On Mon, 16 May 2016 07:32:56 -0700, Ed Jaffe wrote:
>On 5/16/2016 7:23 AM, Jerry Callen wrote:
>> As an example: Does z/OS require that there be sufficient page space
>> available to back all of the space requested for a large memory object?
>
>A virtual page is not backed by a REAL frame until it is referenced, and
>not backed by an AUX slot until that frame is paged out.
>
Some traditional UNIX partisans found this behavior ("lazy malloc()")
a rude shock in AIX:
o They were accustomed at program initiation to malloc() a small work area
only for recovery and orderly termination in case of an unexpected failure.
But if the failure was for backing storage exhausted the recovery area was
unavailable.
o In circumvention, AIX introduced a nonstandard signal, SIGDANGER,
thrown when backing storage was (FSVO) nearly exhausted. But that
might be caught by an innocent, memory-conservative process that
by happenstance caused a page fault at an inopportune time.
I don't know what techniques Windows or Linux uses to avoid these
problems for the OP.
Historically, mainframe programmers overallocate data sets, "Just in case".
But virtual DASD might cause similar undesired benavior by deferring
allocation of backend page frames.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
