|
View previous topic :: View next topic
|
| Author |
Message |
jackzhang75
Active User
Joined: 09 Jun 2014
Posts: 125
Location: US
|
|
|
|
Hi experts,
I am looking at the following assembler and i am not sure how exactly the TM instruction is working.
| Code: |
L R2,0(R1) GET A(ABEND CODE)
LH R1,0(R2) GET ABEND CODE
TM 0(R2),X'F0' GOOD CODE
BZ ABEND YES
LA R1,4095 ELSE SET DEFAULT
ABEND ABEND (R1),DUMP
* SYMBOLIC REGS,R10=YES
YREGS |
I see the content in register
and
So for the mask 'F0' --> 11110000 , is it looking for R2 --> 800B?
When i ran the program , the control BZ is go to ABEND which i think R2 should be all zero? but i don't see R2 is all zero so i am confusing..which part of R2 the mask is used on?
Thanks |
|
| Back to top |
|
 |
steve-myers
Active Member
Joined: 30 Nov 2013
Posts: 917
Location: The Universe
|
|
|
|
| Reg 2 is the address of the data to test. It is not the data to test. The data to test is at B719E. |
|
| Back to top |
|
|
sergeyken
Senior Member
Joined: 29 Apr 2008
Posts: 2018
Location: USA
|
|
|
|
The data to “test under mask” is one single byte addressed by R2, with zero offset from R2 value.
The (uncommented) logic looks like:
- get ABEND code from a halfword addresses by R2 into R1
- verify that the code at location R2 is less or equal to X’0FFF’ = 4095
- if not like this, then replace it in R1 with 4095
For clarity the better way would be
| Code: |
LH R1,0(R2)
CHI R1,4095
JLE *+8
LHI R1,4095
ABEND ... |
|
|
| Back to top |
|
|
steve-myers
Active Member
Joined: 30 Nov 2013
Posts: 917
Location: The Universe
|
|
|
|
The instruction is testing the first 4 bits of the nominal ABEND code. If any bits are on the code is invalid and valid user ABEND code is loaded into reg 1.
Sergeyken's code does not do this. Consider the case when the high order bit is on. For example X'FFFF'. The LH sets reg 1 to X'FFFFFFFF' What's going to happen with the CHI/JLE? |
|
| Back to top |
|
|
sergeyken
Senior Member
Joined: 29 Apr 2008
Posts: 2018
Location: USA
|
|
|
|
| steve-myers wrote: |
The instruction is testing the first 4 bits of the nominal ABEND code. If any bits are on the code is invalid and valid user ABEND code is loaded into reg 1.
Sergeyken's code does not do this. Consider the case when the high order bit is on. For example X'FFFF'. The LH sets reg 1 to X'FFFFFFFF' What's going to happen with the CHI/JLE? |
Agreed.
Better to use
CL R1,=F’4095’
JLE *+8 |
|
| Back to top |
|
|
jackzhang75
Active User
Joined: 09 Jun 2014
Posts: 125
Location: US
|
|
|
|
| Thank you for your replay. now i understand. |
|
| Back to top |
|
|
|
|
