RFR [jdk11]: 8232084: HotSpot build failed with GCC 9.2.1

[Yasumasa Suenaga]

Hi Matthias,

Your change looks good. Thank you for fixing them!
However I'm not sure we can include backport of 8220074 (LCMS) to this (8232084).
Can we merge them to single commit?


Thanks,

Yasumasa


On 2021/05/13 0:37, Baesken, Matthias wrote:
> Hi , here is a new webrev :
> 
> 
> http://cr.openjdk.java.net/~mbaesken/webrevs/8232084.1_jdk11/
> 
>>>
>>>> However, to disable warnings in third party libraries like liblcms, we can disable them at the makefile for them.
>>>> In case of liblcms, it is prefer to disable stringop-truncation at Awt2dLibraries.gmk .
>>>>
>>>
>>> Hi Yasumasa , are you fine with the rest of the patch ?
>>
>> Yes, looks good to me.
>>
>>
>>> Regarding  disabling of stringop-truncation  , I first  disabled it only for liblcms but this was not sufficient.
>>> (there is indeed another change   8220074   for liblcms that disabled this warning for GCC 8.3 )
>>>    There are more issues related to  the no-stringop-truncation  warning and I did not even found a good place to disable them locally in a nice way .
>>>    (some strange jvmti related test coding I think those are single file compilations , no real lib??)
>>
>> I guess the libraries which you mean are located to build/linux-x86_64-server-fastdebug/images/test/hotspot/jtreg/native .
>> If so, do you see stringop-truncation error on JVMTI tests only? Have you considered to use pragma?
>> If you don't want to use pragma, can you use CFLAGS in JtregNativeHotspot.gmk ?
>>
>>    https://hg.openjdk.java.net/jdk-updates/jdk11u-dev/file/0f3f7aa2ef34/make/test/JtregNativeHotspot.gmk#l210
> 
> 
>   Unfortunately  the jvmti_tools.c   is included  into a ton of compilation units  so  setting it  in JtregNativeHotspot.gmk  does not look good to me.
> But the pragma works, did that .
> Additionally the liblcms needs the stringop-truncation  disabling as well .
> 
> Best regards, Matthias
>