iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[PATCH] Corrections to memory barrier doc

Browse Source 
Apply some small corrections to the memory barrier document, as contributed by:

	Christoph Lameter <clameter@sgi.com>
	Kirill Smelkov <kirr@mns.spb.ru>
	Randy Dunlap <rdunlap@xenotime.net>

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This commit is contained in:
David Howells 2006-06-25 05:49:22 -07:00 committed by Linus Torvalds
parent 6f84be84b4
commit 6bc392741d
1 changed files with 9 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
Documentation/memory-barriers.txt
View File

@ -287,7 +287,7 @@ Memory barriers come in four basic varieties:
A write barrier is a partial ordering on stores only; it is not required A write barrier is a partial ordering on stores only; it is not required
to have any effect on loads. to have any effect on loads.
A CPU can be viewed as as commiting a sequence of store operations to the A CPU can be viewed as committing a sequence of store operations to the
memory system as time progresses. All stores before a write barrier will memory system as time progresses. All stores before a write barrier will
occur in the sequence _before_ all the stores after the write barrier. occur in the sequence _before_ all the stores after the write barrier.
@ -418,7 +418,7 @@ There are certain things that the Linux kernel memory barriers do not guarantee:
indirect effect will be the order in which the second CPU sees the effects indirect effect will be the order in which the second CPU sees the effects
of the first CPU's accesses occur, but see the next point: of the first CPU's accesses occur, but see the next point:
(*) There is no guarantee that the a CPU will see the correct order of effects (*) There is no guarantee that a CPU will see the correct order of effects
from a second CPU's accesses, even _if_ the second CPU uses a memory from a second CPU's accesses, even _if_ the second CPU uses a memory
barrier, unless the first CPU _also_ uses a matching memory barrier (see barrier, unless the first CPU _also_ uses a matching memory barrier (see
the subsection on "SMP Barrier Pairing"). the subsection on "SMP Barrier Pairing").
@ -489,7 +489,7 @@ lines. The pointer P might be stored in an odd-numbered cache line, and the
variable B might be stored in an even-numbered cache line. Then, if the variable B might be stored in an even-numbered cache line. Then, if the
even-numbered bank of the reading CPU's cache is extremely busy while the even-numbered bank of the reading CPU's cache is extremely busy while the
odd-numbered bank is idle, one can see the new value of the pointer P (&B), odd-numbered bank is idle, one can see the new value of the pointer P (&B),
but the old value of the variable B (1). but the old value of the variable B (2).
Another example of where data dependency barriers might by required is where a Another example of where data dependency barriers might by required is where a
@ -749,7 +749,7 @@ some effectively random order, despite the write barrier issued by CPU 1:
: : : :
If, however, a read barrier were to be placed between the load of E and the If, however, a read barrier were to be placed between the load of B and the
load of A on CPU 2: load of A on CPU 2:
CPU 1 CPU 2 CPU 1 CPU 2
@ -1466,9 +1466,8 @@ instruction itself is complete.
On a UP system - where this wouldn't be a problem - the smp_mb() is just a On a UP system - where this wouldn't be a problem - the smp_mb() is just a
compiler barrier, thus making sure the compiler emits the instructions in the compiler barrier, thus making sure the compiler emits the instructions in the
right order without actually intervening in the CPU. Since there there's only right order without actually intervening in the CPU. Since there's only one
one CPU, that CPU's dependency ordering logic will take care of everything CPU, that CPU's dependency ordering logic will take care of everything else.
else.
ATOMIC OPERATIONS ATOMIC OPERATIONS
@ -1645,9 +1644,9 @@ functions:
The PCI bus, amongst others, defines an I/O space concept - which on such The PCI bus, amongst others, defines an I/O space concept - which on such
CPUs as i386 and x86_64 cpus readily maps to the CPU's concept of I/O CPUs as i386 and x86_64 cpus readily maps to the CPU's concept of I/O
space. However, it may also mapped as a virtual I/O space in the CPU's space. However, it may also be mapped as a virtual I/O space in the CPU's
memory map, particularly on those CPUs that don't support alternate memory map, particularly on those CPUs that don't support alternate I/O
I/O spaces. spaces.
Accesses to this space may be fully synchronous (as on i386), but Accesses to this space may be fully synchronous (as on i386), but
intermediary bridges (such as the PCI host bridge) may not fully honour intermediary bridges (such as the PCI host bridge) may not fully honour