[LLVMbugs] [Bug 23424] Mesa llvmpipe piglit gl-3.2-layered-rendering-gl-layer regression

[bugzilla-daemon at llvm.org]

https://llvm.org/bugs/show_bug.cgi?id=23424

sroland at vmware.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |INVALID

--- Comment #3 from sroland at vmware.com ---
Ahh this is indeed a bug in mesa with an out-of-bounds index, I'll fix that.

It did not show up in the IR because whenever we dump the IR the out-of-bound
index is already optimized away... At least I don't know of any way to dump the
"raw" IR (we just do a LLVMDumpValue() on the function) which is actually
something I missed before :-(.

(It did show up as a difference to the old IR though, previously it was saying
  %0 = insertelement <4 x i32> undef, i32 %num_prims, i32 0
  %1 = shufflevector <4 x i32> %0, <4 x i32> undef, <4 x i32> zeroinitializer
  %2 = icmp sgt <4 x i32> %1, <i32 0, i32 1, i32 2, i32 3>

whereas now it was
  %0 = insertelement <4 x i32> undef, i32 %num_prims, i32 0
  %1 = shufflevector <4 x i32> %0, <4 x i32> undef, <4 x i32> zeroinitializer
  %2 = icmp sgt <4 x i32> %1, undef
)
That 0/1/2/3 vec4 was initialized with elements/indices ranging from 0-4
before.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20150506/cb93dd7e/attachment.html>