llvm dev - Jun 2016 - Question about changes to 'SelectionDAGISel.h'

It occurred to me that instead of the various breakout 'Select*'
functions
returning the 'SDNode*' result, maybe I should be calling:

 

  ReplaceNode(N, newValue);

  return;

or:

  SelectCode(N);

  return;

 

Perhaps?

 

            MartinO

 

From: Martin J. O'Riordan [mailto:martin.oriordan at movidius.com] 
Sent: 28 June 2016 16:49
To: 'LLVM Developers' <llvm-dev at lists.llvm.org>
Subject: Question about changes to 'SelectionDAGISel.h'

 

Although I would like to track the LLVM head revisions regularly,
unfortunately I only get the opportunity every couple of months or even
every 6 month with a full release.

 

This time I am updating from revision #262824 (8th March) so more than 3
months have elapsed.  For the most part I have completed the changes I need
to make, but I'm stuck on one change that is more significant than it
appears.

 

In the March #262824 revision 'SelectionDAGISel' declared the function
'Select' as:

 

    SDNode *Select(SDNode *N) = 0;

 

But now it is:

 

    void Select(SDNode *N) = 0;

 

The problem is that we have a number of places where we construct DAGs using
the returned 'SDNode*' from explicit calls to 'Select' and I am
wondering
what is the best design approach for me to fix this?  This is old code, so
the architectural changes that lead to the return type for 'Select'
being
changed have probably been long coming, but I was unaware of it.  Should I
be using 'SelectCode' instead for example?  Or is there a good reference
target I should examine to see how they adapted?

 

Thanks,

 

            MartinO

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160628/2895fbb7/attachment.html>

On Tue, Jun 28, 2016 at 8:53 AM, Martin J. O'Riordan via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> It occurred to me that instead of the various breakout ‘Select*’ functions
> returning the ‘SDNode*’ result, maybe I should be calling:
>
>
>
>   ReplaceNode(N, newValue);
>
>   return;
>
> or:
>
>   SelectCode(N);
>
>   return;
>
>
>
> Perhaps?
Yes, I think the core difference is that Select() - not its caller -
now does the replacement, so there's nothing to return.

This is actually mentioned in the release notes (kudos to Justin!):

    SelectionDAGISel::Select now returns void. Out of tree targets
will need to be updated to replace the argument node and remove any
dead nodes in cases where they currently return an SDNode * from this
interface.

http://llvm.org/docs/ReleaseNotes.html

You can look at the changes to the various in-tree targets between
r268693 and r270454 (e.g., r269144 for x86).

-Ahmed
>
>
>             MartinO
>
>
>
> From: Martin J. O'Riordan [mailto:martin.oriordan at movidius.com]
> Sent: 28 June 2016 16:49
> To: 'LLVM Developers' <llvm-dev at lists.llvm.org>
> Subject: Question about changes to 'SelectionDAGISel.h'
>
>
>
> Although I would like to track the LLVM head revisions regularly,
> unfortunately I only get the opportunity every couple of months or even
> every 6 month with a full release.
>
>
>
> This time I am updating from revision #262824 (8th March) so more than 3
> months have elapsed.  For the most part I have completed the changes I need
> to make, but I’m stuck on one change that is more significant than it
> appears.
>
>
>
> In the March #262824 revision ‘SelectionDAGISel’ declared the function
> ‘Select’ as:
>
>
>
>     SDNode *Select(SDNode *N) = 0;
>
>
>
> But now it is:
>
>
>
>     void Select(SDNode *N) = 0;
>
>
>
> The problem is that we have a number of places where we construct DAGs
using
> the returned ‘SDNode*’ from explicit calls to ‘Select’ and I am wondering
> what is the best design approach for me to fix this?  This is old code, so
> the architectural changes that lead to the return type for ‘Select’ being
> changed have probably been long coming, but I was unaware of it.  Should I
> be using ‘SelectCode’ instead for example?  Or is there a good reference
> target I should examine to see how they adapted?
>
>
>
> Thanks,
>
>
>
>             MartinO
>
>
>
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>

