llvm dev - Oct 2016 - Semantics of MachineInstr::findRegisterUseOperandIdx

Hi,

I was wondering what MachineInstr::findRegisterUseOperandIdx "means".
I
have a situation where an instruction writes R0 and another reads R0lo - 
the lower half of R0. Using Use::findRegisterUseOperandIdx(R0) returns 
-1, because it only checks against exact matches and 
"is-Reg-a-subreg-of-MOReg" matches, of which neither are true.

However, the instruction reads _part_ of R0, so should it return true? 
It's really a question of semantics, as either:

1) findRegisterUseOperandIdx returns the index if the instruction reads 
the /whole/ of Reg, or
2) findRegisterUseOperandIdx returns the index if the instruction reads 
/any part/ of Reg

If it's 2) then it appears as if there's a bug in there. If it's 1)
by
design, then it doesn't seem to me as though there's a convenient way of
doing 2).

There is another option, namely:

3) findRegisterUseOperandIdx can do 1) or 2), depending on a flag

There's an "Overlap" parameter to 
MachineInstr::findRegisterDefOperandIdx - would that do the trick, and 
do we want to add it to findRegisterUseOperandIdx?

What do people think?

Cheers,
Fraser

-- 
Fraser Cormack
Staff Software Engineer, Compilers
Codeplay Software Ltd
Level C, Argyle House
3 Lady Lawson St,
Edinburgh EH3 9DR
Tel: +44 (0)131 466 0503
Website: http://www.codeplay.com
Twitter: https://twitter.com/codeplaysoft

This email and any attachments may contain confidential and /or privileged
information and is for use by the addressee only. If you are not the intended
recipient, please notify Codeplay Software Ltd immediately and delete the
message from your computer. You may not copy or forward it, or use or disclose
its contents to any other person. Any views or other information in this message
which do not relate to our business are not authorized by Codeplay software Ltd,
nor does this message form part of any contract unless so stated.
As internet communications are capable of data corruption Codeplay Software Ltd
does not accept any responsibility for any changes made to this message after it
was sent. Please note that Codeplay Software Ltd does not accept any liability
or responsibility for viruses and it is your responsibility to scan any
attachments.
Company registered in England and Wales, number: 04567874
Registered office: Regent House, 316 Beulah Hill, London, United Kingdom, SE19
3HF

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20161004/eb548ea7/attachment.html>

Hi Fraser,

I think we want stick to being explicit about what the method does:
- No TRI, whole Reg
- TRI, Reg or super Reg
I.e., no magic aliases check unless we asked for.

You are right that we miss the any part of Reg option here. If adding an Overlap
flag like we have for the Defs works for you, I would say, that’s the way to go.

I am surprised we didn’t need it sooner.

Cheers,
-Quentin
> On Oct 4, 2016, at 4:00 AM, Fraser Cormack via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
> 
> Hi,
> 
> I was wondering what MachineInstr::findRegisterUseOperandIdx
"means". I have a situation where an instruction writes R0 and another
reads R0lo - the lower half of R0. Using Use::findRegisterUseOperandIdx(R0)
returns -1, because it only checks against exact matches and
"is-Reg-a-subreg-of-MOReg"  matches, of which neither are true.
> 
> However, the instruction reads _part_ of R0, so should it return true?
It's really a question of semantics, as either:
> 
> 1) findRegisterUseOperandIdx returns the index if the instruction reads the
whole of Reg, or
> 2) findRegisterUseOperandIdx returns the index if the instruction reads any
part of Reg
> 
> If it's 2) then it appears as if there's a bug in there. If
it's 1) by design, then it doesn't seem to me as though there's a
convenient way of doing 2).
> 
> There is another option, namely:
> 
> 3) findRegisterUseOperandIdx can do 1) or 2), depending on a flag
> 
> There's an "Overlap" parameter to
MachineInstr::findRegisterDefOperandIdx - would that do the trick, and do we
want to add it to findRegisterUseOperandIdx?
> 
> What do people think?
> 
> Cheers,
> Fraser
> -- 
> Fraser Cormack
> Staff Software Engineer, Compilers
> Codeplay Software Ltd
> Level C, Argyle House
> 3 Lady Lawson St,
> Edinburgh EH3 9DR
> Tel: +44 (0)131 466 0503
> Website: http://www.codeplay.com <http://www.codeplay.com/>
> Twitter: https://twitter.com/codeplaysoft
<https://twitter.com/codeplaysoft>
> 
> This email and any attachments may contain confidential and /or privileged
information and is for use by the addressee only. If you are not the intended
recipient, please notify Codeplay Software Ltd immediately and delete the
message from your computer. You may not copy or forward it, or use or disclose
its contents to any other person. Any views or other information in this message
which do not relate to our business are not authorized by Codeplay software Ltd,
nor does this message form part of any contract unless so stated.
> As internet communications are capable of data corruption Codeplay Software
Ltd does not accept any responsibility for any changes made to this message
after it was sent. Please note that Codeplay Software Ltd does not accept any
liability or responsibility for viruses and it is your responsibility to scan
any attachments.
> Company registered in England and Wales, number: 04567874
> Registered office: Regent House, 316 Beulah Hill, London, United Kingdom,
SE19 3HF
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20161004/d3245b76/attachment.html>