use the bitmask_snprintf info which is more complete, and adjust it for

reality.
This commit is contained in:
christos 2008-12-16 23:19:16 +00:00
parent abe4c2aeb6
commit 930983f9c3

View File

@ -1,4 +1,4 @@
.\" $NetBSD: snprintb.3,v 1.6 2008/04/30 13:10:52 martin Exp $
.\" $NetBSD: snprintb.3,v 1.7 2008/12/16 23:19:16 christos Exp $
.\"
.\" Copyright (c) 1998 The NetBSD Foundation, Inc.
.\" All rights reserved.
@ -27,18 +27,16 @@
.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
.\" POSSIBILITY OF SUCH DAMAGE.
.\"
.Dd July 28, 2000
.Dd December 16, 2008
.Dt SNPRINTB 3
.Os
.Sh NAME
.Nm snprintb
.Nd bitmask output conversion
.Sh LIBRARY
.Lb libutil
.Sh SYNOPSIS
.In util.h
.Ft int
.Fn "snprintb" "char *buf" "size_t buflen" "const char *fmt" "uint64_t val"
.Fn "snprintb" "char *buf" "size_t buflen" "const char *fmt" "u_quad_t val"
.Sh DESCRIPTION
The
.Fn snprintb
@ -53,16 +51,39 @@ of size
.Fa buflen ,
using a specified radix and an interpretation of
the bits within that integer as though they were flags.
The buffer is always NUL-terminated.
If the buffer
.Fa buf
is too small to hold the formatted output,
.Fn snprintb
will fill as much as it can, and return the number of bytes
that would have written if the buffer was long enough excluding the
terminating NUL.
.Pp
The decoding directive string
.Fa fmt
describes how the bitfield is to be interpreted and displayed.
It follows two possible syntaxes, referred to as
.Dq old
and
.Dq new .
The main advantage of the
.Dq new
formatting is that it is capable of handling multi-bit fields.
.Pp
The first character of
.Fa fmt
may be
.Li \e177 ,
indicating that the remainder of the format string follows the
.Dq new
syntax.
The second character
.Pq the first for the old format
is a binary character representation of the
output numeral base in which the bitfield will be printed before it is decoded.
Recognized radix values
.Pq "in C escape-character format"
.Pq in C escape-character format
are
.Li \e10
.Pq octal ,
@ -75,51 +96,131 @@ and
The remaining characters in
.Fa fmt
are interpreted as a list of bit-position\(endescription pairs.
A bit-position\(endescription pair begins with a binary character value
that represents the position of the bit being described.
From here the syntaxes diverge.
.Pp
The
.Dq old
format syntax is series of bit-position\(endescription pairs.
Each begins with a binary character value that represents the position
of the bit being described.
A bit position value of one describes the least significant bit.
Whereas a position value of 32
.Pq "octal 40, hexadecimal 20, the ASCII space character"
.Pq octal 40, hexadecimal 20, the ASCII space character
describes the most significant bit.
.Pp
The remaining characters in a bit-position\(endescription pair are the
characters to print should the bit being described be set.
Description strings are delimited by the next bit position value character
encountered
.Pq "distinguishable by its value being \*[Le] 32" ,
.Pq distinguishable by its value being \*[Le] 32 ,
or the end of the decoding directive string itself.
.Pp
For the
.Dq new
format syntax, a bit-position\(endescription begins with a field type
followed by a binary bit-position and possibly a field length.
The least significant bit is bit-position zero, unlike the
.Dq old
