1. Optional Assembler Features 2
CSCI/CMPE 3334
David Egle

2. More features
Allow for more flexible handling of the source and object programs
the generated machine code appears in a different order in the object program than in the source
the source code may be in independent sections, resulting in independent object programs
these maintain their identity and are handled by the linker
Program block
segments of code that are rearranged in memory
Control section
segments of code that are translated into independent object programs

3. Program blocks Separate in source; contiguous in memory
Primary objective
allow the programmer to keep data sections close to the code where it is referenced
have the data in one place in memory
Need the programmer to specify which sections should be together in the object

4. The USE directive
At beginning of program statements assumed to be part of the unnamed (default) block
If no USE statement is included, entire program belongs to this single block
The USE statement signals the beginning of a block with a specific name
source statements are associated with this block until another USE statement is encountered
USE statement may also indicate a continuation of a previously begun block

5. Handling program blocks
During Pass 1
a separate location counter is maintained for each program block
when new block is started, location counter is set to 0
current value is saved with switching to another block
value is restored when a block is resumed
labels are assigned addresses relative to start of current block; block name/number is included in symbol table
at end, value of location counter is size of block
assembler can assign to each block a starting address in object program

6. Handling program blocks 2
Pass 2
address of symbol relative to start of block is found and added to address of block
this is then used in generating the object code
See Figure 2.11 and 2.12 (p 79 and 81)

7. Blocks table The program blocks table is constructed during first pass
When a new USE directive is encountered, the current location counter is saved in the Length
If the new block is not in the table, it is added
At the end of assembly, the address field is determined by adding the address and length of the previous block
Block name
Block number
Address
Length
(default)
0000
0066
CDATA 1
000B
CBLKS 2
0071
1000

8. Program blocks pros and cons
Reduced use of extended format instructions
No need to use base addressing
Improves program readability
Avoids Jump statements
Can accomplish same thing by rearranging the source statements

9. Control sections
A part of the program that maintains its identity after assembly
separate assembly, loading, and manipulation
think of separate functions in C++
Instructions in one control section may refer to instructions or data in another section
external references
Need to provide information for each external reference which will allow the loader to perform the required linking

10. Control section assembly
CSECT directive signals the start of a new section
Symbols that are defined in another section must be identified as external references
EXTREF directive names symbols used in this control section and are defined in another
Need to specify which symbols may be accessed from other control sections
EXTDEF directive names symbols defined in this control section which may be used in other sections
Control section names are automatically global

11. External reference handling
The assembler cannot assemble the external reference addresses
Need to use the extended instruction format (four bytes) to provide room for the actual address
The assembler uses an address of 0 and passes information to the loader/linker
If a symbol is defined in another section, but not identified in the EXTREF statement, an error will be generated if an attempt is made to reference it

12. Passing information to the linker
Need some additional record types in the object code
To indicate symbols defined here and globally available
Dname1 addr1 name2 addr2 …
e.g., DBUFFER00033BUFEND001033LENGTH00002D
To indicate external references (not defined here)
Rname1 name2 name3
e.g., RBUFFERLENGTHBUFEND

13. Passing information to the linker 2
To allow for expressions, update the modification record
Maddress +/- {external symbol}
e.g, M BUFFER
Note that this is a bit different than the text – but does illustrate the idea
Also note that this still handles relocation, since the control section names may be used
M00007+COPY

14. Handling expressions with external symbols
Earlier expression restriction
all of relative terms must be paired, or all except one that has to be of positive sign
Extended restriction
both paired terms must be relative within the same control section
The assembler cannot in general determine whether or not the expression is legal
it evaluates all the terms it can
The loader finishes the evaluation, and checks for errors