On Wed, 29 Mar 2017 03:48:01 -0700, Ed Jaffe wrote: >On 3/29/2017 2:27 AM, Martin Packer wrote: >> As I understand it the real restriction is not being interruptible (for >> I/O). In which case a clever (and aren't they all nowadays?) :-) JVM might >> well find itself executing machine code that you'd hope wouldn't have to >> be copied below the bar first. > >Actually, that's not a restriction at all. z/OS fully supports executing >enabled code running above the bar. > >What's missing is binder/loader/contents support to load/manage >executable RMODE(64) modules and enough officially-supported SVC and >PC-entered z/OS services to make running programs there feasible. After >that, you would need compiler and HLL run-time language support from LE >to open that "world" up to everyone... > Of course, Java byte code is not executed directly, and could reside above the bar. And the JiT compiler might direct its output above the bar without involving the binder/loader/contents support. And that JiT code could branch back to the interpreter to invoke system services.
(Some OSes separate executable and writeable segments for integrity reasons.) There might be incentive here for LPA above the bar to accommodate large common byte code (data) objects. Or .jars? -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
