mc68ec060 Freescale Semiconductor, Inc, mc68ec060 Datasheet - Page 260

no-image

mc68ec060

Manufacturer Part Number
mc68ec060
Description
Mc68060 Superscalar 68k Microprocessor Including The Lc060 And Ec060
Manufacturer
Freescale Semiconductor, Inc
Datasheet

Available stocks

Company
Part Number
Manufacturer
Quantity
Price
Part Number:
mc68ec060RC50
Manufacturer:
NXP
Quantity:
1 746
Part Number:
mc68ec060ZU50
Manufacturer:
Freescale Semiconductor
Quantity:
10 000
Part Number:
mc68ec060ZU66
Manufacturer:
MOTOROLA
Quantity:
256
Part Number:
mc68ec060ZU66
Manufacturer:
Freescale Semiconductor
Quantity:
10 000
Part Number:
mc68ec060ZU66
Manufacturer:
MOTOROLA/摩托罗拉
Quantity:
20 000
Part Number:
mc68ec060ZU75
Manufacturer:
Freescale Semiconductor
Quantity:
10 000
Exception Processing
Once the write buffers (push and store) are all empty, the pipeline re-evaluates the pending
exception types. If no TEA fault occurred during the emptying of the buffers, the processor
continues with the original exception. If a TEA fault did occur as the buffers were emptied,
the original exception is discarded and an operand data access error exception is taken. The
exception stack for the access error includes indicator bits in the FSLW signalling the occur-
rence of the push buffer TEA or store buffer TEA. Note that both errors may be present
within a single access error exception. The exception stack frame will record the PC value
at the time the exception was detected, but this value has no relationship to the instruction
that caused the push or store buffer entries to originally be made. The stacked virtual
address is meaningless for these two fault types.
There are other non-recoverable write cases which are unrelated to the push and store
buffer cases. The execution of a misaligned read-modify-write instruction which partially
completes the writes before faulting is inherently non-recoverable on a restart machine.
Consider the ADD D0, <mem> instruction. In this instruction, the processor fetches the
memory operand, adds the contents of D0 internally, and writes the result out to memory. If
the memory operand is misaligned and a bus error occurs on the second or later access,
the first part of the memory operand would have been overwritten. If the instruction enters
the access error exception handler, it cannot be restarted because original memory value
has been corrupted. This read-modify-write instruction and others like it can be detected in
the access error handler because the access error frame has separate read and write bits
(RW field). If both bits are set, the instruction is a read-modify-write instruction similar to the
ADD instruction case as discussed.
Another non-recoverable write case is similar to the ADD case above, but is more difficult to
detect. A MOVE <mem>, <mem> instruction in which the source operand and destination
operand overlap may have the same problems as discussed in the ADD instruction if the
destination operand is part of the source operand and a misaligned write occurs, which
result in an access error on the second or later misaligned case. The MOVE <mem>,
<mem> instruction is not normally considered a read-modify-write type of instruction, and is
not detected simply by looking at the RW bits in the FSLW.
An MC68060 system design could implement address/data capture logic to provide addi-
tional information for these bus error scenarios.
8-28
M68060 USER’S MANUAL
MOTOROLA

Related parts for mc68ec060