syntax where it is one.
.Bl -tag -width "xxxxx"
.It Cm b\eB
Describes a bit position.
The bit-position
.Fa B
indicates the corresponding bit, as in the
.Dq old
format.
.It Cm f\eB\eL
Describes a multi-bit field beginning at bit-position
.Fa B
and having a bit-length of
.Fa L .
The remaining characters are printed as a description of the field
followed by
.Sq \&=
and the value of the field.
The value of the field is printed in the base specified as the second
character of the decoding directive string
.Ar fmt .
.It Cm F\eB\eL
Describes a multi-bit field like
.Sq f ,
but just extracts the value for use with the
.Sq \&=
and
.Sq \&:
formatting directives described below.
.It Cm \&=\eV
The field previously extracted by the last
.Sq f
or
.Sq F
operator is compared to the byte
.Sq Cm V
.Pq for values 0 through 255 .
If they are equal,
.Sq \&=
followed by the string following
.Sq Cm V
is printed.
This and the
.Sq \&:
operator may be repeated to annotate multiple possible values.
.It Cm :\eV
Operates like the
.Sq \&=
operator, but omits the leading
.Sq \&= .
.El
.Pp
Finally, each field is delimited by a NUL
.Pq Sq \e0
character.
By convention, the format string has an additional NUL character at
the end, following that delimiting the last bit-position\(endescription
pair.
.Sh RETURN VALUES
The
.Fn snprintb
function returns the number of characters that are required to format the
value
.Fa val
given the format string
.Fa fmt
excluding the terminating NUL.
The returned string in
.Fa buf
is always NUL-terminated.
function returns the number of bytes that would have written to the buffer
if there was adequate space, excluding the terminating NUL, or -1 in case
an error occurred.
.Sh EXAMPLES
Two examples of the old formatting style:
.Bd -literal -offset indent
snprintb(buf, buflen, "\e10\e2BITTWO\e1BITONE", 3)
snprintb(3, "\e10\e2BITTWO\e1BITONE", buf, buflen)
\(rA "3\*[Lt]BITTWO,BITONE\*[Gt]"
snprintb(buf, buflen
snprintb(0xe860,
"\e20\ex10NOTBOOT\ex0fFPP\ex0eSDVMA\ex0cVIDEO"
"\ex0bLORES\ex0aFPA\ex09DIAG\ex07CACHE"
"\ex06IOCACHE\ex05LOOPBACK\ex04DBGCACHE",
0xe860)
\(rA "0xe860\*[Lt]NOTBOOT,FPP,SDVMA,VIDEO,CACHE,IOCACHE\*[Gt]"
buf, buflen)
\(rA "e860\*[Lt]NOTBOOT,FPP,SDVMA,VIDEO,CACHE,IOCACHE\*[Gt]"
.Ed
.Pp
An example of the new formatting style:
.Bd -literal -offset indent
snprintb(0x800f0701,
"\e177\e020b\e0LSB\e0b\e1_BITONE\e0f\e4\e4NIBBLE2\e0"
"f\ex10\e4BURST\e0=\e4FOUR\e0=\exfSIXTEEN\e0"
"b\ex1fMSB\e0\e0",
buf, buflen)
\(rA "800f0701\*[Lt]LSB,NIBBLE2=0,BURST=f=SIXTEEN,MSB\*[Gt]"
.Ed
.Pp
.Sh ERRORS
If the buffer
.Fa buf
is too small to hold the formatted output,
.Fn snprintb
will still return the buffer, containing a truncated string.
will fail if:
.Bl -tag -width Er
.It Bq Er EINVAL
The leading character does not describe a supported format,
or
.Fn snprintf
failed.
.El
.Sh SEE ALSO
.Xr printf 3
.Xr snprintf 3
.Sh HISTORY
The
@ -131,13 +232,13 @@ format string for the kernel
function in
.Nx 1.5
and earlier releases.
It got implemented as
.Fn bitmap_snprintf
for
.Nx 1.6
and this version was used to implement
.Fn snprintb .
.Sh BUGS
.Fn snprintb
supports a new extended form of formatting string, which is not yet
described here.
It was called
.Fn bitmask_snprintf
in
.Nx 5.0
and earlier releases.
.Sh AUTHORS
The
.Dq new
format was the invention of
.An Chris Torek .