Fwd: Re: hsx24 backport: Request for review: 7122222: GC log is limited to 2G for 32-bit

Tao Mao tao.mao at oracle.com
Tue Jul 2 21:09:13 UTC 2013


Thank you for review, Dan.

Tao

On 7/2/13 1:32 PM, Daniel D. Daugherty wrote:
> Adding hotspot-runtime-dev at openjdk.java.net to this thread...
>
> > http://cr.openjdk.java.net/~tamao/7122222_hsx24/webrev.01/
>
> make/linux/makefiles/vm.make
>     No comments.
>
> make/solaris/makefiles/vm.make
>     No comments.
>
> src/os/solaris/vm/os_solaris.inline.hpp
>     No comments.
>
> Also compared the two patch files:
>     7122222GCLogLimitedTo2GFor32Bit_hsx24.patch
>     7122222GCLogLimitedTo2GFor32Bit.patch
>
> In HSX-25, the Makefile construct looks like:
>
> CXXFLAGS/ostream.o += -D_FILE_OFFSET_BITS=64
>
> In HSX-24, the Makefile construct looks like:
>
> ostream.o: CXXFLAGS += -D_FILE_OFFSET_BITS=64
>
> which threw me for a loop for a minute... I realized that
> your construct matches the existing rules for vm_version.o
> in each release so I think the HSX-24 backport is good.
I should've mentioned the point that I just would like to keep 
consistency with the existing way of passing cxxflag to vm_version.o in 
order to make the changeset simple to come up.
>
> Hopefully, you double checked builds logs on both Linux
> and Solaris and made sure that '-D_FILE_OFFSET_BITS=64'
> is only passed where you expect it.
Yes, it is only passed "-D_FILE_OFFSET_BITS=64" to ostream.o

As I mentioned, the changeset passed JPRT.

Also, I manually tested whether the change would solve the CR's problem 
as expected. It's doing its right job on Linux-i586, Solaris-i586 and 
Solaris-sparc.
>
> Thumbs up!
>
> Dan
>
>
>> -------- Original Message --------
>> Subject: 	Re: hsx24 backport: Request for review: 7122222: GC log is 
>> limited to 2G for 32-bit
>> Date: 	Mon, 01 Jul 2013 23:05:46 -0700
>> From: 	Tao Mao <tao.mao at oracle.com>
>> Organization: 	Oracle Corporation
>> To: 	hotspot-gc-dev at openjdk.java.net 
>> <hotspot-gc-dev at openjdk.java.net>, build-dev at openjdk.java.net
>>
>>
>>
>> The format of passing cxxflag to a specific file is different in hsx24
>> than in hsx25. I've changed the setting accordingly in order to pass
>> through the compilation.
>>
>> Please review the new webrev below and ignore webrev.00.
>>
>> hsx24 webrev:
>> http://cr.openjdk.java.net/~tamao/7122222_hsx24/webrev.01/
>>
>> (original) hsx25 webrev:
>> http://cr.openjdk.java.net/~tamao/7122222/webrev.00/
>>
>> Test:
>> Builds were tested on Linux-i586, Solaris-i586, and Solaris-sparc.
>> Builds were successful and they all passed test of the gc-log size limit
>> of 2G.
>>
>> Thanks.
>> Tao
>>
>> On 7/1/13 1:18 PM, Tao Mao wrote:
>> >  The hsx25 fix has been pushed already. Then I got the
>> >  7u40-critical-approved.
>> >
>> >  When I tried backporting to hsx24, it didn't apply to the hsx24 repo
>> >  cleanly. Thus, I made the patch manually (copy-and-paste style) and
>> >  now need some quick reviews to get it in since it's 7u40 related P3 CR.
>> >
>> >  hsx24 webrev:
>> >  http://cr.openjdk.java.net/~tamao/7122222_hsx24/webrev.00/
>> >
>> >  (original) hsx25 webrev:
>> >  http://cr.openjdk.java.net/~tamao/7122222/webrev.00/
>> >
>> >  Thanks.
>> >  Tao
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20130702/5a827677/attachment.htm>


More information about the hotspot-gc-dev mailing list