llvm dev - Dec 2010 - [LLVMdev] About the information about live variable analysis

Hi,

I worked in llvm to get some information from live variable analysis.
If the live range of one variable interferes with another, we could record
the related two variables into a pair. For example, with a function with
local variables 'a', 'b', 'c', 'x' and
'y', the pair set { (a, b), (b,c),
(x,y)} indicates that 'a' interferes with 'b', 'b'
interferes with 'c', and
'x' interferes with 'y'.

The question is: How can I get this information (maybe in other forms) by
using llvm to compile a program?

Could anyone help me?
Thanks in advance.

-- 
Best regards,

Li Qingan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20101219/39f83265/attachment.html>

On Dec 18, 2010, at 5:53 PM, Qingan Li wrote:
> Hi,
> 
> I worked in llvm to get some information from live variable analysis.
> If the live range of one variable interferes with another, we could record
the related two variables into a pair. For example, with a function with local
variables 'a', 'b', 'c', 'x' and 'y',
the pair set { (a, b), (b,c), (x,y)} indicates that 'a' interferes with
'b', 'b' interferes with 'c', and 'x' interferes
with 'y'.
> 
> The question is: How can I get this information (maybe in other forms) by
using llvm to compile a program?
This information is only available during register allocation in the code
generator. Look at LiveInterval::overlaps().

That is quite late in the compilation process, and it is not easy to translate
back to variables in the original program.

/jakob

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 1929 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20101218/937f1212/attachment.bin>

> This information is only available during register allocation in the code
generator. Look at LiveInterval::overlaps().
>
> That is quite late in the compilation process, and it is not easy to
translate back to variables in the original program.
>
> /jakob
>
I'm noticing too that some useful analysis/transforms are only available 
in the code generator, after the IR has been transformed to another 
format.  At least for my application, I'm finding it would be useful to 
have some of these analysis (eg. register allocation) available in the 
optimizer as well, though I'm sure there is a good reason 
architecturally that LLVM is using a separate representation for codegen.

Andrew