llvm dev - Mar 2016 - supporting a type but not supporting an operaion

i'm looking at
https://github.com/llvm-mirror/llvm/blob/master/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp#L2107
but i don't quite understand the comment. how is it possible for a target
not to support an operation and yet support a type? can someone give an
example. i realize the question might sound silly, but i'm not really
getting it.

any help is appreciated.

-- 
Rail Shafigulin
Software Engineer
Esencia Technologies
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160323/10fec255/attachment.html>

On 3/23/2016 1:51 PM, Rail Shafigulin via llvm-dev
wrote:
> i'm looking at
>
https://github.com/llvm-mirror/llvm/blob/master/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp#L2107
> but i don't quite understand the comment. how is it possible for a
> target not to support an operation and yet support a type? can someone
> give an example. i realize the question might sound silly, but i'm not
> really getting it.
A type is "supported" (i.e. legal) if there is a register that can
hold
it.  Specifically, if there is a register class associated with that 
type.  A support of an operation means that there is an instruction (or 
a short sequence of instructions) that implements given operation.
For example, a target can have 128-bit registers, but the only 
instructions that use them may be load, store and an arithmetic "add".
In such a case, i128 is a supported type, but a "sub" operation on 
128-bit values is not.

-Krzysztof

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, 
hosted by The Linux Foundation

>
> A type is "supported" (i.e. legal) if there is a register that
can hold
> it.  Specifically, if there is a register class associated with that type.
> A support of an operation means that there is an instruction (or a short
> sequence of instructions) that implements given operation.
> For example, a target can have 128-bit registers, but the only
> instructions that use them may be load, store and an arithmetic
"add". In
> such a case, i128 is a supported type, but a "sub" operation on
128-bit
> values is not.
>
> -Krzysztof
>
> Oh I see.  Thanks.
> --
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted
> by The Linux Foundation
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>


-- 
Rail Shafigulin
Software Engineer
Esencia Technologies
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160323/e644bc18/attachment.html>