- User Since
- May 24 2016, 8:35 AM (164 w, 5 d)
Thu, Jul 18
Thanks. Looks like a nice change to me, other than one minor modification
Wed, Jul 17
Mon, Jul 15
The original version wasn't getting alignment correct in some cases. For little endian which instruction we choose has more to do with alignment and offset than type being loaded. I've rewritten parts of this to, at least for LE, get this hopefully more correct. I've tried to add BE too, but not added any tests for that yet. We still have a task to sort out BE properly.
Added extra tests
Sun, Jul 14
Sat, Jul 13
Fri, Jul 12
Hello. Can you add some more context? It makes the reviews easier to read.
Tue, Jul 9
Other than a couple of test nits, LGTM.
Now sits on top of D64426, this just handles the MVE side of things.
That test was added in rL365496, to show MVE codegen where no fp instructions are present. This case would be like that without any fpu, just with more __aeabi_fcmpgt and movne's. I think this at least shows the expansion is happening correctly (plus is sidesteps an issues with these being turned into selects, which are not handled yet).
Sure, I can do that.
Mon, Jul 8