x86: Add support for loading 32-bit and 80-bit floats in the x87
Review Request #1910 - Created June 10, 2013 and submitted
| Information | |
|---|---|
| Andreas Sandberg | |
| gem5 | |
| default | |
| Reviewers | |
| Default | |
| gblack | |
Changeset 9757:8b445d3198e0 --------------------------- x86: Add support for loading 32-bit and 80-bit floats in the x87 *** Notes to reviewers *** This changeset requires the presence of the __float80 type which is a GCC extension, I'm not sure if this works with clang. I'll check that before committing the patch.
Quick regressions pass. Solves a bug causing incorrect results in one of the SPEC CPU2006 where the simulated x87 didn't store 32 bit floats correctly. Tested loading and storing of 80-bit floats using a hand-coded assembly test case that switched between a simulated CPU and a hardware CPU to exercise loading and storing using the real thing as a reference.
Review request changed
Updated (June 10, 2013, 7:49 p.m.)
People: |
|
|---|
I don't know about the portability aspects of using __float80, but if others are satisfied with that issue, the rest of this patch looks fine to me.