[Mono-dev] Inlining common calls

Stack Pointer stackpointer at me.com
Tue Nov 27 23:24:03 UTC 2012


Thanks!

I made a quick hack to bypass this and I do see more functions being inlined, so far no "crashes". But at the same time, I see worse performance. So may be there is a correctness + performance issue.

I see a similar thing for "Internal Calls". There are correctness issues there as well?

Is this something on your roadmap to fix? Or should we just "work around" it?


On Nov 27, 2012, at 11:10 AM, Rodrigo Kumpera <kumpera at gmail.com> wrote:

> We don't inline calls due to a correctness issue with stack unwinding.
> 
> 
> On Tue, Nov 27, 2012 at 1:25 PM, Stack Pointer <stackpointer at me.com> wrote:
> At: https://github.com/mono/mono/blob/master/mono/mini/method-to-ir.c:7596
> 
> 
> 	/* Common call */
> 	INLINE_FAILURE ("call");
> 
> 
> Why do you throw a failure directly? Shouldn't there be an attempt to inline methods which only consists calls or other simple body? Is this detrimental to the performance?
> 
> I'd like to understand inlining vs performance trade-offs that mono does.
> 
> _______________________________________________
> Mono-devel-list mailing list
> Mono-devel-list at lists.ximian.com
> http://lists.ximian.com/mailman/listinfo/mono-devel-list
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ximian.com/pipermail/mono-devel-list/attachments/20121127/2fea788c/attachment.html>


More information about the Mono-devel-list mailing list