[LLVMdev] Porting to System z

[Neale Ferguson]

I searched the archives and found this from last month:

I ran into the same problem and fixed it by forcing the
MVT::SimpleValueType enum to be 64 bits so that all of the types
in the union later in the class are the same size.  I tested this
on ppc64 and x86_64.

Index: include/llvm/CodeGen/ValueTypes.h
===================================================================
--- include/llvm/CodeGen/ValueTypes.h    (revision 66504)
+++ include/llvm/CodeGen/ValueTypes.h    (working copy)
@@ -21,6 +21,8 @@
  #include "llvm/Support/DataTypes.h"
  #include "llvm/Support/MathExtras.h"

+static const uintptr_t minus_one = -1;
+
  namespace llvm {
    class Type;

@@ -92,7 +94,10 @@
        iPTR           =  255,

        // LastSimpleValueType - The greatest valid SimpleValueTypevalue.
-      LastSimpleValueType = 255
+      LastSimpleValueType = 255,
+
+      // force the size of the enum to be sizeof(uintptr_t)
+      _BigValue = minus_one
      };

    private:

I added this code but, apart from generating warnings about a negative value
being assigned to an unsigned variable, the problem didn't disappear. It
also resulted in warnings like this:

CodeGenDAGPatterns.h: In member function `bool
   llvm::TreePatternNode::hasTypeSet() const':
CodeGenDAGPatterns.h:187: warning: comparison between signed and unsigned
   integer expressions

Using gdb to see what may be happening shows that the value is llvm::MVT::i8
that it is having problems with. I assume it must be failing the
MVT(VT).isInteger() test, so I added some logic to break down the test:

144        bool isInt = MVT(VT).isInteger();
(gdb) n
145        bool isSim = MVT(VT).isSimple();
(gdb) p isInt
$1 = false
(gdb) n
146      if (MVT(VT).isInteger()) {
(gdb) p isSim
$2 = true

So you'd assume it's the tests in isInteger() that are failing:

    /// isInteger - Return true if this is an integer, or a vector integer
type.
    bool isInteger() const {
      return isSimple() ?
             ((SimpleTy >= FIRST_INTEGER_VALUETYPE &&
               SimpleTy <= LAST_INTEGER_VALUETYPE) ||
              (SimpleTy >= v2i8 && SimpleTy <= v2i64)) :
             isExtendedInteger();
    }

But...

(gdb) p llvm::MVT::i8 >= FIRST_INTEGER_VALUETYPE
$10 = true
(gdb) p llvm::MVT::i8 <= LAST_INTEGER_VALUETYPE
$11 = true

So the SimpleTy variable which is a SimpleTypeValue is behaving strangely.
It's found within a private definition:

    union {
      uintptr_t V;
      SimpleValueType SimpleTy;
      const Type *LLVMTy;
    };
 
The comment in the above fix indicates that in this 64-bit system the enum
needs to be 64-bits is correct, but the fix doesn't seem to do it.

Neale

On 4/7/09 12:36 PM, "Duncan Sands" <baldrick at free.fr> wrote:

> Hi,
> 
>> llvm[1]: Building Intrinsics.gen.tmp from Intrinsics.td
>> tblgen: IntrinsicEmitter.cpp:163: void EmitTypeForValueType(std::ostream&,
>> llvm::MVT::SimpleValueType): Assertion `false && "Unsupported ValueType!"'
>> failed.
> 
> this came up before IIRC, but I don't remember the details - buggy system
> compiler?  Try searching the archives.  Also, if you were compiling with
> optimization, try building without optimization.
> 
> Ciao,
> 
> Duncan.
> 
>> 0   tblgen    0x00000000801be2bc
>> 1   tblgen    0x00000000801bea14
>> 2   tblgen    0x000003ffffad4b84
>> 3   libc.so.6 0x0000020000286ef0 gsignal + 76
>> 4   libc.so.6 0x0000020000288330 abort + 256
>> 5   libc.so.6 0x000002000027ed08 __assert_fail + 228
>> 6   tblgen    0x0000000080111f86
>> 7   tblgen    0x0000000080112afe
>> 8   tblgen    0x000000008011281c
>> 9   tblgen    0x0000000080113354
>> 10  tblgen    0x0000000080113744
>> 11  tblgen    0x000000008019273e main + 2650
>> 12  libc.so.6 0x0000020000270598 __libc_start_main + 256
>> 13  tblgen    0x0000000080009ec2 std::__throw_logic_error(char const*) + 158
>> Stack dump:
>> 0.    Program arguments: /home/neale/LLVM/llvm/Debug/bin/tblgen -I
>> /home/neale/LLVM/llvm/lib/VMCore -I /home/neale/LLVM/llvm/include -I
>> /home/neale/LLVM/llvm/include -I /home/neale/LLVM/llvm/lib/Target
>> /home/neale/LLVM/llvm/include/llvm/Intrinsics.td -o
>> /home/neale/LLVM/llvm/lib/VMCore/Debug/Intrinsics.gen.tmp -gen-intrinsic
>> make[1]: *** [/home/neale/LLVM/llvm/lib/VMCore/Debug/Intrinsics.gen.tmp]
>> Aborted
>> 
>> Is this a known problem? Would it be better to build the tools etc. using a
>> "stable" tarball and just use the svn sources so I can generate the patches
>> for adding System z?
>> 
>> Neale 
>> 
>> 
>> 
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>> 
> 
>