NILH, not NIHL!
Sheesh, assembler is hard :-)
Too bad it's been obviated at this point, I was going to suggest NILF
R0,x'7ffffffe'.
sas
On 4/2/2016 12:05, Ed Jaffe wrote:
On 4/2/2016 6:51 AM, Peter Relson wrote:
And of course none of the IBM-Main readers would ever rely on an
empirical test.
Haha! Of course not! ;-)
(But, ASSEMBLER-LIST readers might ijs... :-) )
The intent of the doc is to show that the 64-bit GR0 can be used as a
BASSM target in all cases. I'll get it updated to make it clear that
bits 0-31 are set to 0's for the AMODE 24 and AMODE 31 cases, and are
not necessarily 0 for the AMODE 64 case when LOAD with ADDR64 was
used. And if RMODE 64 is ever supported for normal LOAD, then the
AMODE 64 case would change too.
Thanks, Peter! That's a big help!
Otherwise, people will be putting LLGTRs into their code and then
removing them (or forgetting to) if/when RMODE(64) becomes reality...
With all of this in mind, I suppose I should amend my example to read:
LOAD EPLOC==CL8'MSGTABLE' Load external message table
TMLL R0,X'0001' RMODE(64) ?
JO *+8 ---------------\ Branch if yes
NIHL R0,X'7FFF' | Turn of 31-bit indicator
NILL R0,X'FFFE' <--------/ Turn of 64-bit indicator
STG R0,MsgTable Save clean table address
