Tom: You're right! I just found the same statement in the basic Assembler Services Guide.
Fortunately, I can't remember any circumstances where I defined an ESTAE(X) in a BAKR routine and expected it to be in effect afterwards. However, I will certainly keep this in mind for the future. Thanks, Art At 12:50 PM 4/12/2011, Tom Marchant wrote: >On Tue, 12 Apr 2011 11:00:33 -0400, Art Celestini wrote: > >>To my knowledge, this is not correct. The only type of recovery routine >>cancelled by a PR is an ARR (Associated Recovery Routine) which is setup >>as part of the PC used to call the routine that issues the PR. (The ARR >>is viable only while in that PC routine.) >> > >My colleague has personal experience with ESTAE(x) routines being >cancelled by PR, and it is documented. > >See "Linkage Stack Consideration" in "providing Recovery" chapter of >the Assembler Services Guide (8.2.3.2) or the Authorized Assembler >Services guide (18.3.2.3): > >"When you issue a PR, the system automatically deactivates all >ESTAE-type recovery routines that were previously activated under >that current linkage stack entry." > >-- >Tom Marchant > >> >>At 08:10 AM 4/12/2011, Tom Marchant wrote: >> >>>On Mon, 11 Apr 2011 15:31:22 -0500, McKown, John wrote: >>> >>>>one could use the BAKR to save the register on the linkage stack >>> >>>A colleague points out that care must be taken when using the linkage >>>stack with ESTAE type recovery routines. For example, if an ESTAE(X) >>>is established in a subroutine, the ESTAE is cancelled when that >>>subroutine exits with a PR.
