFirstClassType(),
+          "Instruction returns a non-scalar type!", &I);
+
   // Check that all uses of the instruction, if they are instructions
   // themselves, actually have parent basic blocks.  If the use is not an
   // instruction, it is an error!

-Chris

-- 
http://llvm.cs.uiuc.edu/
http://www.nondot.org/~sabre/Projects/

I just found "entry". Its the name I gave a basic block at the start
of
the function (name was hard coded in the compiler, not the source).  The
issue here is probably that the block doesn't have a terminating
instruction.  However, that fact used to be caught well by the
validator. Furthermore, the output from the byte code reader is a little
confusing given the source of the problem.  

Has something broken recently?  This has been happening with the 1.2
release. I cvs updated today and rebuild LLVM but it still happens. 

Reid.


On Sun, 2004-03-28 at 16:23, Reid Spencer wrote:
> The XPL compiler produced the attached OneOfEach.bc file without error
> or warning from either LLVM or XPLC. However when I llvm-dis the file,
> I get:
> 
> llvm-dis: Failed value look-up for name 'entry'
> 
> The Bytecode Reader prints this out when it can't find the associated
> value for the name.  Obviously I botched something in the XPL compiler
> but I would have thought the verified would catch my mistake before
> writing the byte code. 
> 
> The patch to the Reader.cpp file that I sent you earlier added the
> "for name 'entry'" part to the output. Unfortunately,
this doesn't
> help because my source input doesn't have a symbol named
"entry".
> 
> Any hints?
> 
> On Sun, 2004-03-28 at 15:58, Chris Lattner wrote: 
> 
> > On Sun, 28 Mar 2004, Reid Spencer wrote:
> > 
> > > Would you like me to send the patches to sabre at nondot.org or
have you
> > > seen them already?
> > 
> > I haven't seen them.  Feel free to send them my way :)
> > 
> > -Chris
> 
> 
> 
> 
> 
> 
> 
> _______________________
> 
> Reid Spencer
> 
> President & CTO
> 
> eXtensible Systems, Inc.
> 
> rspencer at x10sys.com


_______________________
Reid Spencer
President & CTO
eXtensible Systems, Inc.
rspencer at x10sys.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20040328/1d3653bd/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20040328/1d3653bd/attachment.sig>

The XPL compiler produced the attached OneOfEach.bc file without error
or warning from either LLVM or XPLC. However when I llvm-dis the file, I
get:

llvm-dis: Failed value look-up for name 'entry'

The Bytecode Reader prints this out when it can't find the associated
value for the name.  Obviously I botched something in the XPL compiler
but I would have thought the verified would catch my mistake before
writing the byte code. 

The patch to the Reader.cpp file that I sent you earlier added the "for
name 'entry'" part to the output. Unfortunately, this doesn't
help
because my source input doesn't have a symbol named "entry". 

Any hints?

On Sun, 2004-03-28 at 15:58, Chris Lattner wrote:
> On Sun, 28 Mar 2004, Reid Spencer wrote:
> 
> > Would you like me to send the patches to sabre at nondot.org or have
you
> > seen them already?
> 
> I haven't seen them.  Feel free to send them my way :)
> 
> -Chris


_______________________
Reid Spencer
President & CTO
eXtensible Systems, Inc.
rspencer at x10sys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20040328/234503f4/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OneOfEach.bc
Type: application/octet-stream
Size: 1324 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20040328/234503f4/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20040328/234503f4/attachment.sig>