[LLVMdev] SCCP and undef branches

[Nick Lewycky]

I found that "undef" was disappearing early into the optimization chain.
SCCP was the culprit, transforming:

  br bool undef, label %T, label %F

into

  br bool true, label %T, label %F

While that sounds like a great optimization, it shouldn't be happening
that early. I've put together a patch which modifies the behaviour of
SCCP so that it preserves undef in those cases.

I had to modify one of the regression tests which was trying to test
IPSCCP on global variables, but was uses undef and expecting the above
transformation to take place. I changed its global from undef to true,
and wrote a new testcase for the undef-related part.

Patch attached. Comments?

Nick Lewycky

-------------- next part --------------
A non-text attachment was scrubbed...
Name: sccp-branch-undef.patch
Type: text/x-patch
Size: 3800 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20060605/1da45df2/attachment.bin>