On 11/19/2015 09:53 AM, Bernd Schmidt wrote:
On 11/19/2015 04:34 PM, Nick Clifton wrote:
Hi Bernd,
I had a look around. There's code testing HARD_REGNO_NREGS in
ree.c:combine_set_extension. It's inside #if 0, and labelled
"temporarily disabled". See if enabling that helps you? (Jeff, that #if
0 was added by you).
I suspect that the code was disabled because it prevented too many
useful optimization opportunities.
Well, correctness goes first. Since reenabling it fixes your problem,
that change is approved (conditional on testing as usual obviously).
I couldn't convince myself it was a correct check. Having Nick's
testcase should help with the analysis.
Frankly, the overall structure of ree is a mess -- I've found it
incredibly difficult to follow every time I've had to look at it.
Let me take a peek today and see if I can get familiar enough with this
code again to help.
Jeff