On 28 June 2016 at 17:01, Ahmed Bougacha via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> On Tue, Jun 28, 2016 at 8:53 AM, Martin J. O'Riordan via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
>> It occurred to me that instead of the various breakout ‘Select*’
functions
>> returning the ‘SDNode*’ result, maybe I should be calling:
>>
>>
>>
>>   ReplaceNode(N, newValue);
>>
>>   return;
>>
>> or:
>>
>>   SelectCode(N);
>>
>>   return;
>>
>>
>>
>> Perhaps?
>
> Yes, I think the core difference is that Select() - not its caller -
> now does the replacement, so there's nothing to return.
>
> This is actually mentioned in the release notes (kudos to Justin!):
>
>     SelectionDAGISel::Select now returns void. Out of tree targets
> will need to be updated to replace the argument node and remove any
> dead nodes in cases where they currently return an SDNode * from this
> interface.
>
> http://llvm.org/docs/ReleaseNotes.html
>
> You can look at the changes to the various in-tree targets between
> r268693 and r270454 (e.g., r269144 for x86).
Also see http://lists.llvm.org/pipermail/llvm-dev/2016-May/099216.html
for the motivation for the change.

Best,

Alex

Thanks Ahmed and also Alex for your replies.

This is more or less what I was realising, but it is a great confidence booster
to know that it is the correct way also.  I can replace all of my various
'Select*' specialisations with version that use
'ReplaceNode/SelectCode' and return 'void', but what about the
places where I currently call 'Select(N)' directly?  Should I substitute
'SelectCode(N)' instead?

I will examine the X86 implementation as you recommend and hope to glean some
knowledge from that.

All the best,

	MartinO

-----Original Message-----
From: Ahmed Bougacha [mailto:ahmed.bougacha at gmail.com] 
Sent: 28 June 2016 17:02
To: Martin.ORiordan at movidius.com
Cc: LLVM Developers <llvm-dev at lists.llvm.org>; Justin Bogner <mail
at justinbogner.com>
Subject: Re: [llvm-dev] Question about changes to 'SelectionDAGISel.h'

On Tue, Jun 28, 2016 at 8:53 AM, Martin J. O'Riordan via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> It occurred to me that instead of the various breakout ‘Select*’ 
> functions returning the ‘SDNode*’ result, maybe I should be calling:
>
>
>
>   ReplaceNode(N, newValue);
>
>   return;
>
> or:
>
>   SelectCode(N);
>
>   return;
>
>
>
> Perhaps?
Yes, I think the core difference is that Select() - not its caller - now does
the replacement, so there's nothing to return.

This is actually mentioned in the release notes (kudos to Justin!):

    SelectionDAGISel::Select now returns void. Out of tree targets will need to
be updated to replace the argument node and remove any dead nodes in cases where
they currently return an SDNode * from this interface.

http://llvm.org/docs/ReleaseNotes.html

You can look at the changes to the various in-tree targets between
r268693 and r270454 (e.g., r269144 for x86).

-Ahmed
>
>
>             MartinO
>
>
>
> From: Martin J. O'Riordan [mailto:martin.oriordan at movidius.com]
> Sent: 28 June 2016 16:49
> To: 'LLVM Developers' <llvm-dev at lists.llvm.org>
> Subject: Question about changes to 'SelectionDAGISel.h'
>
>
>
> Although I would like to track the LLVM head revisions regularly, 
> unfortunately I only get the opportunity every couple of months or 
> even every 6 month with a full release.
>
>
>
> This time I am updating from revision #262824 (8th March) so more than 
> 3 months have elapsed.  For the most part I have completed the changes 
> I need to make, but I’m stuck on one change that is more significant 
> than it appears.
>
>
>
> In the March #262824 revision ‘SelectionDAGISel’ declared the function 
> ‘Select’ as:
>
>
>
>     SDNode *Select(SDNode *N) = 0;
>
>
>
> But now it is:
>
>
>
>     void Select(SDNode *N) = 0;
>
>
>
> The problem is that we have a number of places where we construct DAGs 
> using the returned ‘SDNode*’ from explicit calls to ‘Select’ and I am 
> wondering what is the best design approach for me to fix this?  This 
> is old code, so the architectural changes that lead to the return type 
> for ‘Select’ being changed have probably been long coming, but I was 
> unaware of it.  Should I be using ‘SelectCode’ instead for example?  
> Or is there a good reference target I should examine to see how they
adapted?
>
>
>
> Thanks,
>
>
>
>             MartinO
>
>
>
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>