Re: An issue of building openjre on CentOS 5.9

2016-05-10 Thread David Holmes

On 11/05/2016 11:56 AM, 31731705 wrote:

Can anybody help me to check this compile error?


OS is CentOS 5.9


[root@localhost openjdk]# uname -a
Linux localhost.localdomain 2.6.18-348.el5PAE #1 SMP Tue Jan 8 18:39:56 EST 
2013 i686 i686 i386 GNU/Linux
[root@localhost openjdk]# cat /etc/redhat-release
CentOS release 5.9 (Final)


--
Configuration summary:
* Debug level:release
* JDK variant:normal
* JVM variants:   server
* OpenJDK target: OS: linux, CPU architecture: x86, address length: 32


Tools summary:
* Boot JDK:   java version "1.7.0_101" OpenJDK Runtime Environment 
(rhel-2.6.6.1.el5_11-i386 u101-b00) OpenJDK Server VM (build 24.95-b01, mixed mode)  (at 
/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.101)
* C Compiler: gcc (GCC) 4.1.2 20080704 (Red Hat-54) version 4.1.2-54) (at 
/usr/bin/gcc)
* C++ Compiler:   g++ (GCC) 4.1.2 20080704 (Red Hat-54) version 4.1.2-54) (at 
/usr/bin/g++)


Your gcc is too old to support the compiler flags that are being used.

David
--



Build performance summary:
* Cores to use:   4
* Memory limit:   16133 MB
* ccache status:  not installed (consider installing)


WARNING: The result of this configuration has overridden an older
configuration. You *should* run 'make clean' to make sure you get a
proper build. Failure to do so might result in strange build problems.




Below is the error information.


/home/rong/openjdk/jdk/src/solaris/native/sun/nio/ch/IOUtil.c: In function 
??Java_sun_nio_ch_IOUtil_fdLimit??:
/home/rong/openjdk/jdk/src/solaris/native/sun/nio/ch/IOUtil.c:132: warning: 
comparison of unsigned expression < 0 is always false
cc1: error: unrecognized command line option "-Wno-error=unused-parameter"
cc1: error: unrecognized command line option "-Wno-error=unused-parameter"
gmake[2]: *** 
[/home/rong/openjdk/build/linux-x86-normal-server-release/jdk/objs/libsctp/SctpNet.o]
 Error 1
gmake[2]: *** Waiting for unfinished jobs
gmake[2]: *** 
[/home/rong/openjdk/build/linux-x86-normal-server-release/jdk/objs/libsctp/SctpChannelImpl.o]
 Error 1
cc1: error: unrecognized command line option "-Wno-error=unused-parameter"
gmake[2]: *** 
[/home/rong/openjdk/build/linux-x86-normal-server-release/jdk/objs/libsctp/SctpServerChannelImpl.o]
 Error 1
gmake[1]: *** [libs-only] Error 2
make: *** [jdk-only] Error 2




Any suggestions are welcome.



Re: RFR: JDK-8155587: Cross compilation may cause compiler warnings for "build" compiler

2016-05-10 Thread Chris Plummer

Hi Erik,

I can confirm the changes resolve the issues I was having. I looked at 
the changes, but don't really have the background to comment on them.


thanks,

Chris

On 5/10/16 9:41 AM, Erik Joelsson wrote:

Hello,

When cross compiling, we compile a subset of the jdk for the build 
host to be able to run jmod and jlink. Configure sets up a toolchain 
for this through variables prefixed "BUILD_" (BUILD_CC etc). The new 
nifty macro we have for conditionally setting compiler flags depending 
on the version of the compiler is currently unaware of this. The 
result is that if the compiler version for the target and build 
toolchain differ, the macro will provide the wrong compiler flags.


This patch makes the version check macro able to handle both the 
target and build toolchain. It also fixes the 
BUILD_CC_DISABLE_WARNING_PREFIX variable which is currently always 
empty. The result is that our cross compile configurations at Oracle 
no longer emits warnings from the build toolchain.


