diff options
author | ru <ru@FreeBSD.org> | 2002-01-10 10:36:04 +0000 |
---|---|---|
committer | ru <ru@FreeBSD.org> | 2002-01-10 10:36:04 +0000 |
commit | b3b66ade4fe7234007fe800e2fc9c0addb7138aa (patch) | |
tree | 497b5f1ccd6f58b2506b8fdcdecfaa5f8f8b0133 /share/man/man9/atomic.9 | |
parent | cf262779561d9da82609da5a4d2ff043b4c00d05 (diff) | |
download | FreeBSD-src-b3b66ade4fe7234007fe800e2fc9c0addb7138aa.zip FreeBSD-src-b3b66ade4fe7234007fe800e2fc9c0addb7138aa.tar.gz |
mdoc(7) police:
- make SYNOPSIS look better with some troff magic
- tidy up the markup
Diffstat (limited to 'share/man/man9/atomic.9')
-rw-r--r-- | share/man/man9/atomic.9 | 151 |
1 files changed, 94 insertions, 57 deletions
diff --git a/share/man/man9/atomic.9 b/share/man/man9/atomic.9 index a07388b..0980f2b 100644 --- a/share/man/man9/atomic.9 +++ b/share/man/man9/atomic.9 @@ -39,77 +39,87 @@ .Sh SYNOPSIS .In sys/types.h .In machine/atomic.h +.\" XXX +.ds LB \f[R]\(lB\f[P] +.ds RB \f[R]\(rB\f[P] +.ds La \f[R]\(la\f[P] +.ds Ra \f[R]\(ra\f[P] .Ft void -.Fn atomic_add_{acq_,rel_,}<type> "volatile <type> *p" "<type> v" +.Fn atomic_add_\*[LB]acq_\*[Ba]rel_\*[RB]\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" "\*[La]type\*[Ra] v" .Ft void -.Fn atomic_clear_{acq_,rel_,}<type> "volatile <type> *p" "<type> v" +.Fn atomic_clear_\*[LB]acq_\*[Ba]rel_\*[RB]\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" "\*[La]type\*[Ra] v" .Ft int -.Fo atomic_cmpset_{acq_,rel_,}<type> -.Fa "volatile <type> *dst" -.Fa "<type> old" -.Fa "<type> new" +.Fo atomic_cmpset_\*[LB]acq_\*[Ba]rel_\*[RB]\*[La]type\*[Ra] +.Fa "volatile \*[La]type\*[Ra] *dst" +.Fa "\*[La]type\*[Ra] old" +.Fa "\*[La]type\*[Ra] new" .Fc -.Ft <type> -.Fn atomic_load_acq_<type> "volatile <type> *p" -.Ft <type> -.Fn atomic_readandclear_<type> "volatile <type> *p" +.Ft \*[La]type\*[Ra] +.Fn atomic_load_acq_\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" +.Ft \*[La]type\*[Ra] +.Fn atomic_readandclear_\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" .Ft void -.Fn atomic_set_{acq_,rel_,}<type> "volatile <type> *p" "<type> v" +.Fn atomic_set_\*[LB]acq_\*[Ba]rel_\*[RB]\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" "\*[La]type\*[Ra] v" .Ft void -.Fn atomic_subtract_{acq_,rel_,}<type> "volatile <type> *p" "<type> v" +.Fn atomic_subtract_\*[LB]acq_\*[Ba]rel_\*[RB]\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" "\*[La]type\*[Ra] v" .Ft void -.Fn atomic_store_rel_<type> "volatile <type> *p" "<type> v" +.Fn atomic_store_rel_\*[La]type\*[Ra] "volatile \*[La]type\*[Ra] *p" "\*[La]type\*[Ra] v" +.rm LB RB La Ra .Sh DESCRIPTION Each of the atomic operations is guaranteed to be atomic in the presence of interrupts. They can be used to implement reference counts or as building blocks for more advanced synchronization primitives such as mutexes. .Ss Types -Each atomic operation operates on a specific type. +Each atomic operation operates on a specific +.Ar type . The type to use is indicated in the function name. The available types that can be used are: -.Bl -tag -offset indent -width short -.It int +.Pp +.Bl -tag -offset indent -width short -compact +.It Li int unsigned integer -.It long +.It Li long unsigned long integer -.It ptr +.It Li ptr unsigned integer the size of a pointer -.It 32 +.It Li 32 unsigned 32-bit integer -.It 64 +.It Li 64 unsigned 64-bit integer .El .Pp For example, the function to atomically add two integers is called .Fn atomic_add_int . .Pp -Certain architectures also provide operations for types smaller than int. -.Bl -tag -offset indent -width short -.It char +Certain architectures also provide operations for types smaller than +.Dq Li int . +.Pp +.Bl -tag -offset indent -width short -compact +.It Li char unsigned character -.It short +.It Li short unsigned short integer -.It 8 +.It Li 8 unsigned 8-bit integer -.It 16 +.It Li 16 unsigned 16-bit integer .El .Pp These must not be used in MI code because the instructions to implement them efficiently may not be available. .Ss Memory Barriers -Memory barriers are used to guarantee the order the order of data accesses in +Memory barriers are used to guarantee the order of data accesses in two ways. First, they specify hints to the compiler to not re-order or optimize the operations. -Secondly, on architectures that do not guarantee ordered data accesses, +Second, on architectures that do not guarantee ordered data accesses, special instructions or special variants of instructions are used to indicate to the processor that data accesses need to occur in a certain order. As a result, most of the atomic operations have three variants in order to include optional memory barriers. The first form just performs the operation without any explicit