On 22/03/2018 12:51 AM, Daniel D. Daugherty wrote:
On 3/21/18 10:23 AM, Daniel D. Daugherty wrote:
On 3/21/18 10:00 AM, Langer, Christoph wrote:
Hi Dan,
that is, you mean the C-code? My original change?
Hmmm... I think I confused myself before I drank enough coffee...
Looking again...
Okay I definitely confused myself... and I clearly don't remember
the attach-on-demand code as well as I used to... sigh...
I think you should keep your original fix since it now properly
handles null arguments at the same attach-on-demand layer as the
Linux code that you quoted.
Handling this in args array processing would also be possible
as David suggests, but it would bother me that Linux and Solaris
lower attach-on-demand layers would have different behaviors.
They already do have completely different behaviours. Linux handles NULL
at the Java layer by inserting empty strings!
David
Hope this is more clear.
Dan
Dan
Best regards
Christoph
-----Original Message-----
From: Daniel D. Daugherty [mailto:[email protected]]
Sent: Mittwoch, 21. März 2018 14:59
To: Langer, Christoph <[email protected]>; David Holmes
<[email protected]>; [email protected]
Subject: Re: RFR(S) 8199924: Solaris: Correctly enqueue null
arguments of
attach operations
Hmmm... shouldn't the inconsistency in the Solaris backend also be
addressed?
Dan
On 3/21/18 8:45 AM, Langer, Christoph wrote:
Hi David,
thanks for looking at this. I currently have no emotions whether to
fix it in C
or in Java - I'll check it out...
Best regards
Christoph
-----Original Message-----
From: David Holmes [mailto:[email protected]]
Sent: Mittwoch, 21. März 2018 10:20
To: Langer, Christoph <[email protected]>; serviceability-
[email protected]
Subject: Re: RFR(S) 8199924: Solaris: Correctly enqueue null
arguments of
attach operations
Hi Christoph,
On 21/03/2018 6:10 PM, Langer, Christoph wrote:
Hi,
may I please ask for reviews of the following small fix.
Webrev: http://cr.openjdk.java.net/~clanger/webrevs/8199924.0/
Bug: https://bugs.openjdk.java.net/browse/JDK-8199924
If one passes null arguments to the varargs of attach operations,
they
get swallowed on Solaris and following arguments will shift to lower
positions.
Other platform implementations handle this correctly, for instance
linux:
http://hg.openjdk.java.net/jdk/jdk/file/f6ad4d73c834/src/jdk.attach/linux/cl
asses/sun/tools/attach/VirtualMachineImpl.java#l178
Wouldn't it be simpler to just handle this at the Java level and
substitute "" for null in the args array? We're only looking at a
maximum of three possible entries.
Thanks,
David
Thanks
Christoph
