On Tue, 27 Feb 2018 15:39:17 -0500, Thomas David Rivers wrote:
>On a typical UNIX platform, the operating system enforces the rule that
>signal #9 cannot be blocked, and that a process that receives
>that signal will be removed from the process table (and thus "killed".)
>
Not just "typical"; it's a POSIX requirement:
http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/signal.h.html
SIGKILL T Kill (cannot be caught or ignored).
The default actions are as follows:
T Abnormal termination of the process.
>I've been setting up a BPX signal handler in some ASM code; using the BPX
>mvssigsetup() function, writing a SIR, etc...
>
>What I have discovered is that there seems to be nothing in z/OS that
>enforces the "signal #9 can't be intercepted" rule... my routines can't
>block that signal, but they are totally happy to receive it, ignore it
>and return back to the executing program. The program isn't summarily
>"killed".
>
This merits a serious SR. Don't let them weasel out with a specious argument
from "compatibility; it would break too much code that has come to depend on
its being wrong," as they are won't to do.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