barriers. -The second form uses a read memory barrier, and the final variant uses a write +The second form uses a read memory barrier, and the third variant uses a write memory barrier. .Pp The second variant of each operation includes a read memory barrier. @@ -118,9 +128,9 @@ effects of any later data accesses. As a result, the operation is said to have acquire semantics as it acquires a pseudo-lock requiring further operations to wait until it has completed. To denote this, the suffix -.Dq _acq +.Dq Li _acq is inserted into the function name immediately prior to the -.Em _type +.Dq Li _ Ns Aq Ar type suffix. For example, to subtract two integers ensuring that any later writes will happen after the subtraction is performed, use @@ -132,9 +142,9 @@ before this operation takes place. As a result, the operation is said to have release semantics as it releases any pending data accesses to be completed before its operation is performed. To denote this, the suffix -.Dq _rel +.Dq Li _rel is inserted into the function name immediately prior to the -.Em _type +.Dq Li _ Ns Aq Ar type suffix. For example, to add two long integers ensuring that all previous writes will happen first, use @@ -146,7 +156,6 @@ To achieve this, one would use a read barrier when acquiring the lock to guarantee that the lock is held before any protected operations are performed. Finally, one would use a write barrier when releasing the lock to ensure that all of the protected operations are completed before the lock is released. -.Pp .Ss Multiple Processors The current set of atomic operations do not necessarily guarantee atomicity across multiple processors. @@ -166,16 +175,16 @@ using a caching policy of either uncached or write back. .Ss Semantics This section describes the semantics of each operation using a C like notation. .Bl -hang -.It Fn atomic_add "p" "v" -.Bd -literal +.It Fn atomic_add p v +.Bd -literal -compact *p += v; .Ed -.It Fn atomic_clear "p" "v" -.Bd -literal +.It Fn atomic_clear p v +.Bd -literal -compact *p &= ~v; .Ed -.It Fn atomic_cmpset "dst" "old" "new" -.Bd -literal +.It Fn atomic_cmpset dst old new +.Bd -literal -compact if (*dst == old) { *dst = new; return 1; @@ -186,10 +195,15 @@ if (*dst == old) { .Pp The .Fn atomic_cmpset -functions are not implemented for the types char, short, 8, and 16. +functions are not implemented for the types +.Dq Li char , +.Dq Li short , +.Dq Li 8 , +and +.Dq Li 16 . .Bl -hang -.It Fn atomic_load "addr" -.Bd -literal +.It Fn atomic_load addr +.Bd -literal -compact return (*addr) .Ed .El @@ -198,8 +212,8 @@ The .Fn atomic_load functions always have acquire semantics. .Bl -hang -.It Fn atomic_readandclear "addr" -.Bd -literal +.It Fn atomic_readandclear addr +.Bd -literal -compact temp = *addr; *addr = 0; return (temp); @@ -208,19 +222,26 @@ return (temp); .Pp The .Fn atomic_readandclear -functions are not implemented for the types char, short, ptr, 8, and 16 and do +functions are not implemented for the types +.Dq Li char , +.Dq Li short , +.Dq Li ptr , +.Dq Li 8 , +and +.Dq Li 16 +and do not have any variants with memory barriers at this time. .Bl -hang -.It Fn atomic_set "p" "v" -.Bd -literal +.It Fn atomic_set p v +.Bd -literal -compact *p |= v; .Ed -.It Fn atomic_subtract "p" "v" -.Bd -literal +.It Fn atomic_subtract p v +.Bd -literal -compact *p -= v; .Ed -.It Fn atomic_store "p" "v" -.Bd -literal +.It Fn atomic_store p v +.Bd -literal -compact *p = v; .Ed .El @@ -230,16 +251,20 @@ The functions always have release semantics. .Pp The type -.Dq 64 +.Dq Li 64 is currently not implemented for any of the atomic operations on the .Tn i386 architecture. .Sh RETURN VALUES +The .Fn atomic_cmpset +function returns the result of the compare operation. +The .Fn atomic_load and .Fn atomic_readandclear +functions return the value at the specified address. .Sh EXAMPLES This example uses the @@ -250,9 +275,9 @@ functions to obtain a sleep mutex and handle recursion. Since the .Va mtx_lock member of a -.Li struct mtx +.Vt "struct mtx" is a pointer, the -.Dq ptr +.Dq Li ptr type is used. .Bd -literal #define _obtain_lock(mp, tid) \\ @@ -279,7 +304,12 @@ and .Fn atomic_subtract operations were first introduced in .Fx 3.0 . -This first set only supported the types char, short, int, and long. +This first set only supported the types +.Dq Li char , +.Dq Li short , +.Dq Li int , +and +.Dq Li long . The .Fn atomic_cmpset , .Fn atomic_load , @@ -288,7 +318,14 @@ and .Fn atomic_store operations were added in .Fx 5.0 . -The types 8, 16, 32, 64, and ptr and all of the acquire and release variants +The types +.Dq Li 8 , +.Dq Li 16 , +.Dq Li 32 , +.Dq Li 64 , +and +.Dq Li ptr , +and all of the acquire and release variants were added in .Fx 5.0 as well. |