Bug: https://bugs.openjdk.java.net/browse/JDK-8155587
Webrev: http://cr.openjdk.java.net/~erikj/8155587/webrev.top.01/

/Erik




RFR: JDK-8155587: Cross compilation may cause compiler warnings for "build" compiler

2016-05-10 Thread Erik Joelsson

Hello,

When cross compiling, we compile a subset of the jdk for the build host 
to be able to run jmod and jlink. Configure sets up a toolchain for this 
through variables prefixed "BUILD_" (BUILD_CC etc). The new nifty macro 
we have for conditionally setting compiler flags depending on the 
version of the compiler is currently unaware of this. The result is that 
if the compiler version for the target and build toolchain differ, the 
macro will provide the wrong compiler flags.


This patch makes the version check macro able to handle both the target 
and build toolchain. It also fixes the BUILD_CC_DISABLE_WARNING_PREFIX 
variable which is currently always empty. The result is that our cross 
compile configurations at Oracle no longer emits warnings from the build 
toolchain.


Bug: https://bugs.openjdk.java.net/browse/JDK-8155587
Webrev: http://cr.openjdk.java.net/~erikj/8155587/webrev.top.01/

/Erik


Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Dmitry Samersoff
Staffan,

Looks good for me.

-Dmitry


On 2016-05-10 17:22, Staffan Larsen wrote:
> Updated webrev with those changes to the hotspot
> makefile: http://cr.openjdk.java.net/~sla/8156658/webrev.04/
> 
> Thanks,
> /Staffan
> 
>> On 10 maj 2016, at 15:50, Dmitry Samersoff
>> > wrote:
>>
>> Staffan,
>>
>>> Maybe. I didn’t want to change existing behavior in this patch. Which
>>> one do you prefer?
>>
>> The code from jdk/test/Makefile (jaxp uses the same code).
>>
>> -Dmitry
>>
>>
>> On 2016-05-10 16:06, Staffan Larsen wrote:
>>>
 On 10 maj 2016, at 14:46, Dmitry Samersoff
 >
 wrote:

 Staffan,

 Long awaited changes, thank you for doing it.

 Is it possible to use the same code for hotspot/test/Makefile:128 and
 jdk/test/Makefile:84
>>>
>>> Maybe. I didn’t want to change existing behavior in this patch. Which
>>> one do you prefer?
>>>

 84   ifdef ALT_OUTPUTDIR
 85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
 86   else
 87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
 88   endif
 89
 90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
 91   ABS_TEST_OUTPUT_DIR :=
 $(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)

 -Dmitry

 On 2016-05-10 15:21, Staffan Larsen wrote:
> For test automation purposes we want a way to run jtreg tests for
> the different components (jdk, langtools, jaxp, hotspot, nashorn)
> in a unified way from the command line. We also want each component
> to be able to define how jtreg is invoked (problemlists,
> concurrency, agentvm, ...).  There is already some support for
> running jtreg tests from the /test folder, but it is not
> general enough since it isn't possible to pick individual tests or
> specific test groups. It is also not possible to direct the output
> to a common location.
>
> The proposed solution is to be able to run jtreg tests using make
> from within the /test folder like this:
>
> $ make JT_JAVA= JT_HOME= PRODUCT_HOME= product path> \
>   TESTDIRS=../
> TEST_SELECTION= \
>   TEST_OUTPUT_DIR= jtreg_tests
>
> This will run the specified tests and collect output in
> TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add
> additional options to the jtreg command line.
>
> No existing behavior should be changed.
>
> webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/
> 
> bug: https://bugs.openjdk.java.net/browse/JDK-8156658
> 
>
> Thanks,
> /Staffan
>


 -- 
 Dmitry Samersoff
 Oracle Java development team, Saint Petersburg, Russia
 * I would love to change the world, but they won't give me the sources.
>>>
>>
>>
>> -- 
>> Dmitry Samersoff
>> Oracle Java development team, Saint Petersburg, Russia
>> * I would love to change the world, but they won't give me the sources.
> 


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Erik Joelsson
There are several issues that I would like to bring up after this, but I 
really don't want to open that can of worms since the goal here is not 
to do a full cleanup. I'm ok with this.


/Erik

On 2016-05-10 16:22, Staffan Larsen wrote:

Updated webrev with those changes to the hotspot makefile: 
http://cr.openjdk.java.net/~sla/8156658/webrev.04/ 


Thanks,
/Staffan


On 10 maj 2016, at 15:50, Dmitry Samersoff  wrote:

Staffan,


Maybe. I didn’t want to change existing behavior in this patch. Which
one do you prefer?

