llvm dev - May 2015 - [LLVMdev] getnode(BB) = 0; block already in dominator tree

Hi
   I run into an issue as part of splitting a critical edge during LICM.
When a new basic block is created and needs to be added into the dominator
tree, the block is already in the dominator tree. I print the dominator
tree and I see it is added into the tree as child of node it is supposed to
dominate.

How do I debug to find out why/when its getting added into the tree. ?
Tips/suggestions on how to debug to find out how the densemap involved is
getting updated will be great.

thanks
Shrey
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150514/92b647a8/attachment.html>

Are you calling splitCriticalEdge?
If so, it will take care of updating dominators for you.


On Thu, May 14, 2015 at 11:55 AM, Shreyas Krishnan <shreyas76 at
gmail.com> wrote:
> Hi
>    I run into an issue as part of splitting a critical edge during LICM.
> When a new basic block is created and needs to be added into the dominator
> tree, the block is already in the dominator tree. I print the dominator
tree
> and I see it is added into the tree as child of node it is supposed to
> dominate.
>
> How do I debug to find out why/when its getting added into the tree. ?
> Tips/suggestions on how to debug to find out how the densemap involved is
> getting updated will be great.
>
> thanks
> Shrey
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

yes, am just using the LICM  that uses splitCriticalEdge  I see the issue
when the dominator tree is being updated

On Thu, May 14, 2015 at 12:20 PM, Daniel Berlin <dberlin at dberlin.org>
wrote:
> Are you calling splitCriticalEdge?
> If so, it will take care of updating dominators for you.
>
>
> On Thu, May 14, 2015 at 11:55 AM, Shreyas Krishnan <shreyas76 at
gmail.com>
> wrote:
> > Hi
> >    I run into an issue as part of splitting a critical edge during
LICM.
> > When a new basic block is created and needs to be added into the
> dominator
> > tree, the block is already in the dominator tree. I print the
dominator
> tree
> > and I see it is added into the tree as child of node it is supposed to
> > dominate.
> >
> > How do I debug to find out why/when its getting added into the tree. ?
> > Tips/suggestions on how to debug to find out how the densemap involved
is
> > getting updated will be great.
> >
> > thanks
> > Shrey
> >
> > _______________________________________________
> > LLVM Developers mailing list
> > LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150514/4a420e55/attachment.html>

Er, LICM currently does not modify the CFG at all.  Are you proposing to 
change that?  And if so, why?  What's the motivating case?

Philip

On 05/14/2015 11:55 AM, Shreyas Krishnan wrote:
> Hi
>    I run into an issue as part of splitting a critical edge during 
> LICM.  When a new basic block is created and needs to be added into 
> the dominator tree, the block is already in the dominator tree. I 
> print the dominator tree and I see it is added into the tree as child 
> of node it is supposed to dominate.
>
> How do I debug to find out why/when its getting added into the tree. ? 
> Tips/suggestions on how to debug to find out how the densemap involved 
> is getting updated will be great.
>
> thanks
> Shrey
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150514/a631f765/attachment.html>

Oh...its machine LICM ...in the following lines...

  CurPreheader = Pred->SplitCriticalEdge(CurLoop->getHeader(), this);
01463       if (!CurPreheader) {
01464         CurPreheader = reinterpret_cast<MachineBasicBlock *>(-1);
01465         return nullptr;
01466       }





On Thu, May 14, 2015 at 1:57 PM, Philip Reames
<listmail at philipreames.com> wrote:
> Er, LICM currently does not modify the CFG at all.  Are you proposing to
> change that?  And if so, why?  What's the motivating case?
>
> Philip
>
>
> On 05/14/2015 11:55 AM, Shreyas Krishnan wrote:
>
> Hi
>    I run into an issue as part of splitting a critical edge during LICM.
> When a new basic block is created and needs to be added into the dominator
> tree, the block is already in the dominator tree. I print the dominator
tree
> and I see it is added into the tree as child of node it is supposed to
> dominate.
>
> How do I debug to find out why/when its getting added into the tree. ?
> Tips/suggestions on how to debug to find out how the densemap involved is
> getting updated will be great.
>
> thanks
> Shrey
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>