Request for Review: 5049299: (process) Use posix_spawn, not fork, on S10 to avoid swap exhaustion

Rob McKenna rob.mckenna at oracle.com
Tue Nov 27 18:10:00 UTC 2012


I'll put a test together for this before sending out another review. I 
was sure I had some sort of issue with this in testing before, but 
perhaps it was IPC related.

     -Rob

On 27/11/12 17:48, Alan Bateman wrote:
> On 27/11/2012 17:45, Martin Buchholz wrote:
>>
>>
>> On Tue, Nov 27, 2012 at 3:06 AM, Alan Bateman 
>> <Alan.Bateman at oracle.com <mailto:Alan.Bateman at oracle.com>> wrote:
>>
>>     On 27/11/2012 06:47, Martin Buchholz wrote:
>>>     :
>>>
>>>     On Solaris bi-arch I think you need only one jprochelper, not
>>>     two, since a 32-bit helper can exec a 64-bit subprocess.
>>>
>>     This is a good point, it needs to know if the target program is
>>     32-bit or 64-bit and chose the appropriate trampoline/helper.
>>
>>
>> ??
>>
>> I believe the architecture of the helper and the target program do 
>> *not* matter, since execv can cross architecture boundaries, so you 
>> only need one helper.
> Ah yes, in any case I think Rob will need a few tests to exercise 
> these cases.
>
> -Alan




More information about the core-libs-dev mailing list