Bug 18347: Sparse switches lowered to an else-if chain

[Alp Toker]

On 11/01/2014 14:33, Jasper Neumann wrote:
> Hi all,
>
> here is the patch as described in
> http://llvm.org/bugs/show_bug.cgi?id=18347
>
> Would someone please review the patch and if OK check it in for me, 
> please?
> For testing purposes it might make sense to temporarily disable 
> dispatching with jump tables by commenting out the two lines 
> containing "handleJTSwitchCase" near the bottom of 
> SelectionDAGBuilder::visitSwitch in 
> llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp and also to 
> activate the logging of the splitting behavior in 
> SelectionDAGBuilder::handleBTSplitSwitchCase.

Hi Jasper,

This patch will need to include a test case under test/ (see the 
documentation and other tests for examples).

It looks like an IR reduction of your C code in PR18347 will do fine for 
that (clang -cc1 -emit-llvm).

It's also worth copying the write-up from your bug report into the patch 
submission email for context.

Just general tips -- handing over to the module maintainers for actual 
review.

Alp.


>
> Cheers
> Jasper
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

-- 
http://www.nuanti.com
the browser experts