Hotspot build stutter

[Kelly O'Hair]

FYI...

The LOG=info,nofile   did not seem to prevent hotspot makefile from being run with the BUILD_LOG_WRAPPER.

I assume that is because the HotspotWrapper.gmk makefile does not include the MakeHelper.gmk file?

-kto

On Sep 14, 2012, at 5:35 AM, David Holmes wrote:

> Magnus,
> 
> I think this is the new HotspotWrapper.gmk:
> 
> $(HOTSPOT_OUTPUTDIR)/_hotspot.timestamp: $(HOTSPOT_FILES)
>         @($(CD) $(HOTSPOT_TOPDIR)/make && $(BUILD_LOG_WRAPPER) $(MAKE) -j1 $(HOTSPOT_MAKE_ARGS) SPEC=$(HOTSPOT_SPEC) BASE_SPEC=$(BASE_SPEC))
>         $(TOUCH) $@
> 
> Here we invoke the sub-make using the build-log-wrapper, but we're already executing in the build-log-wrapper, so any output would, I think, go to both build logs - and as the log file is the same we get the double output.
> 
> David
> 
> On 14/09/2012 10:04 PM, David Holmes wrote:
>> On 14/09/2012 9:12 PM, Magnus Ihse Bursie wrote:
>>> On 2012-09-14 09:32, David Holmes wrote:
>>>> Creating vm.make ...
>>>> Creating vm.make ...
>>>> Creating adlc.make ...
>>>> Creating adlc.make ...
>>>> Creating jvmti.make ...
>>>> Creating jvmti.make ...
>>>> 
>>>> It seems to report everything twice.
>>>> It seems to report everything twice.
>>> 
>>> Is this different in build-infra than in the mainline?
>>> Is this different in build-infra than in the mainline?
>> 
>> Yes it is only in build-infra.
>> 
>>> In any case, there's a lot of code duplication in the hotspot makefiles,
>>> which accounts for at least some duplicate output.
>>> In any case, there's a lot of code duplication in the hotspot makefiles,
>>> which accounts for at least some duplicate output.
>> 
>> No this duplicates every single output line.
>> 
>> David
>> 
>>> /Magnus :)
>>>