Fetch And Add Semaphore Types - Intel ITANIUM ARCHITECTURE - SOFTWARE DEVELOPERS MANUAL VOLUME 3 REV 2.3 Manual

fetchadd
fetchadd — Fetch and Add Immediate
( ) fetchadd4.
Format: qp
( ) fetchadd8.
qp
A value consisting of four or eight bytes is read from memory starting at the address
Description:
specified by the value in GR
sign-extended immediate value specified by
are: -16, -8, -4, -1, 1, 4, 8, 16. The least significant four or eight bytes of the sum
inc
3
are then written to memory starting at the address specified by the value in GR
zero-extended value read from memory is placed in GR
corresponding to GR
The sem completer specifies the type of semaphore operation. These operations are
described in
page 2:82
Table 2-28.
sem
Completer
acq
rel
The memory read and write are guaranteed to be atomic for accesses to pages with
cacheable, writeback memory attribute. For accesses to other memory types, atomicity
is platform dependent. Details on memory attributes are described in
"Memory Attributes" on page
If the address specified by the value in GR
value being accessed in memory, an Unaligned Data Reference fault is taken
independent of the state of the User Mask alignment checking bit, UM.ac (PSR.ac in the
Processor Status Register).
Both read and write access privileges for the referenced page are required. The write
access privilege check is performed whether or not the memory write is performed.
Only accesses to UCE pages or cacheable pages with write-back write policy are
permitted. Accesses to NaTPages result in a Data NaT Page Consumption fault.
Accesses to pages with other memory attributes cause an Unsupported Data Reference
fault.
On a processor model that supports exported fetchadd, a fetchadd to a UCE page
causes the fetch-and-add operation to be exported outside of the processor; if the
platform does not support exported fetchadd, the operation is undefined. On a
processor model that does not support exported fetchadd, a fetchadd to a UCE page
causes an Unsupported Data Reference fault. See
Attributes on Memory Reference Instructions" on page
The value of the ldhint completer specifies the locality of the memory access. The values
of the ldhint completer are given in
affect program functionality and may be ignored by the implementation. See
Section 4.4.6, "Memory Hierarchy Control and Consistency" on page 1:69
3:74
. = [ ],
sem ldhint r r inc
1 3 3
. = [ ],
sem ldhint r r inc
1 3 3
. The value is zero extended and added to the
r
3
is cleared.
r
1
Table 2-28. See Section 4.4.7, "Sequentiality Attribute and Ordering" on
for details on memory ordering.

Fetch and Add Semaphore Types

Ordering
Semantics
Acquire The memory read/write is made visible prior to all subsequent data memory
accesses.
Release The memory read/write is made visible after all previous data memory
accesses.
2:75.
. The values that may be specified by
inc
3
and the NaT bit
r
1
Semaphore Operation
is not naturally aligned to the size of the
r
3
Section 4.4.9, "Effects of Memory
2:86.
Table 2-34 on page 3:152. Locality hints do not
four_byte_form M17
eight_byte_form M17
. The
r
3
Section 4.4,
for details.
Volume 3: Instruction Reference