llvm dev - Nov 2014 - [LLVMdev] Is there a path from MachineInstr to the associated Instruction

Hi,

Recently, I am working on some information collection on the machine
instruction (MachineInstr) level.
But, sometimes, I need check the corresponding IR level instruction of the
a machine instruction.

When visiting a machine instruction which is a call
(MachineInstr::isCall()), I need to check the arguments of this call site.
I know that the CallSite class provides good facility for this work, but
the CallSite can be constructed only fromInstruction rather than
MachineInstr.

Is there any workaround?

Thanks.

-- 
Best regards,

Li Qingan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20141114/407d2787/attachment.html>

On 11/14/14, 6:53 AM, Qingan Li wrote:
> Hi,
>
> Recently, I am working on some information collection on the machine 
> instruction (MachineInstr) level.
> But, sometimes, I need check the corresponding IR level instruction of 
> the a machine instruction.
>
> When visiting a machine instruction which is a call 
> (MachineInstr::isCall()), I need to check the arguments of this call site.
> I know that the CallSite class provides good facility for this work, 
> but the CallSite can be constructed only fromInstruction rather than 
> MachineInstr.
>
> Is there any workaround?
If you're working on MachineInstr's, then you can probably examine the 
MachineIntr's operands to determine which function is being called.  
This probably won't be target independent, but if you're working at the 
MachineInstr level, you're not doing target independent work, anyway.

Alternatively, you should check to see if the MachineBasicBlock has a 
method to fetch the BasicBlock from which it was generated.  Using that, 
you could potentially scan the BasicBlock looking for a CallInst or 
InvokeInst that is likely to match the MachineInstr.

Regards,

John Criswell
>
> Thanks.
>
> -- 
> Best regards,
>
> Li Qingan
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

-- 
John Criswell
Assistant Professor
Department of Computer Science, University of Rochester
http://www.cs.rochester.edu/u/criswell

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20141116/b717c3de/attachment.html>

Hi John,

Thanks so much!
I have checked the BasicBlock. However, there may be multiple call sites in
one basic block, so it cannot guarantee correctness.
Finally, I chose to add the CallInst into VDNode during DAG construction,
and then add the CallInst into MachineInstr during instruction selection.

Thanks all the same!

Best regards,

Li Qingan

2014-11-17 1:44 GMT+08:00 John Criswell <jtcriswel at gmail.com>:
>  On 11/14/14, 6:53 AM, Qingan Li wrote:
>
>  Hi,
>
>  Recently, I am working on some information collection on the machine
> instruction (MachineInstr) level.
> But, sometimes, I need check the corresponding IR level instruction of the
> a machine instruction.
>
>  When visiting a machine instruction which is a call
> (MachineInstr::isCall()), I need to check the arguments of this call site.
> I know that the CallSite class provides good facility for this work, but
> the CallSite can be constructed only fromInstruction rather than
> MachineInstr.
>
>  Is there any workaround?
>
>
> If you're working on MachineInstr's, then you can probably examine
the
> MachineIntr's operands to determine which function is being called. 
This
> probably won't be target independent, but if you're working at the
> MachineInstr level, you're not doing target independent work, anyway.
>
> Alternatively, you should check to see if the MachineBasicBlock has a
> method to fetch the BasicBlock from which it was generated.  Using that,
> you could potentially scan the BasicBlock looking for a CallInst or
> InvokeInst that is likely to match the MachineInstr.
>
> Regards,
>
> John Criswell
>
>
>  Thanks.
>
>  --
> Best regards,
>
>  Li Qingan
>
>
> _______________________________________________
> LLVM Developers mailing listLLVMdev at cs.uiuc.edu        
http://llvm.cs.uiuc.eduhttp://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
>
> --
> John Criswell
> Assistant Professor
> Department of Computer Science, University of
Rochesterhttp://www.cs.rochester.edu/u/criswell
>
>

-- 
Best regards,

Li Qingan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20141118/f74df6ba/attachment.html>