Filter
Top Products
8086
Object
Module
Formats
Version
4.0
(1)
A 5 HORT-JMP
is
being
made
to
an
external
subroutine.
In
this
case,
the
TARGET
should
be
specified
as
TARGET:
EI(subroutine)
,0000H
The
reason
is
that
when
LINK-86
learns
where
the
subroutine
is
located,
it
will
probably
be
a
known
offset
(dl)
within
some
LSEG
A.
Thus,
LINK-8~
will
convert
the
above
TARGET
to
the
form:
. TARGET:
SI(A)
,dl
Now
the
programmer
may
be
correct
in
dknowinq·
that
when
the
proqram
is
eventually
LOCATE'd,
the
TARGET
will
be
within
128
bytes
of
LOCATION;
however,
this
does
not
mean
that
dl
is
less
than
l28!
Thus,
as
LINK-86
maintains
the
(possibly
changing)
value
of
dl
as
various
pieces
of
LSEG
A
are
combined,
it
needs
a
full
word
to
maintain
the
offset.
Since
the
LOCATION
is
a
sinqle
byte,
the
translator
must
provide
an
offset
field
in
the
fixup
record
itself
for
LINK-86
to
maintain
intermediate
fixup
values.
(2)
The
translator
REFERENT.
For
example,
if
external
array
ARY,
and
register
that
will
contain
translator
would
use
wishes
to
reference
·backwards~
from
the
the
TARGET
is
the
word
in
front
of
the
the
reference
is
with
respect
to
a
base
the
address
of
the
LSEG
named
FOO,
the
FRAME:
SI(FOO)
TARGET:
EI(ARY)
,0000H
and
place
the
dneqative
offset-
FFFEH
in
LOCATION.
R&L
will
perform
access
verification
to
the
REFERENT ARY:
however,
access
to
the
TARGET
is
not
guaranteed,
and
is
the
programmer's
responsibility.
Note:
if
Case
3
in
the
above
diaqram
were
available,
the
translator
could
use
FRAME:
SI(FOO)
TARGET:
EI(ARY)
,-2
and
R&L
would
perform
access
verification,
not
to
the
REFERENT
ARY
(as
above),
but
to
the
actual
TARGET
(in
front
of
ARY)!
(2)
(continued)
The
calculation
by
LOCATE-Sfi
involves
3
quantities:
the
MAS-location
of
FOO,
the
MAS-location
of
the
LSEG
(say,
BAZ)
containing
ARY,
and
the
relative
offset
of
ARY
within
BAZ. LOCATE-86
can
enforce
that
the
final
offset,
which
is
the
difference
(location
of
SAZ
plus
relative
offset)
-
(location
of
FOO)
r
is
not
qreater
than
~5535r
provided
that
all
quantities
enterinq
into
this
difference
are
known.--rl~he
translator
had
specified
the
fixup
as
FRAME:
SI(FOO)
TARGET:
EI(ARY)
then
LINK-86
would
have
had
to
maintain
the
(possibly
chanqinq
from
linkaqe
to
linkaae)
relative
offset
of
ARY
within
BAZ.
in
the
LOCATION
itself,
where
it
qets
~added~
to
the
content
FFFEH.
And
because
the
R&L
system
cannot
know
if
the
FFFEH
was
a
neoative
2
or
a
positive
~5534,
the
access
verification
of
R~L
may
thNart
the
translator'S
intentions.
20