The code from jdk/test/Makefile (jaxp uses the same code).

-Dmitry


On 2016-05-10 16:06, Staffan Larsen wrote:

On 10 maj 2016, at 14:46, Dmitry Samersoff  wrote:

Staffan,

Long awaited changes, thank you for doing it.

Is it possible to use the same code for hotspot/test/Makefile:128 and
jdk/test/Makefile:84

Maybe. I didn’t want to change existing behavior in this patch. Which one do 
you prefer?


84   ifdef ALT_OUTPUTDIR
85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
86   else
87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
88   endif
89
90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
91   ABS_TEST_OUTPUT_DIR :=
$(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)

-Dmitry

On 2016-05-10 15:21, Staffan Larsen wrote:

For test automation purposes we want a way to run jtreg tests for the different 
components (jdk, langtools, jaxp, hotspot, nashorn) in a unified way from the command 
line. We also want each component to be able to define how jtreg is invoked 
(problemlists, concurrency, agentvm, ...).  There is already some support for running 
jtreg tests from the /test folder, but it is not general enough since it 
isn't possible to pick individual tests or specific test groups. It is also not 
possible to direct the output to a common location.

The proposed solution is to be able to run jtreg tests using make from within the 
/test folder like this:

$ make JT_JAVA= JT_HOME= PRODUCT_HOME= \
   TESTDIRS=../ TEST_SELECTION= \
   TEST_OUTPUT_DIR= jtreg_tests

This will run the specified tests and collect output in TEST_OUTPUT_DIR/jtreg. 
EXTRA_JTREG_OPTIONS is also supported to add additional options to the jtreg 
command line.

No existing behavior should be changed.

webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 

bug: https://bugs.openjdk.java.net/browse/JDK-8156658 


Thanks,
/Staffan



--
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


--
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.




Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Staffan Larsen
Updated webrev with those changes to the hotspot makefile: 
http://cr.openjdk.java.net/~sla/8156658/webrev.04/ 


Thanks,
/Staffan

> On 10 maj 2016, at 15:50, Dmitry Samersoff  
> wrote:
> 
> Staffan,
> 
>> Maybe. I didn’t want to change existing behavior in this patch. Which
>> one do you prefer?
> 
> The code from jdk/test/Makefile (jaxp uses the same code).
> 
> -Dmitry
> 
> 
> On 2016-05-10 16:06, Staffan Larsen wrote:
>> 
>>> On 10 maj 2016, at 14:46, Dmitry Samersoff  
>>> wrote:
>>> 
>>> Staffan,
>>> 
>>> Long awaited changes, thank you for doing it.
>>> 
>>> Is it possible to use the same code for hotspot/test/Makefile:128 and
>>> jdk/test/Makefile:84
>> 
>> Maybe. I didn’t want to change existing behavior in this patch. Which one do 
>> you prefer? 
>> 
>>> 
>>> 84   ifdef ALT_OUTPUTDIR
>>> 85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
>>> 86   else
>>> 87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
>>> 88   endif
>>> 89
>>> 90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
>>> 91   ABS_TEST_OUTPUT_DIR :=
>>> $(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)
>>> 
>>> -Dmitry
>>> 
>>> On 2016-05-10 15:21, Staffan Larsen wrote:
 For test automation purposes we want a way to run jtreg tests for the 
 different components (jdk, langtools, jaxp, hotspot, nashorn) in a unified 
 way from the command line. We also want each component to be able to 
 define how jtreg is invoked (problemlists, concurrency, agentvm, ...).  
 There is already some support for running jtreg tests from the /test 
 folder, but it is not general enough since it isn't possible to pick 
 individual tests or specific test groups. It is also not possible to 
 direct the output to a common location. 
 
 The proposed solution is to be able to run jtreg tests using make from 
 within the /test folder like this: 
 
 $ make JT_JAVA= JT_HOME= PRODUCT_HOME=>>> path> \
   TESTDIRS=../ TEST_SELECTION=>>> test or jtreg group> \
   TEST_OUTPUT_DIR= jtreg_tests 
 
 This will run the specified tests and collect output in 
 TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add 
 additional options to the jtreg command line. 
 
