[8u] RFR + RFA: 8213419: "C2 may hang in MulLNode::Ideal()/MulINode::Ideal() with gcc 8.2.1" and follow up 8214206

[Aleksey Shipilev]

On 2/8/19 4:40 PM, Aleksey Shipilev wrote:
> On 2/6/19 5:54 PM, Roland Westrelin wrote:
>>
>> I'd like to backport 8213419 and follow up 8214206 to 8u as they can
>> cause c2 to enter an infinite loop in some corner cases when compiled
>> with a recent version of gcc.
>>
>> Neither changes apply cleanly. In particular I dropped the
>> STATIC_ASSERT() from the globalDefinitions.hpp function. My
>> understanding is that keeping them requires a backport of 8181449: "Fix
>> debug.hpp / globalDefinitions.hpp dependency inversion", a big and
>> complicated change.
>>
>> Webrevs:
>>
>> http://cr.openjdk.java.net/~roland/8213419.8u/webrev.00/
> 
> Changes look good.
> 
> Where's the regression test? MultiplyByIntegerMinHang.java is in original change.
> 
>> http://cr.openjdk.java.net/~roland/8214206.8u/webrev.00/
> 
> Changes look good.
> 
> Let me build it on a few platforms (e.g. 32-bits) to check everything is fine.

Builds fine on:
 * linux-x86-normal-minimal1-fastdebug
 * linux-x86_64-normal-server-fastdebug
 * linux-arm-normal-zero-fastdebug
 * linux-x86-normal-server-fastdebug
 * linux-x86_64-normal-zero-fastdebug

-Aleksey