[llvm] r178031 - Make InstCombineCasts.cpp:OptimizeIntToFloatBitCast endian safe.

Ulrich Weigand Ulrich.Weigand at de.ibm.com
Tue Mar 26 10:12:56 PDT 2013


Duncan Sands <duncan.sands at gmail.com> wrote on 26.03.2013 18:00:50:
> On 26/03/13 17:46, Ulrich Weigand wrote:
> > Hmm, it fixed an existing test case failure for me on PowerPC.
> > But I can try and see if I manage to extract a .ll level test case.
>
> yes please, that way people not on PowerPC will notice if they break
this.

Right.  It turns out there was an existing test case for this feature
on a little-endian target; in the attached patch I'm adding a copy of
the test for a big-endian target and verify that we get the correct
element indices there.  Does this look OK?

> > I'm sorry, but I don't see any non-vector case in this routine ...
> > Could you elaborate?
>
> Hrm, you are right.  Sorry about that, I was going by memory and I
remembered
> this routine wrong.

OK, thanks!

Bye,
Ulrich

(See attached file: diff-llvm-instcombine-endianfix-test)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: diff-llvm-instcombine-endianfix-test
Type: application/octet-stream
Size: 2066 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130326/0ea4cd31/attachment.obj>


More information about the llvm-commits mailing list