 No existing behavior should be changed.
 
 webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 
 
 bug: https://bugs.openjdk.java.net/browse/JDK-8156658 
 
 
 Thanks,
 /Staffan
 
>>> 
>>> 
>>> -- 
>>> Dmitry Samersoff
>>> Oracle Java development team, Saint Petersburg, Russia
>>> * I would love to change the world, but they won't give me the sources.
>> 
> 
> 
> -- 
> Dmitry Samersoff
> Oracle Java development team, Saint Petersburg, Russia
> * I would love to change the world, but they won't give me the sources.



Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Dmitry Samersoff
Staffan,

> Maybe. I didn’t want to change existing behavior in this patch. Which
> one do you prefer?

The code from jdk/test/Makefile (jaxp uses the same code).

-Dmitry


On 2016-05-10 16:06, Staffan Larsen wrote:
> 
>> On 10 maj 2016, at 14:46, Dmitry Samersoff  
>> wrote:
>>
>> Staffan,
>>
>> Long awaited changes, thank you for doing it.
>>
>> Is it possible to use the same code for hotspot/test/Makefile:128 and
>> jdk/test/Makefile:84
> 
> Maybe. I didn’t want to change existing behavior in this patch. Which one do 
> you prefer? 
> 
>>
>>  84   ifdef ALT_OUTPUTDIR
>>  85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
>>  86   else
>>  87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
>>  88   endif
>>  89
>>  90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
>>  91   ABS_TEST_OUTPUT_DIR :=
>> $(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)
>>
>> -Dmitry
>>
>> On 2016-05-10 15:21, Staffan Larsen wrote:
>>> For test automation purposes we want a way to run jtreg tests for the 
>>> different components (jdk, langtools, jaxp, hotspot, nashorn) in a unified 
>>> way from the command line. We also want each component to be able to define 
>>> how jtreg is invoked (problemlists, concurrency, agentvm, ...).  There is 
>>> already some support for running jtreg tests from the /test folder, 
>>> but it is not general enough since it isn't possible to pick individual 
>>> tests or specific test groups. It is also not possible to direct the output 
>>> to a common location. 
>>>
>>> The proposed solution is to be able to run jtreg tests using make from 
>>> within the /test folder like this: 
>>>
>>>  $ make JT_JAVA= JT_HOME= PRODUCT_HOME=>> path> \
>>>TESTDIRS=../ TEST_SELECTION=>> test or jtreg group> \
>>>TEST_OUTPUT_DIR= jtreg_tests 
>>>
>>> This will run the specified tests and collect output in 
>>> TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add 
>>> additional options to the jtreg command line. 
>>>
>>> No existing behavior should be changed.
>>>
>>> webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 
>>> 
>>> bug: https://bugs.openjdk.java.net/browse/JDK-8156658 
>>> 
>>>
>>> Thanks,
>>> /Staffan
>>>
>>
>>
>> -- 
>> Dmitry Samersoff
>> Oracle Java development team, Saint Petersburg, Russia
>> * I would love to change the world, but they won't give me the sources.
> 


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Staffan Larsen

> On 10 maj 2016, at 14:46, Dmitry Samersoff  
> wrote:
> 
> Staffan,
> 
> Long awaited changes, thank you for doing it.
> 
> Is it possible to use the same code for hotspot/test/Makefile:128 and
> jdk/test/Makefile:84

Maybe. I didn’t want to change existing behavior in this patch. Which one do 
you prefer? 

> 
>  84   ifdef ALT_OUTPUTDIR
>  85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
>  86   else
>  87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
>  88   endif
>  89
>  90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
>  91   ABS_TEST_OUTPUT_DIR :=
> $(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)
> 
> -Dmitry
> 
> On 2016-05-10 15:21, Staffan Larsen wrote:
>> For test automation purposes we want a way to run jtreg tests for the 
>> different components (jdk, langtools, jaxp, hotspot, nashorn) in a unified 
>> way from the command line. We also want each component to be able to define 
>> how jtreg is invoked (problemlists, concurrency, agentvm, ...).  There is 
>> already some support for running jtreg tests from the /test folder, but 
>> it is not general enough since it isn't possible to pick individual tests or 
>> specific test groups. It is also not possible to direct the output to a 
>> common location. 
>> 
>> The proposed solution is to be able to run jtreg tests using make from 
>> within the /test folder like this: 
>> 
>>  $ make JT_JAVA= JT_HOME= PRODUCT_HOME=> path> \
>>TESTDIRS=../ TEST_SELECTION=> test or jtreg group> \
>>TEST_OUTPUT_DIR= jtreg_tests 
>> 
>> This will run the specified tests and collect output in 
>> TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add 
>> additional options to the jtreg command line. 
>> 
>> No existing behavior should be changed.
>> 
>> webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 
>> 
>> bug: https://bugs.openjdk.java.net/browse/JDK-8156658 
>> 
>> 
>> Thanks,
>> /Staffan
>> 
> 
> 
> -- 
> Dmitry Samersoff
> Oracle Java development team, Saint Petersburg, Russia
> * I would love to change the world, but they won't give me the sources.



Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Dmitry Samersoff
Staffan,

Long awaited changes, thank you for doing it.

Is it possible to use the same code for hotspot/test/Makefile:128 and
 jdk/test/Makefile:84

  84   ifdef ALT_OUTPUTDIR
  85 ABS_OUTPUTDIR = $(shell $(CD) $(ALT_OUTPUTDIR) && $(PWD))
  86   else
  87 ABS_OUTPUTDIR = $(shell $(CD) $(TEST_ROOT)/.. && $(PWD))
  88   endif
  89
  90   ABS_PLATFORM_BUILD_ROOT = $(ABS_OUTPUTDIR)
  91   ABS_TEST_OUTPUT_DIR :=
$(ABS_PLATFORM_BUILD_ROOT)/testoutput/$(UNIQUE_DIR)

-Dmitry

On 2016-05-10 15:21, Staffan Larsen wrote:
> For test automation purposes we want a way to run jtreg tests for the 
> different components (jdk, langtools, jaxp, hotspot, nashorn) in a unified 
> way from the command line. We also want each component to be able to define 
> how jtreg is invoked (problemlists, concurrency, agentvm, ...).  There is 
> already some support for running jtreg tests from the /test folder, but 
> it is not general enough since it isn't possible to pick individual tests or 
> specific test groups. It is also not possible to direct the output to a 
> common location. 
> 
> The proposed solution is to be able to run jtreg tests using make from within 
> the /test folder like this: 
> 
>   $ make JT_JAVA= JT_HOME= PRODUCT_HOME= path> \
> TESTDIRS=../ TEST_SELECTION= test or jtreg group> \
> TEST_OUTPUT_DIR= jtreg_tests 
> 
> This will run the specified tests and collect output in 
> TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add 
> additional options to the jtreg command line. 
> 
> No existing behavior should be changed.
> 
> webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 
> 
> bug: https://bugs.openjdk.java.net/browse/JDK-8156658 
> 
> 
> Thanks,
> /Staffan
> 


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Staffan Larsen
Thanks Erik!

> On 10 maj 2016, at 14:39, Erik Joelsson  wrote:
> 
> Looks good to me.
> 
> /Erik
> 
> On 2016-05-10 14:21, Staffan Larsen wrote:
>> For test automation purposes we want a way to run jtreg tests for the 
>> different components (jdk, langtools, jaxp, hotspot, nashorn) in a unified 
>> way from the command line. We also want each component to be able to define 
>> how jtreg is invoked (problemlists, concurrency, agentvm, ...).  There is 
>> already some support for running jtreg tests from the /test folder, but 
>> it is not general enough since it isn't possible to pick individual tests or 
>> specific test groups. It is also not possible to direct the output to a 
>> common location.
>> 
>> The proposed solution is to be able to run jtreg tests using make from 
>> within the /test folder like this:
>> 
>>   $ make JT_JAVA= JT_HOME= PRODUCT_HOME=> path> \
>> TESTDIRS=../ TEST_SELECTION=> test or jtreg group> \
>> TEST_OUTPUT_DIR= jtreg_tests
>> 
>> This will run the specified tests and collect output in 
>> TEST_OUTPUT_DIR/jtreg. EXTRA_JTREG_OPTIONS is also supported to add 
>> additional options to the jtreg command line.
>> 
>> No existing behavior should be changed.
>> 
>> webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 
>> 
>> bug: https://bugs.openjdk.java.net/browse/JDK-8156658 
>> 
>> 
>> Thanks,
>> /Staffan
> 



Re: RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Erik Joelsson

Looks good to me.

