llvm dev - Sep 2006 - [LLVMdev] Folding instructions on X86

Dear LLVMers,

      I need help with one error in my register allocator. I am getting
errors when folding instructions for the x86 (it does not happen when
I use the PowerPC as target). The problem is that, when I execute
something like: mi = mbb->insert(mbb->erase(mi), fmi), where mbb is a 
MachineBasicBlock, valgrind gives this error message:

Address 0x42C62B0 is 24 bytes inside a block of size 28 free'd

     at 0x401C35D: operator delete(void*)(vg_replace_malloc.c:244)
     by 0x83E4A7B: llvm::iplist<llvm::MachineInstr,
llvm::ilist_traits<llvm::MachineInstr> 
>::erase(llvm::ilist_iterator<llvm::MachineInstr>) (ilist:323)
     by 0x83E464D: 
llvm::MachineBasicBlock::erase(llvm::ilist_iterator<llvm::MachineInstr>) 
(MachineBasicBlock.h:168)

Do you guys know any possible reason of why this would happen when 
producing code for the X86, but not for the PowerPC?

Best regards,

Fernando

Fernando Magno Quintao Pereira wrote:
> 
> Dear LLVMers,
> 
>       I need help with one error in my register allocator. 
Out of curiosity, what kind of register allocator is that?

- Volodya

On Sun, 24 Sep 2006, Fernando Magno Quintao Pereira
wrote:
>      I need help with one error in my register allocator. I am getting
> errors when folding instructions for the x86 (it does not happen when
> I use the PowerPC as target). The problem is that, when I execute
> something like: mi = mbb->insert(mbb->erase(mi), fmi), where mbb is a
> MachineBasicBlock, valgrind gives this error message:
>
> Address 0x42C62B0 is 24 bytes inside a block of size 28 free'd
>
>     at 0x401C35D: operator delete(void*)(vg_replace_malloc.c:244)
>     by 0x83E4A7B: llvm::iplist<llvm::MachineInstr,
> llvm::ilist_traits<llvm::MachineInstr>
>> ::erase(llvm::ilist_iterator<llvm::MachineInstr>) (ilist:323)
>     by 0x83E464D:
>
llvm::MachineBasicBlock::erase(llvm::ilist_iterator<llvm::MachineInstr>)
> (MachineBasicBlock.h:168)
It's hard to say, but it seems that the pointer to erase is most likely 
invalid.
> Do you guys know any possible reason of why this would happen when
> producing code for the X86, but not for the PowerPC?
X86 has many more instructions that can be folded than powerpc does. 
You're probably just happening to get unlucky on x86.

-Chris

-- 
http://nondot.org/sabre/
http://llvm.org/