First build-infra push to jdk8 -- try out the new build system!

[Artem Ananiev]

On 4/27/2012 2:46 PM, David Holmes wrote:
> Could it be that the parallelism of the new build is simply too high for
> your system - hence you get memory exhaustion?

Yes, that's possible, because my desktop has 4Gb RAM. Number of CPU 
cores is only 2, though, so it's not massively parallel. I'm now 
building in 32-bit mode, will check if it helps.

Thanks,

Artem

> David
>
> On 27/04/2012 7:56 PM, Artem Ananiev wrote:
>> Hi, Magnus,
>>
>> thanks for your feedback. Here is what I have already tried:
>>
>> 1. Disable anti-virus
>>
>> 2. Upgrade sh/bash/make to the latest or to the stable versions
>>
>> 3. Disable DEP for sh.exe/bash.exe
>>
>> 4. Run rebaseall/peflagsall
>>
>> 5. Re-run make
>>
>> 6. Reboot
>>
>> It still fails. I cannot say for sure if it fails in the same place, but
>> always in "jdk". All other workspaces are built fine: hotspot,
>> langtools, etc.
>>
>> Thanks,
>>
>> Artem
>>
>> On 4/27/2012 12:36 PM, Magnus Ihse Bursie wrote:
>>> On 2012-04-27 09:38, Artem Ananiev wrote:
>>>>
>>>>
>>>> 15830981 [main] sh 7892 C:\cygwin\bin\sh.exe: *** fatal error - fork:
>>>> can't reserve memory for stack 0x29BCF0 - 0x2A0000, Win32 error
>>>> 48715830981 [main] sh 7892 C:\cygwin\bin\sh.exe: *** fatal error -
>>>> fork: can't reserve memory for stack 0x29BCF0 - 0x2A0000, Win32 error
>>>> 487
>>>
>>> This is a cygwin problem, which pops up every now and then when running
>>> cygwin.
>>>
>>> The basic problem is that cygwin tries to emulate posix behaviour on
>>> windows, but the fork semantics is not available and so it has to be
>>> "faked". Sometimes this fails. It seems to fail more often on Windows 7.
>>> Running make files often causes a lot of forks, so it increases the
>>> likelihood of failure.
>>>
>>> Please just try to re-run make and see if that solves your problem.
>>>
>>> If it repeats and seems to happen at the same place, let us known. If
>>> that is the case, we might find some specific circumstances that makes
>>> it more likely to happen, and maybe we can modify that slightly.
>>>
>>> /Magnus
>>>