This is the mail archive of the
mailing list for the GCC project.
Patch for gcc.dg/tree-ssa/ivopts-2.c failure in LP64 mode.
- From: Adam Nemet <anemet at caviumnetworks dot com>
- To: sje at cup dot hp dot com
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 13 Oct 2006 15:24:52 -0700
- Subject: Patch for gcc.dg/tree-ssa/ivopts-2.c failure in LP64 mode.
- References: <200610132119.OAA17399@hpsje.cup.hp.com>
Steve Ellcey writes:
> I think it fails in LP64 mode because int and long are different
> sizes in that mode and thus the variables 'data_offset' and 'store'
> are different sizes and can't be merged into a single index
As you mentioned it too, this test is meant to be check LP64 targets.
The target that I found it on (mipsisa64-elf) is LP64.
When ivopts estimates the cost of producing the uses from new
candidates it actually generates RTL. So chances are some of those
costs come out differently on your target and hence the different
It could also be that the fix was not generic enough and you're seeing
a slight variation of the original problem.