RFR: JDK-8205942: Build failure on macosx after JDK-8189429

[Tim Bell]

Erik:

Looks good to me as well.

/Tim

On 06/27/18 15:32, Lance Andersen wrote:
> I can verify the patch worked for me Erik
>
> Best
> Lance
>> On Jun 27, 2018, at 6:29 PM, Erik Joelsson <erik.joelsson at oracle.com> wrote:
>>
>> For some users, the build fails with:
>>
>> dirname: illegal option -- s
>> usage: dirname path
>> usage: basename string [suffix]
>>         basename [-a] [-s suffix] string [...]
>> error: cannot read file /Users/danielfuchs/workspaces/jdk/jdk-dev/build/macosx-x64/support/gensrc/jdk.hotspot.agent/mach_excServer.c
>> Gensrc-jdk.hotspot.agent.gmk:52: recipe for target '/Users/danielfuchs/workspaces/jdk/jdk-dev/build/macosx-x64/support/gensrc/jdk.hotspot.agent/mach_excServer.c' failed
>> make[3]: *** [/Users/danielfuchs/workspaces/jdk/jdk-dev/build/macosx-x64/support/gensrc/jdk.hotspot.agent/mach_excServer.c] Error 1
>> make/Main.gmk:112: recipe for target 'jdk.hotspot.agent-gensrc-src' failed
>> make[2]: *** [jdk.hotspot.agent-gensrc-src] Error 2
>> make[2]: *** Waiting for unfinished jobs....
>>
>> I have identified two reasons for failures here and this patch makes it work for at least one reporter of the problem. The missing MakeTargetDir caues a race where the target directory may not exist yet. It seems mig on newer Macos handles this anyway. For some configurations, the SDKROOT variable does not have a value. In that case, we should not set the -isysroot parameter to mig. (I also changed the MKDIR in the other recipe to the now preferred MakeTargetDir macro)
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8205942
>>
>> Webrev: http://cr.openjdk.java.net/~erikj/8205942/webrev.01/
>>
>> /Erik