RFR: 8258856: VM build without C1/C2 fails after JDK-8243205 [v3]

[David Holmes]

Hi Hao,

On 24/12/2020 4:56 pm, Hao Sun wrote:
> On Thu, 24 Dec 2020 06:16:50 GMT, David Holmes <dholmes at openjdk.org> wrote:
> 
>> Just an observation but it seems to me that if we are not including C1 or C2 in a build (ie interpreter-only) then really the build system should not be building the source files that had to be fixed.
>>
>> David
> 
> Hi David,
> I guess "not including C1 or C2 in a build (ie interpreter-only)" refers to the zero build.
> Yes. Regarding zero build, these source files fixed in this patch would not be compiled.
> 
> However, the build failures this patch aimed to address are 'server build without C1 feature' and 'server build without C2 feature'.
> 'C1/C2' in the title of this PR means 'build without C1' or 'build without C2'.

Sorry I did misunderstand the context of the problem. And my comment was 
wrong as JVMCI can be used to replace C1 and/or C2, so a build without 
both them need not be interpreter-only.

David

> As stated in JDK-8258856, the build failures can be reproduced by
> ` ./configure --with-jvm-features=-compiler1 && make images`
> and ` ./configure --with-jvm-features=-compiler2 && make images` respectively.
> 
> Hope that I made myself understood. Thanks.
> Hao
> 
> -------------
> 
> PR: https://git.openjdk.java.net/jdk/pull/1876
>