llvm dev - Feb 2012 - [LLVMdev] how to turn off the tail optimization

Hello,

It seems to me that LLVM is merging/in-lining the tail function to its 
caller.
I am wondering if there is a flag that I can use to turn off this 
optimization?

Thanks!

Wei

Hi Wei,
> It seems to me that LLVM is merging/in-lining the tail function to its
> caller.
> I am wondering if there is a flag that I can use to turn off this
> optimization?
do you want to turn off all inlining, or only inlining of calls that are tail
calls?  Why do you want to turn it off?

Ciao, Duncan.

Hi Duncan,

At this point, I just want to turn off all inlining of calls that are 
tail calls.

I am doing some analysis that pick function calls that only access local 
variables.

Assume (1) A calls B, (2) A accesses only local variables, (3) B 
accesses both
local variables and heap/global variables, (4) B is the tail call, thus B is
inlined in A.

Then my analysis will point out that A also accesses non-local 
variables. So I
want to turn off this inlining.

Thanks!

Wei
On 02/28/2012 04:40 AM, Duncan Sands wrote:
> Hi Wei,
>
>> It seems to me that LLVM is merging/in-lining the tail function to its
>> caller.
>> I am wondering if there is a flag that I can use to turn off this
>> optimization?
>
> do you want to turn off all inlining, or only inlining of calls that are
tail
> calls?  Why do you want to turn it off?
>
> Ciao, Duncan.
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev