[patch] fix hotspot build with small SC_ARG_MAX
Daniel D. Daugherty
daniel.daugherty at oracle.com
Sun Sep 19 16:19:08 PDT 2010
I guess that three line comment before that block doesn't do
the job. :-( Can you please suggest a change to the comment
to make it more clear.
The file lists are generated at variable expansion time so we
have to do the remove at variable expansion time also. The
round 0 fix for 6985848 did the remove as a part of rule
execution, but that's too late and causes the javac to fail.
Dan
On 9/19/2010 2:08 PM, Kelly O'Hair wrote:
> Why are you using
> $(shell rm -f -r ... )
> instead of
> $(RM) -r ...
>
> ???
>
>
> On Sep 17, 2010, at 9:49 PM, Daniel D. Daugherty wrote:
>
>> I messed up the rule expansion... :-(
>>
>> Here's another shot:
>>
>> http://cr.openjdk.java.net/~dcubed/6985848-webrev/1/
>>
>> This time I added a comment so the subtle difference
>> between macro/variable expansion versus rule execution
>> is a little more clear. I hope...
>>
>> Dan
>>
>>
>>
>> On 9/17/2010 2:31 PM, Daniel D. Daugherty wrote:
>>> Greetings,
>>>
>>> The fix for this bug (6561870) has caused the sa-jdi.jar file to
>>> always be rebuilt. I have a minor tweak that fixes that problem:
>>>
>>> http://cr.openjdk.java.net/~dcubed/6985848-webrev/0/
>>>
>>> Thanks to Coleen for spotting this issue!
>>>
>>> The folks on the "To" list were on the original review team
>>> for 6561870. I would like to hear from at least two of those
>>> folks on this tweak.
>>>
>>> Thanks, in advance, for the reviews!
>>>
>>> Dan
>>>
>>>
>>>
>>> On 9/3/2010 6:12 PM, Daniel D. Daugherty wrote:
>>>> Here is the revised webrev:
>>>>
>>>> http://cr.openjdk.java.net/~dcubed/6561870-webrev/1/
>>>>
>>>> One reviewer will do, I think...
>>>>
>>>> Dan
>>>>
>>>>
>>>> On 9/1/2010 12:17 PM, Tom Rodriguez wrote:
>>>>> Can the agent list files go into $(GENERATED) instead of
>>>>> $(TOPDIR)? Otherwise this looks good.
>>>>>
>>>>> tom
>>>>>
>>>>> On Sep 1, 2010, at 10:55 AM, Daniel D. Daugherty wrote:
>>>>>
>>>>>
>>>>>> I'm finally getting back to this thread. Since I'm also applying
>>>>>> Matthias' changes to the Solaris sa.makefile, I figured I would
>>>>>> send out a webrev:
>>>>>>
>>>>>> http://cr.openjdk.java.net/~dcubed/6561870-webrev/0/
>>>>>>
>>>>>> Kelly and Christian, can I get re-reviews from the two of you?
>>>>>>
>>>>>> Matthias, can you verify that I got the Solaris port of your
>>>>>> changes correct?
>>>>>>
>>>>>> Thanks, in advance, for any reviews...
>>>>>>
>>>>>> Dan
>>>>>>
>>>>>>
>>>>>> On 8/25/2010 10:22 AM, Daniel D. Daugherty wrote:
>>>>>>
>>>>>>> On 8/25/2010 10:11 AM, Christian Thalinger wrote:
>>>>>>>
>>>>>>>> On Wed, 2010-08-25 at 10:00 -0600, Daniel D. Daugherty wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>> Thanks for the reviews Kelly and Christian.
>>>>>>>>>
>>>>>>>>> Is there a reason to not apply these changes to the solaris and
>>>>>>>>> windows versions also? I haven't looked yet, but I figured I
>>>>>>>>> would ask...
>>>>>>>>>
>>>>>>>> Yeah, I also thought about this for Solaris and it should work
>>>>>>>> there as
>>>>>>>> on Linux. Don't know about Windows, though.
>>>>>>>>
>>>>>>> Looks like Windows already uses a "make" construct instead of
>>>>>>> running
>>>>>>> into the shell command line limitation:
>>>>>>>
>>>>>>>
>>>>>>>> $(GENERATED)\sa-jdi.jar: $(AGENT_FILES1:/=\) $(AGENT_FILES2:/=\)
>>>>>>>>
>>>>>>> <snip>
>>>>>>>
>>>>>>>
>>>>>>>> @$(COMPILE_JAVAC) -source 1.4 -target 1.4 -classpath
>>>>>>>> $(SA_CLASSPATH) -so
>>>>>>>> urcepath $(AGENT_SRC_DIR) -d $(SA_CLASSDIR) $(AGENT_FILES1:/=\)
>>>>>>>> @$(COMPILE_JAVAC) -source 1.4 -target 1.4 -classpath
>>>>>>>> $(SA_CLASSPATH) -so
>>>>>>>> urcepath $(AGENT_SRC_DIR) -d $(SA_CLASSDIR) $(AGENT_FILES2:/=\)
>>>>>>>>
>>>>>>> I also think this is an "nmake" Makefile instead of a GNU Makefile.
>>>>>>> I think I'll look at applying the changes to the Solaris
>>>>>>> Makefile and
>>>>>>> leave the Windows stuff alone :-)
>>>>>>>
>>>>>>> Dan
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>
>
More information about the hotspot-dev
mailing list