/Erik

On 2016-05-10 14:21, Staffan Larsen wrote:

For test automation purposes we want a way to run jtreg tests for the different 
components (jdk, langtools, jaxp, hotspot, nashorn) in a unified way from the command 
line. We also want each component to be able to define how jtreg is invoked 
(problemlists, concurrency, agentvm, ...).  There is already some support for running 
jtreg tests from the /test folder, but it is not general enough since it 
isn't possible to pick individual tests or specific test groups. It is also not 
possible to direct the output to a common location.

The proposed solution is to be able to run jtreg tests using make from within the 
/test folder like this:

   $ make JT_JAVA= JT_HOME= PRODUCT_HOME= \
 TESTDIRS=../ TEST_SELECTION= \
 TEST_OUTPUT_DIR= jtreg_tests

This will run the specified tests and collect output in TEST_OUTPUT_DIR/jtreg. 
EXTRA_JTREG_OPTIONS is also supported to add additional options to the jtreg 
command line.

No existing behavior should be changed.

webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 

bug: https://bugs.openjdk.java.net/browse/JDK-8156658 


Thanks,
/Staffan




RFR: JDK-8156658 Common way to run jtreg tests

2016-05-10 Thread Staffan Larsen
For test automation purposes we want a way to run jtreg tests for the different 
components (jdk, langtools, jaxp, hotspot, nashorn) in a unified way from the 
command line. We also want each component to be able to define how jtreg is 
invoked (problemlists, concurrency, agentvm, ...).  There is already some 
support for running jtreg tests from the /test folder, but it is not 
general enough since it isn't possible to pick individual tests or specific 
test groups. It is also not possible to direct the output to a common location. 

The proposed solution is to be able to run jtreg tests using make from within 
the /test folder like this: 

  $ make JT_JAVA= JT_HOME= PRODUCT_HOME= \
TESTDIRS=../ TEST_SELECTION= \
TEST_OUTPUT_DIR= jtreg_tests 

This will run the specified tests and collect output in TEST_OUTPUT_DIR/jtreg. 
EXTRA_JTREG_OPTIONS is also supported to add additional options to the jtreg 
command line. 

No existing behavior should be changed.

webrev: http://cr.openjdk.java.net/~sla/8156658/webrev.03/ 

bug: https://bugs.openjdk.java.net/browse/JDK-8156658 


Thanks,
/Staffan

Re: RFR: 8156561: Remove makeClasslist.js

2016-05-10 Thread Claes Redestad

Thanks for the quick reviews!

/Claes

On 2016-05-10 13:50, Chris Hegarty wrote:

On 10 May 2016, at 12:42, Claes Redestad  wrote:


Hi,

the tool makeClasslist.js lost it's purpose after JDK-8150044, so should be 
removed.

Bug: https://bugs.openjdk.java.net/browse/JDK-8156561

jdk repo$ hg rm make/non-build-utils/src/build/tools/makeclasslist

Looks good Claes.

-Chris.





Re: RFR: 8156561: Remove makeClasslist.js

2016-05-10 Thread Erik Joelsson

Looks good.

/Erik

On 2016-05-10 13:42, Claes Redestad wrote:

Hi,

the tool makeClasslist.js lost it's purpose after JDK-8150044, so 
should be removed.


Bug: https://bugs.openjdk.java.net/browse/JDK-8156561

jdk repo$ hg rm make/non-build-utils/src/build/tools/makeclasslist

Thanks!

/Claes




RFR: 8156561: Remove makeClasslist.js

2016-05-10 Thread Claes Redestad

Hi,

the tool makeClasslist.js lost it's purpose after JDK-8150044, so should 
be removed.


Bug: https://bugs.openjdk.java.net/browse/JDK-8156561

jdk repo$ hg rm make/non-build-utils/src/build/tools/makeclasslist

Thanks!

/Claes